How much B+ ripple is too much?

[gdgross]

Hi guys, have been out of the homebrew amp community for probably >10 years; I used to be fairly active on AX84 and apparently here too, as I had an account I logged into!

Quarantine has seen me resurrecting an old amp I never finished with all the equipment from the office I now have at home. It's rather a complex beast: channel switching, EF86 first stage, 2x6V6 power stage with power scaling a la London Power, reverb/tremolo. AND I cant find my old schematic, so I'm tracing everything by hand. (Send me prayers and good thoughts!)

I've got signal all the way through one channel to the speaker and it actually sounds pretty great TBH, but I'm hunting down a couple of buzz sources.

The first one is in either the PA or the poweramp. My question for y'all is, how much B+ ripple is too much on power tube plates/screens?

I measure almost 2-3Vp-p of ripple on each power tube plate. It's in phase, so in theory it cancels, but I'm sure that's not a perfect cancellation. So, how much is too much?

(here's my plates, scaled down to about 250VDC so I don't blow my 300V scope inputs)

[gdgross]

Ok, well it's not the ripple on the B+ going to the plates, I can confirm this.

This new scope trace is of the plate of one side of my PI.



The other side shows similar spikes, but not quite as severe. When I AC ground the PI plates through a big capacitor and 10ohm resistor, the buzz disappears and the plate looks clean. When I AC ground the PI input, it doesn't *quite* go a way, but it gets better. When I hard ground the PI input (which I can do because it's ~0VDC), it does go away entirely. Therefore it's getting in inside the PI.

The PI B+ looks clean.

The period of these spikes seems like 120Hz rather than 60, so it must be from a rectifier somewhere, right?

PS documenting the troubleshooting here helps me work through it logically on my own haha, whether or not ya'll have any insight.

[xtian]

Welcome back!

Based on your clues, the noise is almost certainly a grounding issue. Please shoot a photo of the whole circuit, and indicate where all of the grounds are.

[gdgross]

Thanks xtian!

It's kind of fun getting back into this, especially now that I have actual test equipment at my disposal. And I'm super stoked at how good this amp is sounding already. I did something right ten years ago lol.

FYI I don't think anything is getting in at the PI anymore. grounding the signal at the effects loop output also silences the buzz.

Here's a pic of the amp innards (sorry for the sideways view!)



The grounding and power supplies work like this: the card on the very bottom of the pic is the power supply card, which splits the supplies for the power amp and the preamp stages. The Power amp has B+ and bias voltages scaled by the master volume knob a la london power, and the circuit works great. The preamp stages, including LTPI are separate from the Power stage B+, and not scaled. There are a couple stages of RC to clean up the preamp B+ before the PI, and everything else trickles down from there.

Each supply feed is a twisted pair routed out to it's destination. I've drawn these on top of the picture for hopefully a little bit of clarity on top of this mess. PI/fxloop/preamp routed off to the left, and power supply off to the right. The effects loop send and return wires and controls are in the upper right of the chassis. You can see a bunch of ground wires there, and now that we're talking grounding that does look suspicious to me.

(reverb and trem not populated yet, ch2 populated but not working)

Current signal flow:

input -> EF86 stage -> channel select relay -> tone & volume controls -> effects loop -> LTPI -> power amp

Thanks!

[xtian]

That's a complex build!

Where is the rectifier grounded? Where is each node of the B+ chain grounded? Where is the input jack grounded? Output jack? Is there NFB? Where is the FX loop grounded? How many places on the chassis are ground points?

Have you read up on good grounding schemes? E.g., http://www.valvewizard.co.uk/Grounding.pdf

[gdgross]

Thanks again xtian, and thanks for that reference too. I will have to do some reading up!

There are several rectifiers in this amp: the main B+, the negative bias voltage, and a lower voltage 12V that gets regulated to 5V for my switching relays. All rectifiers live on that card in the bottom section of the picture, and are grounded to the chassis via the mounting screw in the bottom right corner, right near the negative terminal of the first B+ filter capacitor. (in the extreme lower right of the picture.) Ultimately, all supplies are regulated due to the power scaling anyway, but that probably doesn't matter for the buzz question at hand.

There is a twisted wire pair for each card that connects B+ and and the return. They ultimately come back to the main power supply card, but may pass through another card first. There are no chassis connections to ground except for the one near the main filter cap. ie, chassis is not used to carry any return current.

There is a serial effects loop consisting of an effects send buffer and an effects return amplifier (2 halves of a 12a_7). With the effects loop bypassed, which just routs the audio through the buffer and bypasses the return amp, the buzz is present, and disspears when I ground the output of the buffer. When the loop is engaged, there is a horrid buzz from the return amplifier. THis leads me to think the buzz is mainly happening in the return stage and perhaps leaking into the output of the input stage.

Dunno yet, my amp troubleshooting skills are very slowly coming back on line. will report more tomorrow.

Thanks

[Roe]

ripple is essential to some of the classic guitar tones. with a push-pull amp that is balanced well, you can get away with an surprising amount of ripple before the ghost tones become unbearable. the jtm100s or black flags are a classical example - being a 100w plexi with 32uf on the mains and 16uf on the screens.

[gdgross]

Thanks Roe, that's good to know. It looks like I'm at about 2V ripple on a ~380V B+ supply for my power tubes.

I'm now sure that my problem isn't B+ ripple. Since when I ground the output of the effects loop buffer, the buzz disappears. just gotta figure out how that buzz is getting in there.

[martin manning]

It looks like I'm at about 2V ripple on a ~380V B+ supply for my power tubes.

I'm now sure that my problem isn't B+ ripple. Since when I ground the output of the effects loop buffer, the buzz disappears. just gotta figure out how that buzz is getting in there.

So < 0.5%. That's typical, if not on the low side. The +/- spikes look like they are from the input side of a FWB rectifier.

[gdgross]

So < 0.5%. That's typical, if not on the low side. The +/- spikes look like they are from the input side of a FWB rectifier.

Thanks martin - That makes sense, since the period of the pulses in the second scope trace are 1/120Hz, not 1/60Hz (and they go both + and -).

There's only 1 FWB rectifier in this design, for the bias supply. I did put a scope probe on it and I didn't see any ringing on the diodes, was hoping I could cure with some snubbers or something. Looks like maybe it's a ground problem as xtian suggested.

Thanks

[gdgross]

Ok-

I've lifted the B+ return from the effects loop card and verified via DMM that there is no other (unintentional) connection to the effects loop card (it it measures many tens of kohms to over a meg). I then connected an alligator clip to the B+ return on the effects loop card, and connected it to various grounds in the rest of the amp, listening to the amount of buzz in each location. The locations included three different points on the chassis, the power tube plate B+ ground, screen B+ ground, the PI B+ ground, and ground near that FWB rectifier for the bias voltage.

In none of these instances was there any change to the buzz. They all sounded about the same.

Again, the only way I've been able to cure this buzz is shorting the output of the cathode follower effects loop buffer (after the coupling cap of course) to ground. Hard grounding the grid of this CF does not silence the buzz. So it's definitely getting in in this buffer stage. Question is how, and how do I cure it. I'm starting to bang my head on the wall on this one....

I do have a few high frequency bypass caps on the B+ supply in parallel with my bulk caps. (like ceramic 1kV, 1000pF or similar) Is it possible that these could be the issue? I never see them in guitar amps, but the EE in me couldn't help adding them when I built these boards...

Thanks