Gain matching output transistors

Stevescivic

Active Member
Hello everyone I am doing some preliminary research on what things I need to do a full rebuild on a working SX-1050. I have read in past threads that output transistors need to be matched.

MTF has previously suggested that using new modern output transistors will in fact be better than using the stock Pioneer ones that came from the factory. IIRC the stock ones don't have as clean output as the modern transistor.

How does one go about logically gain matching output transistors without a scope ?

The transistors that were suggested to be replaced in quads are
Brand:ON semi
863-mj21193g PNP
863-mj21194g NPN

Does matching infer that I simply replace all 8 at the same time or to replace transistors such that they are all made from the same batch by the same manufacturer?

What happens if I accidentally install one backwards and blow it up? Do I have to replace ALL outputs at the same time again or can I just buy the one that I fry (hopefully that never happens).
Thanks
 
Does matching infer that I simply replace all 8 at the same time or to replace transistors such that they are all made from the same batch by the same manufacturer?
In the strictest sense of what I understand about power transistor matching, no - to me matching implies that you are testing and selecting transistors from a much larger number than is required, and picking the ones that measure the closest for the parameter you are interested in. Note that it is very difficult for the average hobbyist to match power transistors.

However, the manufacturing of such devices has improved so much over the years that essentially you can **buy 4 or 8 (same type) power transistors and be reasonably sure that they are close enough matched for most applications. With the caveat that they should be from the same manufacturing batch for the closest match, and even from different batches the match will very likely be good enough. (**from a reputable source).
What happens if I accidentally install one backwards and blow it up?
It is thankfully quite difficult to mount TO3 style transistors backwards due to the case design, however it is very possible to confuse PNP and NPN devices and put the devices in the wrong place (PNP for NPN etc) - quite easy to do even for pro's, and is a bad mistake. Devices of opposite polarity (PNP//NPN) should be 'Complimentary pairs' - in other words devices designed to work together in a 'push-pull' arrangement, usually having consecutive type numbers - as above '93 & '94. Devices of opposite polarity but not of the same type number family (and thus not complimentary) will very likely be poorly matched.

Regarding replacing all transistors after a damage event, for me it depends how serious it was, maybe if the device went open B-E I might replace just that one, but if it went short circuit C-E - and worse, short circuit in all directions, then I would replace a larger number dependant on that, and how far any collateral damage (blown driver transistors?) had spread.
 
Last edited:
Hmm interesting. Gibmven that I am restoring two SX-1050s that are supposedly fully functional is it even worth the trouble to replace the outputs if they work fine? How much of an improvement will I see from the amplifier by replacing the outputs? Less switching distortion?

Basically all I have is a set of fluke DMMs, a dim bulb setup and a good soldering iron. It would be crazy for me to start investing in a bunch of other equipment such as scopes and such for a once or twice in a lifetime repair job. Is there a transistor matching service from experienced techs here on AK?

I'd hate to get into my unit to realize I am screwing things up. I usually go about repairing one part at a time doing one board at a time. My first and last rebuild was a SX-3800 and that was nasty to work on. Ended up paying a shop to restring and align my FM tuner which cost me a small fortune. Electrically my work on that unit was otherwise perfect from what the tech told me.
 
Hmm interesting. Gibmven that I am restoring two SX-1050s that are supposedly fully functional is it even worth the trouble to replace the outputs if they work fine? How much of an improvement will I see from the amplifier by replacing the outputs? Less switching distortion?
I would see no reason to replace the OP transistors in this case, you are unlikely to see any improvement by so doing unless the old transistors are actually faulty.
Basically all I have is a set of fluke DMMs, a dim bulb setup and a good soldering iron. It would be crazy for me to start investing in a bunch of other equipment such as scopes and such for a once or twice in a lifetime repair job. Is there a transistor matching service from experienced techs here on AK?
Yes, it would be a waste of money for a one time repair, but a good DMM (or 2 or 3) is extremely useful. As regards 'matching services' - you can ask here?
I'd hate to get into my unit to realize I am screwing things up. I usually go about repairing one part at a time doing one board at a time. My first and last rebuild was a SX-3800 and that was nasty to work on. Ended up paying a shop to restring and align my FM tuner which cost me a small fortune. Electrically my work on that unit was otherwise perfect from what the tech told me.
Perfect is good, but receivers have a tuner that is beyond very many people to align, the absolutely essential test gear can be very expensive, and understanding the 'black art' of tuner alignment, with some of the poor alignment instructions given in the Service Manuals is a special gift.

Make sure you heed the words and advice of MTF - he is exceptional.
 
Last edited:
I am skeptical that output transistors have to be matched. This is usually done to ensure equal current sharing in transistors wired in parallel, but if you have emitter resistors, as this amp does, there is far less danger of one transistor taking all the current.
 
That is good feedback. When I am doing the rebuild I have been told to replace all the transistors in the pre-amp board as they are known to get weak and fail. My experiences goes as far as replacing capacitors so I intend to change all of them except for the caps on the tuner board. Transistors I am clueless on which ones should be replaced as well as any resistors or diodes that need pre-emptive servicing.
 
Last edited:
The replacement advice wasn't transmitted to you properly.
Replace all outputs as a set when one of the originals blows up. THIS is to ensure ROUGH matching - same manufacturer / chemistry etc... so Vbe's are closer and loads are shared. Better than with one new and a bunch of originals.
PLUS the replacements will most likely have more power dissipation, higher voltages, different gain and such compared to the originals.

IF the outputs are OK, then it's up to you if you want to replace the outputs with probably more robust devices. You will not gain much in extra power out - that's a function of the power supply and the other circuitry. There will be very little change in distortion either. But the replacements may stand up to overloads better.
.
 
Ah okay gotcha Mark that makes sense. In other words don't mix NEC transistors with ones from on-Semi. If one transistor pops for whatever reason then replace ALL of them from the same manufacturer (and prerably the same batch if possible). Is there any advantages in going to the crazy level of detail to individually test each transistor to get a near perfect matched set? I did some reading on matching and it seems quite involved with suggestions of heat sinking them, using regulated power and being super careful to never overload them such that they blow up in your face...

What do you mean by the outputs will stand up to overloads better? As in driving them really hard and loading them up with some serious power? I doubt I would ever stress an amp to ear piercing limits as I generally find that the 9 o clock position on my sx3800 is ear piercing already! I'm driving a very efficient pair of energy RC-70s. Doesn't take much to drive those towers.

Thanks
 
The MJ-21193 & 94 are heavier duty right out of the box than the Sankens are. The Sankens were chosen to a price and performance point. However, you being the Production, Marketing, and Logistics Depts you can spend as much or as little as you want without external pressures(ie: Penny pinching accountants telling you to spend less and get more performance with what little you have. In most cases it can't be done)

Think of tires on a standard pickup. The Sankens are Factory rated tires and spec'ed for the weight of the truck plus some additional weight, while staying within the tires capacity and the Axles capacity (weight capacity) Plus GCWR.....These are usually oversized Automobile street tires, with 2 plies on the tread and none or maybe 1 on the sidewalls. They have soft sidewalls and don't like heat or shed it well. The OnSemi's would be roughly a Heavy Duty "C" or "E" rated tire with 2 to 4 plys in the sidewalls and tread area (very stiff tire) to be able to take additional heat and load without harm to the tire, where the Sankens would or could blow out or throw a tread from heat generated by towing a trailer that is just below the GCVW (Gross combined Vehicle weight) and the RAWR (Rear Axle Weight Rating), plus the Tire's weight Rating.

You don't have to run the volume up to ear splitting levels to blow an amp. A shorted capacitor, resistors, or a transistor down line that goes Chernobyl on you could very well take out 1 or 3 of the Sankens, while the OnSemi's very well could hold up to it. It's happened, not very often, but it's happened.

I'd put in the On-Semi's and keep the Sankens in a box, pinned to a piece of foam so they don't get bent pins. Use them for spares, or keep them with the 1050, in case you should ever sell it. Do the 6 way test on each of them before storage. http://audiokarma.org/forums/index.php?threads/bipolar-junction-transistor-testing-basics.43186/ In the DIY Forum. Pay particular attention to post #11.
 
The most important reason to replace all the outputs on that channel if one blows is because the ones that test good may still have been stressed by the conditions that blew the one device. In my situation, being a repair tech, I want to avoid a recall.
 
I've replaced outputs many times and never gain matched, if you try to gain match outputs it could become a very expensive practice. Never had an issue with outputs that I replaced without gain matching.
 
The truck tire analogy was great. I am a car guy so that makes the understanding super clear. I guess what I am trying to do is fix what needs fixing and leave well alone things that work perfectly fine. I have two SX-1050s (I just sold my left nut to get one in mint cosmetic and all unmolested condition). I am working my way towards building a BOMbso I can place and order with mouser when I get home so I can dive right into recapping the unit. Both receivers were in a factory like working state prior to me taking ownership so that helps simplify things.

I will go ahead and order complete sets of on-semis for my sx-1050. I want to build a rock solid receiver that I can hand down to the next generation of folks.
 
If you look at the datasheets for MJ2119x, you will see that they have a test setup that measures open loop THDistortion . The devices (npn/pnp) that are gain matched in this push pull config show a lower open loop thd vs ones that are not. You must remember that these devices used in a reg amp are inside the negative feedback loop so it lowers the open loop THD by the a close approx of the amount of neg feedback.
So the conclusion is that if you gain match the ops you can lower the system thd a bit more than if you do not have gain matched pairs.
My suggestion is if you buy a few of sets of them, measure them and try to hfe match them as close as you can.
 
Back
Top Bottom