I tried the PSU without the CRC so using the Juma cap Mx in true star ground topology. I increased the output caps on-board the cap mx to 4700uF and added a 1uF x7R MLCC cap at each of the snubbers (so 1uF plus 0.1uF film cap bypass ea). Connecting the speaker return to the star hub vs the amp PCB gnd helped a little. Even with RCA cable grounded, there is still 1.1mV noise. So I am surprised that this is not working because a SMPS/Cap Mx/CRC worked really well in the MoFo and in the Aksa Lender preamp. But those were not dual rail PSUs and I wonder if that is the difference here. The photo shows a blue wire from the RCA jack gnd the star hub, and that helped a little with the RCA plugged in, but did not make the noise go away. It's not all hum - there is some other noise in there too - need to take a measurement and see where it is.
It might be a noisy or defective DC step-up converter. These were very quiet before but perhaps I have some bad units here, or the cap Mx is misbehaving and adding noise. The o-scope shows a series of pulses at 285kHz and within the pulse trains are oscillations of 4MHz maybe 1vp-p. Looks like the DC-step up is noisy and leaking right through cap Mx. There is actually very little 60Hz "hum" or ground loops.
I used the same DC step up for the Aksa Lender preamp and it is dead quiet. Measurements show that too.
It might be a noisy or defective DC step-up converter. These were very quiet before but perhaps I have some bad units here, or the cap Mx is misbehaving and adding noise. The o-scope shows a series of pulses at 285kHz and within the pulse trains are oscillations of 4MHz maybe 1vp-p. Looks like the DC-step up is noisy and leaking right through cap Mx. There is actually very little 60Hz "hum" or ground loops.
I used the same DC step up for the Aksa Lender preamp and it is dead quiet. Measurements show that too.
Attachments
Last edited:
With +/-25v rails I measured 37.2vpp 1kHz sine into 8ohms before onset of clipping. This 21.6wrms - so the 20w designation for this amp is correct. With a finer resolution volume knob and using Focusrite as DAC I was able to get 38.8vpp into 8ohms before any sign of clipping. That's 23.5wrms.
Attachments
Last edited:
No, nothing capable of the 5amps needed for both channels. When I look at the noise from the DC step up all by itself, it is less than what is coming out of the cap Mx. So there may be some instability in my cap Mx that is amplifying the noise. I added 100R-330pF snubbers at the input to the cap Mx and that cut the 4MHz noise down by half but still there.
Perhaps an NTC8D-20 thermistor would be similar (8ohms cold and capable of 6amps). Do you mean add this between the SMPS and the DC step up or between the DC step up and the cap Mx?
The SMPS are floating relative to earth gnd. The SMPS 24v comes into the amp as isolated from the earth gnd.
Within the amp, 0v is at the star gnd of the cap Mx output.
Perhaps you can indicate on my diagram where you think the 10R should go? I am not sure.
I think the issue is noisy switching so that maybe either the SMPS or the DC step up.
Last edited:
I am not sure I understand your question. Are you asking if the DC to DC converter is grounded? It is connected to the smps at input (floating) and the input pins at the cap Mx. I think the common (0v) connector is grounded to amplifier internal 0v Star. The amp itself is not connected to Earth GND anywhere. I took a Earth grounded wire and touched the chassis and the Star ground and the noise did not go down.
Maybe the common mode noise after DC converters, if passed through common mode choke will go away?
Hi X,
remote debugging is not easy!
VO+ is positive and VO- is negative output (or return) from booster and there´s no connection to châssis or earth (converter or booster not grounded) that´s what I´m mean.
Perhaps common mode noise from AC/DC converter, perhaps from booster, perhaps both?
And, cerise sur le gâteau, cascaded and floating mounting (non isolated, non synchronized ones).
JP
remote debugging is not easy!
VO+ is positive and VO- is negative output (or return) from booster and there´s no connection to châssis or earth (converter or booster not grounded) that´s what I´m mean.
Perhaps common mode noise from AC/DC converter, perhaps from booster, perhaps both?
And, cerise sur le gâteau, cascaded and floating mounting (non isolated, non synchronized ones).
JP
X You see what is the problem it is not allowed me to upload Bitmap better resolution picture.
Even when I create a picture, layout etc with bitmap I have to convert it JPeg which degrades the quality very bad.
Only I can upload less than 200 Kb that is why the quality suffers.
How you post your picture?
Even when I create a picture, layout etc with bitmap I have to convert it JPeg which degrades the quality very bad.
Only I can upload less than 200 Kb that is why the quality suffers. How you post your picture?
Attachments
X You see what is the problem it is not allowed me to upload Bitmap better resolution picture.
Even when I create a picture, layout etc with bitmap I have to convert it JPeg which degrades the quality very bad.
How you post your picture?
While saving your bitmap save it as a png file. And that file you could upload.
No, thank you (and of course whole team!) and no worries
can't speak for everyone else, but atleast I have quite a lot to do before actually needing pcb's. Usually casework takes most time. Haven't even ordered trafo yet, going for Toroidys shielded audio grade 500VA 2x20V and easy cap mx. I guess well see does it need separate CRC supply after cap mx. Hope not.
Cherry on the cake, or "to top things off" as we say in English....
Yes, it may be the whole floating nothing bolted to the chassis that may be getting me.
That gives me an idea to try two of the Prasi/Project16 CRC's after the Cap Mx to see if it cleans anything up?
I have some nice common mode chokes (with airgap ferrite) that may be useful in stomping out this leaky noise.
1. JPS64 suggest swapping out DC step up for the thick heatsinked variety
2. try grounding stuff to earth ground via 10R and anti-parallel diodes GLB
3. add a common mode choke filter (LC)
I am looking over Prasi/Juma cap Mx board grounding path and it looks like the inputs from the SMPS (usually diode bridges +/-) is star grounded to the "clean gnd" 0V star ground point. If I were to do it again, I would break the ground upstream of the cap mx MOSFET and the SMPS and add 10R there - like Hugh suggests. It has a direct noise injection path that goes through no resistance at all.
Hi X,
thick heatsink replacement to see if metal core step up is source of problem (noise coupled capacitively over floating metal plate).
Is metal plate connected to step up ground? If not, then considering it´s the case! CMC on input, capacitor on input to case, capacitor on output to case, toroid and capacitor on output... or simply try to screw it on the châssis.
JP
thick heatsink replacement to see if metal core step up is source of problem (noise coupled capacitively over floating metal plate).
Is metal plate connected to step up ground? If not, then considering it´s the case! CMC on input, capacitor on input to case, capacitor on output to case, toroid and capacitor on output... or simply try to screw it on the châssis.
JP