Introducing GJENGANGAR: a Gate-Delay (with some issues)

[knutolai]

Hi!
This is a circuit I've been working on the last few months. The basic architecture is similar to a MadBeans CaveDweller with "Echo" maxed. However a gate is placed on the delay output for both the feedback and feedforward signals. The gate again is controlled by the input signal.

Needless to say the effect is very interactive with the playing and the addition of a send/return loop is cool for experimentation. I'm quite pleased with the result and the concept as a whole, though not entirely new I've never seen a gate-delay pedal before.  
Oh and the name, for anyone wondering, is in Norwegian. It's a reference to "Gjengangere" by Henrik Ibsen.

The circuit has some issues mainly due to me being a crappy engineer. I've started working on a redesign, but I though I'd chime in to ask for help solving some issues that I can't find the cause of.   (see below the schematic)

Couldn't find a way to reduce the size of the schematic without making the values unreadable. Sorry. It helps to open it in a separate tab.
Edit: schematic correction


1. On start-up the Gate part of the circuit requires 10-15 seconds to stabilize itself. Normally when the THRESHOLD pot is near being maxed (100k) the threshold is always exceeded (gate always open). When stabilized the threshold is automatically exceeded at approx 90% rotation. However at start-up this is reduced to around 20% rotation. Then the "always-exceeding-threshold-point" sinks for 10-15 seconds until it stabilizes at 90% rotation.  
The behavior has a similar feel to a capacitor being charged or discharged. I thought this could somehow have something to do with the SUSTAIN pot and C28, but the circuit requires the same time to stabilize regardless of the SUSTAIN pot being 1M or 0 or anything in between.

2. I've built two of this circuit, same layout and everything. All leads seems to be in order (triple checked everything). One of the circuits is working as intended while the other sometimes works on first start-up (when having chilled out for a good while) then when re-powering the circuit I get clean signal only. The LED-LDR-couple is working. And as I get clean signal with the same intensity as when the pedal is bypassed I assume the onboard op-amps of the PT2399 are working too. Bizarre partial IC latch-up?  
The chip and lm78L05 are bought from Taydaelectronics. The lm78L05 gets just noticeably warm (not really hot) when the circuit is powered.
I guess the circuit might work after the 13. or 27. check though...  

[merlinb]

Try grounding pin 4 of the PT2399. Apparently some people have had problems with leaving it ungrounded, due to internal differences between different manufacturing sources.

Your gate detection circuit needs work. The output of U5 is at 2.5V, but U2 is working with a 4.5V reference, so it's trying to amplify 4.5-2.5=2V by 235 times. I'm surprised it does anything at all...
Incidentally, you don't need D2 or D3 since the LM393 has open collector outputs. I doubt you really need U2 also.

[knutolai]

Try grounding pin 4 of the PT2399. Apparently some people have had problems with leaving it ungrounded, due to internal differences between different manufacturing sources.

Thanks! Ill give that a go.

Your gate detection circuit needs work. The output of U5 is at 2.5V,

whops! Its supposed to be taken from the output of U1. The schematic should be correct now.

Incidentally, you don't need D2 or D3 since the LM393 has open collector outputs

I'll have to read up on that, but if it works that's great. The circuit is really a little too big.

I doubt you really need U2 also.

I think I've tried removing it before and found that the comparator was unable to detect the input signal unless it was really loud. Also the minimum threshold voltage needed for the gate to be in "always open" mode it not equal for the two circuits (resistor and potentiometer tolerance taken into account). Would I be better off adding a full wave rectifier?

[knutolai]

Tried grounding pin 4 and now I'm getting delay. The delay does not stop working however when I lift the ground again. When re-powering the circuit without pin 4 grounded I still get delay. Does grounding pin 4 once somehow "fix" the chip? It's probably a very silly question, but I've been re-powering this circuit about 20 times earlier today hoping for delay, and now I can't tell for certain if it's working because of some cosmic magic or because of grounding pin 4.. I haven't done any other alterations to the circuit.. hmm

Incidentally, you don't need D2 or D3 since the LM393 has open collector outputs

Tried replacing the diodes with jumpers and the result is neither the SUSTAIN or SPEED parameter has any effect on the gate. The idea behind D2 and D3 is to avoid the capacitors from discharging through the comparators. Thanks for the suggestion though!

Also tried putting a 100n capacitor between U2 and U3 (R26 on U2 side, R27 on U3 side) and the threshold seems to stabilize much faster. No clue why I havn't tried something similar already 

[mth5044]

Tried grounding pin 4 and now I'm getting delay. The delay does not stop working however when I lift the ground again. When re-powering the circuit without pin 4 grounded I still get delay. Does grounding pin 4 once somehow "fix" the chip? It's probably a very silly question, but I've been re-powering this circuit about 20 times earlier today hoping for delay, and now I can't tell for certain if it's working because of some cosmic magic or because of grounding pin 4.. I haven't done any other alterations to the circuit.. hmm

Incidentally, you don't need D2 or D3 since the LM393 has open collector outputs

Tried replacing the diodes with jumpers and the result is neither the SUSTAIN or SPEED parameter has any effect on the gate. The idea behind D2 and D3 is to avoid the capacitors from discharging through the comparators. Thanks for the suggestion though!

Also tried putting a 100n capacitor between U2 and U3 (R26 on U2 side, R27 on U3 side) and the threshold seems to stabilize much faster. No clue why I havn't tried something similar already 

Maybe you found the fix for Jimi's in everything PT2399 thread? 

[merlinb]

Does grounding pin 4 once somehow "fix" the chip?

Unlikely; you're probably just getting temporarily lucky. Leave it grounded.

The idea behind D2 and D3 is to avoid the capacitors from discharging through the comparators.

Ah right. It's a slightly convoluted way of doing things, but I see what you're trying to do.

[knutolai]

Ah right. It's a slightly convoluted way of doing things, but I see what you're trying to do.

Yeah you are right. I think I found a way to at least omit D2 and R29.

The inputs of the both comparator are reversed, and the bypass switch and the sustain threshold had to  be changed a little. The sustain is set by the charge time of the cap instead of the discharge time. I don't see how I would go about eliminating D3 and R33 though. Perhaps just one of the cons of using comparators? I would love to somehow be able to control the gate speed with a pot instead of a switch.

I'll try upload a sound sample soon. It's a fun circuit to play around with. Gets very musical coupled with my cool cat chorus in the send/return loop, and the SPEED set to 'slow'.

[merlinb]

I would love to somehow be able to control the gate speed with a pot instead of a switch.

It seems to me that what you're really doing is controlling the attack and release times of the LEDs, not their brightness (sustain). Normally you would do both attack and release with one comparator:

The release time is interactive with the attack time, but the attack time is independent of the release time. No one usually notices this since attack times are invariably much longer than release times anyway.

[knutolai]

I might have mixed up the terminology for the controls. 
To clarify what I mean for the controls to do: The SUSTAIN param. is meant to set the amount of time the gate is open after the threshold is no longer reached by the input signal. By SPEED I mean the time it takes the gate to go from "full-closed" to "full-open" (attack?) and vice versa (release?). I don't think I can accomplish that with a single comparator. Are we on the same page? As it is now the attack is quite shorter than the release.

I'm not eager to replacing R33 and R34 with pots as I expect it would affect with the max brightness of the LED making it interactive with the FEEDBACK param.

Did another test with connecting pin 4 to ground. I let the circuit rest overnight, first tried to power it with pin 4 ungrounded getting no delay, then with pin 4 grounded.. delay! It's weird I got those results yesterday, but now I'm quite certain this was the problem. Thank you so much! Never would have guessed this would be the reason.

[merlinb]

The SUSTAIN is meant to set the amount of time the gate is open after the threshold is no longer reached by the input signal.

That is what I would call the 'release'. It's the time taken for the gate to close again.

By SPEED I mean the time it takes the gate to go from "full-closed" to "full-open" (attack?)

Yes, I would call that the attack.

and vice versa (release?).

But isn't that what you call SUSTAIN? After the signal drops below the thrshold, its the time it takes to go from full-open to full-closed...

I would normally associate 'sustain' with how much signal is allowed through (amplitude), rather than how long it is allowed through for.

[knutolai]

hehe this can get confusing pretty quickly. 

I took a subject on digital synthesis a few years back where you talk about a ADSR curve (attack, decay, sustain, release) in reference to synthesizers. They defined Attack as the time taken for a note to reach maximum amplitude (after pressing the tangent with a finger). Decay as the time taken for the note to reach sustain-amplitude. The sustain amplitude is kept until the note is release (the finger releasing the tangent) and the release as the time taken to go from sustain-amplitude to zero after the tangent is released.

By the definition laid out in that class both my parameters should be called "release" as they both shape the release part of the ADSR curve. Maybe I should rename the SUSTAIN parameter to "hold" or "sustain time" or something. What would you call it? 

[StephenGiles]

I don't get this, what is it supposed to do?

[slacker]

It's a delay with the level and feedback controlled by an envelope generator, play loud you get more level and feedback play quiet and it shuts up, I think. Like the opposite of a ducking delay. Quite a cool idea.

[knutolai]

It's a delay with the level and feedback controlled by an envelope generator, play loud you get more level and feedback play quiet and it shuts up, I think. Like the opposite of a ducking delay. Quite a cool idea.

Pretty accurate description apart from that the level and feedback is either on, off or traveling from one state to the other. Thank you. If anyone wants to give the circuit a go its probably worth it to wait for a updated schematic.

[StephenGiles]

It's a delay with the level and feedback controlled by an envelope generator, play loud you get more level and feedback play quiet and it shuts up, I think. Like the opposite of a ducking delay. Quite a cool idea.

Ah right, a circuit of mine was published in the September 1982 edition of ETI -  CMOS Fuzz/Tremelo, I forget which way round it was but if you played hard you got one effect and if you played soft you got the other!

[merlinb]

By the definition laid out in that class both my parameters should be called "release" as they both shape the release part of the ADSR curve. Maybe I should rename the SUSTAIN parameter to "hold" or "sustain time" or something. What would you call it?  

ADSR is not something we use explicitely in the guitar world very often, so we can be a little more flexible. We don't normally have the 'decay' part, so I still think you have an attack and release control (think of a compressor). If we plot the current through the LED with different control settings, your circuit does this:

Control over the attack and release times can indeed be accomplished by the standard implementation I posted earlier (which would cascade into the second comparator, where you could quite separately control the ramp times too).

[knutolai]

Control over the attack and release times can indeed be accomplished by the standard implementation I posted earlier (which would cascade into the second comparator, where you could quite separately control the ramp times too).

Ah I though you meant for it it to be sent directly to the LEDs. I see what  you mean now.

[deadastronaut]

It's a delay with the level and feedback controlled by an envelope generator, play loud you get more level and feedback play quiet and it shuts up, I think. Like the opposite of a ducking delay. Quite a cool idea.

yeah, sounds like it would be fun to play...nice project knutolai.

[knutolai]

So! Slapped together a quick little demo of the pedal in action. First two segments are with short delay, lots of feedback and a fast switching gate. Third segment has a slower-reacting gate setting. Last segment has a chorus in the loop which IMO just sound plain awesome. Enjoy!  

(excuse my horrible playing!)
https://soundcloud.com/knutolai/gjengangar-demo-1

This thing gets loud pretty quickly if you don't fine-tune the gate and feedback. A drawback of wanting to have a send/return loop that affects both the clean and delayed signal is that you cant control the overall delay level like on a normal delay pedal. A volume pot for the feedforward delay path would tame the delay volume at lower feedback-settings, but has less impact the more you turn up the feedback pot.

Here's a picture of the finished module:

[~arph]

Nice design! I like it sounds good! I agree with you on the chorus part.