(Diode) Bias Tremolo

[samhay]

Something new-ish -- A tremolo based on the DiSCO circuit here:
http://www.diystompboxes.com/smfforum/index.php?topic=110346.0

The only major change was to replace the envelope detector with an LFO. I chose a sine(ish) wave LFO obtaining using a phase-shift oscillator (PSO).

The idea is to capture some of the character of the old valve (tube) amp tremolos that work by modulating the bias of the power valves - tremolo is achieved by driving the valves towards cut-off, which gives rise to some clipping and crossover distortion when the signal amplitude is decreased.

The circuit here uses the PSO to wiggle the forward bias voltage of a pair of diodes (D1, D2). Signal goes through an input buffer (IC1A), is phase-split (IC2A) prior to the diodes and then recombined using a difference amp (IC2B) in an analogous manner to a power amp using an output transformer. The diodes act as directional variable resistors, with each of the signals 'clipped' asymmetrically to (virtual) ground. Much (but not all) of the distortion is removed when the inverted signals are recombined. The clipping gives rise to a reduced signal amplitude - the tremolo effect.

The tremolo depth is controlled by varying the 'bias' voltage of the PSO - the voltage the LFO swings around. This can be varied from ~ 6 V (minimal depth) to ~ 3V (max depth). The swing of the LFO is clamped by D3 and D4 to ~3.5 V peak-to-peak, which gives a good depth of tremolo at higher Bias/depth settings and largely keeps the LFO from hitting the power rails, which probably helps keep the circuit polite - it is fairly quiet for a tremolo.

The schematic:


('view' for larger version)

I built it on vero using this layout, which should be a fairly comfortable fit in a 1590B – that's what I used (the 10u capacitor should be moved 1 or 2 places to the right).


('view' for larger version)

I will try to get a sound clip recorded this weekend.
Enjoy...

[BetterOffShred]

Sounds awesome man, I'm looking forward to some sound clips.   What kind of swells does this get?  And what does it sound like with the LFO toggled off, just a boost?  I know that's a popular mod with the EA tremolo..    Thanks for sharing!
-Brett

[samhay]

Thanks,
The LFO is more sine than triangle, so sounds somewhat like the EA tremolo when the Bias/depth is not maxed. At max depth it gets pretty choppy.
It has a little gain, ~ 5dB, which is clean when the Bias is at minimum (fully ccw).

As drawn the switching is optional. It kills the LFO when the effect is off, which can be a good precaution if the LFO is noisy. As it is, there is no noise with the Bias at minimum, so you could probably leave it running without ill effect.

The only noise comes from the audio path, which may have some LFO bleed-through - I can't see it on the scope, but there is something going on. It is only really noticeable with the Bias turned up and a noisy fuzz (or similar) in front. However, most tremolos suffer under these conditions, so probably not something to loose sleep over.

[BetterOffShred]

Sounds great.   I may build one.. I've been on a tremolo kick.. built a 7 knob Tremulus Lune, EA, a Skippy, A green Currant..   The skippy ticks pretty bad though. 

So the switch you show is part of the bypass switch then?   I guess I'm confused by that part.   

[samhay]

I wired mine true bypass using a 3PDT. The switch wiring shown was on one of the poles.
You could use a 1PDT (if you you can find one, or half of a DPDT) if you don't want true bypass. It might get a bit crunchy when 'off' if the Bias/depth is set near max though.

[BetterOffShred]

Nah I get it, it's just one of the poles of the stomp.   Thanks man.   It looks totally awesome, I'll probably give it a whirl in a couple weeks.   Thanks again!

[Kipper4]

That's neat.
Can't wait to hear it.
Thanks Sam.

[duck_arse]

I like C4. [not the explosive.]

[samhay]

^right, because one doesn't want an explosive C4.
I expect an electrolytic could be used there in either orientation without any excitement, but sometimes belt and braces is not that difficult to achieve.

[samhay]

As promised, here's a sound clip. Recorded directly with a little reverb; apologies for the uninspirational playing.
https://www.dropbox.com/s/aao2o02xkosseke/Bias_Tremolo_1.mp3?dl=0

At faster LFO settings you might be able to hear a little thumping. Looking at the waveform, it is clear that there is some bleed-through from the LFO into the audio path. It isn't crap on the supply, so I think I can probably dial more of it out - For what it's worth, I'm told this is/was a classic problem with the original bias tremolo designs too.

Stay tuned for a revision...

[Kipper4]

I can't hear the thumping. Sounds nice and creamy and mellow. Sweet.

[bool]

Perhaps you could just insert 2 caps at the diff-amp to act as a crude HPF. 30/40Hz? Say 220||15nF / 470nF (235/470nF)?

Maybe you could get by just using 220/470nF? Perhaps such unbalancing the diffamp would result in a more pronounced thump? Hard to tell without further investigation.

[samhay]

>Sounds nice and creamy and mellow. Sweet.
Thanks

>Perhaps you could just insert 2 caps at the diff-amp to act as a crude HPF.
While bread-boarding I originally had caps in series with R14 and R15 to do just this. I didn't seem to to help.
I think a trimmer in place of R15/R18 is worth a try, but that will involve a layout rethink.

[deadastronaut]

cool project sam,  i could hear the thumps though, especially when you sped up......(headphones on)

nice and warm sounding..nice.

[PRR]

If D1 D2 (and most of that network) could be perfectly matched, it would not thump.

In today's economic, a bag of 2k2 and 22k 1% parts is expedient. Diode matching is more tedious; look for dual-diode parts. A single BJT won't have B-E B-C match, but a matched dual transistor may.

Of course if you get much fancier, you might just as well throw in a THAT VCA and get impeccable thump.

[Rob Strand]

Diode matching is more tedious; look for dual-diode parts.

You can put say 5 or 10 diodes in parallel and let statistics pluck-out the lowest Vd.***
It's no match for dual diodes or transistors.

Matched parts is the life blood of all good VCAs and (linear) Mixers.  No one has come-up with practical solutions to avoid it.  Well apart from DSP - haha!

*** I should add this is only recommended for diodes off the same reel.  When picking parts out of your junk bin it's likely to make things worse (because it finds odd parts).  Series diodes will actually help the average better than parallel.

[samhay]

Thanks both.
Yes, it looks like one can't get away from pulling diodes randomly out of the parts bin if you want perfect nulling. It does work quite reasonably if you don't turn the rate and depth too far up, so I am not loosing sleep over this.
Will make a few tweaks and do a little diode auditioning to see if I can get it a bit quieter though - I know I can make it a lot worse, so reason stands I can make some improvement too.

[duck_arse]

maybe, after all these years, a use for my LM194's. hurrah!

[samhay]

I spent a good few hours tweaking this over the weekend. I trimmed the various resistors around D1, D2 and matched these diodes - found 2 on a reel that measured the same forward voltage drop to within a mV.
What did I learn? This is probably all quite unnecessary.
I had an error in my vero layout - now fixed (below). The pull-down resistor on the input was the wrong side of the input cap, so there was a sizable DC offset between the outputs of IC1A and IC2A. This made nulling the LFO quite impossible.
It won't hurt to use 1% resistors and diodes off the same reel to minimise noise, but I suspect you can get quite a usable circuit without going to great lengths to match components - sorry Stephen.
It will be important to use an op-amp with FET inputs at IC1- don't use an NE5532!

As I had the scope out, here a capture showing the LFO (500 mV/division) and the output with no signal (20 mV/division). I think the time axis is 20 ms/division. In any case, the LFO bleed-through seems to be barely above the noise level. You can still hear something when the depth and/or speed are high, but it compares favourably to the (many) other DIY tremolos I have played with.



Here's a new clip:
https://www.dropbox.com/s/ud1vmv66op9d6gi/Bias_Tremolo_2.mp3?dl=0

Fixed/updated layout:

('view' for larger version)

[PRR]

You can in principle add a steep low-cut after the thumper to take the sub-sonic off.

I don't see how to do that well without more chips.