Swr sm-900 dc offset

[nevetslab]

Q5 and Q6 are the current limiting transistors, driven from the current thru R26 & R29 of the output stage.. Lift coupling diodes D2 & D3, so the current limiters are out of circuit. Now, under nominal conditions, you shouldn't see much more than a couple volts across the bias xstr Q4, which sits between the bases of the two driver xtrs Q7 & Q8.

Take a look at your other amp module as a point of reference....it should be like that. If you have way more than a couple volts between the driver bases, turn the bias control down. If the Bias control isn't controlling the DC level between the drivers, maybe the control is bad, or there's a problem with Q4. (or the resistors working with the bias pot R16/R17.

Now, the current path from V+ to V- for the bias xstr is thru Q16 (once turned on), thru R10, R12 & R13, then thru Q4 collector-emitter, then thru R15 and thru Q3 collector-emitter and R9A.

You can check with your good module to see where that amp circuit becomes stable, as Enzo had pointed out, the supply voltage may need to be higher than I suggested. Usually, circuits like these become stable at around 90-100VAC AC mains...sometimes lower. The line voltage may need to be at a certain potential in order to have current source/switch Q15-Q16 turned on all the way.

[bigdrums925]

nevetslab: My left channel amp board has fewer components than the right channel board. Same board number though, just missing a fairly large 35V 1000uf cap on the far right side and a jumper just above it. Any idea why this might be? And could it play a role in this problem? I can see the solder pads there have never been used.

[Enzo]

Only if you think this problem has existed ever since the amp was built.

What is that cap wired to? Could it be involved with the fan circuit?

[MarkusBass]

Exactly as Enzo says: the PC board is universal and only in one channel the fan circuit is assembled. As you can see nothing was ever soldered on the second board. And, I assume, the amp worked for a long time before the failure.

Mark

[nevetslab]

That cap is part of the DC Fan circuit. There ofte is a 300 ohm/5W bathtub resistor there (R35) which is for the Slow Speed Fan mode. There's a thermal swirtch that cuts in somewhere between 50 and 60 Ded C on the heatsink to put the fan into full speed mode. The resistor, if used, runs the fan at slower speed until thermal switch shorts it out.

[bigdrums925]

Across Q4 on the faulty board is 2.53 v Q2, Q3 lifted. bias turned all the way down, and does seem to be working.
1/2 turn changes .4v on meter across Q4 but makes a huge difference on variac.
Across Q4 on the good board is 1.09 v Q2, Q3 not lifted
These measurements were taken at 50% variac because the current draw rises sharply beyond that now

I don't see a Q16 on this drawing

I have quickly pushed the variac until the 1.5 amp meter pegs. This is about 80% variac.
worried to take it past that.

I'm starting to think there is something wrong with the output transistors that's causing the extra current draw.
I've had a lot of problems with them but tested each one of those installed now before I put them in. They are expensive and counterfeits are common.

[Enzo]

OK, turning the bias trim control makes current go up and down? It should. Turn it all the way to the end that does not make current rise. That is the coldest setting. NOW what does the amp do on the variac?

I avoid fake transistors by not buying them from ebay sellers or from China. I use real suppliers like Mouser. Sometimes I find Sanken parts at B&D Enterprises when I cannot find them elsewhere. I have not got fakes from those two suppliers, or from Digikey. Those are popular examples of reliable sellers.

[nevetslab]

I've attached the schematic I use on the SM900 Power Amp modules like you have. This is actually for a SWR 750X, it having 3 pairs of outputs, but the same circuit topology. Q15 & Q16 is the turn-on/current source for the front-end.

Ah...looking at the SWR 2000 heat sink, they have what was called Q15 & Q16 labeled Q13 & Q14. My mistake. I've attached this schematic. The upper power supply rail on the schamatic isn't shown, but looking at the SWR 750X diagram, you'll see the similarities.

When you turn the bias control all the way down, what is the voltage you get across Q4? If it doesn't drop down as low as that of your other module, there's either a problem with the B-C resistor, resistor in series with the pot, or Q4 itself (most likely). With 2.54V across it, you've got the output transistors turned on hard as well as drivers. The circuit behavior should be the same as your good module.

SM900 Sch.pdf

750X Sch.pdf

[bigdrums925]

So I replaced the B-C resistor, it was 2.2k should be 2.7k according to the schematic. No change on the current draw
Looking at the working side, it has only a 1.5k resistor there.
So I changed the 6474 transistor and it helped a little, current draw is not so dramatic. Still its too much for me to push it to 100%
So trying for the resistor in series with the bias pot I guess that would be R4 1k between the emitter of Q4 and bias pot
But when I look at the board I see the emitter of Q4 tied directly to the bias pot on 2 terminals. see below:



and the part of the schematic showing the bias pot and Q4:



What should I make of all this?

[nevetslab]

The bias pot is a 3-terminal device. Safest way to make it an adjustable resistor is to tie the wiper and one end together so it won't go open circuit if the wiper fails to make contact with the track. So, the two terminals (wiper and one end are at the emitter, and the other end of the pot is in series with the resistor, that connects to the base, along with the B-C resistor. Nothing unusual about that. Do you get the same bias voltage range on the two modules? You were having too much voltage when the bias Xstr is being turned off. When it's turned on hard, you'd have little to no voltage difference across it, and hence no output current.

Are the 100 ohm emitter resistors for the two drivers Q7 and Q8 intact? It is a quasi-Complimentary output stage, so there's a third 100 ohm resistor in the bottom half, But, you have the working module to compare against. If it's not the bias circuit, and it's sounding like it isn't, then the stage it is controlling is suspect. If you have Q4 turned on hard, it basically places a short between the two drivers' bases, and there should be NO current flow in the output stage as well as thru the drivers.. If that's not the case, one of the drivers is bad.

BTW, the solder side of that PCB photo is blurry and hard to make out

[bigdrums925]

I gave up on the schematic and replaced that 2.7k resistor with a 1.5 like on the working amp.
That allowed me to power up without excessive current draw
Then attached the heat sink and powered up again. All good, drawing .5 amps full variac
showed a 117mV DC offset though.
Installed the amp unit into the chassis and powered up again. All good
But when I attached an 8 ohm load and ramp up on the variac the current draw is high right away and its that way for either side of the amp loading one side at a time

[bigdrums925]

Disregard the part above where the current is high above. It is under control.
But the just made working side now shows about -115mV DC offset
compared to the been working side having about 12mV DC offset

What do I need to do about this?

And how is the best way to set the bias pots?

[bigdrums925]

OK well at least it is working and I am grateful for all the help
Thanks!

[Jazz P Bass]

A lot of times on dc coupled amplifiers, the slight dc offset can be eliminated by changing out the Long Tail Pair differential input transistors. (Q1/ Q2)

Ideally they should be closely matched for hfe gain, but using 'same batch' parts will usually get you there.

[Enzo]

I wouldn't do anything about it. 115mv of offset into an 8 ohm load is about 14 milliwatts dissipation. The speaker would never notice, and the amp wouldn;t care.