KatieandDad, I guess you need to become a bit more systematic in order to figure out where the problem is. The DC at the speaker output is determined by the balance of the Aleph current source (the "upper" power MOSFETs) and the gain devices (the "lower" MOSFETs). Either the Aleph current source or the gain devices are not doing what they should do, and it would be useful to understand which of the two it is acting up.
Don't just replace "random" parts without clear evidence that they're wrong or broken.
Here's what I suggested earlier:
That said I had somewhat similar symptoms with a recent Aleph build. It turned out that the MOSFET in the Aleph current source (the "upper" MOSFET) had an "on/off tendency" to oscillate. If I touched the pot, it would suddenly go into oscillation, taking the DC output to the rail voltage. The problem was with my choice of the power MOSFET in the Aleph current source, which did not provide voltage headroom for the BJT to bias up the current source in a stable point (ZM was smart enough to figure that out for me).
I don't expect you have the same problem since you used the right MOSFET parts (not like me), but it will help if you are more systematic in isolating the problem (for example by trying my suggestions above).
Do you have a scope? If yes, try looking at the speaker output or the power MOSFET pins if you see any high-frequency oscillation.
I do have a scope.
The weather is awful today so I might not get around to much diagnostic work.
The weather is awful today so I might not get around to much diagnostic work.
They are correct as they are BC. I checked the orientation and pinouts carefully before installation.
I even tested every component before fitting them.
I even tested every component before fitting them.
With R4 disconnected I can just about get the DC offset to stabilise at +4V but it is very unstable.
Q7 and Q8 both have a stable -19.35V on their gates.
Q5 and Q6 have a slowly varying approx. +5.6V (it varies between about +5.4V and + 5.8V.
All voltages are reference to 0V.
Q5 and Q6 have a slowly varying approx. +5.6V (it varies between about +5.4V and + 5.8V.
All voltages are reference to 0V.
Re-assessing the output, there does appear to be a 900mV 100Hz oscillation on the output, a sawtooth waveform.
Attachments
Last edited:
I've just turned the amp off for the time being, there would have been a God-Awful thump from the speakers with the DC offset shooting off to the -ve rail.
Last edited:
Could it be as simple as the length of the power supply cables to the amp board ?
I could try adding some decoupling caps closer to the board. I've got some 4700uF that I could make into a CRC filter with the wires from the main caps forming the R.
I could try adding some decoupling caps closer to the board. I've got some 4700uF that I could make into a CRC filter with the wires from the main caps forming the R.
Ok, let's try to put things together (remember, DC feedback is off):
That looks like noise from the power supply / rectifiers. The oscillation I mentioned would be way up in the kHz / MHz range.
It's expected that DC stability is worse without the DC feedback, since there is no feedback to correct the offset. However, it's interesting that the DC does drift so much, and it approaches a value somewhat close to zero.
Much of the DC drift in the Aleph designs is caused by the temperature change of the BJT (Q4).
--> Check (1): what happens with the DC at the speaker terminals if you blow on Q4 (use a straw or similar to focus the air flow to Q4)?
This means the gate-source voltages (Vgs) are as follows:
Q5, Q6: Vgs is approximately 5.6 V - voltage drop on R16 or R17 (ignoring the BJT/Q4 Vce for a moment)
Q7, Q8: Vgs is approximately 24.33 V - 19.35 V - voltage drop at R18 or R19
If the amp is (would be) working fine, the voltage drops at R16, R17, R18 and R19 would be (almost) the same, and they would be about 0.7 V (Vbe of the BJT) or slightly less (R15 drop):
Q5, Q6: Vgs = 4.9 V (still ignoring the BJT/Q4 Vce)
Q7, Q8: Vgs = 4.3 V
The values don't look awfully wrong. One might think that it looks like the amp wants Q5 and Q6 to turn on harder than Q7 and Q8. Is it possible that either Q5 or Q6 don't do what the amp wants them to do, or do in in a very unbalanced way?
--> Check (2): please measure the voltages across R16 and R17. While you're at it, also measure the voltages across R18 and R19. While you're still at it, confirm the resistor values of R16-19 are correct.
Now, I neglected the Vce voltage of the Q4 BJT, which was probably not right. To improve on this, we'd need to know the voltages across the Q5, Q6, Q7, Q8 pins.
--> Check (3): please measure the voltages between the gate and source pins of Q5, Q6, Q7 and Q8.
That looks like noise from the power supply / rectifiers. The oscillation I mentioned would be way up in the kHz / MHz range.
It's expected that DC stability is worse without the DC feedback, since there is no feedback to correct the offset. However, it's interesting that the DC does drift so much, and it approaches a value somewhat close to zero.
Much of the DC drift in the Aleph designs is caused by the temperature change of the BJT (Q4).
--> Check (1): what happens with the DC at the speaker terminals if you blow on Q4 (use a straw or similar to focus the air flow to Q4)?
This means the gate-source voltages (Vgs) are as follows:
Q5, Q6: Vgs is approximately 5.6 V - voltage drop on R16 or R17 (ignoring the BJT/Q4 Vce for a moment)
Q7, Q8: Vgs is approximately 24.33 V - 19.35 V - voltage drop at R18 or R19
If the amp is (would be) working fine, the voltage drops at R16, R17, R18 and R19 would be (almost) the same, and they would be about 0.7 V (Vbe of the BJT) or slightly less (R15 drop):
Q5, Q6: Vgs = 4.9 V (still ignoring the BJT/Q4 Vce)
Q7, Q8: Vgs = 4.3 V
The values don't look awfully wrong. One might think that it looks like the amp wants Q5 and Q6 to turn on harder than Q7 and Q8. Is it possible that either Q5 or Q6 don't do what the amp wants them to do, or do in in a very unbalanced way?
--> Check (2): please measure the voltages across R16 and R17. While you're at it, also measure the voltages across R18 and R19. While you're still at it, confirm the resistor values of R16-19 are correct.
Now, I neglected the Vce voltage of the Q4 BJT, which was probably not right. To improve on this, we'd need to know the voltages across the Q5, Q6, Q7, Q8 pins.
--> Check (3): please measure the voltages between the gate and source pins of Q5, Q6, Q7 and Q8.
Sorry I've been busy.
OK R16 - R19 all measure at 0.5R within the limitations of my meter.
VR16 = 418mV
VR17 = 407mV
VR18 = 400mV
VR19 = 424mV
With +ve lead to S and -ve lead to G
VGS Q7 = -4.48V
VGS Q8 = -4.46V
VGS Q5 = -4.59V
VGS Q6 = -4.60V
OK R16 - R19 all measure at 0.5R within the limitations of my meter.
VR16 = 418mV
VR17 = 407mV
VR18 = 400mV
VR19 = 424mV
With +ve lead to S and -ve lead to G
VGS Q7 = -4.48V
VGS Q8 = -4.46V
VGS Q5 = -4.59V
VGS Q6 = -4.60V
Last edited: