Anharmonic Tremolo - yes another harmonic tremolo, but with op-amps

[samhay]

It seems that, largely thanks to Jimi Photon, we now have a rash of new Harmonic Tremolo designs (well two). Jon (midwayfair) beat me to it, but here is another approach to getting something that sounds (something)like the old Fender harmonic tremolo. As I am late to the table, I bring an added bonus - you can vary the crossover frequency of this design. It is still on the breadboard - and has been for some time now - and I need a push to get it off the bread board and boxed up. It should fit in a 1590B, but it will be fairly snug. At the moment, it has four pots (rate, depth, crossover frequency and level) plus two switches (treble/harmonic tremolo and squarewave/sinewave LFO). These are not cast in stone.

How it works: U1b and U2a form a fairly typical crossover circuit that splits our signal into bass and treble components. We then pan between them using a quick-and-dirty LDR-based voltage divider, which is modulated using an LFO-driven LED (or a vactrol).

Notes:
- U1b is an almost-Butterworth 2-pole Sallen-Key LPF and the 'frequency' dual-gang pot can vary the frequency of this filter from ~100Hz to ~10kHz. Whether this range is entirely useful is debatable, but I like it. Note that because I can't find a C100k dual-gang pot, this works backwards - clockwise = lower frequency.
- U2a is a differential amplifier that creates our HP (treble) signal by subtracting the outuput of U1b (i.e. the bass) from out clean amplified signal from U1a. This means we can easily control the crossover frequency by only having to wory about varying the LPF corner frequency. This is a compromise as the HPF is not as steep as the LPF, but it keeps us out of voltage-controlled OTA territory.
- The 'treble/harmonic' switch essentially determines the gain of the HPF, which in turn determines whether we have tremolo'ing of both the bass and treble (harmonic) or only the trebble. I think Jon's design does something similar and it is a nice way to add a little shimmer.
- We pan between the LP and HP signal using a voltage divider. One half of the voltage divider is a LDR, which is modulated with an LED being driven by an LFO. This method is discussed a little bit in another thread (http://www.diystompboxes.com/smfforum/index.php?topic=102414.0) and was chosen over a more complicated OTA-based solution. Basically, it is a low parts-count way of doing the job. It isn't perfect, and you do not get a particulalry clean crossover frequency, but it works pretty well in this application. The trimmer is used to set the resistance of the other half of the voltage divider to be similar to that of the LDR when the LFO is a min 'depth'. Note that when you mix the LF and HP signals equally, you get an almost-flat response.
- the LFO (U3a) is the standard Phase 45 LFO design, which is used almost everywhere you want to wobble something. I have used slightly different component values here because I have a lot of 100k resistors and because I'm contrary. If you want more control of the LFO, the 'Tremulus Lune' LFO (among others) is a drop-in replacement.
- U3b buffers the LFO and acts as our depth control - it has gain from 0 to 4.5. I used this approach as, for the LDR to work as a voltage divider, we want it to have an intermediate resitance at minimum depth. To achieve this, we want, at minimum depth, the LED pointed at the LDR to be as bright as it is in the middle of the LFO swing at max depth. I gave U3b gain so that at high gain the LFO is clipped by the op-amp (U3b) rails - the LFO gets progressively more square as you turn up the 'depth'. I like this, but have also added the option of a switch to use a 'proper' square wave LFO, which sounds pretty cool too.
- As seems to be my way, the 4 audio op-amps are all DC-coupled. I have tried uncoupling them and can not hear any good reason to do so.
- I have tried using both a home-rolled LED/LDR combo and a Silonex NSL 32SR3 'vactrol'. Both work equally well and you can get a pretty deep tremolo out of this design.
- The power stage is a little more elaborate than usual as I really don't like LFO noise. The LFO and audio sections are isolated from each other and the LFO ground should be kept away from the audio ground as much as possible too. On the breadboard, I can still hear a little LFO noise, but I think this is a product of the spaghetti of jumpers involved. I guess we won't know until it is boxed up, whether this approach is worth while.

I will post a soundclip this evening and any suggestions are most welcome.

Cheers,
Sam

[midwayfair]

Annnnnnnnnd you're awesome. Can't wait to hear it!

[samhay]

Well at least one thing that is not awesome is my playing tonight. I haven't played much with a tremolo and it would appear that I have a lot to learn. I am also starting to think that I might have to think about make this thing tap tremolo capable.

Here is a quick demo that might give you an idea of what this thing sounds like.
https://dl.dropboxusercontent.com/u/11996927/Anharmonic_Tremolo_demo.mp3

Setup: telecaster(ish) bridge single coil -> optical compressor (A-OK; just a touch) -> tremolo -> amp (Fender Greta; used as a tube preamp) -> line out -> audio interface.
It sounds a little darker than it does through a set of speakers - got to love home recording.

What you might be able to hear -
0:00 - harmonic tremolo, sinewave; crossover frequency at about noon; depth at about 1 o'clock
0:27 - harmonic switched to 'treble' setting
0:53 - 'harmonic'
1:10 - 'treble'
1:33 - 'harmonic'; varying the crossover frequency
2:30 - 'harmonic'; varying the speed then depth then to square wave
3:25 - 'harmonic' then 'treble' then clean

Edit - you might note that it doesn't sound like a harmonic tremolo at each end of the crossover frequency pot range and you would be right. If the crossover frequency is well above or below the notes you are playing, then you are just going to have a tremolo with a slightly funky frequency response. I call that a feature, but if it is not to taste, it is easy enough to reign in the frequency range that the pot sweeps over.

Edit - The depth was at about 1 o'clock from 0:00 to 2:30. This is near the point where U3b starts clipping the LFO.

[Labaris]

Good one!!
What is the Frequency pot for?

EDIT: Forget it, didn't read the description 

[midwayfair]

The frequency pot is very effective. Was this at max depth the whole time?

The stuff in your edit is exactly how these things work.

I did hear the whining sound you were talking about. Those are some pretty hefty bypass caps in the buffers ... Wish I had a good way to help. Hopefully it'll turn out to just be a breadboard issue.

[samhay]

I realised there was a typo in my last post, but the depth was at about 1 o'clock for most of the demo. It doesn't get much deeper from there, but rather more square.

The filter cap values are not cast in stone. I have big caps in the power filtering section on the breadboard, but to be honest, using 22u or 47u in their place doesn't make any audible difference.
The 'Shoot the Moon' tremolo uses a decoupled power supply to clean up the LFO noise relative to the 'Tremulus Lune'. I believe this works quite well, so I am hopefull that it will work here too. I would love to hear any suggestions as to how I can improve this however.

[gritz]

Just a quick post to say that it sounds really good.

[duck_arse]

another tremolo, bloody hell, I'd better get started on mine .......

[samhay]

duck_are - you know what Andy Warhol said: 'three's a party'
Did your design ever make it onto the breadboard?

[duck_arse]

"my design" is so far still spread across 3 or 4 different bits of paper, which have complete circuits on them with names like tremface, or slow gear, or something. my circuit, with "world's best ever tremolo" written on it, is still blank.

but I have a vero for my customised fet-based cardinal drawn .....

[samhay]

"my design" is so far still spread across 3 or 4 different bits of paper, which have complete circuits on them with names like tremface, or slow gear, or something. my circuit, with "world's best ever tremolo" written on it, is still blank.

You're not alone

but I have a vero for my customised fet-based cardinal drawn .....

Now that sounds interesting. I guess if you are at the layout stage then it works - nice job.

[samhay]

I just compared the heap on the breadboard to a modified EA tremolo that I built a good while ago. The EA is a lot noisier than the breadboard, so I guess that means I need to box it up. Only problem is that I am not quite sure how to squeeze the layout into a 1590B. I guess the fun continues.

[midwayfair]

Only problem is that I am not quite sure how to squeeze the layout into a 1590B. I guess the fun continues.

Not even joking, based on the parts count I could get this into a 1590A with a double sided board, probably without resorting to 1/8W resistors. Do you have an Eagle schematic? I could give it a go if you like.

[samhay]

Jon - Initially I thought I might just about be able to squeeze it into a 1590A. While I've got 6 op-amps into one before, I have gone off that idea as I want to be a little careful about the layout to try and keep the LFO noise down. Also, don't discount the extra height you will lose with the dual-gang frequency pot - I guess if you use a double-side board with the back as a ground plane it would help, but there is not likely to be enough room for components on both sides unless you use SMD. I am happy to stand correctly though.

I am sure the circuit will easily fit in a 1590B if I use perf or a PCB. I initially thought about vero, but that is going to be tight. However, this is with a few fairly stringent layout requirements, which you may wish to relax. I was/am planning on keeping the LFO and audio sections completely separate (including ground) on the board and I also am a bit funny about flying connections and jumpers.

Anyway, I have tidied up the Eagle schem a little bit and you (and anybody else who is interested) are most welcome to work your magic on it. I have also updated the schem in the initial post with the new version - only changes are the power filtering caps, which will need to be tweaked once we have a layout.

The Eagle .sch should be here:
https://dl.dropboxusercontent.com/u/11996927/Anharmonic_Tremolo.sch

[midwayfair]

Jon - Initially I thought I might just about be able to squeeze it into a 1590A. While I've got 6 op-amps into one before, I have gone off that idea as I want to be a little careful about the layout to try and keep the LFO noise down. Also, don't discount the extra height you will lose with the dual-gang frequency pot - I guess if you use a double-side board with the back as a ground plane it would help, but there is not likely to be enough room for components on both sides unless you use SMD. I am happy to stand correctly though.

I am sure the circuit will easily fit in a 1590B if I use perf or a PCB. I initially thought about vero, but that is going to be tight. However, this is with a few fairly stringent layout requirements, which you may wish to relax. I was/am planning on keeping the LFO and audio sections completely separate (including ground) on the board and I also am a bit funny about flying connections and jumpers. , I have tidied up the Eagle schem a little bit and you (and anybody else who is interested) are most welcome to work your magic on it. I have also updated the schem in the initial post with the new version - only changes are the power filtering caps, which will need to be tweaked once we have a layout.

The Eagle .sch should be here:
https://dl.dropboxusercontent.com/u/11996927/Anharmonic_Tremolo.sch

Well, I wasn't seriously suggesting a 1590A layout for this. Just pointing it out to indicate that 1590B was more than big enough. I'll see if I can get some time over the weekend to work on a layout. I accept no jumpers or flying leads in an Eagle layout, only on perf. Also, I would hav isolated the LFO even if you didn't ask, and probably will not use a ground plane.

BTW, I use these in some tight builds with dual gangs:
http://www.smallbearelec.com/servlet/Detail?no=1276

And here's the 100KC dual gang if anyone is looking for the reversed taper, just not in 9mm:
http://www.smallbearelec.com/servlet/Detail?no=1032

I bend the solder terminals out slightly on 1590B builds that use those. Works fine.

[Hemmel]

Hey guys,

2 quick questions :

a) Would it be possible to swap both TL072P for a single TL074 ?
b) I don't understand the dotted rectangle marked as "LED/LDR VACTROL" What does this mean ?

Thanks !

[samhay]

Thanks Jon - those links are handy. I hadn't seen a dual-gang 9mm pot before.

[samhay]

Hey guys,

2 quick questions :

a) Would it be possible to swap both TL072P for a single TL074 ?
b) I don't understand the dotted rectangle marked as "LED/LDR VACTROL" What does this mean ?

Thanks !

a) yep, a TL074 would work fine. I left it that way as it is often easier to lay out 2 dual op-amps than 1 quad and it also gives you the option of using e.g NE5532s, which work fine.

b) It is supposed to signify that the LDR and LED are collectively a 'vactrol' or home-rolled LED/LDR combo.

[Hemmel]

b) It is supposed to signify that the LDR and LED are collectively a 'vactrol' or home-rolled LED/LDR combo.

Ooooh... never worked with LDRs before... I think I have one is my sapre parts.
But I have no idea how to breadboard it... anyone have a pic of how to put them on a breadboard ?

Thanks for the info !

[samhay]

b) It is supposed to signify that the LDR and LED are collectively a 'vactrol' or home-rolled LED/LDR combo.

Ooooh... never worked with LDRs before... I think I have one is my sapre parts.
But I have no idea how to breadboard it... anyone have a pic of how to put them on a breadboard ?

Thanks for the info !

The resistance of the LDR is dependent on how much light is shined on it, so we usually point an LED at it. Problem is that unless you work in the dark, the room lights will fully saturate it. I wrap my LED and LDR up together with heat-shrink tubing. Others use insulation tape. At a pinch, you can just throw a black cloth over everything.
One thing to know about LDRs is that they are not all created equal. They have different wavelength responses, so you will have to try different colour LEDs to find the ones that play nice - I usually use yellow ones. Also, the dark and light reistance of LDRs vary wildly. I usually try find ones with the biggest difference between their light and dark resistances, but a good place to start is any with a dark resistance of 1M or more. The trimmer in this design is there to cope with the different LDRs or vactrols one might use.

Happy experimenting.