Eliminating power supply ripple in a G9 clone - suggestions?

Hello all. I have been working on a G9 based clone (made my own board designs with separate boards for each channel and a few other minor additions) and followed the G9 power supply design closely. I get a good bit of ripple in both the B+ and 12V supplies. The B+ has about 375mV ripple (p-p) and the 12V has about 180mV. Accordingly, I get a lot of 120Hz (and up) buzz at all but the lowest gain settings. If I get my digicam working soon, I'll post some scope photos of the ripple. Grounds are starred through the channel one input XLR case ground connection.

I'm using one 220µF/350V cap each for the B+ C14 and C15. I connected a 100µF cap across the 12V supply, and it didn't reduce the noise very much at all. So, what would you suggest I do to reduce this noise in the power supply lines? More caps, a choke? Would additional filter caps on the channel boards help? Or are the regulators suspect culprits as well? Thanks in advance for the help!!

What's your pre-regulator DC reading? I found when testing my G9 with a 12V 30VA power transformer the pre-reg supplies would sag too much and I'd get low readings and excess ripple. Installing the proper 15V transformers fixed the probs I was experiencing. I also used slightly larger VA rated transformers too - 50VA instead of the 30VA ones specified in the schem.

The 12V regulator definitely needs to fitted with a heatsink. The high current draw from the 12V rail in this design can cause the reg to go into thermal shutdown if not sufficiently heatsinked, which could give all sorts of wacky readings and excess noise in the PSU.

Very interesting - thanks Curtis. I used 50VA toroids as well, but one is a 12V secondary and one is a 15V secondary. The 15V is the high voltage trafo. I'll check the prereg voltage and see. What is a good number? The regulated 12v supply comes in around 11.9V. The 12V reg is attached to the rear panel with lots of silicone grease, but I'm wondering if I sent it into thermal shutdown once before I got it adequately heatsinked. (heatsunk??)

Just checked the prereg voltage and I get 19.1V. I scoped the prereg voltage and I get a nearly symmetrical triangle wave - that doesn't seem right to me either. Could I have bad diodes?

I used 50VA toroids as well, but one is a 12V secondary and one is a 15V secondary.

Click to expand...

I think both transformers need to be the same secondary voltage otherwise you end up with too low a step-up on the HV regulator. You need to step down from mains to 12V (or 15V), and step up again to 240VAC using the second transformer. If you've used a 12V transformer feeding a 15V transformer you're only going to have 200VAC at the input of the rectifier feeding the HV reg (transformation ratio is 16.7:1, or 12V *16.7 = 200V), which is probably too low for the TL783 to regulate with.

From memory there is also a minimum load that the TL783 needs to see before it starts regulating properly, so you'll need to have some tubes plugged in to get proper readings.

I scoped the prereg voltage and I get a nearly symmetrical triangle wave - that doesn't seem right to me either. Could I have bad diodes?

Click to expand...

Or dud filter caps? Or a really heavy load on the 12V supply? What happens when you disconnect the output of the 12V regulator from the rest of the G9 circuit - can you make some no-load measurements?

The transformers are hooked up so that the 15V secondary provides the voltage to the 12V regulator circuit and the 48V upconverter. The 12V secondary trafo is connected in reverse to provide the stepup for the B+ supply, so there's plenty of voltage for the TL783. I get 280V for the regulated B+ supply, and that's connected to both channels, under load. I was concerned about that issue and saw in some threads in the G9 meta that using the 15V to 12V connection would boost the B+ to proper levels.

The 12V load may be an issue - in addition to the G9 design, I have additional draws on the 12V supply for some extra switched relays and a board mounted switching power supply that drives the VU meter circuitry. There's a few extra LEDs over what Jakob's design has as well. Not a lot of extra current draw compared to the original, but a bit more. I changed the regulator to a 78T12, which is rated at 3A instead of the 1.5A of the 78S12 in the original. I'll check the no-load levels tonight.