Pioneer SX-1980 Transistor Replacement

[EchoWars]

Anyone who repairs this old Japanese stuff knows that the original transistors (2SD746A's and 2SB706A's) are long gone. Well, that may not be totally true, since I was told that you might be able to get the 2SB706A's from Pioneer still, but that's not a lot of help if the NPN's are not available.

In this thread over at Audiokarma, I'm documenting an attempt to replace the original transistors in a blown SX-1980 with On-Semi MJL4281A's and MJL4302A's. A fellow member machined some 'hold down' brackets that require no modification to the original heatsink, and I have one channel wired and actually running.

However, the bias concerns me. Here's the basic amp setup, and the bias string:

VR2, the 100 ohm bias adjust pot, is set to minimum, and bias is at about 45mA at that point after being on for 10 minutes or so. If I leave the amp on long enough, which I have not done just yet, it may rise to 50mA or better, and that's where I want to set it, but with the pot at minimum I have no adjustment range except up. Seems the On-Semi outputs are a little easier to push into conduction than the original NEC transistors.

The only thing I can think of doing is to replace either the dual-layer diode (D2) with a single diode, or replace the quad diode (D3) with a tri-layer diode (harder to find). In this case, I might even have to increase VR2 to make up for the .55V or so I'd lose by replacing the diode, but then I'm concerned about the amp being undercompensated.

What would you do?

[snkby]

i would ask you what to do as i have no idea and i also have an sx-1980 that has a blown channel !

ive looked at it only briefly but from what i remember there were a couple of fried transistors on a little 2"x2" board right under the top cover.

i can take a pic if you like.

its 1 of the many projects i have lined-up for when i get the garage into a state where i can actually set-up my work/test bench again.

everything on the receiver lights up and looks beautiful.

hope to get mine running as i plan on keeping it for a very long time.

[EchoWars]

What makes you think it has a blown channel? If one channel sounds OK, then the other one likely has a preamp problem, since the protection makes no distinction between L and R channels...if the output transistors blow in one channel, the protection shuts down both channels (and it ought to blow the fuse as well). With the unit I'm working on, the unobtanium outputs are toast.

I don't know what board you are talking about in yours...the three boards in the front of the unit when you open it up are for the tuner. Then there are the driver boards mounted vertically at the rear R and L at the heatsinks. Everything else is buried underneath.

[snkby]

pull the beast from the shelf. lolololo

i have to move a bunch of stuff around it tomorrow any way so i may as well pull it out and take another look and attach a source.

gonna slide something under it in front so i can just slide it out and back.

just made me think of an undertaker pulling a body out of the fridge.

more like suspended animation.

maybe it is a pre problem.

but from what i remember there was no output at the headphone either.

and at least 1 of the transistors is blown apart and the little board is scorched.

i will post some pix tomorrow if all goes well.

[ilimzn]

Glenn, you may try increasing the emitter resistors instead - the MJL parts likely have higher gain which is why you get higher idle current. I would be weary to remove diodes due to thermal comp issues.

[EchoWars]

I appreciate the input.

The emitter resistors are already 0.5 ohm, and you know if they are increased that it will change the point at which the current limiter will kick in.

Consider this about the diodes though...the four-layer diode (STV-4H) is the one mounted to the large main heatsink. The two-layer diode (STV-2H) is mounted to the small heatsink of one of the driver transistors on the driver boad itself, instead of the big heatsink. Perhaps this might be the lesser of any evils concerning diode swapping, since the driver transistors very likely stay cooler than the outputs at higher currents anyway.

Don't get me wrong, I'm not disagreeing with you, rather, I'm just kicking the whole deal around. I'm determined to find the correct solution with a minimum of fiddling and fuddling, and there are six emitter resistors on each driver board. And we know that changing them will interact with the V/I limiter.

Perhaps the thing to do is to swap the '2H' diode with a single diode, and see what it takes to set the bias where I want (might take a 500 ohm pot to replace the 100 ohm). Then, once that is done, heat the amp up real good and see where the bias goes.

[commsysman]

When you say "bias", what are you measuring?

Are you measuring the +V power supply current, the output stage collector currents, or the actual BIAS current to the base of one of the output transistor groups?

If you are measuring the actual idle current from the power supply, and it is stable at 50 ma or so when it is warmed up, that is probably fine; as long as it does not want to run away when it gets hot.

For comparison, my 120W per channel amp is biased at 275 ma
idle current through the collectors with no signal.

It runs cool and stable that way.

[commsysman]

By the way, I would suggest that you just measure the collector current of the output stage transistors when setting the bias; that is what you are really interested in. I do not know the specs on that unit in detail, but based on the 270W per channel power rating, I would say that you want 200 to 400 ma of idle current through the power transistors with no signal. That should translate to about 10W of heat dissipation per transistor with no signal, more or less. The main thing is that they do not "run away"
after 30-40 minutes with no signal, and the heat sinks stay cool. I am assuming that we are talking about a legitimate RMS continuous power rating and not the old mickey-mouse "IHF" power rating, which can be 4-6 times higher than the real RMS power. If the 270W rating is IHF, then 60 to 120 ma of collector current at idle is more like it.

[ilimzn]

[QUOTE]Originally posted by EchoWars
I appreciate the input.
The emitter resistors are already 0.5 ohm, and you know if they are increased that it will change the point at which the current limiter will kick in.
[/b]

True - if you don't like meddling with that part of the circuit, then they must stay the same.

Consider this about the diodes though...the four-layer diode (STV-4H) is the one mounted to the large main heatsink. The two-layer diode (STV-2H) is mounted to the small heatsink of one of the driver transistors on the driver boad itself, instead of the big heatsink. Perhaps this might be the lesser of any evils concerning diode swapping, since the driver transistors very likely stay cooler than the outputs at higher currents anyway.
...
Perhaps the thing to do is to swap the '2H' diode with a single diode, and see what it takes to set the bias where I want (might take a 500 ohm pot to replace the 100 ohm). Then, once that is done, heat the amp up real good and see where the bias goes.

Yes, that seems to be your best and easyest bet. As long as you have no thermal runaway, you will be fine. ALternatively, you may want to fiddle with the emitter resistors of Q11 and Q12, making them smaller. The thing is, though, this increases dissipation in Q11 and 12 for the same output idle current.
The thing with diode strings is, the thermal coefficient is dependant on the number of diodes, but the actual drop has lots to do with diode process used - some of these are rather 'resistive' and drop additional voltage (a good example would be using a 1N4001 vs a 1N4148). To be perfectly honest, I really don't like diode string biassing, replaced them with a Vbe multiplier on several amps to get more stable bias. At least a Vbe multiplier limited to multiplying a diode drop with a whole number

[EchoWars]

Originally posted by commsysman
By the way, I would suggest that you just measure the collector current of the output stage transistors when setting the bias; that is what you are really interested in. I do not know the specs on that unit in detail, but based on the 270W per channel power rating, I would say that you want 200 to 400 ma of idle current through the power transistors with no signal. That should translate to about 10W of heat dissipation per transistor with no signal, more or less. The main thing is that they do not "run away"
after 30-40 minutes with no signal, and the heat sinks stay cool. I am assuming that we are talking about a legitimate RMS continuous power rating and not the old mickey-mouse "IHF" power rating, which can be 4-6 times higher than the real RMS power. If the 270W rating is IHF, then 60 to 120 ma of collector current at idle is more like it.

The SX-1980 is an honest 270WPC, and typical power at clipping is usually a respectable amount over 300WPC (last one here measured 312WPC @ 0.5% THD). I'm surprised that you seem to be unfamiliar with this receiver... There are very few receivers rated at higher power (Sansui G-33000, Marantz 2600, and some ugly-duckling Technics that I can't remember the number of). The SX-1980 sells for stupidly high prices on eBay, thus a method to bring one back to life is a worthwhile endeavor.

I'm measuring bias from the emitter of one of the NPN's to the emitter of one of the PNP's. The emitter resistors are .5 ohm, so my millivolt reading translates directly to milliamps. Stock spec with the original outputs is 30mA. measured in the same way, which would translate to 90mA through all the output devices.

I considered simply having the bias set rather high, but the more I read and the more people I talk to, I come to the conclusion that I'd like to be measuring 50mV from emitter to emitter.

Originally posted by ilimzn


Yes, that seems to be your best and easyest bet. As long as you have no thermal runaway, you will be fine. ALternatively, you may want to fiddle with the emitter resistors of Q11 and Q12, making them smaller. The thing is, though, this increases dissipation in Q11 and 12 for the same output idle current.
The thing with diode strings is, the thermal coefficient is dependant on the number of diodes, but the actual drop has lots to do with diode process used - some of these are rather 'resistive' and drop additional voltage (a good example would be using a 1N4001 vs a 1N4148). To be perfectly honest, I really don't like diode string biasing, replaced them with a Vbe multiplier on several amps to get more stable bias. At least a Vbe multiplier limited to multiplying a diode drop with a whole number

The stock emitter resistors for Q11 and Q12 are 100 ohm, 1W. Perhaps a drop to 82 ohms might do the job... I like that solution. I'd rather not fool with the diode string at all. Seems I just need to make sure that the driver transistors are good ones.

OT...got a spam email this morning from a travel agency. $2800 or some such for 10 days in Croatia. They make it sound pretty good. Didn't know you were living in paradise, did you?