Interesting. There are a couple of things here...
The amplitude difference is all important. Looking at this image it is how much higher in voltage the peek of that blue trace is compared to the yellow one that determines whether the following transistor will turn on. So if you are on 0.1 volts per div (accounting for any divider probe you might use 🙂) then there is indeed around 0.25 volts difference. Same applies to the other image.
That in itself is not enough to turn on a silicon transistor... so is something else going on here?
One question is why are the traces not the same. The only thing I can thing of is that this is oscillation and that the high current you see when the outputs are fitted is due to two possible causes (perhaps both happening together).
1/ There is conduction overlap of the relatively slow outputs and so they both linger in an on condition.
2/ The frequency of oscillation is high enough to cause considerable AC current to flow in the Zobel network at the output.
Even with no outputs fitted you could try measuring from ground to these two resistors (there seem to be two Zobels, one before and one after the relay). The first one has an effective value of 5 ohm. If we know the AC voltage across each resistor we can work out the peak current. With only drivers fitted its going to be low but of course it should be absolutely zero. Fit the outputs and I suspect it will be much higher.
That is all kind of another proof that the issue is instability and not a normal component fault as such.
The amplitude difference is all important. Looking at this image it is how much higher in voltage the peek of that blue trace is compared to the yellow one that determines whether the following transistor will turn on. So if you are on 0.1 volts per div (accounting for any divider probe you might use 🙂) then there is indeed around 0.25 volts difference. Same applies to the other image.
That in itself is not enough to turn on a silicon transistor... so is something else going on here?
One question is why are the traces not the same. The only thing I can thing of is that this is oscillation and that the high current you see when the outputs are fitted is due to two possible causes (perhaps both happening together).
1/ There is conduction overlap of the relatively slow outputs and so they both linger in an on condition.
2/ The frequency of oscillation is high enough to cause considerable AC current to flow in the Zobel network at the output.
Even with no outputs fitted you could try measuring from ground to these two resistors (there seem to be two Zobels, one before and one after the relay). The first one has an effective value of 5 ohm. If we know the AC voltage across each resistor we can work out the peak current. With only drivers fitted its going to be low but of course it should be absolutely zero. Fit the outputs and I suspect it will be much higher.
That is all kind of another proof that the issue is instability and not a normal component fault as such.
Thanks @Mooly
as each division is 500mv, does this mean that the delta between blue and yellow is 1.2v approx?
as each division is 500mv, does this mean that the delta between blue and yellow is 1.2v approx?
@pwdiya12
try to vertical align blue (ch 2) with yellow and you will be sure how much difference you have (as I calculated, you have blue line about 2.2 Vpp, yellow is about 0.8 Vpp), so there is enough difference to cause one transistor to conduct on high frequencies. If I remembered correctly, I think that PNP transistor will conduct and you will see on zobel negative pulses (if you follow Mooly's suggestion).
Now you have starting point to go backwards to the input stage and try to find where oscillation rises with signal similar to negative pulses.
try to vertical align blue (ch 2) with yellow and you will be sure how much difference you have (as I calculated, you have blue line about 2.2 Vpp, yellow is about 0.8 Vpp), so there is enough difference to cause one transistor to conduct on high frequencies. If I remembered correctly, I think that PNP transistor will conduct and you will see on zobel negative pulses (if you follow Mooly's suggestion).
Now you have starting point to go backwards to the input stage and try to find where oscillation rises with signal similar to negative pulses.
It does as long as the coupling was set to DC for both channels and both channels with no signal were aligned to the same point on the screen,
If that is so then that would force a very high current in the output stage.
What was the time/div set at in that shot. One cycle is approximately 2.5 squares in the horizontal plane. 1/T will give us the frequency. For example if the setting is 2us per division then one cycle is approx 2.5 * 2E-6 which is 5uS (for one cycle) and the frequency is 1/5E-6 is 200kHz
If that is so then that would force a very high current in the output stage.
What was the time/div set at in that shot. One cycle is approximately 2.5 squares in the horizontal plane. 1/T will give us the frequency. For example if the setting is 2us per division then one cycle is approx 2.5 * 2E-6 which is 5uS (for one cycle) and the frequency is 1/5E-6 is 200kHz
Do I understand correctly:
All 6 output devices are removed (or have their 4R7 base drive resistors lifted); C17 is shorted; R51 and R52 are connected to the front end; amp input is shorted as a precaution. This is the configuration that produces the scope photos in posts # 140 and 141? Please correct me if wrong.
Assuming the above is correct, I suggest an explanation of what might be happening:
We have 4 b-e junctions pk-pk of crossover distortion between C17 and the speaker terminals. The driving impedance seen by the Zobel network(s) is very nonlinear--- at 0V, it's very large; at 0.6V, Q29 begins to conduct and drive impedance tends to 270 Ohm (R57); at 1.2V, Q31 begins conduction and driving impedance tends toward R62//R57--- say 30 ohms. At audio frequency C24 and C25 look like 0.086uF. And 30 Ohms and 0.086uF form a lowpass pole at 62kHz. At lower amplitudes, 270 ohms and the Zobel caps form a pole at 6.9kHz. I suggest this pole, in cascade with the compensation of the front end, is the probable source of the oscillation. But when the amp is operating and biased properly, at least half of the 6 output devices are active and present a very low driving impedance to the Zobel, pushing the pole to a much higher frequency and stabilizing the feedback.
How to test this theory? Disable both Zobel networks to see if that stops the oscillation.
Another possibility is an experiment conducted on the working LM08: lift the base resistors at the 6 output transistors and install a short at C17. I believe this transforms the amp into the same state as the problem unit. If it presents the same behavior, that gives confidence to the notion that restoring the outputs of the unit under test wouldn't be catastrophic.
Comments?
All 6 output devices are removed (or have their 4R7 base drive resistors lifted); C17 is shorted; R51 and R52 are connected to the front end; amp input is shorted as a precaution. This is the configuration that produces the scope photos in posts # 140 and 141? Please correct me if wrong.
Assuming the above is correct, I suggest an explanation of what might be happening:
We have 4 b-e junctions pk-pk of crossover distortion between C17 and the speaker terminals. The driving impedance seen by the Zobel network(s) is very nonlinear--- at 0V, it's very large; at 0.6V, Q29 begins to conduct and drive impedance tends to 270 Ohm (R57); at 1.2V, Q31 begins conduction and driving impedance tends toward R62//R57--- say 30 ohms. At audio frequency C24 and C25 look like 0.086uF. And 30 Ohms and 0.086uF form a lowpass pole at 62kHz. At lower amplitudes, 270 ohms and the Zobel caps form a pole at 6.9kHz. I suggest this pole, in cascade with the compensation of the front end, is the probable source of the oscillation. But when the amp is operating and biased properly, at least half of the 6 output devices are active and present a very low driving impedance to the Zobel, pushing the pole to a much higher frequency and stabilizing the feedback.
How to test this theory? Disable both Zobel networks to see if that stops the oscillation.
Another possibility is an experiment conducted on the working LM08: lift the base resistors at the 6 output transistors and install a short at C17. I believe this transforms the amp into the same state as the problem unit. If it presents the same behavior, that gives confidence to the notion that restoring the outputs of the unit under test wouldn't be catastrophic.
Comments?
Its bloomin' difficult without it in front of you 😉 good thinking though 🙂
If I remember correctly when the outputs were fitted the current draw was high while the other channel behaves correctly. I think the amp had an unknown previous history, wrong parts and fried...
@BSST - thanks for the explanation and yes your statement as to the condition of the amp is correct
My current plan is to look at the output of the 1st stage(where Q4 connects to R13) of the input section and then second stage prior to R51/R52 to see if the oscillation is there.
If it is not, then I will remove the zobel network as suggested and what is the outcome.
@pitbul - although I checked each of the replacement parts numerous times, I will do so again
I will report back
My current plan is to look at the output of the 1st stage(where Q4 connects to R13) of the input section and then second stage prior to R51/R52 to see if the oscillation is there.
If it is not, then I will remove the zobel network as suggested and what is the outcome.
@pitbul - although I checked each of the replacement parts numerous times, I will do so again
I will report back
If I were in your shoes, I’d experiment with the working amp to develop a feel for how the amp behaves when severely under biased. With no load applied, I’d observe with scope as bias current is lowered via VR2 and perhaps with shorts across D8 and/or D9. And with/without Dim Bulb.
The general goal would be to to gain confidence that both the working amp and the amp under repair are in similar states, and to have rehearsal of how the amp under repair should behave as output devices and Zobels and bias are restored.
The general goal would be to to gain confidence that both the working amp and the amp under repair are in similar states, and to have rehearsal of how the amp under repair should behave as output devices and Zobels and bias are restored.
Thanks again.
I reinstalled C25, which is post the relay bringing R93 back into the circuit. There is no voltage across R62/R65 and no 210Khz on TP 15,16 or 17.
I then installed one of each output (2SA/2SC) with the same result, relay clicks and dim bulb is off.
I reinstalled C25, which is post the relay bringing R93 back into the circuit. There is no voltage across R62/R65 and no 210Khz on TP 15,16 or 17.
I then installed one of each output (2SA/2SC) with the same result, relay clicks and dim bulb is off.
Are you saying that adding that second Zobel flips the amp into oscillation? That was good thinking by @BSST to go down that route.
If so then my first thoughts would be to make sure that C25 really does connect via the series resistor and is not somehow shorted to ground. i.e. not adding a cap directly across the output.
Second thing is to check the ground end of the Zobel is OK and that it is not somehow floating in some way and adding positive feeding to another point in the amp circuit (if that makes sense 🙂) You could try a lashup network across the other one. Also make sure that first Zobel is grounded OK. If both share the same ground and its not correct then it might just be the second network pushes it all over the edge.
If so then my first thoughts would be to make sure that C25 really does connect via the series resistor and is not somehow shorted to ground. i.e. not adding a cap directly across the output.
Second thing is to check the ground end of the Zobel is OK and that it is not somehow floating in some way and adding positive feeding to another point in the amp circuit (if that makes sense 🙂) You could try a lashup network across the other one. Also make sure that first Zobel is grounded OK. If both share the same ground and its not correct then it might just be the second network pushes it all over the edge.
I have ordered new feedback relays ( I don't think it is a problem, but at this stage it can't hurt) and replacements for the 10 ohm 3W resistors (R89,R90,R93) which should be here in a couple of days.
I plan to follow the earthing and make sure that as you say C24 and C25 via their respective resistors are connected to the same ground- TP14 which connects to TP23
thanks again and I will report back
I plan to follow the earthing and make sure that as you say C24 and C25 via their respective resistors are connected to the same ground- TP14 which connects to TP23
thanks again and I will report back
So a work in progress update and an aid memoir for me as to the state and progress.
and a very big thank you for your thoughts, guidance and perseverance
I did replace R89/R90/R93) and the NF relay- no change
also checked all the earth connections around TP 14.
Another small difference on the circuit, on the board it is C25 that goes to earth rather than R93- so output line goes to R93 then C25 then earth.
I take the point that getting the "good" L08M into a state that can be compared with the faulty one is a good way to proceed. I also understand that not having the amp on the bench makes this type of diagnosis very tricky.
So I decided to reinstall some of the components we removed and start to remove some of the shorted components to get the faulty unit to a state where the bias caps are still shorted (zero bias) and C24 is lifted (that is no 210Kz on R62/R65 or TP17). My start point was to have one pair of outputs (2SA/2SC) fitted, C24 lifted, C17,C8,C9 and input shorted, Q25/Q26 removed, D11/D12 removed, Q31/Q32 installed (MJE as below), with zero volts on R62/R65 and with no 210kz on R62/R65, TP15,TP17 and dim bulb off.
"no Change" means no voltage across R62/R65, No 210Khz across R62/R65, no 210Khz on TP15,TP17- C24 still lifted, C8/C9 shorted and input shorted- Q31/Q32 installed- no outputs installed, dim bulb off
Removed the outputs (2SA/2SC)
I reinstalled D11/D12 - relays clicked - no change
I removed the short on C17- again no change
Replaced Q31/Q32 from my substituted MJE15034/MJE15035, back to the original (2SA112/2SC2592). The reason for this is the originals are 200Mhz and the MJE are 30Mhz, and having some familiarity with the APT power amp which also uses a triple darlington I wondered if this was in part causing a problem. Reninstalled C24 and there was a 194KHz on R62/R65 (with 0.27v DC). so no, but have left the originals in place.
Then replaced Q25..Q26 with KSC2383/KSA1013, No relay click- this with Q25 being a KSA1013.
Replaced Q25 with the original 2SC2378- No relay click
Then replaced Q42 after some measuring- and now relay clicks...🙂
I did the test the good L08M, and there is no 210Khz/194Khz on the output, no voltage on R62/R65.
My plan is to get the faulty amp to a state where I can install the outputs with C24 lifted, and short C8/C9 on the good amp, so that both amps are in the same condition (except for the C24 lift) and then compare as much as possible to see why the feedback by C24/R89/R90/R88- (R88 tested good out of circuit), is causing the oscillation
Peter
and a very big thank you for your thoughts, guidance and perseverance
I did replace R89/R90/R93) and the NF relay- no change
also checked all the earth connections around TP 14.
Another small difference on the circuit, on the board it is C25 that goes to earth rather than R93- so output line goes to R93 then C25 then earth.
I take the point that getting the "good" L08M into a state that can be compared with the faulty one is a good way to proceed. I also understand that not having the amp on the bench makes this type of diagnosis very tricky.
So I decided to reinstall some of the components we removed and start to remove some of the shorted components to get the faulty unit to a state where the bias caps are still shorted (zero bias) and C24 is lifted (that is no 210Kz on R62/R65 or TP17). My start point was to have one pair of outputs (2SA/2SC) fitted, C24 lifted, C17,C8,C9 and input shorted, Q25/Q26 removed, D11/D12 removed, Q31/Q32 installed (MJE as below), with zero volts on R62/R65 and with no 210kz on R62/R65, TP15,TP17 and dim bulb off.
"no Change" means no voltage across R62/R65, No 210Khz across R62/R65, no 210Khz on TP15,TP17- C24 still lifted, C8/C9 shorted and input shorted- Q31/Q32 installed- no outputs installed, dim bulb off
Removed the outputs (2SA/2SC)
I reinstalled D11/D12 - relays clicked - no change
I removed the short on C17- again no change
Replaced Q31/Q32 from my substituted MJE15034/MJE15035, back to the original (2SA112/2SC2592). The reason for this is the originals are 200Mhz and the MJE are 30Mhz, and having some familiarity with the APT power amp which also uses a triple darlington I wondered if this was in part causing a problem. Reninstalled C24 and there was a 194KHz on R62/R65 (with 0.27v DC). so no, but have left the originals in place.
Then replaced Q25..Q26 with KSC2383/KSA1013, No relay click- this with Q25 being a KSA1013.
Replaced Q25 with the original 2SC2378- No relay click
Then replaced Q42 after some measuring- and now relay clicks...🙂
I did the test the good L08M, and there is no 210Khz/194Khz on the output, no voltage on R62/R65.
My plan is to get the faulty amp to a state where I can install the outputs with C24 lifted, and short C8/C9 on the good amp, so that both amps are in the same condition (except for the C24 lift) and then compare as much as possible to see why the feedback by C24/R89/R90/R88- (R88 tested good out of circuit), is causing the oscillation
Peter
I am slowly making progress and will report back- I have two outputs installed, C24 reconnected C8/C9 connected (not shorted) VR2 at minimum and no 210Khz, relays clicking and dim bulb off......
so a quick further update- all outputs fitted, relays click, dim bulb off, no 210Khz, no voltage on output.
So this week will partly reassemble (aka reinstall all the broken wires connected by clips) and run some tone through it....
If that is good, then off the dim bulb and see it it biases correctly
Thanks again
Peter
So this week will partly reassemble (aka reinstall all the broken wires connected by clips) and run some tone through it....
If that is good, then off the dim bulb and see it it biases correctly
Thanks again
Peter