The Psychotenuse Tremolo... reverse sawtooth tremolo with very few parts

[moid]

Hello everyone

A new year and a new selection of strange questions from me... hopefully this one is interesting (firstly because it does actually work (mostly) and secondly because it's a really low parts circuit, and thirdly, well who doesn't like a reverse sawtooth tremolo?

Over the Christmas period I watched the below video about making a tremolo pedal from what seemed a stupidly small number of parts... and none of the parts were rare or expensive, admittedly I did have to buy the transistor; I think the one used is not common in effects pedals (BT169), but I put it together on breadboard and after a little bit of adjusting it worked!*

Schematic
https://drive.google.com/file/d/1Fu5m8ZW8VESDOa27-NpzSAE9gNpUIl4e/view


*yeah.... about that. It does indeed produce a reverse sawtooth tremolo. Kind of like the Vox Repeat Percussion but with about a third of the parts... however it does have some issues.

1. Click...at the end of each cycle of the LED pulse, the audio picks up a very short but very audible click noise... which is extremely annoying. If anyone has any idea how to stop it from doing this then please let me know... I can't understand how it does it due to the signal being completely separate to the LFO, and the signal is passive as far as I can tell... I have tried random combinations of capacitors and resistors at different parts of the circuit to take the pulse to ground, but have only succeeded in either achieving nothing, or turning the LED pulse off, or letting the LED pulse once and then not retrigger... so the worst trem ever!

2. Length of LFO cycle. The pot (I used a 50K Lin) does indeed allow you to have reverse sawtooth waves closer together or further apart, but it does not alter the length of each wave very much - slowing the pot down just puts bigger gaps between each wave, with the space between being unaffected guitar signal and each wave only gets slightly longer, wherease the amount of time when the LED is at full brightness is considerably longer - I would say that when the pulse lasts 0.1 seconds, the gap between pulses is 0.9 seconds - so the effect only really works at fairly fast speeds when the pulses are closer together (and if I turn the pot up too fast the LED stays on without ficker and the effect turns off and you just get dry guitar). I tried swapping the 100uF capacitor for a 220uF cap which allows more space between each pulse, and each pulse takes a bit longer to build up but not double the length of time as I was hoping... unless I also adjust a 100K lin pot that I placed in the below position (on a line that connects connects D2, RV1 and D1) - that does allow me to get a bit of variation in pulse length. Swapping in a 470uF cap does give longer pulses, but also even longer spaces between pulses...if the pulse length is say 1 second, there is a gap of five seconds before the next pulse. Ideally I'd like the LED pulse to immediately reset itself after triggering (as you would expect in a tremolo); but it sadly doesn't.

3. I removed a line in the schematic that connects D2, RV1 and D1. If a connection is left at this point, the LED stays on all the time and you get no tremolo (unless I built the circuit wrong?). Removing the line makes the LED pulse. I did try inserting a 100K lin pot in this line to see if I could use varying levels of resistance to create a depth control - and although that does seem to mostly work in terms of the LED pulses not dimming to nothing  / eventually the LED being on all the time etc, it didn't have a huge effect on the sound, certainly not what I was hoping for.

4. Obviously the audio part of the circuit is passive so I'll need to add a boost of some kind to it (knowing me a LM386 amplifier circuit) - if I do this, should I place the amplifier before the R2 10K resistor or after it?

It wouldn't surprise me if there is not a lot that can be done to improve the circuit - it is ultra simple. If the click could be removed that would make it something fun that would fit in a small box easily, but being able to adust the LED pulse cycle so it constantly repeated without waiting would really help - and also being able to make the pulses longer without longer gaps between them! If it's not possible to improve it, it has at least made me realise that theoretically I could replace the circuit with another blinking LED type circuit and use that to control the tremolo... so I may do that if this can't be fixed.

Any thoughts are gladly appreciated, hope you're having a good 2024 so far!

[Mark Hammer]

Doesn't get much simpler.  Nicely done.
As noted, the drawback is that, as a passive attenuation device, it needs some boost to provide reasonable effect<->bypass volume balance.  A simple FET or BJT booster stage between the input and R2/R3 pair would be suitable.  Optimum S/N ratios are always obtained when the signal source is at its maximum clean amplitude, prior to attenuation.

The other concern is whether the signal is loaded down in a manner that loses critical bandwidth.  At the same time, this is of no concern when the circuit is fully bypassed, and tremolo effects can be more "musical" if high frequencies are attenuated a little more, during the "valleys".

[duck_arse]

hello moidy. can I use C106D instead?

[moid]

Doesn't get much simpler.  Nicely done.
As noted, the drawback is that, as a passive attenuation device, it needs some boost to provide reasonable effect<->bypass volume balance.  A simple FET or BJT booster stage between the input and R2/R3 pair would be suitable.  Optimum S/N ratios are always obtained when the signal source is at its maximum clean amplitude, prior to attenuation.

Thanks Mark! I couldn't make up my mind whether it was better to boost before or after attenuation, but what you say makes sense. I'll try that, hopefully at the weekend.I presume this wouldn't fix my clicking issue would it?

The other concern is whether the signal is loaded down in a manner that loses critical bandwidth.  At the same time, this is of no concern when the circuit is fully bypassed, and tremolo effects can be more "musical" if high frequencies are attenuated a little more, during the "valleys".

I'm sorry that goes over my simple brain - I don't think the circuit is attenuating any specific frequencies more than any other...I might be wrong of course, the valleys are so short that it is hard to tell - I'm fairly confident that it is attenuating the volume of all frequencies.

hello moidy. can I use C106D instead?

'Ello your duckness! You can use what you like old boy If yo uare very kindly breadboarding this then thank you very much - but please don't put too much time / effort into this, I suspect the issue would be better fixed by using some sort of sinewave generator circuit to light the LED...


One issue - C106D seem to be hard to come by in the UK though, every seller says they are discontinued... I can find some expensive ones on ebay If you've got one and you're breadboarding this then please try it and if it works without clicking then that would be lovely (also if it can remove the gaps between each pulse that would be awesome) - I will happily buy one if it works, it's just that they appear to be about £5 each on ebay, whereas the BT169's were only 50p each so I didn't mind buying those, but I'd rather not spend £5 on something that might not work... hmmm one reseller offers BT148 as a substitute for C106D - I tried looking at both datasheets, but there was nothing about time in the C106D sheet, however the BTR148 does mention various functions with timings
https://docs.rs-online.com/33db/0900766b800ab59d.pdf in case that means anything useful?

[duck_arse]

moid - you was supposed to say no, use the C106Y instead, then we would both fall about larffing. I think pretty much any SCR could be used in this circuit. you can even make and SCR-alike with 2 transistors, which was a topic some little while back. to replace the PUT in some or other trem circuit thing. I ended up with a noise/oscillator circuit from that convo.

bipolar unijunction oscillators.

[Eb7+9]

no boosting required in a passive tremolo ...
you have the right idea already   ie., oscillator kill instead of bypass

see here: Dallas Arbiter "Trem Face"

one thing thing that comes to mind is applying variable (uni-directional) current shunting to the LED ...
yielding some form of variable DEPTH control



try sticking a 100k resistor in parallel with the photo-element to see if that does anything to the ticking

[Mark Hammer]

Many non-passive tremolos will incorporate some form of variable gain-recovery, simply because average volume level in an operating tremolo can appear to be lower than bypass.  The additional gain is for maintaining perceived average volume level, OR for having the volume level in effect mode be higher, just because the user wants it that way.  But you're absolutely correct that a gain stage is not required for the circuit to function.  My recommendation of simple input buffer/gain stage is more for usability purposes.

[moid]

moid - you was supposed to say no, use the C106Y instead, then we would both fall about larffing. I think pretty much any SCR could be used in this circuit. you can even make and SCR-alike with 2 transistors, which was a topic some little while back. to replace the PUT in some or other trem circuit thing. I ended up with a noise/oscillator circuit from that convo.

bipolar unijunction oscillators.

Well there you go using that complicated humour on me! You know I am a bear of very little brain Thanks for the suggestion though! If you happen to remember the thread you are talking about I'd happily give it a read - I rarely got a chance to look at the forum between October and Christmas.

[moid]

no boosting required in a passive tremolo ...
you have the right idea already   ie., oscillator kill instead of bypass

see here: Dallas Arbiter "Trem Face"

one thing thing that comes to mind is applying variable (uni-directional) current shunting to the LED ...
yielding some form of variable DEPTH control



try sticking a 100k resistor in parallel with the photo-element to see if that does anything to the ticking

Thanks for the diagram! I didn't draw the schematic though - I got it from the youtube link. I added the diode and 5K pot and hte 100K resistor you mentioned and sadly it all still ticks. Turning the 5K pot counter clockwise makes more clicks, whereas clockwise is a few less clicks. I tried using a 10K pot but I think that stopped the circuit from working and a 1K pot made it click even more... I also tried different capacitors in place of the 100uF and found that as I increased the value the clicks became less - at 1000uF there are no clicks at all (yay!) and the pulse of the tremolo is longer (at least half a second long)... but the intervals between pulses are now about 5 seconds (damnit!). The Speed pot also has a tiny range at this point with most positions turning the tremolo off. I did try larger Speed pots - up to 1M but didn't get much ability to improve this. Turnig the 5K pot you suggested also speeds up / slows down the tremolo rate, but only noticeably when there is a 50K pot in the circuit for Speed.

I'm sorry to have wasted everyone's time, I think this is one of those circuits that technically works within very tight parameters, but has a lot of issues that don't want to go away... I think I may be consigning this one to history / experience... and go back to trying to finish an MS20 filter! Thanky you for the help, it's appreciated as always!

[moid]

Many non-passive tremolos will incorporate some form of variable gain-recovery, simply because average volume level in an operating tremolo can appear to be lower than bypass.  The additional gain is for maintaining perceived average volume level, OR for having the volume level in effect mode be higher, just because the user wants it that way.  But you're absolutely correct that a gain stage is not required for the circuit to function.  My recommendation of simple input buffer/gain stage is more for usability purposes.

Thanks Mark - in this case the effect definitely cuts volume though, unless that's an issue with my breadboard (possibly) - if I record the guitar clean into my PC and then through the pedal the waveform when the tremolo is not cutting volume is about 6dB less than going straight in directly.

[moid]

no boosting required in a passive tremolo ...
you have the right idea already   ie., oscillator kill instead of bypass

see here: Dallas Arbiter "Trem Face"

one thing thing that comes to mind is applying variable (uni-directional) current shunting to the LED ...
yielding some form of variable DEPTH control



try sticking a 100k resistor in parallel with the photo-element to see if that does anything to the ticking

Thanks for the diagram! I didn't draw the schematic though - I got it from the youtube link. I added the diode and 5K pot and hte 100K resistor you mentioned and sadly it all still ticks. Turning the 5K pot counter clockwise makes more clicks, whereas clockwise is a few less clicks. I tried using a 10K pot but I think that stopped the circuit from working and a 1K pot made it click even more... I also tried different capacitors in place of the 100uF and found that as I increased the value the clicks became less - at 1000uF there are no clicks at all (yay!) and the pulse of the tremolo is longer (at least half a second long)... but the intervals between pulses are now about 5 seconds (damnit!). The Speed pot also has a tiny range at this point with most positions turning the tremolo off. I did try larger Speed pots - up to 1M but didn't get much ability to improve this. Turnig the 5K pot you suggested also speeds up / slows down the tremolo rate, but only noticeably when there is a 50K pot in the circuit for Speed.

I'm sorry to have wasted everyone's time, I think this is one of those circuits that technically works within very tight parameters, but has a lot of issues that don't want to go away... I think I may be consigning this one to history / experience... and go back to trying to finish an MS20 filter! Thanky you for the help, it's appreciated as always!

I still don't understand why the circuit is able to generate a click in the audio though - to me the audio and the tremolo are completely separate; I don't understand how the click goes from light coming out of an LED into the LDR and that affects the audio... to me that would make sense if the audio path had DC running through it (but it doesn't). Very strange!

[duck_arse]

Well there you go using that complicated humour on me! You know I am a bear of very little brain Thanks for the suggestion though! If you happen to remember the thread you are talking about I'd happily give it a read - I rarely got a chance to look at the forum between October and Christmas.

the skippy, the hummingbird, the Vox repeat percussion, probably others .....
https://www.diystompboxes.com/smfforum/index.php?topic=86385.0
https://www.diystompboxes.com/smfforum/index.php?topic=126057.0
https://www.diystompboxes.com/smfforum/index.php?topic=123344.0
https://www.diystompboxes.com/smfforum/index.php?topic=126049.0

[moid]

Thanks Duck, I will have a read

In the meantime I have changed everything! I built a new LED flashing circuit (see below) and used that to pulse the LED / LDR combo and it actually works as a tremolo!



I took an LED flashing circuit I found online and then changed some values because it generated a loud hum pulse whenever the LED was fully off - so the original circuit had a higher value resistor in place of the 100ohm I used, but the 100ohm lowers the hum pulse to below normal hearing level. I added a pot to control the speed of the LED pulse, and a 25K line pot seemed to go from 4 seconds between pulses up to audio rates. So all that is fine... except... the ratio of audio pulsing in to being off is rather narrow - like the previous circuit the width (in time duration) of each pulse of audio is much narrower than the amount of time that no audio passes through. The wave form is some sort of sine wave I think, but I need to widen the pulse so there is a more even amount of time with the audio gaining volume at a similar rate to it losing volume - so have the LED dim down / be off for longer (when the LED is off the audio passes through the signal). How would I do this? Is this possible with this circuit? I'm not sure what sets the width of the pulse; is it the 7555 chip or the transistor?

Also is there a way to set a different rise / fall length of the pulse? I was thinking if there were two pots, each controlling the amount of time it takes to brighten or dim the LED, one for the rise and one for the fall, it might be possible to make a sawtooth / reverse sawtooth pulse effect?

Thanks everyone

EDIT: Maybe I have found what I need here? http://www.pcbheaven.com/circuitpages/555_Breathing_Pulsing_LED/

Could I run this circuit on 9V instead of 5V? I can stick a 7805 into it if required, I just wondered if it would be safe at 9V. I think I have all the parts for this!

Happy to consider other options if anyone has any.

[duck_arse]

9V is safe, yes. the 7555/555 has a charge time and a discharge time, set by the two resistors and the one cap. it's always a juggle or a trick to get 50% on/off, and moreso if you want variable frequency AND 50% duty cycle. two pots, or dual pot. or different circuit.

[moid]

So you're saying there's a chance

Thanks Duck, I'll try this on breadboard soon. I had a clever idea (ok I stole it from Merlin) about the original circuit I put together (the thyristor one). I was reading Merlin's description of his stompLFO chip and in that he mentioned:

"Wave smoothing
Sometimes when using the LFO output to drive a VCA or similar sensitive audio circuit, the sharp
edges of some waveforms can cause clicks in processed audio due to CV feedthrough.The wave
smoothing feature applies the exact digital equivalent of a simple RC lowpass before the wave
leaves the chip to help prevent clicks in VCAs,VCFs, optical tremolos, and similar vactrol-controlled
circuits.The filter has a cutoff frequency of around 1KHz, and this gives a rise time on sharp edges
of about 10 msecs. Note that since the smoothing is a lowpass filter, the highest frequency waves
produced by the chip will be affected most."

so I thought, aha, I've got an optical tremolo; I'll stick a RC filter between the 100uF cap and the LED (which I did)... and it did not affect the clicking at all. I tried variations of capacitors and resistors to create a low pass filters at 190hz, 1Khz, 3khz and 4khz and none did anything. I think that thyristor circuit is about to become breadboard history (I need the space to make that rise/fall circuit!)

[duck_arse]

moidy - how many tansistors is too many, three? will flash/pulse two leds, and saw/ramp two leds, cause I put pairs of leds in. but the transistors - youse needs low low hFE types to make it go. what types do you have, pnp and npn?

BC54*A and BC55*A are a good start. BC327-16 and BC337-16 maybe. 2N2369 is good. select 2N3906. you tell.

[ElectricDruid]

The 4047 is a better oscillator if you want square waves. The 555 is ok for pulses, but like Duck said, squares is always a bit of a faff, especially with variable frequency.

[duck_arse]

start with the circuit values as shown. if the LOAD leds come on mid-bright and not flashing, R1 is TOO SMALL (10k). if the LOAD leds come on dim-dim and not flashing, R1 is TOO BIG (100k). once you get the right R1 for your Q1, leave it. it should work with all legal led/led/C1 combos.

(if R1 is only a little larger than too small a value, the flash shape steps. it might appeal to the user.)

the TRIG leds will light very dim the whole time, and blink in time with the triggering.

C1 dumps all its charge through the LOAD leds, so a bigger cap will flash brighter. but it will also mean bigger supply spikes. C100 can be as big as you like.

three white or blue leds in either string is probably too much for triggering. special note - LOAD leds can be connected in parallel (and differing colour), and with narry a clr in sight, and will flash about same.

ALL PARTS VALUES will affect the frequency range. the led colours, the number of leds in each string, the C1 size, the R1 size - ALL. and wildly in some cases.

the C106D was used to be called a sensitive gate type. the C122D is more of a brute, and didn't trigger when tested. the BT169 appears to be similar I_GT and V_GT specs to the C106D, so it should work same/similar as tested.

mess about with the leds, colours, pot value, caps and R's as much as you like. I haven't tested with audio, so I don't know how bad it will tik. knock yourself out, and nevermind that previous post. unless this circuit doesn't work.

[moid]

Well thankee kindly you good sir! Sorry about the slow reply (I blame the sheer distance from the UK to the Land Down Under) or maybe just that work has been dreadful this week (and promises to be so until May by the look of it ) Anyway, a progress report! I have rebuilt the circuit based on that drawing from yer own fair hands, and wonder of wonder, fings 'ave got bettah! A certain degree of jiggery pokery and component malarkey (you weren't joking about how each item in the schematic can have a different effect on the sound) and the assistance of moid jnr playing guitar through the circuit while I plugged things in / pulled stuff out have resulted in the following:

R1 is a 7K5
C1 is staying as 47uF, but I will add a SPDT switch to flip between that and a 10uF for super fast sounds.
TRIG LED are both High Brightness Blue
LOAD LED - the first is an Amber /Orange LED, the Second is a Red one, the order actually is important (first being closest to C1)
RV1 has become a 20K lin pot
The LDR is bolted to the Red LED, and is a 8K-24K LDR (this gave the best result)

With this seting we have a reverse sawtooth with very little clicking at all (only present if I set the record volume on my DI box to absolute maximum). If I record at about 3/4 record volume (which is normal for me) then there is very little issue and nothing I would worry about.

Stuff we tried:
Making C1 larger makes the gaps between each pulse of the trem longer, but the pulse doesn't scale to fit that length - there is a minimum setting of how slow the pulse can build up (I presume this is determined by the transistor? Does your transistor behave differently?) so 47uF is the slowest setting that works (it's not slow; it pulses twice a second, but the fastest it goes is four times a second). Any slower and it sounds odd that the volume fades up, then stays at normal volume for too long before the LEDs trigger the transistor into starting another pulse?

Different colour LEDs really effect the sound of the tremolo a lot! On the Trig side, anything really bright (Ultra Bright White or Ultra Bright Blue LEDs cause really loud ticks, anything less bright than the
High Brightness Blue LEDs turns the circuit off. On the Load side, two red LEDs caused a small amount of clicking, two amber/orange LEDs were very soft - there was a gentle tremolo, but not one that really cut much volume out, and trying yellow/green/violet LEDs were all variations of even more subtle (I think violet did nothing). Using brighter LEDs here made the circuit click very badly. The best result was amber LED into Red LED.

The pot and R1 have a close relationship - I got the circuit to work using 10K in R1, but only half the pot worked if i used a 50K (I didnt have a 47K)  - from 12 o'clock to 5 o'clock. Switching to a smaller pot and also lowering R1 got to a point where the entire sweep of the pot affected the speed of the tremolo.

The LDR - I started this with one of my bundle of unknown cheap Chinese LDRs - they work, but they didn't really cut the sound to near 0 dB, but switching to an 8-24K LDR gave much better results.

I did try getting the flash shape into a stepped effect but didn't seem to achieve this - either it would trigger once then not reset until I made a small change to the pot, or the LEDs were just ON for most of the pot travel and then started to blink only at the ends... it's probably something that needs replacing with a pot to see if the exact value can be found... We didnt try for long because we spent over an hour on different LED colour changes... which I thought were the cause of the click, but now I re read your schematic and go d'oh... because I just realised we didn't test C100! OK will try to look at that tomorrow night with a larger cap.


Improvements?
Two items for the request list; one is probably impossible.
Depth pot - is this possible, if so, would this bridge the audio out and audio in part of the circuit? So when set to one extreme of the sweep it would not add any of the original audio to the tremolo signal and at the other end of the sweep it would probably replace the tremolo audio - is that a good idea? Or should it go somewhere else?

Slow the pulse down - is there any way to get the thyristor to slow its pulse down? At the moment the circuit only allows the trigger of the pulse to change in terms of frequency of trigger event; once the trigger happens the length of the tremolo pulse(from audio off to maximum audio) is the same - 0.155 seconds. If the pulse rate is slowed to less than 2 times a second the ramp of the sawtooth is still only 0.155 seconds long, and the rest of the pulse is at maximum volume which doesn't actually sound very good.I assume this is a hardware limitation of the transistor, so probably can only be changed with a different transistor? Perhaps your transistor is a better one? I can buy one if you think it will make the pulse slower?

Thanks again for the use of your grey mattter!

[duck_arse]

what? what do you mean it's not perfect? works fine for me. well, blinks like I expected, and not cennected to any audio.

I won't argue with someone who has done the experiment, but, whether the orange comes first or the red is first in a series string makes no matter, parts in a series string have no order. that's one of the laws. are you changing the led colour that shines on the ldr red for orange? in that case, you might misinterpret a difference in ldr response as being from led order. a reread says no, you aren't. which complicates things.

and your 7k5 bothers me. I couldn't get any response at less than 22k. the led brightnesses matter much less than their forward voltages, and the TRIG leds are not intended for display. they can be as dim as they like, even blank like a zener, and the circuit should still work. I'm not sure about your waveshape notes either - I didn't have anything but boring repetitions that followed the freq pot settings. I didn't notice longer dwells or anything like that. bigger cap made it run slower, too.

which all suggests you may need to pull it down and start again. I will have a quick look on the oscilloscope tomoz to see what the waveshape does, but I'm not expecting anything out of the ordinary.