Amp HF distortion - only with covers on?

This MOSFET AB amp - PMF2150B - measures fine with the top and bottom covers off, within the limitations of my gear anyway. (theres a small DC offset issue, 6mV one channel 40mV the other, but I'm pretty sure I can fix that by replacing the input differential transistors with better gain matched pairs).

BUT When the top and bottom covers are back on, the sine wave on the scope goes really thick and fuzzy, while still visually resembling the fundamental, and on the basic (8bit?) FFT my scope has there is a huge spray of noise around 20kHZ and 220kHZ, and a way high noise floor. I can literally remove the distortion while unscrewing the covers. This is compared with a well functioning Yamaha MX1 AB amp I have here.

Hopefully it is something simple - the service manual under high frequency distortion suggests a few things, one being heatsink coupling capacitors - I checked the heatsink coupling capacitors and they measured 0.01uf where as the schematic says 0.1uf, could this be the problem? Or is it just lack of apparent EMI/RFI filtering on the Perreaux amp power supply vs the MX1? I have both schematics here and can post if need.

20210611_172806.jpg
 
It's hard to make any more specific comments without a schematic, but it seems that your amplifier is oscillating. If it's at the edge of oscillating, just about anything can change its behaviour between seemingly normal and grossly distorting.

Chances are that those capacitors you wrote about are meant to keep the amplifier stable, so it could very well be related to their way too low value. It could also be something else.
 
Last edited:
After deep cleaning, (and a crash course in electronics and amplifiers) I've replaced the heatsink/rail coupling capacitors (C23 in the attached schematic) and it seems* to have sorted the problem. Pinched some good 0.1uf 100V caps out of another old amp until the replacements arrive.

The guy I bought it off was the original owner :eek: 35 years of oscillating much - d*mn it even runs cooler now too :confused: Wrong spec capacitors from the factory? It didn't look like it had ever been worked on, but I've not much experience to say.

*The HF hash/oscillation was pretty random and hard to replicate without the covers on, but for now it seems to be holding will have to keep randomly checking with the scope over the next few weeks.

img-1-hHMAdwyr3Q.jpg
 
MOSFETs can be buggers for this sort of oscillation. It might be worth you renewing the small (too small IMO) rail decoupling electrolytics. The schematic seems to suggest those are 10uF.. personally i would use 100uF-470uF


Yes, and make sure there are some fast decoupling caps as well (MLCC chip capacitors are ideal, but any ceramics are better than none when it comes to HF oscillation) - MOSFETs are unforgivingly fast.

So 100uF electrolytic + 220nF ceramic perhaps? (And check the voltage rating of any ceramics!) The existing schematic shows two each of 10uF and 100nF, which is good, but increasing the values a bit can't hurt and will enhance stability and allow for electrolytics deteriorating over time.


Another observation is that 220 ohm gate-stopper resistors might be a bit small, increasing these a bit may help stabilize the MOSFETs too.
 
Last edited:
Just so I'm clear, you guys refer to C15-C22?

I've replaced Q1 and Q2 with some MPSA92's and the dc offset is now within +-4mV each channel, do you think it is worth replacing Q3,Q4 NPN'MPSA42's and Q5 PNP seeing as they are the same transistors and likely would have degraded also, or just leave them as factory?
 
Working on a unit like this I can confirm the observation in post 1, 4 and 7
with regard to oscillation, so you may want to look at C 23 in both channels.

In this particular amp no. P4943 the caps were labeled 33 nF and surprisingly
enough one had only 3 nF of capacitance - nominal value in the schematic is
100 nF.

I replaced these with two Siemens 100 nF ceramic caps and the tendency for
oscillation went away completely, irrespective of load, without any load and with
different wiring and grounding on the workbench.

Oscillation was mains voltage dependent before also, potentially increasing with
voltage.

The pcb layout of this amplifier is a mess. Since the circuit is close to the original
Hitachi application for 2SJ50 and 2SK135 Mosfet output devices, you probably can
find hints for repairing other faults in the context of similar designs.

The gate stoppers R 18 .. 23 are not close to the gates, but about 1 inch apart.
 
  • Like
Reactions: 1 user