2SB755 Transistor Data?

[Enzo]

B&D Enterprises has them in stock for $9 or so.

www.bdent.com

And IIRC, it first smoked with a load, and then smoked without. Once it blew up with a load, it was not going to heal when the load was removed. Once Q118 shorted it was staying shorted.

I am not too enthused about bringing that CArver circuit up on a variac, but maybe...

[bob p]

When this blew, the fault current followed a path. Clearly it was through R174 and Q118, and since Q124,126 are shot, it went through them. To gwt from one of those sections to the other, the current either ran through R170 and R150 - check them - Or through R170 on through Q122 and R156. (Also maybe R172, Q120, R154) SO check all those.

as a learning exercise i'm trying to trace the fault circuit path, as you recommended. insofar as the out of circuit tests show that Q124 is bad but Q126 is good, am I correct in assuming that the path can not be through R170 and R150, since those would only be suspect if Q126 was bad?

since Q126 is good, does that also rule out the path going through R170, Q122 and R156?

would I be correct in interpreting this to mean that the path had to be Q124 to R172 to Q120 to R154 to Q118 and R174?


i really appreciate your patience as i learn my way around this amp. thanks again. and thanks for the link to BDEnt. I had never heard of them.

[bob p]

You are showing both polarities with your meter. That is good, alwyas worth checking reverse. Q115 is NPN, so the red lead to base is the test. Black to base is reverse. if you are unsure, look at the symbol. B-E looks like a diode, so test in the same direction you would if it WAS a diode. The imaginary diode from B-C poinrs the same way as the E one. They both point away from or both towards teh base. PNP tests with black to base. Red to base is the reverse there.

I have a little sheet that essentially says the same thing about how to test NPN vs PNP transistors, but when I'm looking at the physical transistors in the unit, I don't always remember which ones are NPN and which ones are PNP when I'm doing the testing -- so I just take measurements both ways and jot down the results, and do the interpretation once I have all of the data. I can't say that this is the right way to do things, but it works for me.

By the way, a good xstr tests about the same B-E and B-C. But a common failure mode is a short C-E. The two measures are then in parallel. So they can both look OK, exactly the same in fact, but the EC short is still there. Always check for EC shorts.

thanks for the tip.

OK Q124,126. 124 looks shorted to me too. SOmething I do as policy: when I find a transistor shorted, when I remove it from the circuit, I either bend the legs over, or cut one off. That way I will not install a bad one by mistake.

good idea. i had just marked the transistor as bad with a Sharpie marker, but I can see how clipping lead(s) would make things easy to recognize.

Check R158 at the base leg of Q124. Think of it as checking the screen resistor after replacing a shorted EL34.

R158 = R159 = 5.1R

The good news is you can run the amp with only one of the two power xstrs, at least for test. The amp would only be able to prooduce half power, but without a load that is not an issue.

still don't have a Q118, so I don't think we can do anything until I order parts. once I have parts, I'll have a Q118 and the output transistors. until then, we can't do anything, right?

REmember, R170 and friends are dual resistors - center tapped if you like. From the center leg, 0.22 to each end. 0.44 across the whole thing.

my meter isn't very accurate at sub-ohm values. i'm getting measurements that are about twice what they should be, but i guess that's good news since the ballasts haven't gone open.

R152,150 should connect to the output bus I believe, otherwise they would have used one resistor. Is that the case?

looks like that on the schem, but i can't tell by tracing the board.

If it were me, I'd replace Q118, any burnt resistors, stick Q126 back in, and see what happens. NO LOAD. Monitor the channel out at the ballasts or the coils. SOmeplace convenient.

you mean after i get a new Q118, right?

Since Q118 burnt, I'd check Q112, and compare it to Q111.

readings:

Q111
Red to Base:
BE  .561v
BC  .545v
CE  charge

Black to Base:
BE  .965
BC  OL
CE  .496

Q112
Red to Base:
BE  .555v
BC  .536v
CE  charge

Black to Base:
BE  .965
BC  OL
CE  charge

i don't know what to say about these readings. the schem is poorly marked. Q112 is clearly marked, but Q111 is not. I'm assuming that Q111 is the one that has C118 on its base. If that's the case, the schematic suggests that Q112 is NPN and what I think is Q111 is PNP. Unfortunately, the layout diagram shows the NPN symbol for both Q111 and Q112. something is wrong on either the schematic or the layout diagram. I'm not sure which.

[Enzo]

B & D is like a Sanken outlet, I don't do much with them, but now and then when I need a Sanken something, they usually have it.

Of Q126 is OK, then that does indeed let R160 off the hook. R170 if it measures OK, it is OK. As a dual resistor, clearly then teh upper half shou;d be OK, but the lower half COULD have ben involved via Q122, but I doubt it.

Look at the output bus running across the center as a sort of Grand Central Station. Routes from the two rails can meet there and make transfer connections if you will. Q124, 126 only need one of them to get current to the output bus. From there, there is no cosmic sense of balance that says the fault path has to be symmetric and look like the other half. A shorted D112 - no I am not suggesting it is involved, just making examples - would allow Q126 to burn itself up just as easil;y as Q118 would.

Not sure how you ruled out R150. To me that is the link between the upper and lower halves.

I see the fault path as Q124, R172, R150, Q118, R174.

I could certainly be wrong here, and it doesn't really matter wwhat happened, but I suspect Q118 failed short, slammed the bases of Q120, 122 up against the - rail, which stressed the Q124,126, and Q124 is the one that blinked.

But to me R150 is very suspicious.

You could be right about a path throughQ122, but then I would think the 4.7 ohm R156 would be toast. I don't know the base current rating of Q122, so I f Q122 or 120. R150 might be lower. POOF.

I just checked. The drawing I have been using does not show R150,152 connected to the output bus, merely crossing it. But I know better. And I just looked at two other versions of the drawing, and the connection IS there.

[Enzo]

Ah you snuck another post in there while I was responding to the previous one.

2SAxxxx and 2SBxxxx are PNP
2SCxxxx and 2SDxxxx are NPN

REmember they often leave the 2S off as understood. So 2SB755 might be labelled B755.

marked the transistor as bad with a Sharpie marker

Perfectly good idea. Just want the bad parts easily identified floating around the work area.

my meter isn't very accurate at sub-ohm values.

Most aren't. Your meter leads can add a substantial fraction of an ohm. These are not likely to change in value, they are either open or OK for the most part.

you mean after i get a new Q118, right?

Yes of course, the bad Q118 is useless.

You may not have anything, but I have enough selection here, that I would sub something in its place for testing.

don't know what to say about these readings. the schem is poorly marked. Q112 is clearly marked, but Q111 is not. I'm assuming that Q111 is the one that has C118 on its base. If that's the case, the schematic suggests that Q112 is NPN and what I think is Q111 is PNP. Unfortunately, the layout diagram shows the NPN symbol for both Q111 and Q112. something is wrong on either the schematic or the layout diagram. I'm not sure which.

Mine os pretty clear. Q112 connects to the base of Q118 through R146. It is probably OK, I just like to check anything connected to a bad part.

Q111 is just the equivalent in the other channel, it connets to the base of Q115 through R147. SInce that channel is OK, the readings there serve as a standard.

C118 connects to Q114. The circuit there is the opposite polarity to Q118 but equivalent. I suggested Q111, 115 since it is exactly the same circuit as Q112.118.

Q111, 112 are in similar circuits and are both NPN. I think you were not looking at the real Q111

Your Q112 readings look OK.

[bob p]

Q111, 112 are in similar circuits and are both NPN. I think you were not looking at the real Q111

well, i was looking at the right transistor on the PCB, but I was looking at the wrong Q111 on the schematic. I was not looking in the other channel, I was looking in the top half of the right channel.

Q111 is just the equivalent in the other channel, it connets to the base of Q115 through R147. SInce that channel is OK, the readings there serve as a standard.

I've found the right transistor, though its still not labeled Q111 on my schematic. sigh.

suffice it to say that my readings really were taken off of the real Q111 and Q112 on the PCB. the silkscreened numbers on the PCB pointed to transistors whose printed numbers matched the parts lists. so knowing that my measurements really were for Q111 and Q112, you think Q112 is OK? i was confused by the final reading in my lists, in which the CE reading for Q112 read charge, but the CE reading for Q111 read 0.496v. i couldn't explain the discrepancy.

[Enzo]

Ah. The M500t schematic does indeed leave Q111 unlabelled. The M500 drawing has it added - you can tell it was added by hand, sorta squeezed. The copy I am using is a loose copy from a service bulletin for teh 500. it has the Q111 as well.

ANy time you have a two channel amp and one side works, then it is always handy to compare same position parts between channels, thus the Q112/Q111 comparison.

Well lookee there, my print leaves out the Q116 label. COmparing Q114 to Q112 would work, but all the polarities would be reversed - pain in the ass. Just use the other channel.

Why are the two CE readings different? I think you confused your leads. Look at Q111 up top. C-E path. Q111C goes up through R179 and back down through R175 to Q115C - a total of 32 ohms. Then Q111E goes through R147 to Q115B - 56 ohms. Those two resistances are low enough to ignore. SO when you test EC on Q111 (which normally should be like an open, though charging the power rails caps), with your leads one way you will really be checking the B-C of Q115, hence that junction voltage drop reading. Now reverse the leads, the Q111E-C still acts like an open, and now the Q115B-C junction is reverse biased by the meter, so it does not show the junction.

When you are in a multustage durect coupled xstr circuit like this, ther is almost always SOME sort of parallel path top contend with. You need to get used to keeping that in mind when anomalous results show up. When you get an unexpected junction drop across teh EC of an xstr, then reverse the leads and see if it goes away. If it does, that means you are really measuring a working junction somewhere else. Xstr failures usually result in shorts, and less often opens. I don't recall ever seeing an EC failure that checked like a diode drop.

TUbe amps are voltage based circuits, so all we need to pass between the tubes is signal. Other than teh occasional cathode follower or something, tubes are rarely direct coupled. Xstrs are current based things, and amps like this are going to be direct coupled, and that means what you think you are sticking your probes on is usually not the whole picture.

Q111 is straight inboard from Q121. Q112 is straight in from Q122 across the board.

[bob p]

its been easy enough, i guess, to find the parts on the board. it just requires a lot of tedious inspection. the only problem i had with identifying parts is on the schems where parts are unlabeled.

[bob p]

I could certainly be wrong here, and it doesn't really matter wwhat happened, but I suspect Q118 failed short, slammed the bases of Q120, 122 up against the - rail, which stressed the Q124,126, and Q124 is the one that blinked.

so the driver failed and took out the power transistors? i had always thought that it worked the other way around, where output transistors blow up and take something else with them.

in other words then, when the driver fails spontaneously, it appears to the user that the amp failed for no good reason? i had always thought that you had to screw something up to blow up an amp like this -- by doing something stupid like shorting the outputs, diming the amp and blowing the output transistors, etc.

[Enzo]

Wouldn't the world be a great place if everything happened predictably?

Outputs failing might be more common, and it would be no rare thing for a driver to then fail as a result, but that doesn't prevent the driver or ANY other part on the board from failing on its own. And then zanyness ensues.

[bob p]

Another parts order arrived today so I finally got to work on the amp.

Today I installed the following parts:

Q118 (driver)
Q124 + Q126 (outputs)
Q206 (protection)
C202 (protection timer)

I also reattached R205, which had been lifted previously for testing.

I still don't have replacements for R150 or R174.
R174 is the one that started smoking, but still tests good with an ohmmeter. A replacement part is en route from another supplier.
R150 is a half-watt resistor, and I made the misteak of ordering a quarter-watt 10-pack. Getting the right part will take a while, until I can put together another parts order.

I'm hoping that we can get by without a replacement for R150 and R174 for short term testing until I can order more parts. What do you think?

I plugged the amp in to test the relay with no load. Dead amp, but it was nothing more than a bad AC fuse. The previous smoking power-on episode that burned all of the parts also burned the fuse but I hadn't caught that until now. Murphy's Law prevailing as it always does, I have 200 10A fast blow fuses and not one 10A SlowBlow. Aargh.

The good news is that with a 10A FB fuse, the amp powers on and the relay seems to be working properly. So I thought I'd ask what you'd do next.

Is it too early to feed signal into the amp with no load and monitor the outputs at the ballast resistors?

By the way, our previous testing suggested that Q112 was OK. Even though I hadn't planned on replacing it, I bought a bunch of spare NPN and PNP pre drivers -- just in case.

[bob p]

Well, I got impatient and tried a low level signal into no load. I traced signal through the circuit.

I turned the output of my signal generator down to its minimum setting, to feed a minimum level input into the amp. Both channels produced a clean 1KHz sine wave at the output transistor emitters and at the "center taps" of the ballast resistors. After the relay clicks, the clean signal appears at the speaker terminals.

Next?

[Enzo]

Well, that is the way. If the circuits are both tracking in terms of signal, leave the loads off for now, and advance the signal. The amp runs on the low voltage rails until peaks demand higher. then the higher voltage rails are switched in. Diodes keep it out of the LV supplies. Eventually we will see of the supplies commutate. If not, distortion will be obvious.

I'd also let it idle for an extended period to see if it heats up too much. Monitor bias currents - well, idle currents really - to see that nobody starts climbing.

Got the relay straightened out then, eh?

Once we feel it is not going to overheat on its own, or run away somehow, and when we know it will amplufy signal nicely, THEN we can turn it down, load it, and start incrementally increasing power out to the load. During no load tests, we should try to note the signal level at the input that results in the supply rails commutating. Now under load, see that it still does.

And monitor heat all the while.

The amp is like an athlete coming back from knee surgery. First repair the damage, then work on restoring motion, then work on strength. Finally the amp renegotiates its contract and is traded to Kansas City. Tough break. Maybe that analogy is breaking down a little...

[bob p]

The amp is like an athlete coming back from knee surgery. First repair the damage, then work on restoring motion, then work on strength. Finally the amp renegotiates its contract and is traded to Kansas City. Tough break. Maybe that analogy is breaking down a little...

I thought it was a pretty good analogy and pretty funny too. I've said it before -- you could have a second career as a comedian.

I was thinking along the lines that you had recommended about gradually dialing the amp up under no load, and then starting over again at lo levels with a load attached. But I didn't consider a temp monitor or watching the commutating transistors. That seems pretty obvious now that you've mentioned it. I have a few questions about that.

I'm not sure of he best way to watch temperature, other than to feel the amp with my hand. I don't have thermometer probes or anything like that. The M-500t has temperature sensors wired to each channel's heatsink. I don't know if we can take advantage of them in some way, or if we can rely on them to shut the amp down if things go bad while I'm feeling for heat with my fingers.

What's the best way to check for commutation? I'm guessing that we'd need to monitor the switching transistors on the scope. IIRC there's a photo of a multichannel scope watching the commutation transistors in one of the M-1.5 manuals. That manual explains the commutation switching while the M-500t manual does not. Another limitation is that I only have one scope, with two channels.

When we get to the point of testing it with a load, is any particular load resistance better than another? I guess I'm really asking if testing at higher load impedances is any less dangerous than testing at lower load impedances. OF course the amp is rated at 8R and 4R, and any extremes would be bad. I'm thinking that 2R would be really bad. 4R would be OK, 8R would be better. What about 16R?

[bob p]

Got the relay straightened out then, eh?

It looks like Q206 and C202 took care of the problem. I also replaced D201 with a 1N4007 back on Page 3. IIRC that's all we did with the relay circuit.