PT2399 Stutter questions/ideas

[forsakenrider]

So I've got a dream. I want to build a momentary auto stutter and I've been gathering ideas and info. I was first thinking of something along the lines of a simple tremolo with an Attiny programmed to do a pulse wave. I have tried square wave and found to mimic the switch on a guitar the way people generally using it, the off time is quite a bitt shorter then the on time.... ANYWAY moving forward.

I found the MWFX Mini judder and realized that would basically do what I want AND more. So after hours of browsing and seeing all the different revisions I ended up finding this statement:

"The Judder effect produces short, near infinite loops of audio through a tweaked feedback loop in an analogue echo chip.  It does this by constantly sampling your input signal in the background in a 'ring buffer' state
To make it easier to tap in and cut up short bursts of repeats, the pedal uses a soft touch momentary footswitch.  This mutes the input signal and directs a looped version of the sample to the output.
The rate control varies the duration of the recording and varies the pitch of the sampled sound.  The Princeton PT2399 has been carefully selected for it's analogue type noise artifacts and loft 'stomp box' character.
-Repeat length 10ms - 300ms"

So now to figure out how its done with the PT2399. I've found many posts saying a 2399 looper is "NOT POSSIBLE".

Here are My thoughts and problems:

• Momentary switch to engage effect
• Probably hard to do true bypass as it will need to be "sampling" the incoming audio before you push the switch
• how does the loop "clear" when you release so it doesn't get into self oscillation, (I think by cutting the feedback loop when the effect is not engaged)
• mode one: cuts input and plays chunk on repeat (100% wet)
• mode two: keeps sampling input and adding but its not going straight through (100% wet)
• mode three: guitar passthrough over top of sampled loop without adding to it (wet/dry mix)

Any ideas, advice, or examples would be really appreciated. I will start working on a block diagram then schematic now.

[Josh?]

I've played around with the PT2399 a bit but I'm by no means an expert, so take what I say with a grain of salt. That said:

A PT2399 isn't "analog" in the way that advertisement implies. It still sounds great, but it bugs me when people try to sell stuff on buzzwords or half truths.

With enough feedback and a loud enough input signal, a pt2399 will oscillate, but it's got a sort of white noise behind it. It's not a looper in the traditional sense, but it can be used really well for the type of glitchy infinite stutter thing you seem to be going for. The eqd space spiral has infinite feedback and uses a pt2399, so you should check that out.

You can make the "loop clear" by decreasing the feedback when the effect isn't engaged, or maybe by setting the feedback to slightly below oscillating.

Modes one and three can be reached by using a dry mix or a switch to cut out the dry signal, just like in a regular diy pt2399 delay or chorus. I'm not sure what you mean by mode two, though, sorry. If it's the dry signal without the stutter, then that can be covered by having the momentary foot switch cut the wet signal when it's not on.

I'm pretty sure you can't have true bypass, but I'm sure someone here can help you with ways to minimize any tone alterations from that.

[forsakenrider]

Thanks for the reply Josh.

Mode two, is basically a standard delay/echo with no dry guitar signal, only the delayed "chunk" repeating, and if more sounds are played while the switch is held down, they would be added to the loop.

Mode three is the standard delay style, but new guitar notes are passed to the output through the mixer and NOT added to the repeating loop.

After drawing it on paper it seems 4 things need to be switched, When engaged the Input to the Delay and the straight through need to be disconnected, and the output from the delay needs to be connected to the input of the delay chip (feedback) and the output of the effect.

[anotherjim]

You can't stop a PT2399 from recording it's input to its memory and overwriting what is already in there. So you cannot recycle a recording exactly. If you cut the live input and only record the delay output to get a loop, you ideally need to maintain unity gain and flat frequency response around the loop exactly, which is tricky. Greater than unity and it builds up to clipping, less than unity and it fades into noise. It will be better than a BBD delay at this since the delay memory itself is digital and doesn't degrade the signal once its inside. Probably, it can loop well enough and for long enough to be used shortly after the loop is first recorded and if the loop is refreshed often enough. Might have to have at least 2 delays taking it in turns.

[forsakenrider]

Well I've uploaded a Kicad Schematic. It's basically just valve wizards Small Time with added jfet switches.

The PDF is HERE

Here's how I figure it works:

In regular playing (bypass) mode, Q1 and Q2 are "on" so signal passes through the buffer to the output and is continuously going into the PT2399 to be sampled and overwritten each cycle. Q3 is off so no signal is coming out.

When Normal mode is engaged, Q3 is on, and now the "loop" in the  PT2399 is passed to the output and back into itself via RV1 (repeats), Q1 and Q2 are off so no dry signal is passed through and nothing new is added to the loop.

Another mode would be the same as "Normal" mode, but Q2 is on, so dry signal is getting passed to the output with the loop but is NOT being added to the loop.

And the last mode, Q1 and Q3 are on, signal is continuously added to the "loop" and then passed to the output, no "dry" signal is mixed in.

The reason why I chose to use JFETs for switching is so I can slow down the switching slightly to avoid pops, but this could all probably be done with a momentary 3pdt stomp and a few toggle switches, maybe, I think....

What I am mostly wondering, is the placement of Q3 going to do what I want? is there as better solution? Realistically, I probably don't need very long delay times, so noise shouldn't be much of a problem.

[bean]

R17 should come from OP2-out (pin12).

Two suggestions: put a buffer in the feedback path to keep the continuous feedback from getting too dark or filtered and make R7 a trimpot so you can dial in the precise amount of feedback without self-oscillation.

One more suggestion - to get the really long delay times, add a 100n cap in parallel with C21, tie it to lug3 of RV2 and have R20 off the RV2 wiper. Make RV2 100k. So, when you dial in those really super long sample times it adds another LP filter at the PT2399 output.



This is a cool idea, BTW!

[bean]

I lied, one more suggestion. And this is just because it would be fun in this application.

Put a 22R (1/4W or 1/2W) between pin1 of the PT2399 and another momentary stomp. Connect the other end of the momentary to ground. When the switch is engaged, it will continuously pitch detune the output of the PT2399. It works for about 2 seconds until the PT2399 stops operating. But, when you release the switch the PT goes back to normal operation. I did this on my Loophole project a few years ago - it does not damage the PT insofar as I can tell and it sounds rad.

[forsakenrider]

Thanks for the suggestions Bean! I fixed the first error.

I don't really need longer then 200ms delay times I think as I am going for more of a stutter, but its definitely and idea to play with.

The trimmer to avoid oscillation is a good idea too. I was originally thinking of using a trimmer instead of RV1 and set it at always near "infinite" repeats, but then I got to wondering.... what about when you stack some other pedals in front of it and they start pushing a hotter signal in. Or even, what about when you roll back the volume and push a weak signal in. Wont the initial signal effect how hot it is the first time coming out?

For the buffer, I'm assuming a simple jfet buffer would do, but where to put it? I'm guessing right after the repeats pot (RV1).

[forsakenrider]

Two suggestions: put a buffer in the feedback path to keep the continuous feedback from getting too dark or filtered and make R7 a trimpot so you can dial in the precise amount of feedback without self-oscillation.

One more suggestion - to get the really long delay times, add a 100n cap in parallel with C21, tie it to lug3 of RV2 and have R20 off the RV2 wiper. Make RV2 100k. So, when you dial in those really super long sample times it adds another LP filter at the PT2399 output.

Here is my updated schematic with what I believe you were suggesting (Its getting a little messy). I didn't quite understand the "long delay times" filter, so have a look.

[anotherjim]

It's pretty clever. RV2 is doing 2 things at once with its wiper towards ground. The top end is controlling a high cut filter  that reduces the treble when the delay time is longest and will reduce digital artifacts that increase in that case. It will sound duller overall and the repeats will decay faster because you can't really cut the high treble noticably without some of the mid frequencies also getting cut a little with the simple filters used.
I'm not sure if it's drawn correcly though. One of those caps at "C21", maybe the 100n, should be going to the top of RV2 and not ground as shown?

[bean]

It's pretty clever. RV2 is doing 2 things at once with its wiper towards ground. The top end is controlling a high cut filter  that reduces the treble when the delay time is longest and will reduce digital artifacts that increase in that case. It will sound duller overall and the repeats will decay faster because you can't really cut the high treble noticably without some of the mid frequencies also getting cut a little with the simple filters used.
I'm not sure if it's drawn correcly though. One of those caps at "C21", maybe the 100n, should be going to the top of RV2 and not ground as shown?

Credit goes to culturejam on that. It's his trick

But, yeah, that new 100n cap should not go to ground, but lug 3 of Rv2. If you are going to do that, I suggest moving R20 to between pin6 and lug1 of RV2 then connect the wiper directly to ground.

On your buffer, I can't see what the value of C23 is but you probably don't need it since you already have C8. Also, if you are going to include those 2M2 resistors, you should put a coupling cap between the wiper of RV1 and the bias resistors. You could simplify though - make RV1 50kB and connect it directly to the gate of Q4, eliminate R22, R23 and C23. It'll work.

(sorry if I'm sounding bossy - just putting in my $0.15 on the idea).

[forsakenrider]

Ya I figured I had that 100n cap wrong, but now that both of you have chimed in I get it. I'll get that fixed up when I get home.

Also the simplified buffer sounds better, I wasn't really sure what was needed there.

No ones sounding bossy to me! Ideas from more experienced people help me learn more and more, that's exactly what I love about this hobby!  All ideas are welcome and appreciated, keep them coming!

[forsakenrider]

Well here's the updated schematic, with suggestions including the Detune mod (I believe it's correct).

Only thing I'm not sure how to do is the switching options. I have DPDT Momentary switches I was planning on using for the main stomp, and a toggle of sorts for the modes (On-Off-On, single or dual pole).

The Four mode now looking at the schematic:
Bypass = Switch A and Switch B HIGH, Switch C LOW.
Normal = Switch A and Switch B LOW, Switch C HIGH.
Addition = Switch A LOW, Switch B and Switch C HIGH.
Play Over = Switch B LOW, Switch A and Switch C HIGH.

I'm sure there's a way that isn't too complex I just haven't wrapped my head around it. Any ideas welcome.

[forsakenrider]

Here's an updated schematic with a ON-OFF-ON to bypass the fet switches to get the two other modes (I hope this works, seems simple), and another fet switch with an ON-OFF toggle to do the bean-detune. I think it should all work like this, please let me know if you see something amiss, I will be prototyping this shortly!

[BluffChill]

As a PT2399 fiend I'm very interested in this!

[bean]

Your Q5, soft turn-on, is a cool idea but might cause problems. If using this, I suggest socketing the 22R resistor. You may have to go down in value due to the slight added series resistance from the FET.

[forsakenrider]

Your Q5, soft turn-on, is a cool idea but might cause problems. If using this, I suggest socketing the 22R resistor. You may have to go down in value due to the slight added series resistance from the FET.

I was thinking about that too, I will definitely be prepared to try different values for r24. I will probably have to tune all of the slow switch on R/C components, but I figured I'd start with something I know works.
I was wondering if maybe I should use a mosfet for Q5 instead of the jfet, but that should be easy enough to swap around.

Hopefully I will get this thing making noises this weekend, but I've come down with a cold so that's definitely going to dampen my progress. I also just received my Uber Tuber and Dreamtime boards so it's looking like a busy weekend!

[forsakenrider]

wellllll, my jfet switches don't work as I hoped they would. I guess I kinda figure that because I've never used jfets as switches before and obviously don't quite understand whats going on. If anyone has some ideas then let me know, otherwise I'm headed for google.

Hard wiring the switches does work, however there is a massive POP when switching, which I was expecting. The "LOOP" seems pretty quiet also.....

[[WZ]]

I'm relatively new to this hobby, but once I get comfortable enough to work on PT2399 stuff, I want to try a different-but-related idea (Sorry for hijacking your thread, but considering your stutter idea maybe you'll get some inspiration from it!).
Here it goes:

Multiplexer Delay Idea:
>Take a simple delay capable of short loops/bursts (like yours), then use a CMOS Multiplexer (CD4051 or CD4052) to automatically switch between different delay times, or to automate the activation of the stutter (basically what you're gonna do manually with a momentary stomp switch).

>The Multiplexer analog switching chips can select between (up to 8 ) different channels based on (up to 3) binary "channel select" inputs.
So, now imagine driving those binary inputs with some LFO (or more complex clock division pulses).
The delay could do automated, rhythmic delay time changes; should result in weird pitch shifting and/or glitching..
Or it could simultaneously kill the Input and set Repeats to infinite, alternating every 5000ms or so; automated stutter patterns (think Malekko's Charlie Foxtrot or Hexe Revolver, but then Analog, no DSP)

===

Now before you guys are going "...the hell is this guy smoking? " I'll say that I've already had great results with a CD4051 Multiplexer Tremolo driven by a CD4040 Binary Ripple Counter, which doesn't just do the old boring on/off-type tremolo, but does all kinds of rhythmic bursts and patterns and goes all "pa-pa-pa-paaaaa-pa-paaaa" and "kirrrrrrrrrr-ak-ak-ak-ak-kirrrrrrrrrr-ak-ak-ak-ak", it's great fun! (Will start new thread soon!)

Again, sorry for hijacking, but maybe some inspiration for you to help explore the stutter idea!

[KoenT]

was this project ever finalised? Looks interesting.