You are on page 1of 63

Understanding amplifier power ratings

There are different methods for measuring the power ratings for amplifiers and speakers. And
different measuring methods give different values so it is vital to understand the difference
between these different power ratings to be able to make at least some comparison between
different power ratings.

RMS power

To make it short, an RMS power value is directly related to perceivable energy (acoustical, heat,
light - or what else applies).

"RMS" is really a rather meaningless figure, when measuring power. R.M.S. is useful for
measuring the "power-producing equivalent" voltage. Thus 10 Volts RMS will produce the same
power into a given impedance that 10 Volts DC would produce (onto a resistance) Any waveform
of 10 V R.M.S. will produce the same power into that impedance. This is because it's the root of
the mean of all the average squared voltages to which Norbert Hahn referred in the prior post. It
is if little meaning to compute the mean of squares of all the power values in a wave.

RMS, when applied to power measurements, has come to mean "sine-wave power." A 100 Watt
"RMS" amplifier can produce a 100 Watt sine-wave into its load. With music, the total actual
power would be less. With a square-wave, it would be more.

DIN power

The DIN 45000 defines different methods to measure power, depending on the device under test.
Well, this is what I remember from reading the DIN some 25 years ago.

For home appliances there are three different numbers for power: Continuous power, Peak power
and power bandwidth; the latter does not apply for speakers.

Power measurement of an amplifier requires that the amplifier is properly terminated by Ohmic
resistances of nominal value both at input and output. The continuous power is measured when
the amplifier is supplied by its normal power supply. It must then be able to deliver the rated
power at 1 kHz for at least 10 minutes while the maximum THD does not exceed 1 %. To
measure the peak power the normal power supply is replaced by a regulated power supply and
the time for delivering the power is reduced. Thus, higher values for peak power are obtained.
You may skip measuring the peak power by simply multiplying the continuous power by 1.1.

The power bandwidth is defined as the bw for which 1/2 of the rated continuous power can be
obtained.

Actually, DIN 45 500, CNF 97-330, EIA RS-426 and the encompassing IEC 268-5 specify not
pink noise, but pink noise filtered by a filter that provides significant attenuation in the low and
high frequency region of the spectrum to more closely model the long-term spectral distribution of
music. Pink noise itself does not accomplish this

PMPO (Peak Music Power)

So called "music power". This power figure tells the power which the amplifier can maximally
supply in some conditions. PMPO rating gives the highest measuring value, but this info is quite
useless, because there is no exact standard how PMPO power should be measured.
The reason for this power rating was to show the max capability of equipment for recreating
strong musical transients like kettle drums and the like. Similar thing (music power rating) was
used in the sixties, and I think it assumed a square wave that swung the whole supply range of
the output stage. This alone gives them a factor of two over a clean sine wave note. But the
ugliest thing they did was to assume that the high power lasted such a short period of time that
the power supply caps would hold the voltages steady without any drooping. In the real world, an
under powered PS could be hidden by this ruse and the PMPO might be a factor of 10 or higher
than what could be sustained on a nice instrumental performance.

Forget what adverts say about peak power or other "power terms" because they are not
standardized and anyway comparable between equipments. Just look for "RMS continuous
Power" or other reliable power rating (like DIN power).

Speaker power ratings

The nominal power for speakers is defined quite differently: The continuous power is measured
by pink noise rather than a sinusoidal signal and it is applied for 24 hours. Bandwidth of the noise
is as required / specified by the speaker. Thus the nominal power is applicable to both a single
chassis/driver and complete box. And the THD is not the limiting factor: It is replaced by the term
that the speaker should by no means be damaged. Rhe requirement is that the speaker meets
the manufacturers performance specification after the power cycle.

The maximum power is defined for woofers and boxes only. It is measured by applying sinusoidal
signals of 250 Hz and lower such that the speaker is neither damaged nor produces unwanted
output.

The AES/ANSI spec provides for two power measurements: thermal power, as you describe
above, and excursion limiting, which is determined by either the hard mechanical limits afforded
by the suspension, or the difference between the length of the voice coil and the length of the
magnetic gap.

Other amplifier specifications

Speaker impedance the amplifier is designed to drive

Many amps manufactured these days are rated only for 8-ohm-and-above loads, and not for 4-
ohm loads. This is done largely as a cost savings by the manufacturer. Amps which are capable
of driving 4-ohm loads to the same output voltage require heftier power supplies, heat sinks, and
(often) output-stage transistors: they'll be delivering twice as much current into the load, and will
be dissipating roughly twice as much heat within their output stages.

If a manufacturer chooses to quote a power rating at 4 ohms in their advertising, the amp must be
capable of delivering this much power after a 'warm up' period of operation at 1/3 power (which
level actually dissipates _more_ heat in the output stage than full-power operation).

In order to save money during manufacture, manufacturers often use skimpier power supplies,
heat sinks, and output stages - and as a result, the amps may have a 4-ohm power rating which
is _less_ than the 8-ohm rating. This is somewhat embarrassing for the manufacturer to advertise
- and, so, they often do not quote a 4-ohm power rating at all, and state that the amp is designed
to be used only with loads of 8 ohms or above.
With many such amplifiers, you can drive a 4-ohm load safely, as long as you don't try to drive it
too hard. If you drive a low-Z load to too high a volume, one of several things may happen: the
amp may begin to "clip" (sounds very harsh and distorted, may damage the tweeters), or may
overheat and shut itself down, or may overheat and burn up (all the magic blue smoke leaks out).

Methods for making 4 ohm speaker to appear as 8 ohm

 Wire a 4-ohm power resistor (10-20 watt) in series with each 4-ohm speaker. This makes
the system to be appear as 8 ohm load and is inexpensive. The cons are that the resistor
wastes power, may cause frequency response go bad because speakers do not have
constant resistance with frequency. When you play at high volumes the resistor may get
hot and burn thing or itself.
 Using 4 ohm to 8 ohm matching transformer will not waste much power, but the
transformer will be heavy, expensive and hard to find. Transformer has also problems in
playing back lowest frequencies (saturation causes distortion in high levels) and in higher
frequencies the inductance in the transformer will cause phase shifts.
 You can wire two 4-ohm speakers in series if you have two identical speakers. Problem is
that if the speakers are not identical type the frequency response and power distribution
will be uneven.
 Most "8-ohm" amplifiers can drive a 4-ohm or 6-ohm load as long as you don't try to get
full power out of the amp (if you do, it may overheat and shut down).
 Buy yourself a decent power amplifier whose output stage and power supply are capable
of handling a real honest low-impedance load. Good amplifier will be expensive but gives
best sound quality and reliability.

Dampling factor

The output impedance of an amp should be extremely low. If it's .8 Ohms, then an 8-Ohm
speaker has a damping factor of 10. If it's .08, then the amplifier provides a damping factor of
100, etc. Don't confuse the actual output (source) impedance with the load impedance that is
recommended for the amp (4-Ohms, 8-Ohms, etc).

The idea is that if the speaker is 8 Ohms, and the amplifier has a source impedance of .08 Ohms,
then the amplifier "damps" the motion of the cone by a "factor" of 100. In reality, the true damping
that the cone "sees" is determined by many things, part of which is the damping limitation
imposed by the resistance of the voice coil, usually around 5 Ohms or so for an 8-Ohm speaker.
You can see that if the speaker has 5 Ohms of resistance, the internal (source) impedance of the
amplifier (.08 Ohms for a damping factor of only 100) doesn't add much to the total resistance in
the voice coil circuit, hence has very little effect on total damping. So any modest change in the
amplifier damping factor correlates to virtually no change in total damping.

A speaker designer shoots for a certain damping (same as 1/Q) to achieve a certain desired type
of low-frequency roll off. The assumption is that the source impedance of the amplifier is 0 Ohms.
If the source impedance is .08 Ohms (damping factor of 100), very little error is introduced into
the system. Higher damping factors are getting into diminishing returns in terms of the total
damping. In practice we want a certain, relatively low damping figure for the whole speaker
system, (1.414 for a maximally flat bass response).

What is amplifier "bridging" or "monoblocking"?

When you're told a stereo power amplifier can be bridged, that means that it has a provision (by
some internal or external switch or jumper) to use its two channels together to make one mono
amplifier with 3 to 4 times the power of each channel. This is also called "Monoblocking" and
"Mono Bridging".
Bridging typical HIFI amplifier involves connecting one side of the speaker to the output of one
channel and the other side of the speaker to the output of the other channel. The channels are
then configured to deliver the same output signal, but with one output the inverse of the other.
The beauty of bridging is that it can apply twice the voltage to the speaker. Since power is equal
to voltage squared divided by speaker impedance, combining two amplifiers into one can give
four (not two) times the power.

In practice, you don't always get 4 times as much power. This is because driving bridging makes
one 8 ohm speaker appear like two 4 ohm speakers, one per channel. In other words, when you
bridge, you get twice the voltage on the speaker, so the speakers draw twice the current from the
amp.

Another interesting consequence of bridging is that the amplifier damping factor is cut in half
when you bridge. Generally, if you use an 8 ohm speaker, and the amplifier is a good amp for
driving 4 ohm speakers, it will behave well bridging.

Also consider amplifier output protection. Amps with simple power supply rail fusing are best for
bridging. Amps that rely on output current limiting circuits to limit output current are likely to
activate prematurely in bridge mode, and virtually every current limit circuit adds significant
distortion when it kicks in. Remember bridging makes an 8 ohm load look like 4 ohms, a 4 ohm
load look like 2 ohms, etc.

If your amplifier does not have built-in bridging option built in you can use an additional stage to
invert the signal for one channel but drives the other channel directly.

Classes of Amplifiers

Amplifiers do not actually increase the strength of an electronic signal. What


happens instead, the signal is copied and enlarged. There are different schemes
for amplifying the signal. There are different classes of amplifiers. These classes
are A, AB, and C. There have been some special classes such as G, created by
Hatachi. Class H created by Soundcraftsman. Class D for the so-called digital
amps and Class T for Tripath's digital amplifiers.

Class A amplifiers use one or more transistors that conduct during both the
positive and negative cycles of the signal. This Class of amplifier has the lowest
distortion but it is very inefficient and generates a lot of heat. A Class A amplifier
requires that the amplifier generate the full current no matter what the output is. If
you were simply listening to FM or watching a movie, the amplifier would be
consuming as much power as if you had it turned up to full volume.

In order to increase efficiency, Class B amplifiers use one transistor to conduct


the positive portion of the waveform and another transistor to conduct the
negative portion of the waveform. 99% of all audio amplifiers today are Class B.
Class B amplifier can be built today so that its distortions are well below what the
human ear can detect and nearly to the point where it is unmeasurable.

Many amplifiers call themselves Class A/B. In reality, very few are. Early Class B
amplifiers had a problem known as switching delay. In a class B design, a
transistor works 50% of the cycle while another transistor works 50% of the
cycle. In early class B amplifiers, there was a distortion created between the time
the devices were switching back and forth. Some people referred to this
distortion as notch distortion because there was a notch appearance on an
oscilloscope between the two waveforms.

Class A/B was created to leave the transistor conducting while the second
transistor was conducting. This created an overlap between the two signals. The
problem with this approach is that it created its own distortion called gumming.
This means that the signal would get a little fatter where the two devices were
both conduction.

Today, if you look at a properly designed Class B amplifier on a scope, you will
see no switching distortion.

Class D amps are sometimes called digital amplifiers. There is really no such
thing today as a digital amplifier. A Class D amplifier uses transistors that are
either switched on or off to represent positive or negative values. The transistors
are either on or off. The advantage of such a system is that it is highly efficient
and generates very little heat. The disadvantage is that there can be a distortion
caused between the switching of the positive and negative transistors as the
positive and negative transistors can not be on at the same time.

Many Class D amplifiers are finding their way into Subwoofers. They are
inexpensive to build and the logic is that the switching distortion is not important
in a subwoofer.

Class T amps are a more refined switching amplifier developed by Tripath. It


uses signal processing to eliminate the switching distortion of Class D. nOrh is
currently working with parts from Tripath to determine the sonic merits using the
Tripath parts. Our current view is that advantage to using Class T and Class D
amps is not to achieve better sound than can currently be achieved with standard
A or A/B amplifiers. Rather it is an attempt to create a lower priced amplifier that
offers good performance.

Probably 90% of the amplifiers on the market are designed to get a particular
RMS rating and little consideration is provided to provide any extra power from
the power supply. A properly designed power supply requires three times the
output from the power supply than required to drive the amplifier to its maximum
RMS rating. Buying and shipping large transformers is very expensive.
Manufacturers are looking for ways to cut these costs.

At nOrh, we over design and overbuild our power supplies. Our new subwoofer
amp has a massive 600 V/A R-Core transformer. The Multiamps have two
massive R-Core transformers that are specially built to offer the highest speed
possible. It is important to remember that classes of amplifiers do not describe
quality but rather topology.

Tubes vs. Transistors


Transistors are far more popular than tubes. However, there are a group of
audiophiles who continue to swear by tubes. Solid State amplifiers are more
reliable, test better and are usually less expensive. None-the-less, many
audiophile believe that transistors sound clinical. There has been a resurgence of
tube amplifiers in the past few years. The main reason is that Russia and China
never stopped producing tubes. These tubes are entering the market and
creating a new source for tubes.

Tube amplifiers are said to sound more


musical. The reason is that tube amplifiers
produce even ordered harmonics. Musical
instruments give off harmonics in even
orders. Transistor amplifiers tend to give
off harmonics that are odd ordered. These
harmonics are not pleasing to the ear as
second order harmonics are. Modern solid
state amplifiers have very low distortions
but their distortions are less tolerated by
the ear than even ordered harmonics.

One should note that while most solid


state amplifiers have very low distortions(An SE EL34 based integrated tube amp)
(Total Harmonic Distortion) for the left and
right channel, other channels are often
much higher as these specifications are
rarely noted.

One should note that while most solid state amplifiers have very low distortions
(Total Harmonic Distortion) for the left and right channel, other channels are often
much higher as these specifications are rarely noted. Subwoofer amplifiers are
particularly bad at creating odd ordered harmonics. I believe that the best tube
and solid state amplifiers sound amazingly alike. Bad tube amplifiers sound tubby
and slow. Bad transistor amplifiers sound harsh, bright and strident.

Just like you can't judge a good book by its cover, you can learn very little about
an amplifier without digging in and seeing what is inside. Generally speaking, the
most important component of any amplifier is its power supply. It is sufficient? Is
it accurate? Is it fast? Unfortunately, almost no amplifier company talks about
their power supplies or what transformers they use. I think most manufacturers
would prefer you not ask.
Balanced Inputs - I get lots of questions about balanced inputs. Most people want
to know do they sound better or not? We have to look first what true balanced
inputs do. Balanced inputs are used for professional gear. What happens is that
the signals between two amps are compared. One signal is in positive phase and
the second is in negative phase. Noise will appear to be in the same phase to the
noise component can be eliminated.

Some very expensive high-end equipment features balanced inputs. Most of


these are gimmicks. The reason is that in order to compare the two signals, you
literally have to have two amplifiers for every one. What most high-end
equipment manufacturers do is they provide balanced connectors (called XLR
connectors) and then convert this to a standard signal. The benifity is that it
allows you to use XLR connectors instead of RCA connectors. It doesn't offfer
the ability to drive longer cables and eliminate noise.

nOrh doesn't believe in gimmicks so if we ever see the need to offer balanced
inputs, it will be using a true balanced system. Also, check out "advanced
amplifier concepts" linked below.

How Audio Amplifiers Work. (by the usual design)

Here's an ultra simple description of the usual design for audio amplifiers. The
complexities are infinite. As one expert said, you might design an audio amplifier and
find that it is a dud. In other words, the problems might be unsolvable. The primary
problem is instability due to high voltage gain at the output. Noise pickup is also a
problem due to the gain.

(fig)

The output transistors are Q5 and Q6. To increase curent handling, boosters are
added. Darlingtons can be used for Q5 and Q6, and then boosters are not needed.
There must be a small amount of quiescent current through the output, typically 5-
8mA. This means Q5 & Q6 are slightly open. They are opened by Q3, which has
approximately 1.2V across it, as set by R1 and R2. Both resistors are about the same
size, but a trimmer is used with R2 to set the quiescent current. The size of R1 and
R2 is such that about one fifth as much current goes through them as through Q3. If
darlingtons are used at the output, there is 2.4V on Q3, and R2 is about three times
as large as R1. The purpose of Q3 is to function as a voltage regulator.

Q2 sends the signal up the column. Its collector must get close to the rails so as to
not waste voltage which would increase heat. The system is tested by rolling the
voltage up and down on Q2 and watching for linearity. There should not be more
than about 10 or 20% variation in output quiescent current from rail to rail.

A major problem is in the fact that Q2 must be close to the lower rail so as to not
waste voltage. The result is that it produces very high voltage gain. The voltage gain
is approximately the resistance at the collector divided by the resistance at the
emitter. The resistance at the collector is replaced by a current source (Q4), which is
the equivalent of infinite resistance. The components will reduce the gain from
infinity to something like 50,000. The goal is to reduce it some more, which is what
R3 does. It would typically have 1 or 2 volts across it. The current through the
source might be 5-8mA

Q1 not only functions as an input transistor, it also creates inverting feedback, which
centers the output. (The arrows from it and the output are supposed to be linked
together.) R4 might typically be 15K and have 5V across it, as determined by the
size of the resistor at the collector of Q1. So the biasing circuit at the input of Q1
would then set the input voltage at about 5.6V bellow center. If for example, the
supply voltage was 24V, and R4 had 5V across it, the input bias would be set at
6.4V, and the output would center at 12V.

C1 hardens the emitter creating some gain on Q1. The gain can be reduced by
adding resistance in series with the capacitor. The cap might be 100µF.

Keeping the output from oscillating is usually difficult. Capacitors are often used in
various places to reduce the speed for that purpose. Sometimes inductors are used
at the output to increase stability.

The Current Amplifier.

The current amplifier drives the speakers. It has no effect upon the signal. It is the
perfect black box which allows the voltage amplifier to drive a heavy load as if it
were no load at all.
 
Basically, the current amplifier consists of four transistors. Two divide the signal at
the input, and two create a push-pull output. The result is a totally symmetrical
circuit, as shown in Figure 1.

Figure 1.   Current Amplifier.


Current can be supplied to the first stage through two resistors (at X and Y), when
low voltages are used, but usually current sources will replace the resistors.

The current which flows through the output transistors when there is no signal is
called "output quiescent current," and it is being represented by the symbol Ioq. It's
value must be controlled when booster transistors are added, as with audio
amplifiers.
 
The input and output transistors must consist of two matched pairs (NPN=NPN and
PNP=PNP), because they must have the same base-emitter voltages. Their betas
(hFE) must be similar, because it determines the amount of current that flows
through them and thereby determines their base-emitter voltages. The betas can
easily be matched within 5 or 10%, which is adequate, by using a digital voltmeter
with a beta tester. In the absence of a beta tester, a breadboard can be used, as
shown in this test circuit.

Betas do not have to be matched between opposite type transistors, but if they are,
no input current will be seen under quiescent conditions, because all of it will flow
out the base of one input transistor and back into the base of the other.
 
The power transistors are used as boosters, as shown in Figure 2. The boosters do
not influence the output voltage; they increase current only.

Figure 2.   40V Current Amplifier for Audio.


This current amplifier is functional down to about 5V and up to 40V or ±20V. It has a
ripple rejection of 68 dB or better, increasing with the load.

The transistors Q5 and Q6 are current sources. They supply a constant current
through their collectors, in this case about 2.7mA. More current is OK, if heat is no
problem.

To calculate the current through the sources, there will be about 0.6V across the
resistor at the emitters (220), because there are 2 diodes from base to rail, each
having 0.6V, and the base-emitter junction has about 0.6V.

The current which flows through the output transistors when there is no signal is
called "output quiescent current," and it is being represented by the symbol Ioq. It's
value must be controlled when booster transistors are added, as with audio
amplifiers. It is measured as the voltage across either of the 120 resistors.

The Ioq of this current amplifier must be atleast 1 mA. It will be approximately the
same as the current through the sources (2.7mA), when the input and output
transistors have the same betas.

When the input and output betas are not matched, the Ioq will not be the same as the
current through the sources. If the input transistors have the higher betas, the Ioq
will be less than the current through the sources.

For example, consider these betas: Q1=135, Q2=125, Q3=185, Q4=165. The inputs
are both higher than the outputs, so the Ioq will be less than 2.7mA - maybe 2.0mA.
Q3 and Q4 should be reversed, which would increase the Ioq to something like
3.0mA.

If in doubt, the entire current amplifier can easily be set up on a breadboard for
testing. Boosters can be omitted, but they should be safe if located close to the
output transistors.

The size of the 120 resistors must be tailored to the Ioq, so the boosters are off
during quiescent conditions. That means about 0.3-0.4V across those resistors.
When the booster transistors heat up, their base-emitter voltage decreases; so there
must be a wide margin in closing them.
 
Ringing.

The booster transistors are attached to aluminum for cooling. Wires going to them
cause them to ring at 20-40MHz. Since ringing depends upon energy being stored in
wires due to their inductance, it cannot occur at low frequency. Using short wires
between boosters and the rest of the circuitry reduces the tendency to ring, but a
complete absence of wiring is not totally feasible, and therefore capacitors (150pF)
are used to limit the speed.

Those capacitors produce a very linear slew rate for the current amplifier, because
they are acted upon by current sources. To determine the slew rate, the formula is
V/S = I/C. It reads: volts per second equals current over capacitance. Slew rate is
expressed as volts per microsecond; so the result is divided by 106. The current in
the calculation is that produced by the sources.

For example, in the above circuit, the current is 2.7mA, and the capacitor is 150pF.
So the slew rate is (I/C) 2.7x10-3/150x10-12/106 = 18V/µS. If need be, the slew rate
could be set as low as 5V/µS and still be fast enough for audio purposes.

If wires to the boosters are 5 inches (12.5cm) long or more, only trial and error can
determine the requirements for controlling the ringing. Bunching of wires to the
boosters increases ringing, and therefore, headers should not be used. The
recommended procedure is to put the entire current amplifier on the same aluminum
plate as the boosters, as shown later. Besides making ringing easy to control, it
eliminates a clutter of wires.

Long power supply wires also contribute to ringing; but they only become significant
when long wires go to the boosters. Because of such complexities, precarious circuits
should only be tried when a scope is being used.

The ringing must be thoroughly controlled, because it will cause the boosters to lock
open and burn out. Under precarious conditions, the ringing might appear to be
controlled when testing with a resistive load; and then speakers with long wires will
cause ringing to break out. But with the conservative designs shown, the ringing is
easy to control.
 
The Basic Audio Amplifier.

Figure 3. shows the simplest form of the audio amplifier. It uses a single op amp
and a moderately high impedance (220K) feedback resistor, so input current is low.
The input current is calculated as the maximum output voltage divided by the
feedback resistor. The feedback resistor could be reduced to 47K and still only draw
0.35mA of input current, which should be acceptable.

Figure 3. Basic Audio Amplifier.


The current amplifier can be included in the feedback loop as shown, which removes
offset voltage at the output. Or feedback can go from the op amp output back to the
input. There would be a few tenths of a volt across Rx creating output offset, but
that amount should be no problem; so the feedback resistor can be connected to the
output of the op amp (removing the C.A. from the feedback loop), which increases
stability, if circuits are precarious, like a lot of wiring to the boosters.

The volume control is used on the inverting input, which minimizes gain and
simplifies. A noninverting volume control could be used. With the inverting volume
control shown, the potentiometer should be about twice the size of the smallest input
resistor, and never more than five times as large. The reason is because it produces
its own logarithmic taper in proportion to the input resistor. Everything above the
wiper is added to the input resistance in determining the gain. So if the
potentiometer is five times large than the input resistor, the gain is reduced to half,
when the volume control is only 20% down.

The gain formula for the inverting amplifier is the feedback resistor divided by the
input resistor. The gain on the offset voltage is the feedback resistor divided by the
size of the volume control potentiometer. It is low enough that no trimming of offset
voltage is necessary.

The feedback capacitor compensates for input capacitance and stray capacitance
which produce overshoot. Overshoot shows up as a spike on the leading edge of a
square wave. It does not show up on a sine wave; but removing it seems like the
prudent thing to do, particularly if the amplifier is going to be used for laboratory
purposes.

If a scope is not being used, the following formula indicates the approximate amount
of feedback capacitance required to compensate for input capacitance. If there is
stray capacitance on the board, an additional picofarad or two might be needed to
compensate for it.

1x10³
 
picofarads (feedback)  =    
  feedback resistor
 
If your browser doesn't read tables, it says "picofarads (feedback) equals 1000
divided by the square root of the feedback resistor."

Stray capacitance is caused by the proximity of other voltages, including ground, to


the high impedance area of the inverting input. The high impedance area is that part
of the wiring which is isolated by large resistors. The mass of wire on the inverting
input also contributes to stray capacitance.

Stray capacitance should be minimized by keeping at least 2mm of space around the
inverting input, and even more around wires connected to it, and by minimizing the
length of wire connected to it. When the volume control connects to the input, the op
amp is located near the potentiometer, and a stiff wire goes to it for straight line
routing.

The noninverting input is not affected by stray capacitance. Sometimes a resistor is


connected to it for protection, as shown; but it should not be necessary with FET
inputs.

The feedback capacitor reduces the bandwidth in proportion to its size and the
impedance. The following table shows the largest feedback capacitor that can be
used with a particular feedback resistor, while maintaining a flat bandwidth up to
20KHz.

feedback resistor 18k 47k 100k 220k


100p
feedback capacitor   27pF 10pF 5pF
F
 
Above 220k, the bandwidth depends upon stray capacitance. With a clean board, a 1
meg resistor will allow a 20kHz bandwidth.

When the C.A. slew rate is reduced to 5V/µS, the overshoot spike is increased; and
then the feedback capacitor is increased by 50% or more. Current through the
sources should also be increased when increased stability is needed.

In determining the number of gain positions to use, here are some general concepts.
The lowest gain is most convenient around 3 to 5. The highest gain should be able to
produce full output voltage from signals as low as ±0.2V. A typical signal is ±0.5V.
Since a bridge amplifier produces four times as much wattage from any voltage, its
gain should start lower; but the size of increments should be the same as for other
amplifiers.

The smallest step desirable for each gain increment is a factor of 2, the largest, a
factor of 4. When a rotary switch is used, I would suggest a factor of 3, which
reduces switching compared to a factor of 2. With the 10W amplifier, a toggle for
gain select is highly convenient. The low gain is then set at about 10 or 15, and high
gain at about 50. That much jump between gains is not ideal; but the convenience of
a toggle justifies it.

With high gain and low impedance circuitry, the input resistor gets small; and
additional resistance on the line must be taken into account. For example, I put a 1k
resistor on each line as it comes in to protect sources from shorting in the mono
mode. The 1k resistor becomes significant when the input resistor is small. If the
input resistor is supposed to be 4.7k, substitute 3.9k; and the total with line
resistance will be 4.9k.
 
Op Amp Current Amplifier.

Nearly all op amps would benefit greatly from current amplifiers at the output. This
not only hardens the output but creates stability and symmetry while reducing rail
voltage. This would allow low current and low drift op amps to drive difficult loads.

There is a real need for this, which shows up with ac loads. The wave shape is often
distorted due to the lousy outputs which normally exist. To cope with this problem,
designers are forced to use hard driving op amps which do not have the best drift
characteristics. With the current amplifier, drift does not have to be sacrificed for
demanding ac loads.

Figure opc.   Op Amp Current Amplifier.

This current amplifier has low current levels for integrating into the chip. If you add
this circuit to an op amp using discrete transistors, you can add more current by
making the resistors smaller.

Be sure to match betas on input and output transistors. On an IC, they should match
very well.

When this current amplifier is integrated into the op amp chip, the normal output
protection can be eliminated. The diodes used with the current amplifier are much
superior to normal protection, being faster and more stable.
Output Protection

Output protection by current limiting is more convenient than usual, consisting of


two diodes from the protection resistor going back to the input of the current
amplifier. The diodes are fast and stable.

The voltage across the protection resistor at cut-in is a little less than 0.6V, because
there is a slight increase in the base-emitter voltage of the output transistors when
loads are heavy. For example, the 10W amplifier has about 0.57V max across the
protection resistor, while the other two amplifiers have about 0.50V.

The current limiting must be set for peak voltage, not rms voltage. (Rms is 0.707
times peak voltage.)

The resistor at the output of the op amp (Rx) is used to prevent the op amp's
protection circuit from cutting on and off creating an oscillation, when the protection
diodes are conducting.
 
Over-wattage Protection.

With current limiting at the output, low power amplifiers are virtually indestructible.
But with high power amplifiers, current limiting creates so much over-wattage on the
boosters during short-out to ground that it might do no more than create a false
sense of security. When short-out to ground occurs, all of the heat that would be on
the speakers and about 50% more gets dumped onto the transistors.

Therefore, I put over-wattage protection on the 90V amp, but not on the 10W or
bridge amps.

Wattage is viewed as the equivalent of heat in electronics. The heat is located where
resistance or impedance creates voltage from current flows through it.
 
Below is an over-wattage protection mechanism which removes heat from the
circuits by cutting the signal for a timed interval when the current exceeds a certain
level. (Figure 4.)

Figure 4.   Over-wattage Protection.


A capacitor delay holds the signal off for about 1.5 seconds. Diode current limiting
must be used with it, because there is about a millisecond of delay in the response of
the relay. The diode mechanism is set here at 6 amps (for the 90V amplifier); and
the other mechanism at 5 amps. The latter must be set lower than the first, or it
won't cut in. A reed relay is used to cut the signal because of its absence of voltage
and two directional conduction. A 15 resistor is used on the relay line, because it
improves stability by allowing about 0.5A to be conducted to the load. A few
milliamps of current flow through that resistor from the output of the voltage
amplifier.

Since there is so much variability in resistors and junction voltages, the cut-in
currents should be tested. Parallel additional resistors if needed to make
adjustments.
 
A low Impedance Audio Amplifier.

Low impedance amplifiers are preferred, because they are slightly more resistant to
noise pick up, considerably more resistant to radio frequency interference (RFI), less
sensitive to stray capacitance and much faster making designing easier. The
feedback capacitor can be quite a bit oversized with low impedance.
 
Below is an example which uses a noninverting op amp as a preamp allowing a high
impedance input with low impedance elsewhere. There is no problem with noise or
RFI with this type of high impedance, because the input creates a low impedance
path. (Figure 5.)

Figure 5.   Low Impedance Audio Amplifier.


The gain for the main amp should be set at about 3 or 4 to minimize the probability
of clipping by the preamp. In other words, the preamp only needs 1/3 or 1/4 of full
volume that way.

The preamp should not be what is called a preamp, because they are designed for
picking up very weak voltages; and they, are not easy to use for this purpose. Op
amps are used for all signal handling, because they are designed for that purpose.

The feedback resistor for the main amp is 18K; so it creates a load of 1mA at full
volume for the preamp. The preamp is an LF411 because of its low offset, which
eliminates the need for a coupling capacitor between the two op amps. The input
coupling capacitor (0.022µF) is tailored for the 10M resistor, as described later for
coupling capacitors. The gain select resistors are switched to ground; and the gain
formula changes to that of a noninverting amplifier, which is this:

Gain = R-in + R-feedback / R-in.


 
Nonsplit Power Supplies.

If split power supplies are not possible because of battery powered devices, the
same circuitry can be used; but centering and output capacitors must be added.
These capacitors reference the signal to a voltage halfway between the ground and
the positive voltage. The centering capacitor creates the equivalent of a false ground
for the volume control and the noninverting input. The circuit below shows how to
create a very high ripple rejection by using two centering capacitors creating a slow
line and a fast line which are linked together through diodes. The fast line takes the
voltage to within 0.6V of center (at turn on); and then the diodes cut it off. The
voltage for the centering lines should be the same as that of the op amp, so the op
amp is always operating half way between its two voltages. (Figure 6.)

The slow line shown here has a ripple rejection of 80dB, which is quite a bit. If the
supplies are quiet or regulated, one line between the extremes should suffice.

Figure 6.   Nonsplit Power Supplies.


The size of the output capacitor is determined by the amount of current involved and
its ripple current rating. Using an oversized one allows lower frequencies to be
conducted.
 
Coupling capacitors.

Coupling capacitors serve as automatic voltage level shifters. Each side seeks the
average voltage that it sees. The rate at which it moves to that voltage at start up
depends upon the line resistance impeding the flow of current.

However, it is not exactly the average voltage that is centered; it is the voltage
integrated with time. This quantity is about the equivalent of average power. This
technicality shows up when the signal is a square wave with something other than a
fifty percent duty cycle. The peak voltage will not be the same distance from the
reference voltage on both sides of the wave, because the area under the wave curve
must be the same for both halves of the wave. Here is what one sees:

The coupling capacitors must have the right relationship to the impedance. If they
are oversized, they take too long to center when they start up from an off voltage.
With these amplifiers, the voltages are usually near ground level at start up; so the
capacitors can be oversized without much slide in voltage.

The minimum size for the capacitors is dependent upon the lowest frequency and
impedance. An undersized capacitor will not conduct low frequencies at full volume,
because the voltage slides toward center during each half of the cycle.

For audio frequencies, a suitable combination of coupling capacitors and line resistors
is this:

1µF - 220k 4.7µF - 47k 22µF - 10k 100µF - 2.2k

2.2µF - 100k 10µF - 22k 47µF - 4.7k 220µF - 1k


 
The resistors are those on the signal line such as the ones which set the gain. When
a coupling capacitor connects to very high impedance, such as the input of a FET or
closed switch, a resistor to ground references the voltage on the high impedance side
of the capacitor.
These capacitors are a little larger than is sometimes used, which allows them to
produce a flat response down to 5Hz.

When switching gain positions, more than one coupling capacitor might be used. I
often switch one coupling capacitor for two gain positions, except with the ultra
simple design, where one capacitor for four positions is adequate, because it can be
considerably oversized with no problems.
 
Switching for 10W Amplifier, Mechanical and Electronic Forms.

(Fig)

With electronic switching, the input capacitors are moved ahead of all of the
switches, so the voltage is centered at ground level before going through the
switches. Four capacitors are then used.

 
CMOS Switching.

CMOS switches (4066) have a maximum voltage of 15V. It is used as ±7.5V. The
incoming signal should not exceed the supply voltage; so protection diodes are used
infront of the first switches on the lines. The protection diodes take off at 0.6V inside
the supply voltages (±6.8V), so the signals max at the supplies. If zeners are used
for the supplies, they have diodes with them to set the 0.6V.
(Fig)

Zener diodes have quite a bit of variation; so they should be pretested at the current
levels used, or the next size lower should be used. Small regulators could be used.

The incoming signal would be switched in this manner:

The 1M resistor to ground defines the voltage on the capacitor when it is isolated by
the switch, so turning on the switch does not throw an offset voltage on the line. Its
size can be optimized by making it ten times larger than the largest resistor on that
line. It then produces no more than a 10% alteration of the capacitor's effectiveness.
Making it much larger would allow it to create an offset voltage of its own, because
the leakage current through the caps can be significant. The CMOS switch produces
no significant current or voltage when off.
 
The FET Switch.

FET switches work fine; but they are a little harder to construct than CMOS switches.
Their only advantage is that they handle more voltage; and that is only relevant
when tone controls are used. Even then, CMOS switching can be used with tone
controls, when they are tailored appropriately, as was done with the 90V amplifier.

The most commonly available 40V FET is the 2N5460 PFET, which was tested for
switching on a 10W amplifier. Its design is this: (Fig)
The on resistance of this FET is about 500 or more, which must be taken into
account for high gains. There are 40V NFET choppers which have a very low on
resistance, such as the PN4393 or PN4092.

The pinch off voltage should not be high, because it is wasted voltage. If an NFET
circuit is used, it would require opposite voltages.

The 10M resistor is what opens the FET when it is conducting. The shutoff
mechanism requires the 1M resistor to ground whenever the incoming side might be
isolated, such as a disconected input line or a coupling capacitor might cause. Here,
it is used only on the four lines coming in. Otherwise, the signal defines the source
voltage allowing shutoff to occur. An improper shutoff puts a biase voltage on the
line. Even a complete shutoff will put a biase voltage on the line when exposed to it
as occurs when switching the gain setting resistors. Therefore, a coupling capacitor
must be located after the gain setting resistors and just before the volume control.
The incoming capacitors might be eliminated, because the FETs handle a lot of
voltage.

Digital CMOS circuits open and close the FET switche through a bipolar transistor.
The negative voltage for the CMOS must be the same as the emitter voltage, which
in this case is about -20V. A zener diode then sets the positive voltage for the CMOS
at 5V higher.

The LED indicator should have a diode with it in series, because LEDs are not
specified for 20V reverse which can cause them to conduct.
 
Digital Switching.

When these circuits are switching 4066 CMOS, they should use the same voltage as
it.
(fig)

Flip flops function as two position counters for the line select and stereo-mono. A
4017 counter is used for gain select. It recycles from the high position to the low
position, when the 15 pin is connected to the count above the highest used. So to
get to the low gain, one must go through the high gain. There is no difficulty in doing
this with medium powered amplifiers; but for high powered amplifiers, an up-down
counter is better. Separate swithces are then used for up and down. Such a circuit is
shown later for the 90V amplifier.

Using more than two lines coming in for each channel might often be desireable; and
then the line selector would have to be a 4017 set up like the gain select.

Digital counters must be operated with bounceless switches, or switch bounce would
often cause them to move more than one position. The bounceless switch shown can
be used at any CMOS voltage without much change in characteristics. The output of
the schmidt trigger goes into the clock pin of the counters.

The LEDs can be controlled with 74C901 inverting buffers. These buffers sink more
current than they source; so they are operated on the negative side. Their inputs
tolerate overvoltage; so the plus supply only has to go up to ground level. Infact, it
is two diodes below ground; so the reverse voltage on the LEDs does not go over
specifications.

In the 10W amplifier which used digital switching, all of the switching circuits were
put on a lower board; and only the op amps and current amplifiers were put on the
upper board.
 
Features of the 90V Amplifier.

A lot of features were added to the 90V amplifier in order to evaluate them. Eight
gain positions were used with up-down counters and a clock for the motion. Besides
two push buttons for the gain of each channel, there are two more for controlling
both channels simultaneously. The latter could be omitted.

Two bargraphs were used to indicate the gain position of each side. The unused LEDs
on the bargraphs were concealed behind the panel. If the amplifier were used for
stereo purposes only, one set of switches and one bargraph would probably be
adequate. Construction would be a little easier if the bargraph were replaced by eight
small LEDs. Infact, the number of gain positions could be reduced to five. The
bargraph had a high visibility; so a lens was not required. Pretesting the bargraph for
visibility is advised. If contrast needs to be enhanced, windshield tinting might be
used rather than a lense.

If rotary switches are used, six gain positions would probably be more desirable than
eight for the 90V amplifier.

A gain limiter was put on the front panel. It consists of a rotary switch which stops
the counter at a selected upper limit.
 
A good feature is switching of the volume controls allowing them to be used
separately or combined. At a flip of a switch, the left side becomes a balance control,
while the right side controls both volumes. Two dual 50K potentiometers are
required. (Controls)

The Balance Control.


The black boxes are electronic switches such as 4066 CMOS.

The balance control has ideal characteristics. First, it is nondistorting, as it functions


as a resistor divider on the signal line; and then an op amp returns the signal to
unity gain. The center half of the potentiometer motion produces a linear response,
as one side decreases at the same rate the other increases. Then towards each edge,
the boosting discontinues, while the cut drops to zero. The total boost is about 1.25.
To create these characteristics, the resistor divider should have a ratio of 1 or 2 to 1
at center position. This ratio is the incoming resistance divided by the resistance on
the potentiometer, or one half its size. The op amp then has resistors which set its
gain equal to the reduction, or about 3. In order to maintain unity gain in both the
on and off position, a parallel resistor is switched also. If the balance control is not
switched, only one resistor is needed 39K. (Controls)
 
Tone Controls.
The tone controls were tailored to do more cutting than boosting, so the resulting
signal would not have too much voltage for the CMOS switching. The user does not
see that fact, since the flat response centers half way between cut and boost.
However, the net gain is about 1/2 at center position. This result is only detectable
as a change in volume when the tone controls are switched. If it mattered, an op
amp with a gain of 2 could restore unity gain; but it was not done here.

The treble centers around 1kHz. It boosts to 1.8 and cuts to 0.13. The bass produces
a strong effect at 300Hz and lower. It boosts to 2.0 and cuts to 0.12. The center
position is flat and gains at 0.5.

A bypass switch should always be included with tone controls, so distortionless sound
is available without having to watch the position of the tone controls. A mechanical
switch was used for the bypass, since that switch is not often used. A header was
used for that switch; and it does not produce too much channel crossover, though
the wires have thick insulation to minimize the problem by keeping wires separated a
little.

Separate op amps were used for the tone controls, so spacial arrangements could be
controlled easily. The entire tone control circuitry was located on the edge of the
board; and stiff wires went up to the potentiometers. A header might have been
used without too much pickup or stray capacitance. Dual potentiometers were used
for tone controls; so both channels could be controlled simultaneously.
The edge of the board looked like this: Fig.

The Quad Input Amplifier.

High voltage amplifiers are not commonly available on ICs. In the past, constructing
them from discretes was a little cumbersome. But this new voltage amplifier, made
possible by the current amplifier, reduces the problem to simplicity.

An important characteristic of this amplifier is that temperature drift is minimal,


because it occurs by about the same amount in both directions. For this reason, FET
inputs are feasable, even with their high drift. However, FET inputs are not needed,
because another important characteristic is that the input current is gained through
two stages rather than the usual one. If the betas are similar for the transistors, the
quiescent input current dissapears by going through both bases. (Fig)

Quad Input Amplifier Circuit.


The internal gain (analogous to open loop gain) can be quite high for this amplifier.
However, it reduces rapidly with a load; and even the feedback resistor will bring it
down to 10,000 or less. If FETs are used on the input, the internal gain will be a
maximum of 1,500; and the feedback resistor will reduce it to about 1,000. That is
still sufficient gain for audio purposes. Infact a gain of 500 is sufficient. However, in
these circuits the feedback resistor does not load the output, because it connects to
the current amplifier.

Another important characteristic of this amplifier is that it has a very high slew rate,
which is needed when high voltages are used. I have measured the slew rate as high
as 115V/µS. That's five times faster than the fastest commercially available op amp
of this sort. A small feedback resistor, 22K, and a gain of 5 are required for that slew
rate. Under the conditions used here, the slew rate peaks at 40V/µS. However,
unlike most amplifiers which have a fixed slew rate, the slew rate here increases as
the output supply voltage increases. The net result is that the maximum possible
frequency will be about the same at any output supply voltage.

The voltages supplying current to the first stage must be stable. Variations in the
current will influence the Ioq. If one side only is varied, the offset voltage changes. So
a trimmer is used to adjust the offset in that manner. But since the Ioq also varies,
minimal trimming should be done. To minimize the trimming, the zeners should be
pretested for similarity within atleast 2%. Otherwise, voltage regulators should be
used.

The input transistors do not have to have a high voltage rating, provided that some
of the voltage on the right side is taken up with resistors. Here, the voltage rating is
high enough; but resistors (27k) are added to take some of the heat off the
transistors minimizing drift. The current through those resistors is 1/2 the current
supplied to both sides of the input stage, which is 0.55mA through the resistors. If
the power supply has a lower voltage rating, the 27K resistors should be smaller or
removed. Since the power supply voltage drops with a load, the resistors cannot be
marginal in size.

The output of the quad input amplifier might use TIP61C & 62C for handling more
heat. The quiescent current could then be increased. The only reason for using such
a low quiescent current was to minimize the heat.

The thermal noise of the quad input amplifier is about twice that of a one sided
amplifier, which amounts to about 3mV. The 475 resistors between the emitters of
the inputs hold down the thermal noise. The output noise is not gained; and at 3mV,
thermal noise is not audible on speakers.

You won't find a higher quality voltage amplifier in terms of stability, speed and
absence of resonance. There are three reasons: One, there are only two gain stages
instead of the usual three. Two, there are no internal capacitors. And three, the
emitters are close to the rails.
 
High Voltage Current Amplifier.

(Fig)

Power transistors are used throughout the high voltage current amplifier to dissipate
heat. The input and output transistors should be attached to the same aluminum to
equalize temperatures and minimize drift in Ioq. The current sources do not have to
be attached to aluminum.

The power transistors can be attached to a plate with the boosters. With the plate
shown, the p.c. board is above the plate, with wires extending over the edge for
attachment to the transistors. The copper side is down. An over-wattage mechanism
can be crammed onto the same board, with careful designing. The plate is attached
to the back panel for cooling.

The ratings for TIP33C & 34C are 10A, 100V, 80W. You can use TIP35C (NPN) and
TIP36C (PNP), rated 25A, 100V, 175W. They are presently available at Mousers.

The voltage across the protection resistor is 0.50V (6 amps) at cut in, depending
upon the nature of the protection diodes.
 
Gain Chart for 90V Amplifier.

gain, gain, R to
total preamp R in R feedback caps ground

380 21 4.7k 100k      \


47µF 100k
180 10 10k 100k      /

82 4.5 22k 100k      \


10µF 470k
38 2.1 47k 100k      /

18   15k 270K      \
22µF 270k
10   27k 270K      /

5.7   47k 270K      \


4.7µF 1M
2.7   100k 270K      /

 
The preamplifiers are op amps. They do not have to be the things called preamps.
The specially designed preamps have the purpose of picking up weak signals as
noiselessly as possible. These signals are not weak; and they are handled more
easily with op amps.

A 5pF capacitor was found to be suitable for the feedback of the preamps used here.
Feedback capacitors can usually vary by as much as 50% without producing much
distortion.

The highest gain shown on this chart could be omitted, particularly if tone controls
are not reducing the gain. A gain of 180 will reach down to ±0.2V signals while
producing a maximum output voltage, which is quite adequate. A typical signal is
±0.5V or more.
 
Switching for 90V Amplifier.

The black boxes are 4066 CMOS switches.

(fig)

Each line coming in has capacitor coupling on it, so the signal is centered before
going through the CMOS switches. Where a switch is on, the ground referencing is
provided by a feedback resistor which holds the inverting input of the op amp near
ground level. Where a switch is closed, ground referencing is provided by a 470K
resistor. The resistor sizes vary for the gain setting circuitry, as indicated in the
righthand column of the gain chart.
Protection diodes for the switches are placed infront of the four switches at the input
only. Elsewhere, signal voltage is limited by the supply voltages for the op amps.
Two separate diodes must be used on each signal line.

At the balance control, all four CMOS switches are switched simultaneously using the
same signal. A couple of wires with heavy insulation go from the CMOS switches to
the potentiometer for the balance control.

At the same time the balance control is being switched on, the volume control for the
left channel moves from the left control to the right control. A mechanical switch is
used to minimize stray capacitance. It must be a 3PDT. Two of the poles are used to
isolate both sides of each potentiometer. The third pole is used to,control the CMOS.
The right channel volume control does not have to be switched.

Digital Circuits for 90V Amplifier.

Most persons won't have time to add digital switching. Instead rotary switches can
be used for variable gain. A 6P2T rotary allows 6 gain positions, which is adequate.

Digital circuits are on page 5. If you want to skip over it, you can proceed to page 6.

Digital Circuits for 90V Amplifier.

The numbers inside the gates are the chip numbers that I used, as indicated in the
sketch for the board. Since I did a lot of modifying, some of the chips aren't located
ideally, and they could be reevaluated, particularly since a few gates might be omitted
for some things.
 
Below is the gating for the preamps. The gates look for a high signal in the first four
gains. If a high is present, the preamp switch stays closed; and only when all four are
low does the preamp open. (Fig)
 
General Scheme.
(Fig)

A clock allows gain positions to move with one push of the button. It does get used.


Microprocessor control wasn't needed. The clock oscillator frequency adjusts from 12
to 35khz. It is then divided by a factor of 4096 in the 4040.

The oscillator was moved to the power supply board, so the trimmer would be
accessible. Its negative voltage is connected to the gates preventing it from
functioning except while switching and thereby keeping the oscillator noise out of the
rest of the circuitry.

The gating has some tricky requirements, because some gates have to be activated
before the signal goes through. The signal was therefore delayed with a flip flop (chip
20). The delay circuitry is shown below. The NOR gate going into it is a leftover gate
functioning as an inverter.

The 5 pin of the 4029 prevents further counting when it is high. It rests high; and only
when a count is being added does it go low. One of its functions here is to prevent the
counts from going beyond the desired end points. The end point signals come out of
the decoders as lows, when the highest or lowest gain positions are reached. With a
rotary switch being used as a gain limiter, the upper signal comes from the center pin
of it. But it would normally be omitted; and the upper signal would come from the 9
pin of the decoder, when eight gain positions are used. The upper and lower end point
signals go into different NOR gates, because one is reactivated by an up signal, and
the other by a down signal, coming from the switches.

The reset mechanism jams in a predetermined count, as shown later. The transistor
circuit attached to it forces the reset during turn on.
Gating the clock is tricky, because it could make a count in one half of a time unit. To
prevent that, it should not make a count in its first change of state. The first count is
forced by the switch, when the 9 pin of the delay comes down.
 
Here is the delay mechanism:

(Fig)

This delay mechanism requires a clock which sets the delay at one pulse plus a little
additional time which occurs betwen the start and the first up motion by it. If the
oscillator is omitted, the delay could be produced by running the signal through several
gates. About six inverters would probably be the best method. Four might suffice; but
it would be risky.
 
A simplified form of the gating is shown below. This circuit has not been tested as
such; so one might leave a little extra space on the board for debugging.
Here's a general delay mechanism.

(Fig)

The general delay mechanism can be used with any gates. The AND gates are not an
essential part of it. The resistor and capacitor would substitute for the six inverters.
 
To use just one up and one down switch for both channels, omitt one counter and one
decoder, and fan out from the remaining decoder. Here are the decoder pinouts:
(Fig)

The gain position that occurs upon reset is determined by a binary code on the jam
inputs. The left side picks up the code from the right side. A quad pc switch was used
for setting the code. The switch was located near the edge of the board for accessing,
since there were two boards above it. When the switches are open, resistors to the
negative voltage create low signals. For example, to reset the counters to the fourth
gain position, the binary code for three is used (0011).

If reset is not used, the 1 pins of the 4029s are fixed low; and the four jam inputs are
stabilized by connecting to either rail.
 
Here are the pinouts for the line selectors. When 74C901s are driving LEDs, their
positive voltage is two diode levels (1.2V) below ground - that is, when the negative
voltage is -7.4V. There is then not an excess of reverse voltage on the LEDS.
(Fig)
 
Here are the pinouts for gain switching. These are 4066 switches located on the
amplifier board, not the digital board. The linked lines connect one capacitor to two
resistors.

(Fig)
 
The Digital Board.

The headers are machined pin DIP sockets or similar for both males and females. They
have #30 kynar wires soldered into the top.
(Fig)
 
Below are header pinouts. Header #26 and #28 go to the bargraph LEDs. Headers
#27 and #29 are for the control pins that operate the CMOS switches on the next
board up.

Coming out of the decoders are inverters and inverting buffers for driving LEDs. Some
of these pins have additional connections which are not shown. For example, the
preamp gating takes off from the same points as headers #27 and #29.
(fig)

#32 header,   #31 header,


gain limiter sw. & LEDs
pins                from pins                       from
  1 - ¥4L - - - 4[3]   1 - AL LED - - - 1[18]
  2 - ¥5L - - - 5[3]   2 - BL LED - - - 3[18]
  3 - ¥6L - - - 6[3]   3 - St LED - - - 11[18]
  4 - ¥7L - - - 7[3]   4 - AR LED - - - 5[18]
  5 - ¥8L - - - 9[3]   5 - BR LED - - - 9[18]
  7 - ce L - - 11[16]   6 - 150 LR - -  res
  8 - ce R - -  5[16]   7 - 270 St - -  res
10 - ¥5R - - - 5[9]   8 - 2.2k sw - (-)7.4V
11 - ¥6R - - - 6[9]   9 - St sw - - -  5[25]
12 - ¥7R - - - 7[9] 10 - ABR sw - - - 3[25]
13 - ¥8R - - - 9[9] 11 - ABL sw - - - 1[25]
14 - ¥4R - - - 4[9]
 
#30 header,
sw.etc
pins                            from
  1 - L up - - - - - - 5[24]
  2 - L down - - - -  9[24]
  3 - combo up - -  11[24]
  4 - combo down - 13[24]
  5 - bal & res + - - +7.4V
  6 - 150 bar gr. - - res
  7 - bal cent - - - - 4[13]
  8 - 2.2k sw - - - (-)7.4V
  9 - reset - - - - -  1[10]
10 - 270 bal - - - - res
11 - bal LED - - - -  5[13]
12 - bal LED - - - -  3[13]
13 - R down - - - -  3[13]
14 - R up - - - - - - 1[13]
 
¥ is gain position.
ce is center pole on rotary switch.

These are female headers attached to the board. Header #32 is for the gain limiter
only. Both channels use the same rotary switch for gain limiting. The signals come
directly off the decoders. So there is a fanout on those outputs which is not shown on
the earlier chart. The signal comes back from the center of the rotary switch to pins 7
& 8 of the header.

Header #31 is for the left side of the front panel including the LEDs and the 3
pushbutton switches. Pin 6 comes from a 150 resistor to ground to supply current to
both the left and right LEDs. Four of them have anodes linked; but only two are on at
a time. Pin 7 is a separate resistor for the stereo, since it is not always on, and the
current varies. Pin 8 connects to a 2.2K resistor from the negative supply to operate
one side of the 3 switches. All 3 switches have one side linked together for that
connection. The other side of those switches goes to 3 separate schmidt triggers for
bounceless switching.

Header #30 is the 6 gain select switches and a few other things on the right side of
the panel. Again, all 6 pushbuttons have one side linked for the 2.2K resistor to -V;
and the other sides of the switches go to various schmidt trigger inputs. For the
bargraphs, all 16 anodes are linked together; and they connect to the 150 resistor
from ground through pin 6 of the header. Only 2 LEDs are on at a time. Two LEDs
indicating balance position have anodes linked for connecting to the 270 resistor to
ground through pin 10 of the header. Only one of those LEDs is on at a time. Their
cathodes are controlled through pins 11 & 12, which connect to the outputs of two
buffers. At pin 5 of the header is +7.4V for both the balance switch and reset switch
which are linked on the panel. Pin 9 is the other side of the reset switch; and it links
back to the 1 pin of the right counter. Pin 7 of the header connects to the center of the
balance switch, as shown earlier. On the board it goes to several things including the 8
pin of header #29 for going to the upper board and controlling all four CMOS switches
at the balance control. It also connects to pin 4 of buffer #13 for operating the balance
LEDs. And it connects to a 47K resistor to -7.4V creating the low signal when the
switch is down (toggle up), so a separate line didn't have to go up for the low signal. It
was because all 14 pins of the header were used up that the 47K resistor was used.
Otherwise a 16 pin header would have to be used.
 
Board for 90V Amplifier.

The dimensions of the amplifier board are 5.25" x 9.5". It must be on printed circuit
board, not perf board, atleast if it is to be gotten in that amount of space. The
unlabeled chips (4,5,6,7,8,9 & 10) are 4066 switches. The round things are capacitors.
The resistors near the edge of the board go to the volume controls. (see "Switching for
90V Amplifier) The op amps use ±8V, or one diode level above the CMOS, because
their outputs waste about 1/2V or more on each rail. So their output voltages are
within CMOS levels. There are several kynar wires, #30, on top of the board for long
jumps.

 
(fig)
Voltage Header and CMOS Control Headers.

#1 header
pins
1 -   +8V
2 -   -8V
3 -   +7.4V
4 -   -7.4V
5 -   -45V
6 -   +6.8V diodes
7 -   -6.8V diodes
8 -   +45V

#2 female header #2 male header


pins    to (this board) pins    from (lower board)

1- 6[5] preamp 1- 7[27]

3- 5[9] ¥3 3- 12[27]

4- 6[9] ¥2 4- 11[27]

5- 5[4] B in 5- 6[27]

6- 13[4] A in 6- 5[27]
   

8- 12[7] ¥6 8- 3[27]

9- 13[7] ¥7 9- 2[27]

10 - 6[7] ¥5 10 - 14[27]

11 - 5[7] ¥8 11 - 1[27]

12 - 12[9] ¥1 12 - 4[27]

13 - 13[9] ¥4 13 - 13[27]
90V amplifier Board, CMOS Control Headers.

#3 female header #3 male header


pins   to (this board) pins   from (lower board)

1- 6[10] ¥3 1- 6[29]

2- 6[4] BR in 2- 12[29]

3- 12[4] AR in 3- 13[29]

4- 5[8] ¥8 4- 1[29]

5- 6[8] ¥5 5- 4[29]

6- 6[6] st 6- 10[29]
   

8- 12[6] preamp 8- 11[29]

9- 13[6] 4 bal 9- 8[29]

10 - 12[8] ¥6 10 - 3[29]

11 - 13[8] ¥7 11 - 2[29]

12 - 13[10] ¥1 12 - 14[29]

13 - 12[10] ¥4 13 - 5[29]

14 - 5[10] ¥2 14 - 7[29]

 
¥ means gain position.

page six  |   page one  |   contents

Bridge Amplifier.

A bridge amplifier produces about four times as much wattage from the same
voltage as a regular amplifier. It does that by controlling both sides of the speaker
instead of just one side. Regular amplifiers operate with one side of the speaker
grounded; so only half of the power supply is being used at any time. The bridge
amplifier applies that unusued voltage to the other side of the speaker. It does that
with another amplifier which is inverted from the first. The voltages on the two sides
of the speaker then move in opposite directions.

The only advantage of the bridge amplifier is lower voltage. But that's enough of an


advantage to justify it at high wattages, particularly since home builders cannot get
parts for systems above 100V. Op amps can then be used at fairly high wattages. In
the past, amplifiers were too difficult to produce to put twice as many in a box for
bridging. Now they are easy enough to produce to make bridge amplifiers practical.

Mechanical switching was used with the bridge amplifier in order to evaluate it. The
rotary switch for gain select is not convenient; but considering that there was no
gain select on previous amplifiers, it's acceptable. The rotary switches are 2P6P
allowing one side to be used for LEDs. The LEDs could be omitted, if the rotary
switch position is visible enough.

The op amps were again zener regulated to ±18V, because the transformer voltage
peaked a little over their specifications. I try to keep circuits within specifications, but
you could get by with some overvoltage on the op amps.

A preamplifier was used for gain select, so the main amplifiers could be lower in
impedance, and the volume control would be easier to design. Using a preamp is
easier than not for these reasons. The preamps and incoming signals were put on a
lower board, so they could be gotten close to the volume controls easier without
crowding.

The current amplifiers were put on the back panel to reduce the maze of wires to the
boosters. Their quiescent current is 4mA; and the slew rate is 18V/µS.

The output jacks for the bridge amplifier have to be the insulated type; so leave
plenty of space for them. The input jacks take up less space if they are the individual
type which attaches directly to the aluminum.
 
The bridge amplifier does not need output capacitors, even if the power supply is
nonsplit. However, differences in reference voltages would create an undesireable
offset; so the same centering capacitor should be used for both halves of one
channel, but use a different one for the other channel to prevent crossover. (fig)

Bridge Amplifier Circuits.


 
Power Supplies.

Transformers produce a peak voltage of a little more than 1.4 times their rated
voltage which is rms. The filtering capacitors capture the peak voltage; but that
voltage drops rapidly to the rms level with a load. There is some ripple at the rms
level, because filtering capacitors are not thorough at heavy loads. A typical ripple on
the filtering capacitors would be 10 to 20% at full load. Oversize capacitors reduce
the ripple somewhat. The capacitors should be atleast large enough to handle the
current according to their ripple current ratings, which are given in some catalogues.

Voltage regulators are not generally used for the main load with audio amplifiers,
because minimizing the heat is desireable, and regulators function by burning away
some of the power. However, it is common to use regulators for the light circuitry.
 
The three power supplies used for these amplifiers are shown below. (fig)

Power Suppy Circuits.

10 Watt Amplifier.

Bridge Amplifier (32W).

90V Amplifier (75W)


The transformer for the 10W amplifier is underloaded, as it only sees about I amp at
full load. With the light load, the voltage does not drop all the way to its rated level;
and the speakers see about ±13V peaks, even with a volt across each booster
transistor.

The bridge amplifier uses about the same power supply, except that two
transformers are-paralled for doubling the amperage. The polarity must be watched
when paralleling transformers. If in doubt, apply voltage and compare outputs. The
15,000µF filtering capacitors are a little larger than they would have to be. A size of
10,000µF would have been within specifications for the 4 amps of current.

These two transformers produce a lot of heat, which is a problem for the bridge
amplifier. I had to put fins on the sides of the lid for additional dissipation. For this
reason, it would be preferrable to use a toroid transformer for the bridge amplifier.

The 90V amplifier uses a toroid transformer (from Active Electronics). At high
wattages, toroid transformers should be used, because they are far more efficient
than rectangular transformers. A lot of transformer heat under the aluminum is not
easy to handle; but here it is no problem. The toroids waste almost no energy with
light loads; so using a high powered amplifier on a table top for quiet listening is
very practical.

Again, this transformer is a little underloaded; and the speakers see ±35V peaks
with full load on both channels. The transformer load is a little more than 3 amps. So
there is more than twice as much capacitor filtering as would have been required,
unless parallel loading of the speakers increases the amperage.

Voltage regulators are used for the light circuitry. They are set at ±8V for the op
amps. Diodes then drop it to ±7.4V for the CMOs. Another 0.6V drop is used for the
protection diodes infront of the first four CMOS switches. But instead of just using
another diode on each side, which would have allowed the signal to pull against
resistors, transistors buffered the ±6.8V, as shown.

Ahead of the regulators, the voltage was dropped to ±l5V with transistor buffered
zeners. The regulators could have handled the higher voltage; but the heat was
reduced on them, because the lower one handles quite a bit of current for the LEDs.

The drift on the regulators probably would not have precluded their use for setting
the input current for the quad input amplifier. But since the regulators are
adjustable, excessive or unwitting variations in the amplifier input current could have
resulted; and therefore, zeners were used along with the regulators for the quad
input amplifier.
 
Shut Off Mechanism.

The on-off switch controls the ac line voltage; but not the dc. Therefore, the filtering
caps are drained through a power resistor which is switched with the ac. Reverse
diodes from supplies to ground protect against reverse voltage. Regulators are
protected from reverse current with diodes, as shown; or the current can be drawn
through their outputs.

(fig)

 
Headphones.

The headphone jacks were usuallly on the front panel. A switch near the speaker
outputs allows the speakers to be turned off for headphone listening. An 1/8" stereo
jack was used for the lightweight headphones; and a 1/4" stereo jack for regular
headphones.

The wattage for the lightweight headphones was resistor limited to 100mW for
protection. Their expected impedance is 35. For the regular headphones, the
wattage was limited to 500mW. Their expected impedance is 8. Current limiters
were not used for the headphones, because they would have peaked out too fast
resulting in nothing for range on the volume or gain. The protection circuits are
aligned upon peak voltages rather than rms voltages, because clipping results in a
squaring of the wave.

With the 10W amplifier, a 270, 2W, resistor for each channel went from the output
of the current amplifier to the 1/8" jack. A pair of 56, 5W, resistors are used for the
1/4" jack. The bridge amplifier uses the same procedure; the headphones are not
bridged.

For the 90V amplifier, separate current amplifiers were constructed for the
headphones, so the full 90V range would not be required. Otherwise, the resistors
would have had too much wattage on them. The current amplifier uses ±8V; and the
incoming signal is further limited to ±5.7V with zener diodes back to back. The
resistors to the 1/8" jacks are 75, 1W, and to the 1/4" jacks, 15, 2W. Two jacks
for each size were used. The tip of phone plugs is left side.

The power transistors (boosters) are placed on the board without heatsinking.

(fig)
 
Headphone Current Amplifier.

 
Calculating Wattages.

Almost all watts in electronics are converted to heat; and that heat has to be
accounted for.

Wattage calculations for the output analyze one half of the wave. The effective
voltage (called rms) is 0.707 times the peak voltage, since audio signals are sine
waves. The peak voltage is analyzed from one side to ground; and the voltage drop
across the transistor is subtracted. It is usually 1 or 2 volts with the current
amplifier. After the rms voltage on the speaker is determined, it is divided by the 8
ohms of the speaker to determine the current. This current determines the wattage;
but it is not the current that the transformer sees in total, because it only goes
through half of the transformer. So the transformer current is one half this amount;
but it is again doubled because of the other channel. The wattage for the channel is
the current times the rms voltage.

For example, the 10W amplifier produces ±14V under a full load. The speakers see
±13V peaks, because there is a volt across the boosters. The rms voltage is 0.707 x
13 = 9.2V. The current is 9.2/8 = 1.15 amps. The wattage per channel is 9.2 x
1.15 = 10.6W.

With the bridge amplifier, both sides of the power supply operate simultaneously.
The amplifier voltage under full load is about 25V. The speaker sees about 23V, at an
rms of 16V. Therefore, an 8 speaker uses 2 amps. Two amps at 16V is 32 watts per
channel. The transformers get 2 amps from each channel for a total of 4 amps.

The 90V amplifier at full load ripples between ±38V and ±44V, depending upon
frequencies. The speakers see a peak of atleast ±35V. The rms is 24.7V. The rms
current for an 8 speaker is 3.1 amps; and the wattage is 76.5W per channel.

Actually, wattage calculations apply to the resonant frequency of the speaker. At


other frequencies, the current lags or leads the voltage, which lowers the wattage.

This result shows up when the current limiter is allowed to clip the output. The waves
are not flattened on top but are sloped along one side. Since the voltage is not
maximum while the current is maximum, the actual wattage is not as high as the
calculations indicate. But the calculations provide relative values in a simple way for
making comparisons.

PC Boards.

Complex digital circuits are generally put on p
erforated boards, because many wires have to
cross. But where discretes are used, etched
boards are incomparable for space saving and
clean circuitry.

The techniques that I use create high density


boards. If other techniques are used, more
space would usually be required. I don't use
wire wrapping for digital boards. Instead, I
run the wires on top of the board, between
the sockets, and attach them to solder tail
sockets by making one loop around the pin.
Good pointed tweezers are required. Two or
three connections can be made to a single pin.
There is a hazzard; prying off the wires with
an iron for debugging can break pins. Copper
clad perf board would be preferable. The wire
could then be soldered to the copper rather
than the pins. Of the myriad of patterned perf
boards that are available, none are quite ideal
for the purpose. They should have one or two
holes beside each pin, and two lines down the
center of the chip for supply voltages.

With etched boards, there are a variety of


possible techniques. Here's the way I do it at
this time. First I punch mark the holes by
tapping on a large sewing needle. Perf board
can be used to mark socket patterns by
scraping a needle in each hole. Then the
marks are punched. Then drawing is done
with a black Sharpie pen. Then etching in
ferric chloride. Then drilling with a #70 bit. It
can be used with a regular drill by wrapping
aluminum foil around the shank. Then coating
the surface with solder. Then redrilling with
#68 or larger as needed.

Drill bits are becoming more available, but if


you can't find some, try Tools Unlimited 1-
800-537-1993.
 
Below is a representation of the board pattern
used for the current amplifier. Also shown is a
30 x 70 mm board with two current amplifiers
on it, as it was used for the bridge amplifier.
Two of these boards were attached to the
back panel.

Current Amplifier Board.


(fig)

Notice that there are two places where wires


cross near the input. That is done by drilling
holes and jumping wires over the top.

The resistors coming from the bases of the


current sources attach directly to the
collectors of the input transistors, and from
there to the opposite rails. Attaching those
resistors to ground would allow less voltage
and less voltage change. But it doesn't really
matter; and this way the ground doesn't have
to be brought onto the board.

Sequence is important for positioning


properly. First, the output transistors are
marked; then the input transistors; then the
current source transistors; and then the rest.
 
The board pattern for the quad input amplifier
is shown below. The inverting input is watched
for stray capacitance by leaving plenty of
space between it and the collectors of the
input transistors. Drill the hole patterns a little
wider for those transistors. Two resistors jump
over a line; and they are drilled wider also.

Board Pattern for Quad Input Amplifier.


(fig)
 
Soldering.

Core type solder should be avoided at all


times. Besides being messy and ineffective,
it's slow promoting heat damage. Liquid rosin
is very fast and efficient. It's applied with a
toothpick, while the solder is picked up on the
tip of the iron, which frees one hand.

However, for most purposes, citric acid


disolved (saturated) in distilled water is much
preferred over rosin. It doesn't create burnt
carbon to coat surfaces, and it is easier to
remove. It can conduct some current when
humidity is high, so it needs to be removed
with water. It can be found at pharmacies.

The board surface needs to be pretinned


before adding components. Using citrate is not
as easy as liquid rosin, but it works if the
surface is cleaned well and buffed with a
kitchen scratch pad. Then rub on the citrate
with a cotton swab or paper towel. The citrate
is self-cleaning and will eventually coat the
surface uniformly.
 
Switches.

The electronic switching is controlled with soft


touch momentary switches. Nowdays, various
types of posts are available on those switches.
Back in the old days (1986), I had to glue
pieces of quarter inch rods (from
poteontiometer shafts) onto flat swithces
using epoxy.

Where there are only two positions, such as


the lines in and stereo, toggles are as
convenient as electronic switching. It is
primarily the multipositions of the gain select
that make electronic switching valuable.

The soft touch switches are attached to perf


board or pc board. The latter is stronger.
Usually, more than one switch is attached to a
single board. With the 90V amplifier, the
bargraphs, a reset switch and six gain select
switches were all attached to the same board.
The board is attached to the front pannel with
#2 or #4 screws. All of the wires coming off
the board should go through a hole on the
edge of the board for gathering them. Kyrnar
wires, #30, are used. I tie them with string at
2" intervals. It's hard to make a gripping knot
with string, so I make a half knot and add
some instant glue. The wires than attach to
headers on the main board. Machined pin
sockets make the best headers. Wires solder
into the top easily; and they have less
exposed area than regular headers. The same
are used for both male and female headers.
 
Back Panel.

It is recomended that the entire current


amplifier be constructed on an aluminum plate
which is attached to the back panel for
cooling. Then there is not a clutter of wires
going back to the boosters; and the absence
of long wires makes ringing easier to control.
With the bridge amplifier, four boosters for
each channel were attached to one aluminum
plate.
 
The aluminum plates are elevated about 8mm
or more by making a double bend on the
sides. The boosters are attached directly to
the aluminum using mica insulators, while
everything else is on a pc board.

Aluminum plates for cooling the current


amplifier can be made flat without bending by
putting strips of quarter inch aluminum along
two edges. There is then enough height for
screws and fiber washers; but the heads of
the screws must be downward. Without the
need for bending, heavy aluminum can be
more easily used. Use thickness of 1/8 or
3/16 inch, 3 or 5 mm, for high wattage. Drill
the quarter inch aluminum, tighten it
thoroughly and do not use spacers, because
good heat conduction is needed. Use heatsink
grease between metal slabs.

The leads for the boosters are protected from


bending by attaching them to a board which is
directly below them, as shown below. But
rather than attaching the leads directly to the
board, they are soldered to wires which come
off the board. The ends of the wires then
attach to the lines or loop over to other
components. Two boosters have their leads
pointed towards each other, because the
collectors are linked, and the wires should be
short.
 
Plates for Current Amplifiers.

(fig)

Two boards were used - one under the main


board for insulating it. The lower one was also
used for stabilizing the booster leads. The
lower one was copper side up (mostly etched
away), and the upper one was copper side
down. The lower one was attached by screws
on the ends; and the upper one was held in
place by the wires only. A few wires were
extra heavy for that purpose. The bridge
amplifier could have been put in a little larger
box; and then the dimensions of the plate
could be a little larger.
 
Boxes.
Box for 10W Amplifier, Compact (Lab
amp). 7 x 5.5 x 2.5 inches (W D H).
 

 
Box for Bridge Amplifier. 12.25 x 7.5 x 3.25 inches. (W D
H).

Box for 90V Amplifier. 15.5 x 10 x 4.5 inches (W D H).

          dual headphone outputs, lower left.


 
End View.
 
(Fig)

There are few altern
atives for the design 
of the boxes. They 
must be rectangular
with fixed front and
back panels; and the
sides should open
with the lid for
accessibility. The
few with such a
design being sold in
the catalogues are
not very large; so
they might have to
be constructed from
aluminum sheet
metal, which is
available from the
heating and cooling
shops, if a metal
shop is not nearby.

Aluminum is sold as
if it were fixed sizes
(0.030, 0.040, 0.050
and 0.060 inches),
but it has a
continuous gradient
of sizes and is put in
one of those
catagories, because
the measuring
devices have those
notches on them.
Size 40 is used for
small boxes, 60 for
large ones.

The edges for the


main chasis must
have a flange of
about 1.5cm to
strengthen the metal
and for attaching the
lid with sheet metal
screws. The lid is a
flat sheet with the
ends bent down.

Fins are put on the


back occupying all of
the space that is not
used for jacks. The
bridge amp needed
more cooling
because of two
inefficient
transformers; so fins
were also added to
the sides of the lid.
The 90V amp has 9
fin units (18 fins) on
the back and none
on the sides. Putting
fins on the sides
(lid) would allow the
box to be smaller;
but it was packed
with components.
When the switching
complexity is
reduced, the box can
be smaller.

Fins are constructed


by bending
aluminum into a
squared U shape.
Each unit can be
attached separately;
or small ones can be
placed inside large
ones for fewer
attachment points.
Attach with 2
screws, #6 or #8.

With the 90V amp,


most of the fins
were 4" high; but
two units were 3"
high, so the input
jacks could be
placed above them.
The output jacks and
a switch went along
the edge. With the
bridge amp, fins
were omitted in the
center for the input
jacks.

The outward length


of the fins is 3", for
the 90V amp, and
2", for the bridge
amp. To determine
their effectiveness,
feel the
temperature. The
transformers are
placed on the left
side, and the volume
controls on the right
side, to keep them
separated as much
as possible. The
transformers do not
generally have to be
shielded, because
these amplifiers do
not pick up noise
easily. The toroid
transformers do not
emit much anyway.
The bridge amp,
having two
transformers, picks
up 3mV of hum at
full volume. It can
very faintly be
detected at the
speaker, because
transformer hum is
more audible than
most types of noise.
But it is so faint that
shielding was not
necessary. If
shielding a
transformer
becomes necessary,
putting roofing tin or
aluminum around it
is an improvement;
but some noise will
go around it.

Signal ground should


be handled as a
signal, which means
not connecting the
ground to just
anything. The
incoming lines use
shielded wire within
the box to prevent
channel crossover, if
not hum pickup. The
shielding can be
connected to the box
where the lines
come in. But ideally,
the input jacks
should be insulated,
because more than
one path to ground
results in current
edies which resonate
with transformer
hum which
sometimes increases
hum pickup a little.
Metal RCA jacks can
be insulated with
shoulder washers.

The other end of the


shielded wire for the
lines in connects to
the ground for the
noninverting input of
the amplifier and the
volume control. If
hum pickup is heard
on the speakers,
check the grounding
procedure.

The LEDs on the


front panels are
small 3mm ones.
Larger ones could
usually be used. I
attach them by
carefully drilling
holes of the right
size, so they stay in
with force. The holes
must gradually be
increased in size
with the side of the
drill bit. Making the
holes larger and
glueing the LEDs in
with instant glue is a
little easier; but to
remove them, they
must be drilled out.
I use toggles rather
than slides, because
they are easier to
install and use than
slides. Also, slides
tend to be
unreliable. A 3P2T
toggle can be found
at Mousers for the
balance control.

After putting
markings on as dry
transfers, I cover
them with clear
tape, which works
very good. The tape
is nearly invisible
and never gets
damaged.

 
Harmonic
Distortion.

Harmonic distortion
is so low in these
amplifiers that it is
not a significant
consideration. The
primary reason why
HD is used as a
statistic with audio
amplifiers is because
the output of the
usual design
oscillates. Removing
the oscillation is
difficult and
expensive. If it is
not entirely
removed, the
residual oscillation
creates harmonics.

Tone controls can


also create
harmonics, but the
problem primarily
occurs with passive
tone controls and
should not be
serious with active
ones, if high
impedance is
avoided.

The current amplifier


cannot produce
harmonics, because
it has no voltage
gain, and it is too
high in speed.

The quad input


amplifier can be
known to be
extremely low in
harmonics based on
the circuitry. It has
only two gain stages
rather than the
usual three; it has
no internal
capacitors, it is very
low impedance and
it is very high in
speed. For these
reasons, there is no
better voltage
amplifier for the
purpose than the
quad input amplifier.

Labor Saving.

To minimize labor,
potentiometers can
be used for variable
gain. To prevent the
need for multiple
input capacitors, the
potentiometer can
replace the feedback
resistor rather than
the input resistors.
The current load is
then calculated from
the input resistor.
Assuming a normal
signal to be ±1 volt,
and an optimum
load to be 1mA, a
1K input resistor
could be used, and
the coupling
capacitor would be
220µF.

Parts.

If you are wondering


where to get parts,
the largest selection
is Digi Key. The
second is Mousers.

After getting their


catalogs, you can
verify availability of
parts by searching the
part number on their
websites.

You might also like