Help eliminating ICL7660SCPAZ Whine

[kenjib]

Hello, I am building the Geofex ABY splitter with isolation transformers. I also modded it for polarity switching on both channels. I am also adding an ICL7660SCPAZ circuit to get the -9V for the op amp that the schematics require. I seem to be getting the dreaded 7660 high pitched whine and can't figure out how to get rid of it. I have tried a few things. Can anyone help explain to me what I can do to eliminate the noise?

Here is the noise I am getting:
https://www.aronnelson.com/DIYFiles/up/Aby_noise.mp3

Here is the original circuit:
http://www.geofex.com/FX_images/TransformerSplitter.pdf

Here is the vero layout I made for the circuit (warning -- there is a lot of off-board wiring):
https://www.diyschematics.com/schematics/upload/2023/06/26/20230626030506-9f7d831c.png
(Note that Pin 6 of the ICL7660 is lifted off the board so that it does not connect, so this pin does *not* connect to ground)
I also used TY-250p for the transformers instead of the specified ones.

The two things I added in the vero are:
1. dpdt switches for the outputs of the transformers for the purpose of phase switching.
2. The ICL7660SCPAZ circuit to invert the 9v power supply.

Some things that I have noticed:
1. Other than the whine, the pedal seems to work in every other respect.
2. The whine is present on all three outputs: direct out, output A, and output B.
3. If I plug the pedal output into an amp with nothing plugged into the input, the whine gets really loud. It quiets when an instrument is plugged in but is still faintly present.
4. I checked for any solder overflow issues or bad cuts and the connectivity on my board all seems correct everywhere.
5. To try to limit the number of variables and make sure that the ICL7660 was the problem I tried lifting both of the connections circled in green so that the audio signal never gets to the switching and only goes to direct out. I then test using the direct out line after the 1k resistor and get the same sound that I used to get out of the complete circuit. So I think the whine is coming from the ICL7660 and going out through the op amp.
6. I have tried subbing out for a different ICL7660 chip to make sure the chip isn't bad. I still get the whine after swapping. I suppose it could be an entire bad batch though...
7. I also swapped out the LM833 op amp and put in an OPA-2134 to see if that was the issue and the noise was still there after that swap as well.

Any help would be greatly appreciated! Can I get rid of this whine or should I just go ahead and order some MAX1044's and see if that helps?

Thanks,

-Kenji

[idy]

Before you despair of silencing the charge pump check the power supply you are feeding it with. I have found that 9v supplies that say they are "for guitar pedals" can add terrible whine, especially to charge pumps (heterodyne.)
I never have these problems with OneSpots.

[Krystal]

Have you tried connecting the 7660S pin 1 to pin 8 ? It enables a frequency boost that should take the whine out of audio range.

[FSFX]

Do you have a good decoupling capacitor right on the input to the 7660S power pins.
There is lots of advice regarding why you need this in documentation from the charge pump manufacturers like Maxim and TI.

https://pdfserv.stg-maximintegrated.com/en/an/AN2027.pdf

[kenjib]

Awesome. Thanks everyone. I will try all three of these tonight when I get home from work. For the decoupler do I just put a 10nf box cap from pin 8 to ground or do I need to run it with a resistor as an LC filter per the article?

[FSFX]

For the decoupler do I just put a 10nf box cap from pin 8 to ground or do I need to run it with a resistor as an LC filter per the article?

The capacitor from pin 8 to ground should be larger than 10nF. It is there to provide a 'reservoir' of power for the charge pump to move into its transfer capacitor and so reduce any instantaneous current draw through the power feed to the pedal from the external power supply. It needs to be at least 10uF. The additional resistor and capacitor on the power feed is there to provide some EMI filtering to stop the higher frequencies from the switching transients getting back to your power supply. A resistor there will obviously drop some voltage to the charge pump and so it may no be desirable but if used needs to be low enough to drop only a small voltage at the working current.
   

[thomasha]

I think almost everything I would suggest was addressed:
- Boost pin if it is the ICL7660S
- Large supply capacitor, close to pin 8
- Check wall wart (prefer the heavy ones with a transformer, instead of switched power supplies)

Another layout suggestion is to use connections as short as possible for everything that involves the ICL7660.
That is also valid for the unconnected pins.
Ideally, do not solder them to the board. You don't want the copper traces working as an antenna.

[kenjib]

Alright so I went through these step by step:

1. I tried 4 power supplies first, including one that has a class 2 transformer. The whine was still there. Hard to tell if there was less noise or not.
2. I put a 10uF cap from V+ to ground. I didn't have any unpolarized caps that big so I put neg to V+ and pos to ground, assuming that it was bleeding off AC. Was that correct? I am not sure without being able to A/B compare but I think some of the static noise decreased even though the whine remained. Very nice though!
3. The boost from pin 1 to 8 removed the whine completely. My understanding is this pushes the frequency up out of hearing range @30kHz rather than removing it. Is there any chance this has unintended side effects like, say unexpected aliasing distortion when going through an A/D converter?

In any case. The pedal works perfectly now and sounds great! Here's my updated vero in case it is of use to anyone:
https://www.diyschematics.com/schematics/upload/2023/06/27/20230627041452-6cd514b8.png

Thank you so much for your help. I use this pedal to run wet/dry in a single amp. The wet channel is basically chaotic noise with high feedback reverse delay, PLL, a rainbow machine, a slow gear, univibe, tremolo, etc. Fun stuff and this pedal is the center of it all so it's nice to get it working cleanly.

[idy]

The cap is upside down and could explode.
+ to V+, - to ground.

If you are using an invertor to create -v then you use - to -v and + to ground.

Yes it is "bleeding off AC" is a sense; it passes any signal, any variation from DC, to ground. But you have to respect polarity. But its main function FSFX explained, it is the reservoir you are charging up to the voltage you want.

[kenjib]

Yikes! I must have gotten confused with clipping diodes or something. I don't know what I was thinking. Thanks again.

Fixed:

https://www.diyschematics.com/schematics/upload/2023/06/27/20230627052600-94c59379.png