Weak bias wiggle trem

[cdemike]

Just done with preliminary testing on a series of mods for a Marshall-style build that involved adding in a standalone board for bias wiggle tremolo connected the bias for the cathode of the second preamp gain stage. The signal path is basically the same as any late 60s 4-hole Marshall, and the LFO's speed circuit is borrowed from the Supro-style version of Sluckey's Trem-o-Nator. My trouble is that I'm finding that the bias wiggle is really weak. I based the initial anode-driven circuit off the 6G2 Princeton setup with the LFO output coming from the anode via a resistor equal to the anode load resistor and a 100nF capacitor (https://schematicheaven.net/fenderamps/ ... _schem.pdf). This setup produced no audible tremolo, which I thought may be due to the unfavorable voltage divider ratio formed by the resistor on the LFO output (R8 below) and V2's cathode resistor, which led me to try decreasing values down to 3k3, at which point adjusting the intensity control would stop the LFO altogether.

Bias wiggle anode.PNG

After thinking over alternatives, I figured that the relatively high output impedance would be a significant limitation to the tremolo's effective output, so I looked through other single-triode bias wiggle circuits (as opposed to circuits like the Vibrocamp, which require an additional triode which I don't have in this amp). I noticed that circuits like the Airline Ward 9058 (https://el34world.com/charts/Schematics ... _9058a.pdf) were able to retain the cathode bypass capacitor on the cathode getting wiggled, and with the bypass corner frequency being well above the ~20hz maximum frequency from the LFO, I figured the stock 680nF cap shouldn't represent an AC path to ground for the LFO. This line of thinking led me to try another version where I substituted the old circuit's ground with the input wiper of the trem's depth control. Unfortunately, this also didn't produce any audible tremolo.

Bias wiggle cathode.PNG

Lastly, I tried to bypass the diode altogether to try to get maximum signal strength at the intensity control, but again without success.

Bias wiggle cathode no diode.PNG

After looking over the Airline Ward schematics again, I see that they split the cathode and injected the LFO signal partway down the total cathode's resistance to ground, while the cathode bypass cap of the gain stage attaches to the top of the stacked resistors. However, if this were the solution, I'd expect that I'd get some signal mid-way up the intensity control's travel, but this was not the case.

I was able to confirm the LFO is working both visually via the LED biasing the LFO and with my multimeter which showed 2.7VAC signal swing, which I'd anticipate should be more than enough to wiggle the 2V at that gain stage's cathode.

Can anyone help me figure out what's going on here?

[cdemike]

Whoops, forgot the attachments. Edited above.

[LOUDthud]

In the last two schematics you posted it looks like the wiper of the intensity pot is just grounded. Is that really what you did ? Please confirm.

[cdemike]

In the last two schematics you posted it looks like the wiper of the intensity pot is just grounded. Is that really what you did ? Please confirm.

Sorry for the confusion — the Spice potentiometer symbol isn’t the most elegant. The wiper is not grounded without the foot switch disengaging the term (planned to just use a typical Fender style SPST). The intensity pot is 1K linear standing in for the cathode resistor for the second gain stage with a 680nF bypass capacitor across it. The wiper is not grounded when the foot switch isn’t plugged in, but to your point I didn’t put together than the foot switch shorts Rk to ground when disengaging the tremolo. I might put a large resistor in parallel with a capacitor to ground across the foot switch jack to avoid audible bias shift when using the foot switch.

But right now there isn’t a foot switch and I can see the LFO working via the LED, and the amp sounds like I’d expect with the 1K cathode resistor and 680nF bypass capacitor, so there’s not an accidental ground impacting it as far as I can tell from that perspective.

[Helmholtz]

The plate output of the LFO works with high impedance circuits like power tube grids.
Feeding the plate output to a low impedance cathode circuit will cause the LFO signal to collapse.
The cathode of the LFO cannot be used as output when using the LED, because the LED has low AC impedance, not allowing for a significant signal there.
OTOH, the LED is used to generate the necessary cathode bias voltage of around 3V.
You might try something like the Wards circuit with a partly bypassed cathode resistor, but it will be tricky to adapt and probably not very effective.
Best would be using a cathode follower (or FET source follower) after the LFO.

[lonote]

Looking at a VibroChamp schematic (& similar on the Airline), the tube being wiggled has a dedicated cathode resistor to establish the bias, then the LFO signal is applied to the cathode side of that resistor. Looks to me like you have the pot itself establishing the bias.

[cdemike]

Redid the switching arrangement. Thanks for pointing that out, LOUDthud.

The plate output of the LFO works with high impedance circuits like power tube grids.
Feeding the plate output to a low impedance cathode circuit will cause the LFO signal to collapse.

This makes sense to me. Seems like the intensity control would work basically like a very low impedance speed control in parallel to the actual speed control.

The cathode of the LFO cannot be used as output when using the LED, because the LED has low AC impedance, not allowing for a significant signal there.
OTOH, the LED is used to generate the necessary cathode bias voltage of around 3V.
You might try something like the Wards circuit with a partly bypassed cathode resistor, but it will be tricky to adapt and probably not very effective.
Best would be using a cathode follower (or FET source follower) after the LFO.

Looking at a VibroChamp schematic (& similar on the Airline), the tube being wiggled has a dedicated cathode resistor to establish the bias, then the LFO signal is applied to the cathode side of that resistor. Looks to me like you have the pot itself establishing the bias.

Certainly open to adding a source follower (no spare triode with the current setup), but I guess I'm still unsure why feeding the ground of the LED to the wiper of the intensity control has a nearly-inaudible effect. The pot itself is establishing the bias, so it seems to me that inserting the LFO signal on the wiper should have a similar effect as the Vibrochamp's setup. If anything, the higher resistance to ground from the LFO relative to the Vibrochamp should increase the LFO signal magnitude, right? Does the difference in function between my circuit and the Vibrochamp again come down to having a low impedance buffer in the cathode follower? Or maybe it has something to do with the cathode bypass cap not also connecting where the LFO signal enters as in the Vibrochamp?

[Helmholtz]

You're forcing the LFO cathode current to run through the intensity pot.
This lowers the gain of the LFO triode and varies the bias of both V4A and V5A.
Availabe LFO signal is only 1k/270k of the plate signal.

The cathode follower in the VibroChamp is fed from the LFO plate, so has lots of output signal.

[lonote]

I built a MOSFET version of a typical LFO which worked well & tried various ways to tie it in, but never really followed through to any acceptable design. I wish I had taken some notes, which I didn't but, I was in a similar place to where you are & was coming to the conclusion that there just wasn't enough LFO signal to work with without a second stage follower/driver.

You are awfully close to building a Sluckey Trem-O-Nator, which would manipulate the volume pot feed to the grid of your V5a.

[cdemike]

Thanks for helping me figure this out! It's becoming clearer to me that I might need to change directions with this; seems like bias modulation probably isn't the way forward with this circuit unless I wiggle the output tubes, and I'm hesitant to do that on a cost basis since this amp runs KT88s; one of my design goals was to get the most out of them both sound wise and in terms of longevity. I tried connecting this to either end of the tail resistor at the LTPI, which also didn't work, so I'm considering just going full-blown Trem-O-Nator. My only hesitation is that it'd require a major rework over again to make room for the optocoupler, and I already put in a lot of time in terms of getting noise down viz lead dress. I do have enough board space right now to try out a similar concept in using a transistor as a switch like in the original Marshall tremolo circuits (for reference this is a JTM45/100T schematic: https://www.drtube.com/schematics/marshall/1959t-66.gif). I have a 2N2222A and a 2N3906 on hand, so I'll see how those work out shunting the signal after the tone stack (i.e., shunting signal the same way as a JCM800-style master volume).

I'll report back on how it goes, but I have been reading about others who have tried similar circuits that the transistor makes a noticeable difference. However, I haven't been able to find anything on what specifically to look for in a transistor's datasheet. I'm hoping to keep this tremolo sounding smooth, which is why I originally wanted to try bias wiggle trem, so I gather that slower switching would make some difference. Is that correct? Is there anything else to look out for?

[LOUDthud]

You don't see this often, but you could feed the LFO into the phase inverter. Either right into the cathodes or into the tail somehow. Feed it in a way that balances both phases so the LFO cancels in the OT.

[lonote]

I have only mid-level tube knowledge & pretty much zero regarding SS. The appealing thing about the MOSFETs was that they were kind of a plug & play replacement with a tube, relying on similar outboard components/functions.

I think that this drawing below (not mine, but from the smart guys at Hoffman) was pretty close to where I was when I bailed on my experimenting.

I was using a little reverb Champ build as the guinea pig (AB763 pre, 2-tube reverb, SE 6V6) & if I remember, I was tapping the Trem-O-Nator in on either side of the 3.3M/10P combo between the 2nd & 3rd gain stages (same topography as a Deluxe at that point in the circuit). I think I also split the 3.3M using two series resistors to try an insertion point that was slightly attenuated but didn't screw up the reverb.

Wish I could share/remember more.