Help sizing psu smoothing cap

[riotubes]

Hi,

Please help size a smoothing cap in my psu. Here's what it looks like

Traffo >> schottky bridge >> 1uF film cap with a 100K ohm bleeder strapped across it. This powers a series reg (the Pimm revision to John Swenson's HV reg based on Pimm's SBCCS board).

I can get the correct 70V output on the swenson reg under NO load.
Here's the voltages
AC: 122 VAC
DC on smoothing cap: 182V
DC on output of reg: 70V

Voltage when the reg is loaded by 2.25K resistor (which approximates the load of the actual analog output state this reg is supposed to drive which is 70V at 33ma

DC on smoothing cap: sags to 147.3
DC on output of reg: sags to 58.4V

Its not the reg. I built multiple versions of the orig Swenson version, and now today I built the beefier Pimm revision. Same story every time. It's either the traffo or the smoothing cap is too small. I'd like to start with replacing the 1uF cap. What would be a more conventional size to identify whether the problem is the cap?

Thanks!

 

[jitter]

If the transformer has the correct ratings for the application and the smoothing cap is fine, my money would be on the bridge rectifier. The symptom of having good voltage unloaded but dropping voltage loaded is something I've seen several times in my work as tester/repairer of industrial electronics (dealing with tried and tested designs, of course).

A quick look with an oscilloscope will often reveal excessive ripple voltage when a diode inside the bridge rectifier has failed. If you don't have a scope, take out the bridge rectifier and use the diode-check on a digital multimeter. All four diodes should measure the usual voltage drop in the forward direction. If one reads very low (short), or very high (open), it's failed.

 

[cbdb]

I dont know these regulators but a 70V regulator that needs 180V input seems strange. Do they really need such a high input to regulate? At 147 V in thats still 80V! for the reg to work with.

 

[riotubes]

Thanks for the input. This is a reg designed by John Swenson and optimized by Gary Pimm. Yes, this is a fairly unconventional approach where the reg does the lion's share of the smoothing and the cap in front of it is sized small to reduce the peak currents through the diodes thereby reducing high frequency noise. The issue with the small cap is that the ripple is fairly large such that the bottom of the ripple must be sufficiently high above the dropout voltage of the regulator. I posted a more general question on my ills with this project in the Tube forum and the spotlight was cast upon this small cap issue. Gary replied by modeling out the ripple in PSUD2 and determined that this cap needed to be increased to at least 4.7uF such that the bottom of the ripple never drops below 125V providing a minimum of 50V to regulate which he says is optimal. Here's the link to the graphic which is quite interesting. Post #3.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=147041

 

[cbdb]

If you have a scope, you can see if its the cap, the tranny or the bridge. If a diode is gone the waveform on the cap will be at 60hz instead of 120hz. If the cap is to small the waveform bottom will fall excessively when the load is switched in, if the tranny is to small the waveform top will fall excessively.

 

[riotubes]

That's very educational for me. Thank you. Someday I'll invest in a scope...I think my learning would be accelerated considerably.