Tremolo XVIII - how does this work?

[soggybag]

I just built Tremolo XVIII from the Stompboxology newsletter. An interesting design. The audio path is built from two inductors and to JFETs.

I have a good idea of how the control path works but the audio path is very confusing. Can anyone shed some light on this?

It sounds pretty good but has a couple issues.

- The quiet part of the tremolo sweep doesn't seem to get fully silent. Or at least that's what sounds like.
- There is some distortion. A crisp edge to the sound. The notes mention "Both transformers have been left unloaded to facilitate increased gain and ringing."  I'm not sure if this is what I'm hearing or if the JFETs are being overdriven.


Here is the original schematic. I found this hard to follow.


This is my schematic drawn in Eagle.








This is a build post on my blog: http://www.super-freq.com/tremolo-matic-xviii-build/

[Mark Hammer]

Hmmm, I don't have the rest of the documentation for this, from either thew scanned book, or the newsletters that Moosapotamus posted.  Is this somehow attempting to mimic bias tremolo by use of the transformers?

[idy]

Yes, mimicking tube bias trem. It's in the newsletter issue about "going discrete."

[soggybag]

Correct, this is from the going discreet newsletter page 8. It is described as based on a tube tremolo. I'm guessing the JFets replace a pair of triodes.

Here's a link to the newsletter: http://moosapotamus.net/files/stompboxology-going-discrete.pdf

For a discreet transistor project it's only got two transistors!

It does work, it's quiet. The tone is good.

The problem is the depth of the tremolo effect. It's not as deep as I feel it should be.

[PRR]

...I'm guessing the JFets replace a pair of triodes....

Could be, but it is more in the line of "a pair of 6V6". A JFET is a pentode except G2G3 are internal and internally "powered" (by doping).

It can't go "to zero" until the top of R7 is driven ALL the way to ground. FWIW, the tube trems also didn't go true-zero; they'd thump.

Are you using the specified opamps and JFETs? Since it all has to go way-close to zero, the opamp may be critical and the JFET Vto may matter.

> For a discreet transistor project it's only got two transistors!

Yes but that and the two transformers is the ENTIRE audio path. No opamps to your ears, only as control signals.

EDIT: no, the gate drive has to go BELOW zero. And for 2N5457, maybe several volts to get "near-silence".

[soggybag]

...I'm guessing the JFets replace a pair of triodes....

Could be, but it is more in the line of "a pair of 6V6". A JFET is a pentode except G2G3 are internal and internally "powered" (by doping).

It can't go "to zero" until the top of R7 is driven ALL the way to ground. FWIW, the tube trems also didn't go true-zero; they'd thump.

Are you using the specified opamps and JFETs? Since it all has to go way-close to zero, the opamp may be critical and the JFET Vto may matter.

> For a discreet transistor project it's only got two transistors!

Yes but that and the two transformers is the ENTIRE audio path. No opamps to your ears, only as control signals.

EDIT: no, the gate drive has to go BELOW zero. And for 2N5457, maybe several volts to get "near-silence".

Thanks for the reply. I am using all of the parts specified in the original project. TL064 and 2N5457.

[amptramp]

This appears to be as much of a ring modulator as a tremolo.  This is how I would design a singly-balanced mixer for radio frequencies.  As long as the FET's are balanced and operated in Class A (FET's never driven to cutoff), it should work without introducing much distortion.

[PRR]

> This is how I would design a singly-balanced mixer for radio frequencies.

It's also the core of every good tube recording limiter. And most of Fender's middle "Tremolos" (after the Harmonic patent, before opto-resistor).

"Offness" does depend on how far off you drive the devices. On tubes and FETs, "cut-off" is relative; you can never really cut them off, just starve them.

[soggybag]

How does this work?

My amateur eyes see the control voltage going into the center tap secondary of T1. The input signal appears at the T1 pin 6 and inverted at pin 4. These signals are offset by the control voltage at the center tap.

The two outputs go to the Drain of Q1 and the source of Q2.

And... then something unexplainable is going on where the Source of Q1 and Drain of Q2 do some mysterious voo doo magic at the primary of T2.

I'm guessing there's something going on where the output phase signals cancel each other and somehow the control signal makes this happens?

Or... the offset from the control voltage and somehow that "turns on" and "turns off" (allows the these to conduct?) Q1 and Q2.

I could come up with some "theories" but I'm just making this up.

[Rob Strand]

The basic idea:
- The LFO output appears at IC1a pin 1.
- Think of the LFO output as slowly varying "DC" voltage.
- The LFO voltage applied to the center-tap of transformer T1
   and appears on the gates of Q1 and Q2.
- The varying gate voltage changes the drain bias current of Q1 and Q2.
- When the drain current varies it varies the AC gain of the JFET.
   (This is the called the transconductance, transconductance is proportional
    to the square root of the drain current.)
-  The net effect is the LFO modulates the gain and produces a tremolo.

More details:
- The main idea of using the transformer T2 is to prevent thumps feeding through
   into the audio.  The transformer has a low DC resistance and prevents large "DC" shifts
   in the drain voltage as the JFET drain current is varied.  JFETs are inherently poorly matched
   so the drain currents don't track - feeding a low DC resistance hides the imbalance.
   Another bonus the transformer doesn't pass low frequencies from the LFO very well.
- The transformer T1 allows the bias voltage the gates without affecting the AC signal.
  The center-tap allows ac input signal to be applied to the gates differentially.

[PRR]

The two outputs go to the Drain of Q1 and the source of Q2.

Are we looking at the same page?

[Rob Strand]

- The quiet part of the tremolo sweep doesn't seem to get fully silent. Or at least that's what sounds like.

Not much you can do about that the JFETs would need to get to a low bias voltage on the source.
A higher source resistor might help but that will go against when the JFETs are on.

Another trick would be to take a resistor from where the sources connect to the + rail.  Ideally it should go to a nice clean + rail.

If you consider tweaking the DC offset from the LFO, tweaking the value of R7 and tweaking the resistor to the +rail you have a lot of options.  The trick is to find the best set of values for your JFETs.

- There is some distortion. A crisp edge to the sound. The notes mention "Both transformers have been left unloaded to facilitate increased gain and ringing."  I'm not sure if this is what I'm hearing or if the JFETs are being overdriven.

If it's overly crisp you might be seeing some response peaking from the transformers.   Try loading the 600 ohm winding on T2 with 1k to 100k.   The load here will also tune the overall gain.  If there's not enough load the JFETs could clip.   The right load will also depend on the bias current as mentioned above.   

Another loading point is the output of T1, you can also try putting a large resistor across, say 220k,  between the gates - I suspect that's going to very subtle.

Normally you would measure the frequency response to get an idea how flat it is and what the overall gain is.

I haven't analysed to circuit to narrow down any of the part choices.

[soggybag]

Thanks Rob, that was a great explanation. It's starting to make sense to me.