Yet another (*yawn*) PT2399 question: Where to input my modulation signal?

[KarenColumbo]

Different layouts, different techniques:

Some feed the LFO signal into pin 6 ...


... some seem to prefer pin 2 ...


... while some use pin 3 - I forgot which circuit that was, but it exists  Ah! This one: http://www.diystompboxes.com/smfforum/index.php?topic=120165.0

Why? What are the differences? Could some of you PT2399-experts shed some light on this?

[Kipper4]

Thats pin6 in the vomit comet mate.
pin3 is Gnd

[KarenColumbo]

Yeah, wrong schematic, sorry. I stand corrected.

[ElectricDruid]

Pin 6 is the VCO control pin. It seems to be the case (Princeton Tech don't actually tell us this) that the current drawn from this pin to ground sets the VCO frequency and hence the delay time. There are various ways to get modulated control over that current, including vactrols or basic LED/LDR, FETs, or current sinks. The Jenny Greenteeth BJT can be regarded as a very basic current sink. But note that only one of the transistors is there for modulation - the other one is to prevent the chip latching up at start-up. The PT2399 can be fussy about having a low resistance on pin 6 initially and if the resistance is too low, the VCO clock doesn't start-up correctly.

Pin 2 is completely different. It's supposed to be a 2.5V reference voltage for the internals of the chip. For some reason I don't understand, modulating that voltage also alters the delay time.

That's all I know.

[KarenColumbo]

Thanks, Tom, for bearing with me  It seems, this PT thing is a pandora's box. Well, off to the breadboard. I already killed a few, so what the heck ...

By the way (and a bit off-topic): If I want to phase-reverse the Stompflo modulation signal, will an inverting opamp buffer suffice?

[Kipper4]

I just have one thing to add.
If using mod on pin2 of pt2339, you can leave the smoothing cap in (47uf Jenny greenteeth).
Im using a 10k+ series resistor from StompLFO pin3 to pt2339 pin2. Anything smaller than 10k it starts to thump a little.

[mth5044]

... while some use pin 3 - I forgot which circuit that was, but it exists  Ah! This one: http://www.diystompboxes.com/smfforum/index.php?topic=120165.0

That's also not pin 3. Hard to tell because the schematic is not high resolution enough to read the PIN numbers, but because of the surrounding components and the gray VCO label, that is pin 6. The bottom side of the pt2399's in that schematic are not in numerical order for some reason.

Electric Druid nailed it, you basically have the LFO wiggling delay time, or the relatively unknown effects of pin 2 modulation.

[ElectricDruid]

By the way (and a bit off-topic): If I want to phase-reverse the Stompflo modulation signal, will an inverting opamp buffer suffice?

You might struggle a bit, actually. The reason being that your average common-or-garden op-amp won't be able to go right down to 0V, whereas the StompLFO's input signal does go right down to 0V. So you need an offset to get it into a better range for the op-amp. If we can shift our signal 2V upwards, our 0-5V input will produce an output from 2V to 7V. That's within the range that the op-amp can easily cope with and there's no problem. We can easily stick the bias voltage into the +ve input of the op-amp. But somewhat counter-intuitively, the required bias is *not* 2V, or even 4.5V.


This paper from MIT about biasing an op-amp tells us what we need to know. We're dealing with a DC-coupled signal, and an inverting configuration. The equation for it is the following:

Vbias = Vout + (G x Vin) / (G + 1)

Vin and Vout are the input and output voltages, and G is the gain. Our op-amp is inverting and has a gain of 1. When we put 0V in from the StompLFO, we want 7V out (it's inverting, remember). When we put 5V in, we want 2V out. Either of these points can be used to calculate the Vbias:

Vbias = 7V + (1 x 0V) / (1 + 1) = 7/2 = 3.5V
Vbias = 2V + (1 x 5V) / (1 + 1) = 7/2 = 3.5V

So stick a potential divider that gives you 3.5V on the +ve input, and you should be fine. As a bonus (?) you finish up with a signal which is more like the standard schmitt/integrator LFO, so it'll be a drop-in replacement. Put a little cap on your feedback resistor too (cutoff around 500Hz maybe?) to help smooth the LFO signal.

I've got a half-finished article about this for my website. I should really try and get it online.

[Kipper4]

Can it get near enough with an lm358?

[PRR]

What opamp?

StompLFO output is PWM, zero to +5V. To get the inverse, rig a PNP off the +5V to make a +5V to zero version of the StompLFO's main output. One transistor, two resistors. (Or a CMOS inverter, if you have such onboard already.)

The opamp is mildly stronger but not simpler, and most "audio" opamps don't swing to zero well. It can be done (LM324), but the PNP is sure to work.

[KarenColumbo]

Paul: thank you! That answered a lot of unasked questions, too.

[KarenColumbo]

By the way (and a bit off-topic): If I want to phase-reverse the Stompflo modulation signal, will an inverting opamp buffer suffice?

You might struggle a bit, actually. The reason being that your average common-or-garden op-amp won't be able to go right down to 0V, whereas the StompLFO's input signal does go right down to 0V. So you need an offset to get it into a better range for the op-amp. If we can shift our signal 2V upwards, our 0-5V input will produce an output from 2V to 7V. That's within the range that the op-amp can easily cope with and there's no problem. We can easily stick the bias voltage into the +ve input of the op-amp. But somewhat counter-intuitively, the required bias is *not* 2V, or even 4.5V.


This paper from MIT about biasing an op-amp tells us what we need to know. We're dealing with a DC-coupled signal, and an inverting configuration. The equation for it is the following:

Vbias = Vout + (G x Vin) / (G + 1)

Vin and Vout are the input and output voltages, and G is the gain. Our op-amp is inverting and has a gain of 1. When we put 0V in from the StompLFO, we want 7V out (it's inverting, remember). When we put 5V in, we want 2V out. Either of these points can be used to calculate the Vbias:

Vbias = 7V + (1 x 0V) / (1 + 1) = 7/2 = 3.5V
Vbias = 2V + (1 x 5V) / (1 + 1) = 7/2 = 3.5V

So stick a potential divider that gives you 3.5V on the +ve input, and you should be fine. As a bonus (?) you finish up with a signal which is more like the standard schmitt/integrator LFO, so it'll be a drop-in replacement. Put a little cap on your feedback resistor too (cutoff around 500Hz maybe?) to help smooth the LFO signal.

I've got a half-finished article about this for my website. I should really try and get it online.

Ha! Very cool, as usual, Tom! Will try this today!

[Kipper4]

Something like this Paul?

[PRR]

> Something like this Paul?

It's only a brain-spam. Let's see what Tom W thinks.

[ElectricDruid]

Personally, I'd have done it with a NPN and a pull-up (so the other way up) but I don't suppose it makes much odds. Will the transistor handle the switching speed? The output pulses are at 2MHz, and slow edges show up as LFO waveform distortion.

T.

[KarenColumbo]

I have postponed this now - just did the whole thing twice, two StompLFOs, two PTs. Sounds fine as it is. Another time I will switch between 2nd LFO or added phasereversed LFO.

[merlinb]

Different layouts, different techniques:
... while some use pin 2 - I forgot which circuit that was,
Why? What are the differences?

Pin-6 is a curent source whereas pin-2 is a resistive input, so you have to think about different approaches to driving the modulation signal into either pin.
Modulating pin-2 will eat headroom. If you want modulated clipping to go with your modulated delay then this could be an interesting effect. If not, it's a defect, so modulate pin-6 instead.

[telebiker]

I just have one thing to add.
If using mod on pin2 of pt2339, you can leave the smoothing cap in (47uf Jenny greenteeth).
Im using a 10k+ series resistor from StompLFO pin3 to pt2339 pin2. Anything smaller than 10k it starts to thump a little.

I believe you're talking regarding pt2399, not the pt2339? Just want to be sure.

Also, I'm thinking, will this work with Echo Base Delay from musicpcb?

Not sure what does "Analog reference voltage (1/2VCC)" mean (from the chip's datasheet).

[ElectricDruid]

I just have one thing to add.
If using mod on pin2 of pt2339, you can leave the smoothing cap in (47uf Jenny greenteeth).
Im using a 10k+ series resistor from StompLFO pin3 to pt2339 pin2. Anything smaller than 10k it starts to thump a little.

I believe you're talking regarding pt2399, not the pt2339? Just want to be sure.

Yes, that's a typo. He means PT2399.

Also, I'm thinking, will this work with Echo Base Delay from musicpcb?

Should do. All the stuff everyone has described here would work on any PT2399 delay.

Not sure what does "Analog reference voltage (1/2VCC)" mean (from the chip's datasheet).

The PT2399 runs on a 5V supply, not 9V like the rest of your typical stompbox circuit. So the PT2399 can't use the usual 4.5V bias/reference voltage. That might be a midpoint voltage for 9V, but for 5V it only gives you half a volt headroom! So instead you need a 2.5V reference voltage - that's what they're talking about. It's 1/2 of Vcc, so 0.5 x 5V = 2.5V.