blue lander said:
Yeah, as I said above, zero bias can open the loop and cause the amp to oscillate as it tries to close the loop and reach DC balance.
The output is not looking so good. Sometimes an amp with problems needs a little nudge to show off its issues. That's why I suggested you try an input signal before hooking up the output transistors.
It looks like from the scope photos the problem is asymmetrical and it doesn't look like normal high frequency oscillation (it would be useful to know what you had the scope set to for time and voltage per division?). If you have the scope hooked up with ground to ground and it's not set to invert, the amp is only unstable towards the negative rail which might be a clue.
The problem, to me, looks like a bad transistor or other error in the circuit somewhere. It's hard to tell from the pics you sent, but it seems when the audio signal exceeds a certain threshold the amp is spiking towards the negative rail. So I would suspect a problem with Q101/Q105, Q102/Q104 or Q107/Q108 or any of the circuitry and wiring surrounding them.
Was this with only the drivers (Q107/8) hooked up and no output transistors? If you're certain all the transistors are healthy, the correct parts, with the correct pinout (i.e. installed correctly), and everything on both driver boards matches between the channels, I'd measure the DC voltages between the channels at some key points in the circuit.
With the output transistors disconnected, first adjust the bias, if you haven't done so already, to be the same in both channels so you have about 0.1 volts across the 10 ohm resistors R128 and R130. Verify it's the same in all 4 resistors.
With both inputs disconnected hook one DMM probe to ground and carefully probe the following locations with the amp running and compare it between the channels:
The speaker output terminals (should be .05 volts or less).
The collector of Q106/emitter of Q108
The emitter of Q106/R125
The emitter of Q105/R117
The collector of Q101/R102
Any difference between the channels indicates a DC operating point problem. And that's most likely caused by an error, bad part somewhere, bad solder connection, etc.
If the above all checks out (i.e. everything matches), it might still be a stability problem. It's possible, if you have the resistors in place of the fuses, the instability is being caused by that. If you're running the amp with no output transistors, you can *shoiuld* be able to have the fuses in temporarily bypassing the resistors and R120 and R130 *should* protect the drivers from brief problems. But take the fuses back out before trying to run the amp with any output transistors installed.
If it's still having problems even with the fuses in place. I'd try adding 100pf to C113 (in parallel making it 200pf total). If that fails, I'd lower R127 and R129 to 100 ohms Next, I'd try adding 100pf to C106. To make things more complicated, some of the above techniques (especially C113 and C106) can be tried together and one at a time (i.e. remove the last tweak before adding another).
What might be going on is the newer transistors are likely much faster than their 1977 counterparts and may have more gain as well. Both of those things can change the open loop behavior of the amp which can change the compensation needed for stability. Unfortunately, in your situation, adjusting the compensation can be a hit or miss thing.
Others following this thread might have some other ideas based on the asymmetrical instability in your scope photos?
I'll make all those readings tonight and try to get better scope pictures. Only Q107 and Q108 were in when I took that photo, and I haven't put the resistors and caps in on the fuses since as of right now I don't have any output transistors to proceed with. Digikey still hasn't responded to my e-mail, so I'm going to try calling them.
Oh boy... on the left channel r130 reads between -55 and -60 volts depending on how I have the pot set, and r128 reads 0 no matter what I do. On the right channel r130 reads roughly the same and r120 reads -.4 volts. This is with no output transistors except for the 4 drivers. Do I need to take those out too?
No, you want the voltage *across* each resistor--i.e. put one DMM lead at each end of R128 and then each end of R130. It sounds like you're measuring the voltage to ground? If R130 really had 60+ volts across it, it would have vaporized long ago. All the measurement points I listed were with respect to ground, *except* the bias across R128 and R130.
Okay, in the left channel across R130 I can only get it to go as low as .274 volts. If I push the pot all the way to the other extreme, it goes as high as 2 volts. R128 goes to .1 volts and slowly climbs. I watch it get to .15 before I turned the power off, I don't know how high it'd go.
In the right channel, I get 0 in both resistors. BUT, I discovered I have the wrong transistor in Q102. The schematics calls for PN3569, but I put in a 2n4401. I'm not sure if Digikey sent me the wrong part or if I ordered the wrong part, I'll have to check the invoice.
At the output on the left channel I get -160mv, 288mv in the right channel
R125 is -59 volts left, -.2mv right
R177 is 264mv left, and -160mv right
R102 is 59 volts on both sides.
In the right channel, I get 0 in both resistors. BUT, I discovered I have the wrong transistor in Q102. The schematics calls for PN3569, but I put in a 2n4401. I'm not sure if Digikey sent me the wrong part or if I ordered the wrong part, I'll have to check the invoice.
At the output on the left channel I get -160mv, 288mv in the right channel
R125 is -59 volts left, -.2mv right
R177 is 264mv left, and -160mv right
R102 is 59 volts on both sides.
I just tried playing some music through the amp now that I've adjust the pots to give the smallest voltage across r130, and now I don't see those strange peaks on the good channel. I don't see any weird distortions or clipping or dropouts or whatever no matter how high I turn the volume up. The bad channel looks the same, but I bet it's because of that wrong transistor.
Yeah, don't even try to go further with the wrong transistor.
The climbing voltage is probably OK as long as it settles down and stays under about 1 volt. Just an FYI, ohms law is i = v/r. So in this case, 1 volt divided by 10 ohms is 0.1 amps or 100ma. At that current the driver transistor will be dissipating about 6 watts of heat which, as long as it's properly mounted to the heatsink, should be OK. When the amp is working properly with the output transistors connected and biased properly, there will be about 0.6 - 0.7 volts across those resistors. But you'll be using R133, R134, etc. to set the final bias.
The climbing voltage is probably OK as long as it settles down and stays under about 1 volt. Just an FYI, ohms law is i = v/r. So in this case, 1 volt divided by 10 ohms is 0.1 amps or 100ma. At that current the driver transistor will be dissipating about 6 watts of heat which, as long as it's properly mounted to the heatsink, should be OK. When the amp is working properly with the output transistors connected and biased properly, there will be about 0.6 - 0.7 volts across those resistors. But you'll be using R133, R134, etc. to set the final bias.
Well, Digi-Key's going to replace the wrong transistors, so I should have them by the 10th. I also ordered the bias transistor I had wrong, some mica pads and grease, as well as new caps for the PL36 board while I'm at it. Nothing left to do until then but wait, I guess... unless I can replace the PN3569 transistor with something from Radio Shack while I wait for the parts? The closest part they have is the MPS2222A, is that close enough for testing at least?
I think I'll just cool my jets until I have the right parts rather than risk damaging anything.
I'm curious, can the light bulb limiter be used as protection when working on other amplifiers? A little while ago I bought an Onkyo M-5150 amplifier from a thrift store that keeps blowing its fuse. I'd turn it on, all the VU LEDs in the front would light up, and the device would let out a humming noise. I was going to wait until I had access to a variac to run any tests on it, but I'm curious to see what would happen if I plugged it into the light bulb limiter.
I'm curious, can the light bulb limiter be used as protection when working on other amplifiers? A little while ago I bought an Onkyo M-5150 amplifier from a thrift store that keeps blowing its fuse. I'd turn it on, all the VU LEDs in the front would light up, and the device would let out a humming noise. I was going to wait until I had access to a variac to run any tests on it, but I'm curious to see what would happen if I plugged it into the light bulb limiter.
blue lander said:
If you've already powered up the Onkyo any damage that was going to be done has likely been done. But yes, the lightbulb trick is a good idea working on any power amp. It doesn't come close to replacing a variac, however.
A variac lets you bring up the voltage slowly and immediately crank it back down if you see too much current. The light bulb can let the amp reach full operating voltage where it's much more likely parts will be damaged if something is wrong. But a lightbulb is much better than no protection at all.
Even with a variac I like using the power resistors in place of the DC fuses with the caps in parallel as I mention earlier. That trick will probably work with the Onkyo as well. But you should still use the lightbulb even with the resistors--in part because it provides immediate visible feedback when something goes wrong. The resistors won't last long if they have most of the power supply voltage across them during a fault condition.
This doesn't apply to the Phase Linear as it has no output relay but may apply to the Onkyo and any other amp with a relay. If you are using a Variac to power up an amp make sure you monitor the output BEFORE the relay with an o'scope. If you wait for enough voltage or time for the relay to energize you might be back where you started or worse. Using the scope in this way you can see DC offset and other nasties long before that relay goes.
Craig
Craig
By repair of a "Phase Linear 200" some years ago I observe also problems with oscillation, that was the reason for blown away the output power devices from NEC.
I simulate the circuit. The other types in the output stages creates no large simulation errors. The result of frequency response give me confirmation for the oscillation effects - see pdf attachement.
A very complicated and cad supported re-compensation was necessary in order to remove the oscillation. I think, you must do similar work.
I simulate the circuit. The other types in the output stages creates no large simulation errors. The result of frequency response give me confirmation for the oscillation effects - see pdf attachement.
A very complicated and cad supported re-compensation was necessary in order to remove the oscillation. I think, you must do similar work.
I got the new mj21196 transistors in the mail today, so I plugged 4 into the working channel. I was able to power the system up with no problem! I then connected my multimeter across R128 and adjusted the bias until I got about .375 volts (this is after letting it warm up for 10 minutes and disconnecting all inputs and outputs). The service manual suggested between .35 and .40 volts. I then plugged my laptop into the inputs and started the sine wave generator and looked at the output on the oscilloscope. I didn't see any distortion and it seemed to reproduce the sine wave faithfully. It's centered at 0 volts, and when I turn it up full blast it clips just a tiny little bit on the top and bottom, but it looked completely symmetric.
Tomorrow I'll be getting the new protection transistor for the bad channel as well as some new grease and mica pads. Are there any further tests I need to do on the working channel at this point?
Tomorrow I'll be getting the new protection transistor for the bad channel as well as some new grease and mica pads. Are there any further tests I need to do on the working channel at this point?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.