My power amp's DC Bias is pretty stable and to spec, but the DC Offset bounces constantly. The swings are not large (+/-15mV left, and 0-8mV right) but it never settles and bounces through 3 random values every second.
Am I right in thinking this constant bouncing is not normal, let alone the left channel of +/-15mV? Is this a clue that all is not working optimally and I should have concern to get it resolved if I want this amp to perform to its potential?
The trim pots are a bit sensitive and fiddly (Bourns 3299) so don't know if they are going bad after 25 years.
Model 70 Service Manual
Am I right in thinking this constant bouncing is not normal, let alone the left channel of +/-15mV? Is this a clue that all is not working optimally and I should have concern to get it resolved if I want this amp to perform to its potential?
The trim pots are a bit sensitive and fiddly (Bourns 3299) so don't know if they are going bad after 25 years.
Model 70 Service Manual
That kind of offset won't do any damage, and it's doubtful it would effect the sound. Having said that it should be more stable. Turn the pots back and forth a few times see if that makes a difference, also see if it's stable at a higher offset
These are definitely worth replacing. While you are at it replacing C104 may also be a good idea, as it may have become a bit leaky.
Service manual prescribes 1mV...a bit optimistic perhaps.
Service manual prescribes 1mV...a bit optimistic perhaps.
If Q101/3 were matched, there will be no need for a DC offset control.
I would check one of them is not noisy.
I would check one of them is not noisy.
You're concerned about Q101/3 but not Q102/4 or Q105 for that matter?
My DMM will test transistors, but no scope or test circuit to check for noise. It might be just easier to replace them preemptively if I replace the trim pots. Yes?
Yeah, leaning this way on the trim pots. I already replaced all the electrolytics five years ago, so C104 should be great.
I don't think it is noise per se. If it is not the caps, nor the trimpots, you may have a leaky transistor. No easy way to measure this, a scope will certainly be no help.
This is called 'motorboating'. It results from two interacting RC time constants in the DC part of the feedback loop. Like a DC servo and an AC coupling with similar time constants. Pull them apart and it will stop.
Jan
Jan
I agree with the above posters re: input paired transistors. If the issue gets worse as things warm up more then this is probably an indicator of this. It may indicate that your input pairs are drifting in their values. Changes to this extent however arent much to worry about.
OK, I'll order the bits, and probably install in the next month, see what results.
Thank you for all of your help.
Sent from my GT-I9505 using Tapatalk
Thank you for all of your help.
Sent from my GT-I9505 using Tapatalk
Don't change transistors yet. Don't change non electrolytic capacitors.
Try multi-turn pot first.
Try anything before you start changing transistors.
Try multi-turn pot first.
Try anything before you start changing transistors.
Because..?
It's a real pain every time I get access to the under side of the pcb with several un-soldering and re- soldering on an old pcb just to get in there.
Sent from my GT-I9505 using Tapatalk
It's a real pain every time I get access to the under side of the pcb with several un-soldering and re- soldering on an old pcb just to get in there.
Sent from my GT-I9505 using Tapatalk
Because we have no data on what Classe did during selection of the transistors.
You could ruin the performance of the amplifier by changing any transistor to a device that does not meet Classe's selection criteria.
You could ruin the performance of the amplifier by changing any transistor to a device that does not meet Classe's selection criteria.
Don't overlook popcorn noise in the input devices, very slow pops like that have been observed many times. One of the heatsinks on the cascodes is misplaced (at least on the schematic)? Classic op-amp on steroids. 🙂
surely some mistake in the circuit? MPSA56 and MJE350 positions swapped?
In circuits like this (asymmetrical) I have seen thermal oscillations cause DC bias drifts. Plastic transistors are worse than metal cans (longer time constants) but metal cans need isolating from each other when using a common heat clip. If circuit behaved better in the past then I would suspect the electroltyics.
In circuits like this (asymmetrical) I have seen thermal oscillations cause DC bias drifts. Plastic transistors are worse than metal cans (longer time constants) but metal cans need isolating from each other when using a common heat clip. If circuit behaved better in the past then I would suspect the electroltyics.
Thermal oscillations could also be it. Kind of thermal 'motorboating' (I think it should be clear where the term comes from ;-).
Edit: Scott, Jon, maybe the heatsinks on the cascode are on the correct transistors but they swapped the transistor positions on the schematic?
Jan
Edit: Scott, Jon, maybe the heatsinks on the cascode are on the correct transistors but they swapped the transistor positions on the schematic?
Jan
Last edited:
Yes after looking that is probably it. I also noticed a couple of spots where a DC level is determined simply by base current through a very large resistor. This makes substitution difficult since that voltage is beta dependent. One of my fellow engineers (very serious audiophile) had all Classe electronics, not sure if any of the design eccentricities had sonic implications.
The curse you hear is over the realisation. But thank you for warning me, makes things less simple but less likely to end in tearing of garments.
I did turn the pots back and forth, had no real effect on settling the Offset readings.
I don't see a heatsink misplaced, though the schematic could have swapped the MPSA56 and MJE350 positions.
To be clear, the sound was pretty good and I only did a check of the Bias & Offset on a lark while I had it open looking at replacing the big cans (caps). Come to think of it I've never checked B & O until 2013 after changing all the board electrolytics. My notes from then show the Bias was at 27mV which I left, and Offset was bouncing around. Changing to 21mV per the specs actually produced a result that sounded like a decrease in performance, but I still have some experimentation yet.
If thermal is an issue, I could only see it likely as to the position of Q106 (MJE350), but it seems that might only be addressed by a larger (taller?) heatsink. There's not a lot of room to play with there. See attached photo B4 cap change.
Attachments
- Status
- Not open for further replies.