Testing power transistors

[Mick Bailey]

I have a bag of power transistors, MJ15022 & MJ15023, all pulled from Dynacord powered mixing desks. They were hired out and when a power transistor failed the whole lot on both channels were replaced. There are 5 complete sets (80 transistors), though only 10 shorted, so it looks like just one pair went on each unit.

The problem is, the transistors from the good channels were never marked, but they're in there somewhere.

I'm undecided whether test, match and re-use them, or bin them. What would you do?

[Leo_Gnardo]

test, match and re-use them

Right there, after all the good ones are well 'broken-in." Your matching will make for better replacement sets as well. Rare that MI manufacturers bother with matching. The transistor that draws more current than the others becomes the first to go into thermal runaway then bang goes your amp. When you have a well matched team, you can expect fewer failures.

I had a good jaw wag with the amp tech at Clair Bros in the early 80's. When they got a shipment of new power amps, he removed all the output transistors, ran 'em one by one thru a curve tracer, selected matched sets, reinstalled those sets & adjusted bias accordingly. When you have big crowds to entertain & major acts paying the bills, reliability is a very good thing. Helps maintain a good reputation. Heck, it's a good thing no matter what!

[Mick Bailey]

Increasingly I've found many power transistors to be an expensive lottery draw and that's why I'm tempted to re-use. It's difficult getting matched sets anyhow and the emitter resistors should take care of any minor differences. The problem comes where any have been stressed and may give rise to an early bath. A curve tracer would be a nice thing to have.

[Enzo]

My thinking is this. In a parallel string of output xstrs, one will short E-C, and that then relieves all the others from carrying current. If the parallel transistors have a shorted one in parallel, then all the current flows through the short. In other words, a shorted parallel does no harm to the others. Usually the high current there does stress somebody on the other polarity side. And it usually shorts too. The danger is not finding any failed xstr on the other side. Then it is hard to tell which one, if any, was stressed.

If a row of outputs, if one shorts B-C, then in my experience, usually the whole row goes up. They will all be bad. If I see a row of six and five are shot, then I assume the last one is also, evn if it still works.

So I do usually reuse the "good ones", usually by not removing them in the first place.

Unmatched transistors? I am sure they were never matched at the factory. That is the purpose of those ballast resistors, those half an ohm jobs in the emitter legs. Those act like mixing resistors.

[Mick Bailey]

Thanks guys, that's reassuring. I have 5xPNP and 5X NPN shorted, so the probablility is one pair from a failed channel on each mixer. Think I'll match them anyhow for re-use, at least to group them. A nice little job for Sunday morning instead of reading the paper.

My thought has always been that with just two failed opposite transistors there's no evidence which one shorted first and no way of telling how close to failing the rest got on the opposite side to the short when one of those failed too. When transistors were much cheaper I just replaced the lot on a failed channel, though why these got completely pulled on both channels escapes me. Maybe they uprated them. I bet a powered mixed on hire gets the same treatment as a hire car.

[nickb]

Increasingly I've found many power transistors to be an expensive lottery draw and that's why I'm tempted to re-use. It's difficult getting matched sets anyhow and the emitter resistors should take care of any minor differences. The problem comes where any have been stressed and may give rise to an early bath. A curve tracer would be a nice thing to have.

I do a fairly crude match but it seems to work well in practice. I use a 10V power supply, place a 12K resistor between C & B and measure the collector current. I buy the trannies in batches and write the measured collector current on each one (on label) so I can sort them. It's not at all unusual to see a 2:1 variation across a batch. The drop across the emitter ballast resistor is fairly small compared to the swing in Vbe ( say 1V cf 0.2V kind of order) so it only partly mitigates the HFE variations.

For me, if one transistor is blown then they all get changed. Partly 'cos I don't trust 'em, partly because of rough match issues as the replacement may be a different brand so I'd rather not mix and sometimes because I have to substitute as the original is no longer available and I won't risk grey market sources.

This is my personal take on it. Enzo's been doing it for long time and if he says that it doesn't matter, then I would not disagree.

[olddawg]

Under warranty I would always replace sets. (Usually caused by a shorted load condition) It becomes more problematic with out of warranty repairs. Is it ethical to use "used" parts at all? Many times what you use determines wither it's a fix or no fix because of a cost threshold with the customer. I always kept the "good" transistors from the warranty repairs if I wasn't required to return them. Many times I have used one or more of those to repair a unit out of warrant or in the case of obsolete parts years later. BUT... I made sure to cook the living f@@k out of the unit and explain to the customer what I did. I know other techs that took the "what they don't know won't hurt them" approach or just charge them for labor with no warranty.

[Leo_Gnardo]

I bet a powered mixed on hire gets the same treatment as a hire car.

Truer words was never typed. All other hire gear too. Many people do respect the equipment & the fact they get to use it without having to buy it. Others, not so much. All it takes is one bad one.

American band on tour in the UK, had an Ampeg endorsement deal. Every Ampeg SVT-CL that was provided no cost on the endorsement was AFU. Finally got an amp from an out-of-town hire outfit, paid for it & that one ran properly. A conversation with the London gear outfit that provided dud amps, their tech finally revealed the truth: "If you paid for the hire gear you'd get something that works." So ... what's the point in having the endorsement deal, load of crap, why bother filling in forms etc etc. Just pay for the hire & surprise, the gear works. "You get what you pay for." Even if you're a world class band. Phooey on endorsement gear.

[Enzo]

There are exceptions to anything. I used to see a zillion solid state Peaveys. And when they were blown up, it would usually be one pair of push pull transistors. For the reasons stated, I usually expect the remainder to be OK. I had very few amps come back with a refailed power amp. But there is no demerits for being cautious. There is a national act from this area, and they brought me their gear before going on tour. it was endorsement gear. Crate even. So it was warranty claims for me. "Dude, we're going to Germany next month, here's my stuff, check it all out." I don't recall any of it coming back with complaints, and they were repeat customers. SO at least in their case, endorsement gear worked out. Repairs and maintenance are only as good as the shop providing them. and that is for rental gear, warranty repairs, paid repairs, you name it.

When matching transistors, I think it important to check them at high currents, like they face in use. Gain varies over the range of currents. Note for example the Peavey number system. 70483180, a house number for MJ15024, means gain specified at 3A and 180v. The MJ15024 is a 250v part rated at 16A. So they are not testing them at a few milliamps, they are spec-ing 3A conditions. A lot of the low voltage and low current variations in power transistors even out at high levels.

[Mick Bailey]

.....Is it ethical to use "used" parts at all? Many times what you use determines wither it's a fix or no fix because of a cost threshold with the customer. I always kept the "good" transistors from the warranty repairs if I wasn't required to return them. Many times I have used one or more of those to repair a unit out of warrant or in the case of obsolete parts years later.....

Good point. It all depends on the condition/age of the item being repaired and how much the customer wants to pay. Often a used part can bring life back to something that may otherwise be scrapped. Other times the only way is to use a salvaged part because there just aren't any new items, or the cost of adapting the equipment to take a modern equivalent isn't worthwhile.

Other times a used part can be a better bet than a new replacement. I had a bad run with new EF86 tubes and lost on every one. I had some old Mullard pulls that I was a little reluctant to use, but they became a last resort and they've been in service for years with no trouble. The same thing with 2N3055 until I re-used some old RCA ones.

So long as the customer is happy with the repair and used components I think it's an equitable situation.

[Enzo]

On my shelf I have an old Soundcraftsman power amp. It is four channels, and each channel has pairs of J49 and K134 MOSFETs. (Or is it J50 and K135, I forget now) One channel is blown and the amp was abandoned due to estimate. I will never fix it because all those good MOSFETs in there are hard to find and expensive, so every one of them represents a repair I can accomplish that would otherwise have been a pain. Like some great old Trace Elliot head.

Of course now that I have closed the shop, I guess I need to sell it off.

[Leo_Gnardo]

On my shelf I have an old Soundcraftsman power amp. It is four channels, and each channel has pairs of J49 and K134 MOSFETs. (Or is it J50 and K135, I forget now) One channel is blown and the amp was abandoned due to estimate. I will never fix it because all those good MOSFETs in there are hard to find and expensive, so every one of them represents a repair I can accomplish that would otherwise have been a pain. Like some great old Trace Elliot head.

Of course now that I have closed the shop, I guess I need to sell it off.

Those Soundcraftsmen were a major PIA due to wire-wrap construction and their "revolutionary" class H (for Hell) operation. After seeing only one and giving up on it I refused all others. Hafler amps use those JFETs too - the only thing good about it. Might consider prying out the transistors, a lot cheaper to ship. The rest, to the dump dump dump.

[Mick Bailey]

I had a stroke of good luck with those transistors. They grouped up in sets of 4 based on the code printed on them - maybe a manufacturer's batch code. The codes are unique to each batch. By deduction I could work out which had come from good channels.

[gbono]

If you are going to test these devices at or near the application bias be careful not to overstress the device. The Tek curve tracers have a pulsed mode for looking at HFE at high power levels. I also like to look at the breakdown voltages (you can do this on a scope but a curve tracer is much safer) - must limit current under .2mA when doing this type of test.

The really old semiconductors from the 60's and 70's did not enjoy the benefits of later break throughs (improvements) in fabrication which are common place today. It is unfortunate that counterfeit material is common place since modern semiconductors are always a more reliable choice IMHO.

[km6xz]

When stocking power transistors, I always got them directly from the manufacturer, usually Motorola/ON Semi and never had to worry about new transistors failing to meet spec. Those who buy from eBay or other discounter had better have a quality high current curve tracer and test every one to verify if can handle the rated max. I bought one batch of "MJ15024/25", flat of each from Thailand and received them a few weeks later. Testing them I discovered they matched the spec of a much cheaper TO-3 transistor, but not the demands placed on the 15024/25 series.
Transistors made by the major companies in the last 25 years are very close in matching random devices from the same factory, even with different batches. It is not like the old days when matching was needed during the equipment manufacturing and by any tech working on it before.
If there is a simple CE short on both P and N side, replacing just the impacted emitter resistor and transistors was enough. Replacing perfectly good power transistors only runs up the cost and does not influence the reliability. Having a power amp return after a repair with more problems was just about NIL in a high volume shop. Very old amps with early technology transistors needed closer attention to matching.