PT-80 delay insert/mods, help

[naja]

hello dear people, i need your help. of course i did some research, but now i am stuck and i need some clarification.

i have decided to do some mods on the PT-80 delay;
i purchased a PCB from a guy who already did some small mods (i am attaching his schematic), and i want to add an fx loop in the regeneration/feedback path (send and return),
as well as low pass and high pass in the wet signal chain

this is the the original schematic including some notes/ideas:


and here comes my modded schematic, which includes a simplified representation of the original schematic.
here, all the parts that were already in the original schematic are labelled exactly the same, all new components are labelled with a consecutive index starting at 101.
the original schematic i cannot change too much, because i already have the PCB - i am putting my mods where there are cables anyway, and on an external little board.
this is it:


for the low pass and high pass:
what i want is to change the tone of the repeats,
so i thought about putting the filters before the signal splits to output and feedback path
i did some research, i considered the "New AMZ Tone" control for a one-knob control, i also considered an active Baxandall circuit,
yet now i am using a simple low pass and high pass, buffered and in series.
doing so, i hope to be able to also bandpass-filter the signal and change the cutoff frequencies of high or low pass - the Baxandall and the New AMZ Tone both have some sort of fixed center frequency.
i placed those op-amps so that both filter stages don't interact.
1) i wonder, if i need those two buffering op-amps and if they are sitting at the right spot.
     also, hope i didn't do anything stupid...
ah, by the way, i used capacitor C29 from the original schematic as part of the high-pass  that's the reason, i put the high pass first

concerning the send and return path:
first of all, i want to have the possibility to get the wet signal out of the send jack and the dry signal out of the output jack (therefore i would lower the "Wet Level" pot to zero),
i could use this for a stereo slapback effect or other means.
therefore, its directly connected to the send jack and is not disconnected when plugging in a cable.
the other thing i want is the typical insert: put some fx unit into the regeneration/feedback path, for various effects, so the return jack is disconnected when pluggin in a cable.
i don't know if i need buffers here for the send and return, i read different opinions. so i placed these two.
2) do i need both? also: are they at the right position?

i have to explain why in the low pass op-amp feedback, i do a gain of x2: this is for some +6dB, which i can attenuate with the "Repeat" pot to go back to 0dB,
but once i heavily filter in the high pass and low pass stage, i will have to make up some of the loss.
3) don't know if the resistors in all those buffers/op-amp configurations have the right values and if forgot about anything?
4) i also wonder about the input impedance of the return buffer: is it too low?
     i'm thinking: no guitar will be plugged into the return, just the outputs of whatever fx unit i put into the insert, so it should be fine, right?

then there is the thing with the pre-/de-emphasis. as far as i understood, input buffer is raising the highs, output buffer is lowering them, typical pre-/de-emphasis situation.
i don't know how hard the pre-emphasis really is, and when using send and return jacks, it will get de-emphasized again.when putting the inserts like this,
5) i will put the pre-emphasized (high-shelved) signal out of the jacks. that may be a problem, right?
     but when using send as a wet-out, will the wet signal sound way too bright?
i mean, again, i could attenuate this with my filter stages.

another question i have is about the phase/polarity:
the original schematic has a non-inverting input buffer and an inverting output buffer,
and then there's stuff happening with the inverting/non-inverting inputs ("RESR3"?) and the outputs of the compander unit.
i don't know how the compander works, so i decided to not change anything here and, in my modded schematic, draw it simplified.
6) is it true, that the wet signal is out of phase with the dry signal here? and that the dry signal is inverted?
7) does it matter?
i read a post of a guy who was wondering if the author of the original schematic made a mistake at the output de-emphasis.
i substitued the 47p (C32) of the original schematic with a 470p (C104), to match the pre-emphasis.
8 ) thats fine, right?

as a little extra, i added an input-kill-switch and a momentary "dub" delay switch,
the idea behind those is that if the delay is switched on (there will be a simple on/off true bypass, also), i can mute the delay input with the input-kill-switch
and then with a momentary footswitch be able to just let certain parts of the input signal into the delay (for example, just a snare hit, hence the "dub" name)
9) i hope these are at the right position and the dub momentary switch won't make pops/clicks too easily!?

the last questions: i am using a lot of coupling capacitors. the pots always reference to GND, the op-amps all reference to VGND, of course (single-supply amps).
10) is there any mistake i'm doing? using too many coupling caps?
and:
11) is there a reason i cannot simply reference all those pots to VGND and spare me some coupling caps?

i will also be adding an LFO, but that is another story and there are already some posts discussing that

i am very thankful for any help/suggestions, and to those who read this long story,
best wishes, jan

[ElectricDruid]

1) i wonder, if i need those two buffering op-amps and if they are sitting at the right spot.

Yes, the buffers look fine. But R104 is redundant, since you've got the 10K pot doing the same job. There's no need for a separate bias resistor.

i don't know if i need buffers here for the send and return, i read different opinions. so i placed these two.
2) do i need both? also: are they at the right position?

Whether you "need" them depends what you plug into the loop, and that's an unknown, so the safe thing is to include them. Why use inverting buffers though? Non-inverting unity-gain buffers would give you a better input impedance.

i have to explain why in the low pass op-amp feedback, i do a gain of x2: this is for some +6dB, which i can attenuate with the "Repeat" pot to go back to 0dB but once i heavily filter in the high pass and low pass stage, i will have to make up some of the loss.

Yes, seems reasonable to me. You *will* lose signal when you start hitting it with the filters.

3) don't know if the resistors in all those buffers/op-amp configurations have the right values and if forgot about anything?

Your 10K/470p combination gives a roll-off above 33KHz, so that's safe enough. I can't see anything that immediate makes me shout.

4) i also wonder about the input impedance of the return buffer: is it too low?

Aah, now I mentioned this earlier..maybe swap both send and return buffers for non-inverting.

i'm thinking: no guitar will be plugged into the return, just the outputs of whatever fx unit i put into the insert, so it should be fine, right?

"Probably..." Like I said, you don't know what will be in the loop, so it's best to make sure it's safe for anything.

then there is the thing with the pre-/de-emphasis. as far as i understood, input buffer is raising the highs, output buffer is lowering them, typical pre-/de-emphasis situation.

Yes, exactly.

i don't know how hard the pre-emphasis really is, and when using send and return jacks, it will get de-emphasized again.when putting the inserts like this,
5) i will put the pre-emphasized (high-shelved) signal out of the jacks. that may be a problem, right?

This is where I'd say that a specific design decision is "character" rather than a "problem". Yes, the send signal will have boosted treble, but who's to say whether that's a problem? It'll change the character of effects in the loop (and probably give good results with passers or flangers which like plenty of high frequencies) but the highs will also get trimmed again by the de-emphasis in the output mixer. I'd give it a try. This is an art as well as a science!

but when using send as a wet-out, will the wet signal sound way too bright?[/b]
i mean, again, i could attenuate this with my filter stages.

Exactly! So why worry?!

6) is it true, that the wet signal is out of phase with the dry signal here? and that the dry signal is inverted?

Since the wet signal is delayed, the phase is a bit of an open question. A delay of 1msec is 360 degree phase shift for a 1KHz frequency, after all. So when you start putting 100's of milliseconds of delay in things, what "phase" the output is depends entirely on the delay and not much on how many times you flipped it.
The dry signal *is* inverted with respect to the input though, and that *could just about* be serious if you were to put this pedal in parallel with something else that didn't invert the phase.

7) does it matter?

I'd say it's "good practice" to not invert the phase, but I wouldn't lose sleep over doing it if it saved me sticking an extra op-amp chip in a circuit.

i read a post of a guy who was wondering if the author of the original schematic made a mistake at the output de-emphasis.
i substitued the 47p (C32) of the original schematic with a 470p (C104), to match the pre-emphasis.
8 ) thats fine, right?

It looks ok to me, but I haven't checked it in detail. I'd expect to see the two stages being "mirror images" like you have it, but sometimes people use different-but-equivalent values in the two stages which makes that much less obvious.

Ok, I've got to stop for a break!

T.

[naja]

Ok, I've got to stop for a break!

Wow, thanks a lot, ElectricDruid! The break you truly deserve

Yes, the buffers look fine. But R104 is redundant, since you've got the 10K pot doing the same job. There's no need for a separate bias resistor.

yeah, i see. as they are parallel, the 1M won't do much...

Whether you "need" them depends what you plug into the loop, and that's an unknown, so the safe thing is to include them. Why use inverting buffers though? Non-inverting unity-gain buffers would give you a better input impedance.

you're right, i did this to have the output buffer for the send the same as the output mixer, just in case i would ever mix a wet signal from both, output and send, together and to have them in phase in that case. but that will be the case very rarely, or more like: never. and it may even be something nice: to have a phase-inverted and non-phase-inverted output, maybe in some case it can be used for something. i just changed it for non-inverting unity-gain buffers - i still need R106 though, don't i?

with R105 = 1M, can i swap C105 for something smaller now? like, make it 47n instead of 1u?

also, i changed the position of the long connection to the return buffer: its now taking the signal -before- the send buffer, not after the buffer. i can do this now, because of the non-inverting buffers.
but will it still work the same? i mean, the send buffer now is driving only what is connected outwards, so it should be fine, right?

i also made the return buffer non-inverting, can i wire it up the same as the the input (except for the feedback loop, of course)?

i have to explain why in the low pass op-amp feedback, i do a gain of x2: this is for some +6dB, which i can attenuate with the "Repeat" pot to go back to 0dB but once i heavily filter in the high pass and low pass stage, i will have to make up some of the loss.

Yes, seems reasonable to me. You *will* lose signal when you start hitting it with the filters.

i was just thinking: would you suggest 2nd-order-filters or would you go with these simple, 1st-order -6dB/oct? i'm asking because i'd like to have the option of making the repeats quite dark.

I'd give it a try. This is an art as well as a science!  

true! will try

this is the new schematic so far:

does it work like this?

the "only" questions that are left now are:

as a little extra, i added an input-kill-switch and a momentary "dub" delay switch,
the idea behind those is that if the delay is switched on (there will be a simple on/off true bypass, also), i can mute the delay input with the input-kill-switch
and then with a momentary footswitch be able to just let certain parts of the input signal into the delay (for example, just a snare hit, hence the "dub" name)
9) i hope these are at the right position and the dub momentary switch won't make pops/clicks too easily!?

i am using a lot of coupling capacitors. the pots always reference to GND, the op-amps all reference to VGND, of course (single-supply amps).
10) is there any mistake i'm doing? using too many coupling caps?

11) is there a reason i cannot simply reference all those pots to VGND and spare me some coupling caps?

i will also be adding an LFO, but that is another story and there are already some posts discussing that

yet, i think i might take the chance to ask about this as well as i modified a little bit the tremulus lune LFO,
so now i dare to ask about this as well, hoping it won't scare you all away, asking so many questions...
just shortly:
sw1 - choose between triangle/up ramp/down ramp
sw2 - choose the LFO range. this is important to me, because i want to go very high also, to get the LFO into audio range, like 60 Hz or something (AM effect).
sw3 - choose between triangle and rectangle
this is the circuit:

will it work like this? i will breadboard it soon, anyway, but if you see anything problematic please let me know.
i had the feeling it got unstable when i once breadboarded it, but maybe i made some mistakes...

thanks a 100 times

[naja]

concerning question 9): of course it may pop like this, i'm sorry i even asked. i added two 1M resistors to ground to prevent any pops that may occur if there was a DC buildup one one of both ends once i have opened the kill switch.
is R110 redundant?

updated schematic:


if someone could take a quick look on my last post i'd be very happy because i could start building right away,
best wishes, jan

[naja]

ok, i will narrow down my question to the last two most important ones:

1) is there a reason i cannot simply reference all those pots to VGND and spare me some coupling caps?

2) i'd like to use ValveWizard's pin-6 latch-up protection together with one of the standard approaches of LFO modulating, including a "Time" potentiometer
http://www.valvewizard.co.uk/jennygreenteeth.html
my goal: to have delays somewhat between 20ms and 700ms
one thing i did was raising the output voltage of 78L05 from 5V to 6V (to achieve faster delays generally):

the other is said latch-up protection approach that i cannot yet realise because i don't fully understand the pin 6 current modulation
i posted a similar question here already:
https://www.diystompboxes.com/smfforum/index.php?topic=120528.0

my LFO:


schematic including a latch-up protection approach (bottom right): does it work like this?:


or maybe, alternatively, somehow like this?:


(you see i don't yet fully get it, but i'm trying)

[ElectricDruid]

1) is there a reason i cannot simply reference all those pots to VGND and spare me some coupling caps?

I'm guessing you mean Vcc/2 and no, there's no reason. You'd be better off keeping the same bias level all the way through the circuit, and then you don't need to bother with the coupling caps until you get to the output (which will have to be referenced to ground).

2) i'd like to use ValveWizard's pin-6 latch-up protection together with one of the standard approaches of LFO modulating, including a "Time" potentiometer
http://www.valvewizard.co.uk/jennygreenteeth.html
my goal: to have delays somewhat between 20ms and 700ms
one thing i did was raising the output voltage of 78L05 from 5V to 6V (to achieve faster delays generally):

PT2399 datsheet says maximum Vdd of 6.5V, so you're still safe at 6V. No problem there.

the other is said latch-up protection approach that i cannot yet realise because i don't fully understand the pin 6 current modulation
i posted a similar question here already:
https://www.diystompboxes.com/smfforum/index.php?topic=120528.0

I put a brief answer in the other thread, but to add to that, my understanding of the latch-up problem is that the PT2399 VCO doesn't start up correctly if the pin6-ground current is too high (e.g. if the delay is too short). This can be a problem for chorus designs and similar projects which are trying to use the chip for very short delays. For delay pedals, you can just put a resistor in series with your delay pot so the minimum delay is still safe and there's no problem.
The Jenny Greenteeth design uses a transistor to make sure the current is switched off initially. There will be no base current on the transistor, so no current will flow from collector to emitter. When the power comes on, the cap will charge up gradually, and the transistor will switch on, allowing the other transistor to modulate the current. It's just a way of making sure there's no big current draw on pin6 until it's got running.

HTH,
Tom

[naja]

hello Tom, thanks again. i'm sorry to stress your helpfulness, but unfortunately i'm not there yet.

i understand what you wrote already, i could read through a lot of forum posts where people talk about this matter and i read your very useful "Useful design equations for the PT2399". in most posts, people are discussing some aspects, not analyzing the circuits in detail, which is maybe a bit what i need due to a lack of experience

what i know so far:
- current out of pin 6 determines delay speed
- pin 6 provides 2.5V, but not always, as it can get dragged down with increased current (~ 150mV)
- if PT2399 is supplied by 6V, pin 6 will provide 3V (Vcc/2)
- on power-on, pin 6 must not exceed 2.5mA

The Jenny Greenteeth design uses a transistor to make sure the current is switched off initially.

this also i understood and tried to add to my schematic

For delay pedals, you can just put a resistor in series with your delay pot so the minimum delay is still safe and there's no problem.

my delay should have the shortest delays possible. so the "safe minimum delay" is no option in my case, ill go with the latch-up protection. i'd like to go lower than 30ms, but i don't know if that's possible

my problems:
- which approach to choose: bipolar transistor as current sink or applying a voltage directly. pros/cons?
- once i decided one of these two approaches, how to set it up properly
- to really understand what currents/voltages i want and where

lets say i'd like to use only one transistor for the latch-up protection. so i'd choose the voltage solution.
let me try to think this through (i will probably fail), using this latest version of my circuit:

say my LFO is outputting a signal between 2V and 3V (with "Depth" on 100%), then 2V-3V will also appear right to R122 (only high-pass filtered?).
if PT2399 pin 6 has 3V and the LFO is at 2V, then
the current out of pin 6 will be 0.02mA ((3V-2V)/50kOhm) in case either the transistor is still closed,
0.04mA ((3V-2V)/25kOhm)  in case the transistor is open the "Time" pot is at 100% (because R123|R124 = 25k),
and 6.6mA ((3V-2V)/150Ohm) in case the transistor is open and "Time" is at 0%.
the current will be 0mA if the LFO is at 3V ((3V-3V)/x Ohm).
did i get this right so far?

- if i want to go as short as possible without damage: what is the max current i should have on pin 6? 10mA? 20mA?
- do i have to make sure the LFO is not giving more than 3V and not much less than 2V?
- now if i turn the "Depth" pot down i will go close to 0V, so this whole approach doesn't work anymore?
  for example if LFO output at low "Depth" is around 20mV, "Time" at 0%, the current will be around 20-30mA?
  what to do? how to solve?
- what are C112 and R122 doing exactly? just high-pass filtering? current limiting?
- do i have to care about R121?

i hope to accomplish a better understanding of this part of the circuit soon,
both to get started building and to finally not steal any more of your time,
being a bit uncertain about whether i can ask for so much detailed help,

best wishes

[ElectricDruid]

my delay should have the shortest delays possible. so the "safe minimum delay" is no option in my case, ill go with the latch-up protection. i'd like to go lower than 30ms, but i don't know if that's possible

The shortest delay possible seems to vary a little bit from chip to chip, but it's generally in that area. Much shorter won't happen, or everyone would be making simple flanger pedals using the PT2399 and not a lot else.

- which approach to choose: bipolar transistor as current sink or applying a voltage directly. pros/cons?

Voltage directly is simpler if you can get the LFO voltage in the right range and it works. I haven't tried that approach, so I don't know.

- once i decided one of these two approaches, how to set it up properly
- to really understand what currents/voltages i want and where

If you've read my page on the topic, you already know what I know about it. There's not much I can add to that.

lets say i'd like to use only one transistor for the latch-up protection. so i'd choose the voltage solution.
let me try to think this through (i will probably fail), using this latest version of my circuit:

say my LFO is outputting a signal between 2V and 3V (with "Depth" on 100%), then 2V-3V will also appear right to R122 (only high-pass filtered?).

If the LFO is in the right voltage range (2-3V) then why do you need the capacitor at all? We only use coupling caps when the bias level differs between one part of the circuit and another. In this case, you need to avoid that.

if PT2399 pin 6 has 3V and the LFO is at 2V, then
the current out of pin 6 will be 0.02mA ((3V-2V)/50kOhm) in case either the transistor is still closed,
0.04mA ((3V-2V)/25kOhm)  in case the transistor is open the "Time" pot is at 100% (because R123|R124 = 25k),
and 6.6mA ((3V-2V)/150Ohm) in case the transistor is open and "Time" is at 0%.
the current will be 0mA if the LFO is at 3V ((3V-3V)/x Ohm).
did i get this right so far?

It all seems reasonable to me. A couple of queries though: What is the purpose of R123? Why is it there? And secondly, have you thought about the 150R? The LFO can put 3V across it, which is 20mA - not a huge amount, but a reasonable draw nonetheless.

- if i want to go as short as possible without damage: what is the max current i should have on pin 6? 10mA? 20mA?

The VCO will stop before you get to either of those currents, I would expect.

- do i have to make sure the LFO is not giving more than 3V and not much less than 2V?

Why do you need to make sure it doesn't go below 2V? Above 3 would be bad, I agree, but since the pin expects to be grounded, I don't see why taking the voltage below 2V is a problem.

- now if i turn the "Depth" pot down i will go close to 0V, so this whole approach doesn't work anymore?

See my previous answer. What's the problem with being close to ground? But yes, if the "depth" depends on the distance from 3V, then you need to think about how the depth pot is wired.

for example if LFO output at low "Depth" is around 20mV, "Time" at 0%, the current will be around 20-30mA?
  what to do? how to solve?

I don't know. I don't think that's going to work, because there's no way the PT2399 is going to be happy with 20-30mA coming out of pin 6. It'll stop working and lock up, if the magic smoke doesn't escape.

- what are C112 and R122 doing exactly? just high-pass filtering? current limiting?

I don't know. I don't see the point of them.

- do i have to care about R121?

I would be. It's loading the pot. When the LED is on (e.g. when the LFO is above the Vf of the LED) then R121 is effectively in parallel with the bottom half of the pot. And that's a simplification!

i hope to accomplish a better understanding of this part of the circuit soon,
both to get started building and to finally not steal any more of your time,
being a bit uncertain about whether i can ask for so much detailed help,

You're welcome. We're all here to learn, and if I can help a bit, I'm glad to do it.

T.

[naja]

@ElectricDruid: i've re-read your answers and questions couple of times, thanks a lot! - but some i just didn't understand.
i went with the Dreamtime Delay circuit, because when breadboarding, all these circuits that apply LFO voltages directly to pin 6 via resistors didn't work

is it possible to look at my LFO circuit and compare it with the approaches i posted and then tell me what i didn't fully understand?
or, say, if i wanted to do it without any transistor, how would i do it, using also a latch-up protection (okay, i would need a transistor here at least)?
if i don't use a capacitor from LFO to PT2399, i will have to get my LFO output voltage into a 0-2.5V range, which is impossible with a voltage divider, because the voltage divider again would change the current from pin 6...

[naja]

btw i built the "Dreamtime Delay" LFO mod circuit, which works,
but the shortest delay times i get, even when changing the 560 ohm to 100 ohm, is 60 ms, which is way too long.
any idea?