Why this noisy delay is clean only when it's plugged in the effect loop ?

[eh la bas ma]

Hello,

I built a Pirate Shift some time ago (march 2019). The delayed signal is clean at short range, but gets noisy when the Delay control is set past 9 o'clock. It is like a "pssshhhhh" sound mixed with the repeats.  Here is a demo I recorded to be more specific :

https://soundcloud.com/ehlabas-ma-chaussure/pirate-shift-low-input-test-3wav

building docs : https://pcbguitarmania.com/wp-content/uploads/2018/02/Pirate-Shift.pdf

I asked about this issue to pcbguitarmania : I've been answered that the PT2399 is clipping the signal because it gets "too much input ".

I guessed it means the guitar signal is too loud. I tried with the guitar's volume set on minimum, I tried with a passive electric guitar and a bass at low volume without success. I even tried 3 other PT2399 chips, unfortunatly the delay is still noisy.

Then I thought "too much input " might mean an impedence problem. If I understand buffers correctly, it is designed to correct impedance issues, so I tried the pedal with an OCD, a Klon clone (both buffered effects) and a Ge-7 (Boss EQ) before the Pirate Shift, but it's still noisy.

Recently I tried to plugg my guitar (active hambuckers pickups) directly in the amp ("High" input), with  the Pirate Shift in the effect loop, and I finally got rid of some of the background noises. In the effect loop, the pedal works almost perfectly, even with the guitar's volume control set on high values.

You can hear it here : https://soundcloud.com/ehlabas-ma-chaussure/pirate-shift-in-effect-loopwav

My questions are :

What could possibly happen with the amplifier's effect loop that corrects the issue I have with the Pirate Shift 's delay ?

Is there anything I can do to "repair" the circuit so it can work perfectly when I plugg it in front of the amp ?


Thank you very much for your help.

[eh la bas ma]

I found this document by the Valve Wizard about Pt2399 :

http://www.valvewizard.co.uk/PT2399_Data_Notes.pdf

"Noise is minimised by heavily filtering the audio, particularly at the output of the chip. For the longest delay times it may be necessary to severely limit the bandwidth to 1kHz or less"

Does someone has some idea about how I can " limit the bandwidth to 1kHz or less " ?

I guess I should use a capacitor in series, added between the pt2399's output and the next component in the circuit : between pin 12 (OP2-Out) and R2 on the Pirate Shift.

Which value would " limit the bandwidth to 1kHz or less" ?

I also found an alternative solution on an old thread from madbean's forum. Some people managed to fix the issue by:

"(...) connecting the delay resistor between Pins 6 and 4. That's it. No more "SSHhhhh" until the signal dies out. Awesome stuff!!!"

https://www.madbeanpedals.com/forum/index.php?topic=9041.0

I don't understand what is the "delay resistor". On the Pirate Shift's schematics, pin 6 (VCO) goes to Depth and Delay pots...pin 4 (DGND) is connected to pin 3 (AGND)...

[bluebunny]

"(...) connecting the delay resistor between Pins 6 and 4. That's it. No more "SSHhhhh" until the signal dies out. Awesome stuff!!!"

I don't understand what is the "delay resistor". On the Pirate Shift's schematics, pin 6 (VCO) goes to Depth and Delay pots...pin 4 (DGND) is connected to pin 3 (AGND)...

The delay time is determined by a resistance from pin 6 to ground.  Since pin 4 is more or less ground, some schematics connect the resistance between pins 4 and 6.  (Anything else connected to pin 6 is likely to be modulation circuitry.)

[Elektrojänis]

I'd say the problem is not too much input when you plug your guitar in to the effect... If it would be, it would distort.

I think the difference here with the effect in front of the amp vs. in the effect-loop is gain staging.

When you have the effect in front of the amp, all the gain (amplification) of the amp comes after the effect. So all the noise the effect outputs will be amplified by all the gain of the amp. When you place the effect in the loop, the preamp in the amplifier amplifies the signal before the effect. The effect is then fed a bigger signal (more input, louder). The noise level from the effect will be the same but the actual signal will be bigger, so the ratio between the actual signal and noise will be better. After the effect loop the signal and the noise will go to the poweramp and will be amplified more, but the total amplification for the noise will be less than when the effect is in front of the amp.

[percyhornickel]

I found this document by the Valve Wizard about Pt2399 :

http://www.valvewizard.co.uk/PT2399_Data_Notes.pdf

"Noise is minimised by heavily filtering the audio, particularly at the output of the chip. For the longest delay times it may be necessary to severely limit the bandwidth to 1kHz or less"

Does someone has some idea about how I can " limit the bandwidth to 1kHz or less " ?

I guess I should use a capacitor in series, added between the pt2399's output and the next component in the circuit : between pin 12 (OP2-Out) and R2 on the Pirate Shift.

Which value would " limit the bandwidth to 1kHz or less" ?

I also found an alternative solution on an old thread from madbean's forum. Some people managed to fix the issue by:

"(...) connecting the delay resistor between Pins 6 and 4. That's it. No more "SSHhhhh" until the signal dies out. Awesome stuff!!!"

https://www.madbeanpedals.com/forum/index.php?topic=9041.0

I don't understand what is the "delay resistor". On the Pirate Shift's schematics, pin 6 (VCO) goes to Depth and Delay pots...pin 4 (DGND) is connected to pin 3 (AGND)...

The way I see this:

-There is a 3rd order Multiple Feedback Low-pass Filter (pins 15/16)  before entering from the feedback filtering (pins 13/14) which is...     ...another 3rd order Multiple Feedback Low-pass Filter.

You can calculate the frecuencies using this tool:

http://sim.okawa-denshi.jp/en/Multiple3tool.php

C1 is 10n in this case, and 27n in rebote 2.5 which uses the same topology...   ...some others uses a little les complicated topologies (one resistor less):

http://sim.okawa-denshi.jp/en/OPtazyuLowkeisan.htm

So you can calculate the frecuencies you are working with and the cuts you will enter in the PT2399 from the feedback line and the frecuencies comming from the first buffer.

Saludos

[Radical CJ]

I'd say the problem is not too much input when you plug your guitar in to the effect... If it would be, it would distort.

I think the difference here with the effect in front of the amp vs. in the effect-loop is gain staging.

When you have the effect in front of the amp, all the gain (amplification) of the amp comes after the effect. So all the noise the effect outputs will be amplified by all the gain of the amp. When you place the effect in the loop, the preamp in the amplifier amplifies the signal before the effect. The effect is then fed a bigger signal (more input, louder). The noise level from the effect will be the same but the actual signal will be bigger, so the ratio between the actual signal and noise will be better. After the effect loop the signal and the noise will go to the poweramp and will be amplified more, but the total amplification for the noise will be less than when the effect is in front of the amp.

This would be my guess as well. Depending on the filtering in the pre-amp it may be emphasising the frequencies at which the noise is sitting.

I have a similar situation with a NE555 based squarewave tremolo I concocted that has a slight ticking problem when plugged into the pre-amp of my Marshall, but when placed in the effects loop (and blended in the with effects loop's dry/wet blend) has no tick.