You are on page 1of 1

gerard’s columns

Reliability
By Gerard Fonte (USA) the best choice. For example,
why use a 1% resistor when
Arguably, the biggest difference between a hob- a 5% resistor will work? The
byist-designed project and a commercial product reason is that 1% resistors
is reliability. Hobbyists don’t mind if they have to are so commonplace now
tinker with it from time to time. After all, we like that the price is virtually
to tinker. But you don’t want to tinker every morning to get the same. My Mouser catalog shows
your car started, or with the toaster, or with your computer. A reli- 1% (through-hole, 1/4 watt) resistors at $0.06 each and
able product is tinker free. 5% resistors at $0.12 each! In hundreds the 1% resistors
are $0.04 each and the 5% are $0.01 each (more sensible).
Analyze This But in hundreds, the difference is still only 3 cents per resis-
It cannot be overstated that reliability comes from attention to tor. If you use 30 resistors in your design, that’s less than a dol-
detail. It’s the small things that count. You need to take the lar increase. For surface mount resistors (1206 size) there is no
time to examine all the aspects of any circuit the first time you difference for singles ($0.05) and only about a penny for 100’s
design it. After the first time, you will know the important ($0.02 vs. $0.032).
areas to look at. But for the first time, you have to consider Obviously, for a pull-up resistor where any value from 1K to 100K is
every part. usually sufficient, there is absolutely no requirement for a 1% resis-
My favorite example is the lowly unregulated, power tor (the once standard 20% resistors don’t exist anymore). But for
supply. It’s just a transformer, bridge rectifier and filter capacitor. any application where a specific resistance is needed (voltage divid-
Simple, right? If you choose a 24 volt transformer, then a 35 volt ers, gain-setting, etc) the question is reversed. Why shouldn’t you
capacitor should be fine, shouldn’t it? That’s a 50% safety margin. use a 1% resistor? Additionally, 1% resistors usually have better tem-
The answer is a resounding, NO! perature stability.
The transformer is rated in RMS voltage, under load; not peak volt- It’s a similar story for generic power rectifiers (1N400x). Different
age. The peak is 1.414 times the RMS or about 34 volts. Add about manufacturers have different basic prices for the diodes (I saw $0.05
10% for being unloaded and the actual DC voltage at the filter to $0.15 each). But within the series (1N4001 to 1N4007) there was
capacitor is over 37 volts. Capacitor confetti, anyone? And if you often no difference in price. The 1N4007 has a PIV (Peak Inverse
plan on using a common 3-terminal voltage regulator (LM78xx Voltage) of 1000 volts where the 1N4001 has a PIV of only 50 volts.
type) you should know that they have a maximum input of 35 volts According to my Motorola Data Book, every other specification is
(although there is some variation from manufacturer to manufac- the same. So, there is simply no reason not to buy the better diodes.
turer). Another failure in waiting. You are getting much more for your money.
You choose to use half-amp rectifiers because your design only
draws 0.1 amps. To be extra safe you use a 2-amp transformer Look Before You Leap
and high value capacitors you have on hand. There shouldn’t be a These examples show the need to know what your design requires
problem here, should there? Yes there is. During turn-on the capaci- and what parts are available. Using parts that exceed the opera-
tors act as dead shorts across the transformer. During a short, the tional needs of the circuit will improve performance and reliability.
transformer’s output is only limited by the resistance of the wind- And we’ve seen that this can be achieved with a minimum increase
ing, which is typically about 0.5 ohms. In theory, a surge current in cost or no increase at all.
of about 75 amps could occur (37 volts into 0.5 ohms). In practice The point is not to focus only on your needs when looking for parts.
it’s quite complicated, but the usually accepted value (not guaran- You might be able to find parts that are better than what you need
teed!) is 10 times the transformer rating. In this case, that’s 20 amps for a small difference in price. And there are trade-offs. Which is
(10 × 2 amps). Most silicon rectifiers are robust and can take very more important in your design, the actual value of the capacitor or
short surges up to 30 times their continuous rating or 15 amps here its working voltage? Will the lower on-resistance of the power MOS-
(30 × 0.5 amps). A surge of 20 amps into a 15-amp-surge rectifier FET switch reduce heat build-up? Is it worth spending an extra dol-
won’t go. Also, the length of the surge depends upon the size of the lar for a relay with 20 amp contacts versus 10 amp contacts? Does
capacitors. The bigger the capacitors, the longer the surge. So, the it make sense to use a solid-state relay, instead?
reliability of the rectifier may depend more on the transformer and There is clearly much, much more to reliability than I can possibly
capacitor rather than the down-stream circuit. cover here. There are many fine books on the subject. But funda-
It’s important to see the critical need for proper circuit analysis for mentally, designing reliable products is the result of a mind-set. It’s
every part under every condition. You can’t guess at reliability. You not enough to make it work, it has to work well. And in order to do
have to be sure. this, you have to consider every aspect of your design as well as the
practical considerations of the parts that go into that design. After
More for Your Money all (one of my favorites), engineering is just common sense with
Hobbyists (like me) are perpetually short on cash. So, we tend to attention to detail.
look for the minimum requirements for a part. Sometimes this isn’t (120236)

76 05-2012 elektor

You might also like