JMI AC30/6 - Vibrato woes

[four_corners]

Hey everyone,

Last year I built a JMI AC30/6 with integrated top boost with tons of incredibly helpful info from this forum. Everything was built to the Normal channel schematic and Top Boost schematic (attached), other than I added push/pull pots for On/Off-Speed, and Vibrato/Tremolo-Depth.

Everything sounds really awesome, except the Vibrato. The tremolo is fine, as it only uses one triode of the LFO and Mod valves, and while the vibrato "works", when I switch from the trem to vibe, it almost sounds double the speed, and is pretty thin, and not very vibey.

I scoped all around, and have decent looking sine waves in all the right spots, except at the anodes of the ECC82 (pin 1 and 6 of V8). They are correctly out of phase, but the positive side of both  sine waves look they have been run through a wave folder. I thought it might be that the ECC82 triodes were poorly balanced, so I found a NOS Tung Sol that is balanced down to 0.8%, but that didn't seem to help. I also thought it might be tolerance mismatch on some of the matching 0.1uF caps, but I feel like I'd be seeing wackier looking sine waves if that was the case. I can order tighter tolerance ones though if you think it could be that.

Voltages look good for V7,V8,V9. I've attached a bunch of scope images.

Any advice would be greatly appreciated!

[four_corners]

V9 images:

pin 1/6 image, and 3/8 image have the depth pot all the way down. Sorry the scales aren't exactly matching, but as you can see they are about 10x of each other at the lowest depth setting.

[four_corners]

The last 2 images are also of V9 at pin 1/6, and 3/8, but with the depth setting turned much higher. Can't remember it they were all the way up, but quite a bit.

[pdf64]

V9 pins6,7,8 is a cathodyne, providing opposing phase modulation signals to V8 grids.
What are the waveforms like at the coupling caps C34&37 between those stages, eg are they well balanced sine waves?

[four_corners]

Thanks again for your help on this!

The first image is of the V8 side of C34 and C37 (I just probed where R49/R50 and R52/R53 meet, respectively).

The second image is of the V9 side of C34 and C37, which is essentially probing the plate of triode A, and the cathode of triode A after a 1.5k resistor.

They don't have exactly the same amplitude, which I'm assumed just comes down to C34 and C37 tolerances being slightly off, but is that enough to significantly distort the signal as much as it did in my original scope images?

(again, sorry for the scale mismatches)

[four_corners]

V9 pins6,7,8 is a cathodyne, providing opposing phase modulation signals to V8 grids.
What are the waveforms like at the coupling caps C34&37 between those stages, eg are they well balanced sine waves?

Meant to click quote on my last message, oops.

[Merlin]

It looks to me like you are applying too much bias modulation to V8, producing a glitch when one of its triodes goes beyond cut-off. Does the signal clean up if you turn down the depth pot? I suspect all you need to do is adjust the end-stop resistors to limit the depth range a bit.

They don't have exactly the same amplitude,

They do have the same amplitude. In a PI the things to compare are the opposing peaks (mirror image), not one peak with the following peak.

[four_corners]

It looks to me like you are applying too much bias modulation to V8, producing a glitch when one of its triodes goes beyond cut-off. Does the signal clean up if you turn down the depth pot? I suspect all you need to do is adjust the end-stop resistors to limit the depth range a bit.

They clean up somewhat, but still aren't very pretty all the way down.

Here are some images of 0% depth, 25% depth, 50% depth, 75% depth.

Since I don't have another AC30 to compare mine to, I found this video that has it dialed in pretty well, and it just sounds a lot better and way more "woozy" vibrato than mine does. I guess I could plug into my reactive load and record a bit of audio to post.

[Merlin]

Hmm curious. I would be tempted to increase R51 (the shared cathode resistor). When I built my vibrato I found that a colder bias in the mixer stage, with a bypass capacitor, gave a thicker effect. I know that deviates from the original AC30, but perhaps worth a try?

[four_corners]

Hmm curious. I would be tempted to increase R51 (the shared cathode resistor). When I built my vibrato I found that a colder bias in the mixer stage, with a bypass capacitor, gave a thicker effect. I know that deviates from the original AC30, but perhaps worth a try?

Worth a try, as both the normal and flipped signal are distorting on the same side it seems like it would be something right around there. I’d assume if the distortion was mirrored in the negative swing then it might be upstream around the LFO valve.

[four_corners]

Hmm curious. I would be tempted to increase R51 (the shared cathode resistor). When I built my vibrato I found that a colder bias in the mixer stage, with a bypass capacitor, gave a thicker effect. I know that deviates from the original AC30, but perhaps worth a try?

increasing R51 ended up making the distortion worse unfortunately. When I reduce R51, the distorted + side starts to smooth out some, but the negative side starts getting sharp and lopsided, more like a sawtooth wave.

I lifted R49/R52, and R50/R53, and tried different values in those spots (which was a bit of a pain) but it was a bit like whack-a-mole, changing one set made one side better but the other worse, and changing the other set would fix the previous thing but make the original thing worse! Obviously it is a delicate ratio.


Lastly, I looked back at the scope images for C34 and C37, and on the V8 side, the waveforms actually aren't mirrored, so this most likely comes down to those 0.1uF cap tolerances. If you check the image named C34-C37_V8.jpg you will see what i mean. Is that enough to cause such distortion?

[four_corners]

Sorry, I know it looks like I'm just talking to myself after posting 3-4 times in a row, haha.

Any time I was able to figure out a way to smooth the positive side of the waveforms a bit, the scope looked better, but the vibrato sounded worse actually.

The more and more I listen to the actual sound of the vibrato, the more I think this might actual be correct?

Looking at the schematic a bit more, it seems I've been ignoring this RC network that feeds into the grid of the mod valve.

The two RC networks in the screenshot I've just posted are obviously filtering the signal for each grid, I'm guessing one more so on the low frequencies, the other on high frequencies. Maybe I just need to tweak C22 (200pf), C23 (100pf), and then C24 (750pF), C25 (500pF). Might make sense to inject some audio into the Vib/Trem Input jack, and tweak these values to taste.

[Merlin]

When I reduce R51, the distorted + side starts to smooth out some, but the negative side starts getting sharp and lopsided, more like a sawtooth wave.

Could there be an incorrect resistor value somewhere around V9? Especially R40, R41, R42, R43, R44

I looked back at the scope images for C34 and C37, and on the V8 side, the waveforms actually aren't mirrored

They look close enough to me.

Looking at the schematic a bit more, it seems I've been ignoring this RC network that feeds into the grid of the mod valve.

That shouldn't matter. You're applying a sine wave from V8, so the outsome should still be a fair sine wave, regardless of that network. But you could lift C20 & C21 to be sure.

I just noticed a dodgy bit on your schematic that makes it look like C19 is in parallel with R40, when C19 should be grounded. I assume you built it the correct way?

[four_corners]

I just noticed a dodgy bit on your schematic that makes it look like C19 is in parallel with R40, when C19 should be grounded. I assume you built it the correct way?

Yeah I have C19 grounded, a little confusing in the cropped schematic shot I sent, but C19 is indeed grounded, as right above your red circle, that connection to R36 is a ground bus.

Could there be an incorrect resistor value somewhere around V9? Especially R40, R41, R42, R43, R44

Yesterday I put a 50k trimmer in parallel with R41 (47), and lowering that resistance, almost to zero significantly smoothed the waveform. It wasn't perfect, but much smoother. This is what I was saying when it sounded like there was barely any depth to it, almost like while making the LFO smooth out also cut out a lot of frequencies to make both sides too similar?

Maybe it is a matter of smoothing out the waveform at R41, and then increasing the caps at C24/C25, and lowering C22/C23 (i think?) to make the filtering more extreme?

[four_corners]

Could there be an incorrect resistor value somewhere around V9? Especially R40, R41, R42, R43, R44

Okay! Got some audio and new scope shots:

First scope image is before any changes (aka, yesterday amp state).

Second scope image I swapped R41 from 47K to a 15K resistor. Not perfect but much much better looking.

[four_corners]

Now onto the audio. All clips follow the same format. They all start with the depth all the way down on Vibe, slowly turn the depth all the way up, then switch to Trem, and turn depth all the way back down. I never mess with the speed.

Here are 4 clips:

http://sndup.net/c57sx
Original.mp3 - this is how the amp was up until yesterday (first scope image).

http://sndup.net/z625d
just_15k.mp3 - this is with R41 swapped (second scope image). Basically doesn't end up having the positive swing fold back over and hit the 0V line, so it never falls apart. Sort of just reduces the depth I guess?

http://sndup.net/3w7n7
just_1nF.mp3 - this is with R41 still at 47K, but swapped C25 from a 500pF cap to a 1nF cap. A lot more liquidy sounding! Since it still has the 47K cap at R41, it gets messy at full depth.

https://sndup.net/gwv94/
All_mods.mp3 - this has both the 15K at R41, and the 1nF at C25. This sounds the best to me, and the waveform is the same as the second scope image.

The original is much brighter, and all the mods have some of the tone sucked out, but I'm assuming this is fine, as this isn't just an LFO strictly modulating the pitch, you
have to have the filtering for it to sound vibe-y, right?

I know it isn't to the original specs, but I'm curious what your thoughts are when you hear these 4 clips?