mitigating noise via ground plane isolation

[jessenator]

So I was initially working out noises in this thread a while ago, and mauman helped me through the initial problem of MASSIVE noise.

I made the change and put the diode and cap on both the boost board and the main effect board. There was still an eeeensy bit left over at higher gain levels, but while you're playing it's not really noticeable. It's still enough where I don't feel comfortable sending the pedal to anyone without investigating further. Is "needs isolated power supply" a reasonable requirement?

I have put it into an aluminum enclosure, for one. And having seen what it and the power bank difference has made (it sill makes the itty bitty noise with a generic wall wart), I wonder if there's even more I could be doing, or if it's "as good as it gets"
Based on that previous topic I started, the reason for the heterodyne is probably to do with the boost (LPB-1 workalike) combining with the switching oscillator in the 7660H on the Death to the King board. I don't know if it's hairbrained or not, but I was wondering if there's any benefit to isolating the ground planes of the DttK as much as possible, with a small connection to the main ground plane (rest of the board)(option B shown in the image).

sorry, ignore the difference of pad isolation between the two versions...

OR, am I just trying to solve an impractical (not impossible) problem, and the signal will be there regardless?

Thanks.

[mjg]

An approach I've seen is to not put any ground plane underneath the charge pump.  Keep that area of the board completely clear. 

You could also try the LT1054 instead of the 7660, but beware the difference on pin 1.

Not sure if it will help, but if you're willing to try out ideas put them on the list?

[jessenator]

An approach I’ve seen is to not put any ground plane underneath the charge pump.  Keep that area of the board completely clear. 

You could also try the LT1054 instead of the 7660, but beware the difference on pin 1.

Not sure if it will help, but if you’re willing to try out ideas put them on the list?

Thanks. So just run a ground net to pin 3? Yeah, I need to order this board again anyway, so I can work that out. Wtih the 1054 I just need to leave pin1 unconnected in this case, right?

Something like this?



Also, the cap coming off of pin5: the schematics as far back as the Chittum trace have it as 1uF, while the "example" layout in the LT1054 datasheet has a 100uF cap. I'm guessing there's a raisin for it being 1uF, but thought since I have the hood open I'd ask. Again, this is probably something an EE would actually understand, but I'm a hack 

[mjg]

 Yeah, that screenshot looks good.  Worth a try anyway!

For the cap on pin 5, given that's to keep the output negative voltage steady, it probably wouldn't hurt to put a 100u and a 1u in parallel. Looks like you'd have room to fit both on your layout. 

Anyway, just another hack here, but hopefully this helps