Pass F5m

Jump to Latest
#45
Glennkaspar

Regarding your question, try Digikey to obtain the Linear Systems equivalents (LSK170 and LSJ74). Each is around $14 - 15 each, you will need 2 of each. Digikey has free shipping for orders over $60 to Australia, so when you include the GST you are there. I don't think you're going to get away with spending less on the jfets, from personal experience.
 
#47
I added Store SMPS filters between each SMPS and its RC filter. There was no significant change in the low level noise in the 55kHz region of the spectrum.
The second image below is the noise with the SMPS filters. The third image is without.
I explored some other mechanisms for the noise coupling but could not find anything that changed the things.
Bottom line: The noise level of very low and mostly outside the audio spectrum.

1702750404931.png


1702750489750.png


1702750586512.png
 
#50
The 4kHz spike is still there. My earlier screenshots did not accurately resolve the height of the peaks. I do some more spectral measurements, both with and without the SMPS filters.

The first image is WITHOUT the filters, showing both the 60Hz noise and the 4kHz noise, which is at about -125.5dB.
1702754869830.png


The second image is WITH the SMPS filters, showing the 4kHz peak at about -134dB, which is a reduction of around 8dB.
1702754883188.png


There was no improvement in the 40kHz-60kHz range.
 
#53
Nelson Pass said:
I seem to see a 4 kHz spike? I am accustomed to spikes above 20k with switchers.
Click to expand...
I am unable to find an explanation for that 4kHz spike. It is 100% repeatable and only occurs in the channel with the SMPS. I have looked for instrumentation problems with no success.
I need to do some spectral measurements of the supplies in isolation to directly see the noise on their outputs.
 
#57
The loop gain would increase with higher frequencies. The impedance of capacitor C at frequency f is:
Z = 1/(2*pi*f*C)
Examples:
C= 0.1Farad f=20Hz Z=0.08 Ohms
C= 5643uF f=60Hz Z=0.47 Ohms
I could go into more details of this affect on loop-gain, but the bottom line is that there are probably better ways to increase loop-gain.
 
#58
I didn't mean as a goal to increase loop gain, but whether that consequence is a bad thing, and whether it might provide another PS filter. That's a rather big cap, yes, to extend down that far. But would it need to have a corner that low if you just want to filter 4kHz?
 
#59
That 4kHz spike is still a mystery, but is at such a low level that it should not have an audible affect.
With only the RC filters, the spike is 103dB below the 1 Watt into 8 Ohms, resulting in an amplitude of about 20uV at the amplifier output.
With the DIYA Store filters and the RC filters, the spike is 110dB below the 1 Watt into 8 Ohms, resulting in an amplitude of about 9uV at the amplifier output.

Perhaps my spectrum analyzer has too low a noise floor, allowing measurements that are of no importance.
 
#60
👍 No worries from me on that.

Though for grins I did a quick sim and it seems 1000uF from output sources to ground shouldn't wreck anything. OL gain would go up in middle frequencies, so perhaps some extra PS filtration in that range. Larger C would extend OL down further but any capacitance here shouldn't change overall FR.
 
You must log in or register to reply here.
Top Bottom