Another approach to the Marshall style tremolo

[cdemike]

I came back to a single-channel Lead style project that I've posted about before. A previous iteration of this amp was an all-access style build which used a parallel triode input stage which later I ditched once I started finding I only really used a handful of settings. I also picked up a particularly nice NOS UK-made RCA 12AX7 that led me to try to place as much of the signal chain as I could through that tube. Reassigning the roles of the preamp tubes to have both the first and second gain stages going through the RCA left a spare triode open at V2, and I had been wanting to add tremolo to the amp anyway.

My initial effort aimed to make a preamp bias wiggle tremolo, which didn't work out since a single triode was loaded down trying to wiggle bias at the 1k cathode resistor at the second gain stage (https://ampgarage.com/forum/viewtopic.p ... f8e54780ea). So it seemed like the best next steps would be to try 1 of 3 options: a sharing the cathode of the LFO and the gain stage getting wiggled like in the Supro 1690t, use a source follower in a Vibrochamp or Supro 6424 style arrangement, or shunt signal like in Sluckey's Trem-o-nator or in Super Tremolo style circuit. I didn't want to change the operating point of the second gain stage since I like the present voicing of the amp, ruling out the first two options. Since I already had two spare BJTs in my spare parts drawer from a previous fuzz pedal project, I thought I'd try something like a Super Tremolo circuit. The two BJTs are a 2N2222A and a 2N3906. I'm less knowledgeable about solid state design than I'd like, but a previous threads here and elsewhere about people trying to replace aged-out germanium transistors in original Super Tremolos helped guide me. As I understand it, it shouldn't matter whether the transistor is PNP or NPN since the LFO signal has both positive and negative components, and the LFO signal reaching the base should theoretically be a sine wave centered around ground and therefore have both positive and negative components (assuming minimal distortion at the LFO). After reviewing the datasheets, I decided to give the 2N3906 a try since it slower switching time should theoretically sound a little gentler.

To my surprise, the circuit worked right away using similar values connecting the LFO's plate to the base. I originally used a 100nF capacitor and 750k resistor in series to the base, where a 10k resistor ran to ground. I omitted the 10k, 33k, and 100nF network to ground to keep the amp sounding as close to unchanged without the tremolo engaged as well as to maximize the potential depth of the tremolo. While this setup worked, there was a very audible thumping even with the volume at zero which I gathered to be a noise coming from a combination of harsh operation of the transistor and signal cross-talk with the signal chain. Lead dress adjustment helped some, but there was still low-frequency thumping, especially at higher tremolo speeds even with the volume control at zero. Seeking to minimize disruptions at the B+ node sneaking into the signal chain, I lowered my LFO's plate resistor from 470k to 270k and revised my oscillator network values to allow more loss of LFO signal. This also helped, but only marginally, so the transistor's operating conditions seemed like a reasonable place to turn next. Tuning the resistor and capacitor values linking the LFO's plate to the 2N3906's base helped, especially finding the right values to only just activate the transistor as a switch, though there was still an unacceptable amount of tremolo thump. Drawing inspiration from Sluckey's diode fix for thumping 6G2-style tremolo, I placed a diode allowing current to flow from ground to the base, which finally reduced the thumping to acceptable levels. It's still audible, especially with the tremolo depth turned up to maximum, but it gets about as loud as conversation level with the amp dimed (not likely audible in a mix). Adding a capacitor between the base and ground helped tune the shape of the tremolo, smoothing it out to something closer resembling a sine wave.

In addition to the noise, I'm currently having a strange issue where I'm struggling to nail down footswitch arrangement. I have tried something similar to the usual Fender SPST to ground in after the second capacitor, which rather than muting the tremolo made it oscillate extremely quickly (sounded like a woodpecker -- about 50hz). I'm currently working out an arrangement where a footswitch grounds the transistor's base, which I predict should work. I'll report back when the switch arrangement gets better fleshed out, but I wanted to post since I'm mostly happy with how this is working out.

Is there anything else I could try to reduce the tremolo's noise? I'm on the lookout for a scope and signal generator which I anticipate would help identify strangeness in the LFO and switching circuit, but haven't found any used ones come up locally quite yet.


As-built with Fender-style switching arrangement:

Signal shunt trem.PNG

Original Super Tremolo schematic:
https://el34world.com/charts/Schematics ... _1959t.pdf

[cdemike]

Got the footswitch working using the approach mentioned in the original post (grounding the base). Strangely, the footswitch kills the tremolo sometimes when plugged in at low speeds, which I think will require chasing down some fault in the shunt-to-ground scheme. I'm not quite satisfied with the available depth and noise associated with the tremolo, but I also suspect both of these might come with the territory in this topology. With the tremolo shunting signal in advance of the cathode follower, I suspect the depth is being limited by the compression intrinsic to the DC coupling in the cathode follower. Similarly, having a potential noise source relatively early in the signal chain would seem to be a recipe for noisy tremolo operation. With that combination of factors, I decided to try moving the tremolo insertion point to the treble pot's wiper so it'd work similarly to a JCM800's master volume. Didn't love the result, but the noise issue was improved.

Overall, I'd say this is a success for anyone trying to implement a Marshall-style tremolo, though to be honest I don't really see the utility over going with an vactrol-based solution (read: Trem-O-Nator) outside of the marginal cost difference. I'll be ordering a VTL150 with my next parts order.

[Reeltarded]

The 1959/T models normally sound like a raygun tatatatata pretty fast, like vox fast speed. chop chop chop..

A strong tube (stops at low speed) to anyone not opening it up to tweak around makes tgem work, although I have never opened one up to fix the trem.

Fun though!