Hello Mr. Pease:
Glad to see my EE hero is still around in semi-retirement. In fact you are so much appreciated by me, I posted your autographed pic on my Facebook business page. I wish you lived in Fresno and could have been my mentor. I would have preferred that over college. (Next time you drive into San Francisco, we could get together for a bull session and beer. /rap)
We have met and have spoken before, but not for several years. I kind of need some rudimentary advice for a problem that seems to be getting worse as professional power amplifier designs get more complex. Any advice you could provide would be very much appreciated.
Besides design, which I haven’t done for several years now, my little company services pro-audio equipment, which is kind of lucrative and rivals profit margin against manufacturing. In my pro-audio service division, I service all sorts of professional audio equipment. You can see the details on my Facebook page if you are interested. Search “Dynastar Electronics” and then click “Like.” (I understand that the audio equipment may need some special repairs. /rap)
I have a longwinded technical question that your Troubleshooting book doesn’t really address, so perhaps you might have an article on the subject or can shed some light.
Commercial power amplifiers have become quite sophisticated: Class G, H, etc. They are not always worth servicing if it takes more than four hours to repair. I think other “repair shops” tend to just replace a channel PCB, stick a large bill to the customer, and be done with it. I think the truth is a lack of knowledge about repairing.
Since even AB-type amps nowadays can provide so much power (1 to 2.5 kW), they utilize a lot of paralleled power output transistors that become very difficult to find shorts on due to inherent low resistances in the circuit by design. I recently purchased a Tek DMM4050, so I do have the six digits I need to look for extra-low resistances, but PCB traces interfere with this technique of looking for shorts. (At least five digits is a huge help. /rap)
I often cannot use a variac effectively, or the ole light-bulb shunt on the ac line to bring up the voltage and dig around due to other complications, especially on more advanced amplifier classes. Often manufacturers are not much help, because they just replace PCBs and don’t want to hassle with repairing them. They can afford to do that. (I do understand that getting a number of “paralleled” transistors to test out can be grueling, but necessary… /rap)
So here is what I end up doing to find shorted output transistors: I desolder B and E, then painstakingly go through each device with the diode test function and look for shorted devices while isolated. Occasionally, I must remove the devices to fully isolate it from PCB traces, then re-apply grease, mica washer, etc.
By not doing a repair in this tedious manner, I take a risk of re-assembling the amp with a shorted device remaining, risking re-damage, or having to dismantle the unit again for re-diagnosis. I don’t always have access to lifting emitter resistors and associated support parts with dense integration these days. It takes a long time to this, and there is a point where I have to start eating my time, otherwise the end customer could just purchase a new power amplifier. Some of the brands I work with are Peavey, QSC, Crown, Carver, and others.
Continue on next page
(Maybe you should warn all these makers that they are risking a lot of customer dissatisfaction if they don’t make better layouts that can be tested and troubleshot. It’s a bit late, but start straightening them out. /rap)
Given the dozens of paralleled output transistors on a typical unit, what is the best and/or most efficient way to find a shorted device(s) amongst the group? (Wow, I think you are doing about the best you can. /rap) What am I missing here? Many times I have had to let the repair go, because I passed the estimate, so I took a loss. (Sometimes you have to keep on investing a little more time to find out what is really wrong. Maybe that can help teach you what to do next time. /rap) I’d like to change my ways, mentality, or whatever it takes. I have been doing this method for years. Any help you can advise me with would be very much appreciated.
I’m afraid you are using some of the best T/S approaches you can. These equipment makers are not helping you. They really are paralleling the devices? Bastards! I’m putting five PNPs (TIP-42s) in ~ parallel, but they each have a base and an emitter resistor. I used to work on 12-packs of 2N4860 FETs that were paralleled, but it was easy to swap them out as a 1- by 0.8-in. submodule. No bolts needed.
Your digital-to-analog converter (DAC) article really started me browsing and travelling down memory lane (see “What’s All This DAC Stuff, Anyhow?”). At one time I had acquired a Philbrick applications manual. (Yeah, the big 1966 one. If you go to the Analog Devices Web site, you can find the whole book, 140+ pages. Of course, it will take you some time and paper to print it out. But ADI did a nice service. If you can’t find it, philbrickarchive.org should help. If not, holler to me. /rap)
At one point I used information from it to build a linearizing network for a voltage-controlled crystal oscillator (VCXO) I needed for a test. Shortly after that I loaned it to someone and never got it back. It was a great manual and I have missed it ever since. (Now you’re in luck! Now, everybody has a printing press! /rap)
I also clicked on one of the links in your article, which took me to the Philbrick archive site. That prompted me to go to the garage and check my junk box. I found six PP55 modules. I don’t remember where or how I got them. They look like pulls. I probably should power one up and see if it still works. (The PP55s are 99.9% likely to be good. Good old machines. Comparable to a ’741. /rap)
Ditto for the CK722 that is in the same box. My dad got me some for Christmas when they first came out. I think he paid $9 bucks a pop for them! (I don’t know that much about the CK722, but it’s a good historical artifact, even if it just barely wiggles. Those old pieces of germanium were not hermetically sealed, nor planar passivated. If you wanted to use it to build a one-device radio, your chances are about 50%? /rap)
Continue on next page
Just for laughs, you can see my first computer at www.retrotechnology.com/aux/swtp6800.html. I also have www.bmblack.org (mostly personal) and www.bmblack.com (consulting business). I was laid off 4/15/2010, so I am trying to get the consulting moving again.
Will view later. Thanks for writing. Lunch time. Beast regrds.
Okay, Bob, kiddie Spice is for the rest of us. But surely the smart analog designers now have special CAD that models all the thermal coupling on the die. (Largely not. Relying on symmetry and ’centroid usually works much better. /rap) Or is all that symmetrical/diamond layout stuff still just guesswork ?
(A mix-master that can run Spice and do thermal analysis at hundreds of thousands of points, on and in one die, is a slow machine—and I have seen machines these days make the most horrible errors. Have you ever run OP-07s? Sometimes I do, and sometimes I don’t. Most of the guys who can run a computer that big, complex, and powerful have no idea what the results mean, especially if the output involves a stupid error. A little thermal analysis may be done well. But why bother to use a digital computer to show you how to find an orgasm, when you still have to know how to urinate? Do you need a computer to teach you that? /rap)
I am sending you a nice, simple low-dropout regulator (LDO) cct with a –90-dB V power-supply rejection ratio (PSRR). (I’ll look for it. But be careful of its weaknesses. I’ll assume you’ll tell me what you think it’s good for. Then I’ll have to tell you what it’s bad for, if anything. /rap) Since you know the most about the LM317, would you please point me to the best Spice model you know of?
I still use Pease-Spice, which is, “Back-Of-Envelope Spice.” I don’t know anybody who has a good model for the LM317. Many of its errors and imperfections are thermal and not easy to model.
I just read an article claiming that “signals from seismic sensors left on the lunar surface by Apollo astronauts in 1971 have revealed that the Moon has a liquid core similar to Earth’s.... The Apollo Passive Seismic Experiment consisted of four seismometers deployed between 1969 and 1972, which recorded continuous lunar seismic activity until late 1977.”
When I read this, I recalled your article on “What’s All This 2401BG Stuff, Anyhow?” and I’ve got to believe this is all part of your handiwork. You can read the very interesting article on the data collected from the seismic sensors.
Arlie Stonestreet II
Yeah, I heard that yesterday on NPR. Thanks for hollering. I’ll check it out.
When we originally acquired those data, we found some interesting info. But now we have much better computing capability, so we are starting to find out new nuances about the internal structure of the moon. Solid core, liquid core, too?
Yes, I’m sure all those data came from 2401BGs.
Thanks for writing.