Skip navigation
Electronic Design

Bob's Mailbox


On the subject of TO-3 power transistors (, steel and aluminum packages were mentioned. I think y’all were referring to all steel (mounting plate and cover) or all aluminum. (Actually, the “steel” TO-3s had a nickel cap. /rap) Those aluminum ones were just too high and always alien-feeling to me. (Yeah, but they may have had lower thermal impedance! Aluminum has better thermal conductance than iron by four times. /rap) The TO-3 packages that I remember weren’t built that way. A file to the mounting flange revealed a thick slab of copper with a steel cover. (I am ignorant of copper TO-3s. Maybe there were some. I never heard of any. /rap) Those TO-3 packages of the 1960s and 1970s were heavy and I’d imagine had the best thermal impedance specs of the lot. It may be that the all-steel, all-aluminum packages appeared when IC dies were being installed in them. Who wanted the package to cost 10 times what the electronics cost? Would it be that the packagers of the mundane crap at Wal-Mart would take the hint?

Dean Huster

Dear Dean,

I know nothing of Wal-Mart. I never go in there.



Hi Bob,

I know I tend to be a little thick-headed at times, but I must be having a particularly tough time lately. I’ve been staring at your circuit, which measures really low levels of distortion ( and have a couple of questions.

To quantify the amount of distortion in the amplifier, would you feed the signal at the false summing junction (labeled 1000 VE) into a spectrum analyzer and subtract 60 dB from each of the values at the harmonics? (That is one good way to do it. I’ve done that and it works. But actually, you can just barely see the second harmonic, slightly under the noise. /rap)

Would the lower spur values then represent the performance the amplifier would have with a gain of –1 at the test frequency? (Yeah. You might be skeptical of your powers of analysis, but actually you did figure it out just right! /rap) I can’t quite grasp what the scope display would represent when connected to the vertical and horizontal test points.

(Ahh, then you should build up this test circuit someday! I mean, you barely have to solder four resistors! Then look at the waveforms. Also, plug in some mediocre amplifiers. After studying them, you will say “Ah, that does make sense!” /rap)

The horizontal channel would have an inverted input signal plus distortion (In theory, yes, a little distortion. But the amount of distortion is so tiny, well below 1%, there’s nothing visible. /rap) while the vertical channel should only have distortion plus some low-level residual input signal. If there were zero distortion, I guess you’d see a horizontal flat line. (Or at least a slightly titled straight line. Or an ellipse. /rap) With some distortion, there should be some height to it. But how do you interpret what you see?

(You subtract all the linear terms, and what is left, p-p, is usually quite small. That is the distortion. Within a week, I can send you a big app note, AN-1485, which is almost finished. You will see dozens of examples! /rap)

Sorry for being a doofus!

Robert J. Barden

Dear Robert,

No, you are just cautious when somebody tells you four potentially unbelievable things before breakfast. You got it right!



Dear Bob,

There is a thing more annoying than a circuit done fast that is not working—a circuit that is known to not be properly working but going to production anyway. (Ouch! /rap) I am just now sitting over a design that I pointed out to have a bad printed-circuit board (PCB) layout several years ago. Nonetheless, the customer has been producing this circuit unmodified since it works most of the time. Now they have found a situation where it does not work at all and they need a fix fast. (Circuits that are intermittent will drive you crazy, but things that are really not working are much easier to fix! /rap)

The problem in this case is that the designer of the PCB used an auto-routing tool for everything. (I will not let auto-routers screw up critical connections like that. I keep critical components in good places. /rap) The result (among other flaws) is a power and ground layout where the capacitors and the chip are electrically connected, but traces between them are too long and the main blocking capacitor is outside the current path. (Obviously, this is disastrous! /rap)

Also, the hint to have a direct connection between the two ground pins of the chips has been ignored, and the actual connection goes through several vias and way around the chip instead of straight under it. Unfortunately, I do see a lot of this kind of PCB layout. Many people do rely on their CAD system instead of their experience. (Maybe there is nothing to rely on.) I wonder if they ever heard of parasitic effects and RF behavior. My first rule for checking PCB layouts is that if they look chaotic, they are usually not good.

An anonymous engineer

Dear anonymous,

I usually look at it the other way. If it is too neat and orderly, I expect that to be no good!



TAGS: Components
Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.