Ripple in unregulated supply for class-A amp

[IG81]

Two ~1.3mH 16awg air-core inductors seem to be the best so far, taking ripple down to ~20mV. Four gets me to ~10mV, pretty good, can still hear faintly in not sensitive speakers. They do get hot after a while and I wonder just how much they can take. Could a single 16awg 9.1mH Solen coil take ~8A DC on a continuous basis, DCR is 1.18ohm, maybe a bit high? Price goes way up moving to larger gauges. Not sure if there is a cored choke out there that won't saturate with this current.

I think I really have to go CLC instead of CRC, any resistance that reduces ripple enough takes B+ down too much and dissipates an obscene amount of heat, current stays the same since my output biasing is through CCS.

IG

 

[AndrewT]

8A across 0r1 = 6.4W of dissipation.

Do the arithmetic and there are no surprises.

 

[agdr]

They Hammond chokes have too much DCR for my application, I would have to had planned for them from the start or to lower my power output as it stands.

0.044 ohms for that 10A DC rated Hammond 159ZL? That is (8A)(0.044R) = 0.35Vdc drop. You can get that much DC voltage fluctuation to your load just from the line voltage fluctuations during the day feeding the transformer primary. You may have been looking at the wrong resistance number. Or I may be missing something.

But that Hammond choke is (just) 2.5mH and it sounds like 1.3mH x 4 with your air core chokes is still not doing the job. Have you tried distributing those chokes between several banks of 2 x 10mF like I originally posted (use the air core chokes in place of the 0.1R resistors). You will probably get more ripple suppression from a multi-stage filter like that.

Interesting and very clever about using the old microwave transformers as chokes!

Another thought - you have have the power supply hum reduced to the point where demeterart's capacitor bypass trick on the transistor base biasing may be worth a try to nail the rest of it.

 

[IG81]

I had only looked quickly at the Hammond choke series, there are indeed a few candidates with DCR below 1ohm, but as you noted, a bit low inductance or current rating. I was considering two of the 159ZJ, but I'd have to split my cap bank into two, feeding two channels each, to reduce current.

I guess I'll see in simulation and then perhaps practice if CLCLC would work better. If I can get a ripple figure under 10mV, the cap-bypass trick might be good to finish this off.

Thanks for the tips to all who posted.

IG