Professional Documents
Culture Documents
Specifications
NORMAN H. CROWHURST*
20 30 40 50 60 70 80 100 200 300 400 500 voltage drop and thus the current in each
FREQUENCY KILOCYCLES tube. This is then used as a means of
adjusting the bias, or differential bias,
Fig. 3. Response curves illustrating typical results of poor relative coupling in a on the tubes so as to ensure equal op-
transformer intended for ultra-linear operation. Dashed line represents measured erating conditions. If the circuit is in-
characteristic of the transformer In a typical ultra-linear circuit, but does not indi- tended for this purpose, not only should
cate what happens to the waveform. the d.c. resistance, be closely matched be-
Practical Cases
Someone is going to notice at this
point that we have not specified fre-
quency response as such. Even before Fig. 4. The standard method of specifying the response of an audio transformer,
the advent of feedback amplifiers, the in the absence of other information. But this does not tell how it will behave in
published frequency response for an out- any practical circuit.
put transformer bore no direct relation- transformer, but of the amplifier using sary in the rest of the amplifier to suit
ship to the frequency response it would this transformer for the output. It is "for the time being."
achieve in a given amplifier. This is therefore, suggested that the foregoing Maybe at this stage the amplifier man-
because the frequency response in the electrical specification should be sub- ufacturer decides he would like to see
transformer varies with the different stituted for the frequency response what some other transformer manufac-
circuit conditions to which it is con- normally called out for output trans- turer can do, so he submits a sample of
nected. formers. this deteriorated transformer (although
The transformer may be specified, for Such a specification as we have out- the deterioration has not been identified),
example, as matching 6000 ohms plate- lined in this article provides a useful together with a prototype amplifier and
to-plate on the primary, to 16 ohms on tool for determining, to the satisfaction specification, to another transformer
the secondary. This specification tells us of both transformer and amplifier manu- manufacturer. This man makes an ap-
that terminating the secondary with 16 facturers, whether a given product meets proximate copy of the poorer trans-
ohms will cause the primary to look like its specification. All these electrical former and proceeds to supply to this
6000 ohms plate-to-plate. Now, how do quantities can be measured and thus the specification, possibly at a slightly lower
you measure the frequency response of facts accurately determined as to whether cost. The amplifier manufacturer thinks
the transformer? the transformer does or does not meet he is doing well on the deal.
The generally recognized method is to its specification. It is not dependent upon This kind of process can go on, grad-
feed from a source resistance of 6000 a prototype amplifier and upon possible ually relaxing the tolerances until the
ohms on the primary side, maintaining deviation due to tolerances of compo- amplifiers no longer anywhere near meet
the voltage constant at the input to the nents in other parts of the amplifier. the original specification, developed in
source resistance, as shown at Fig. 4, The over-all effect of using this ap- the original prototype amplifier. This
and then measure the output across the proach should be a considerable im- can explain why many amplifiers do not
16 ohms. This may give an excellent fre- provement in the general quality of am- perform according to the specifications
quency response of the transformer. plifiers and a more satisfactory control laid down in the catalogue. With this
But very few amplifiers, even without over the performance of output trans- method of working it is extremely diffi-
feedback, operate under this condition: formers for them. But some transformer cult for the quality-control department
the average triode amplifier provides a manufacturers dislike working to this of an amplifier manufacturer to secure
source resistance considerably lower than kind of specification. They would prefer satisfactory results and the transformer
the optimum load resistance—in this case working the other way—even though manufacturer can easily supply com-
6000 ohms; whereas the average pentode eventually it may cause trouble due to ponents which do not satisfactorily meet
or tetrode provides a source resistance the differences discussed earlier. the requirements, although a reasonably
in the region of five times the optimum It is difficult to see why there should satisfactory argument can be put up to
load resistance. Either of these, condi- be any objection to using this kind of show that his transformers do meet the
tions will produce a frequency response specification, unless it is that the more specification as provided.
quite different from the one published, generally accepted method allows a more
taken under the condition of Fig. 4. sloppy operation. What happens seems Development Procedure
Now we add feedback. This compli- to go something like this: transformers Finally, we should have a few words
cates matters even further, because the deviate by some factor or other; maybe about the development of amplifiers.
parameters of the transformer contrib- a certain wire gauge is not available, or Under the present method the usual pro-
ute to the over-all feedback stability the quality of laminations may vary. cedure seems to be to utilize a trans-
characteristic, and determine whether This alters the performance of the trans- former that approximates to the require-
closing the over-all feedback will cause former slightly, but there is not definite ments of the amplifier, or to obtain a
the response to drop off, peak, or evidence that the transformers are not sample from a manufacturer that he
achieve maximum flatness. This variety meeting specification. When this devia- predicts will give the required perform-
of possibilities is quite independent of tion is combined with the available toler- ance.
the frequency characteristic of the trans- ances in other components, in some sam- The amplifier is then built around this
former by itself. It depends entirely on ples of production amplifiers, the trans- transformer and the circuit adjusted to
the way the relevant parameters in the formers may still work, while in other obtain the required performance. If it
transformer combine with time constants samples the combined effect of the toler- proves impossible to achieve the desired
through the rest of the amplifier circuit ances results in an unsatisfactory per- performance with this particular trans-
to produce a resultant closed-loop formance. former, an estimate is made of the
characteristic. After making checks that discover this changes required—for example, better
Obviously, then, specification of the fact, the manufacturer probably comes primary inductance to improve the. low-
frequency response of an output trans- to the conclusion that the trouble was a frequency performance, or tighter coup-
former for a feedback amplifier is com- matter of component tolerances and so ling to improve the high-frequency per-
pletely meaningless—unless the amplifier he virtually sets up new tolerances for formance—and a new sample is sub-
is specified in complete detail as well. the transformer, by adjusting other com- mitted.
This is not then the response of the ponent values and tolerances as neces- {Continued on page 51)
AUDIO • jUNE, 1957 23
TRANSFORMER