What is wrong with my summing op amp mixer on this delay pedal?

[BluffChill]

I've been coming up with a new version of my Shagpile pedal. It's a PT2399 based modulated delay. The modulation comes from an envelope follower rather than an LFO.

The first version I came up with was cool, but it did have some problems - namely noise, so I've re-engineered it to be less noisy. It works great and sounds exactly like I wanted to...with one exception. As soon as I add in the dry signal, it either goes crazy noisy, picking up whistles, whines and RF noise. It uses one half of an LM358  or 4558 (the other half is used elsewhere in the circuit, for the envelope follower). Here's how it looks:




As you can see I've got a switch for a 'dry cut' which I was using to test. But as soon as I put the dry signal back in, i get mega-noise. What am I doing wrong here? Any help appreciated.

[patrick398]

You could try a 50pf between pins 6 and 7, should eliminate any RF at least

[BluffChill]

Tried it - did nothing. Even went up to a 470pF. It's somehow amplifying all the whistling that the noisy old PT2399 can do! Thanks though.

[anotherjim]

Notice that the dry feed side will have the PT2399 2.5v DC reference level. I don't know if the wet feed has 2.5v or is AC coupled. Also, don't know what VR for the opamp is, but that 2.5v from the PT2399 is probably going to crank the opamp close to its limits with some gain applied.

[BluffChill]

VR for this is 4.5V. HOw does the dry side have the PT2399's 2.5v ref level? Is that caused by R18?

[ElectricDruid]

I think AnotherJim might be right that the bias levels are causing problems. The mixer seems to be DC-coupled, but it's fed from the PT2399 which has a 2.5V bias...except the mixer itself has a 4.5V bias.

Can we see the rest of the circuit please to make sure?

[BluffChill]

Here's the whole thing. I think Jim's right too. Forgotten to account for different bias levels for the PT2399 v op amp.



I was trying to get away with just using the one dual op amp for this but think I might need another. I've got a deluxe version of this too which uses a quad op amp, and I set the mixer up in a similar way to a lot of other PT2399 delays, like the Small Time, but clearly I've missed something by omitting that first buffer used in those examples.

[anotherjim]

If the opamp is an LM358, pin5 VR could be 2.5v from the PT2399 pin2, and that would be a simple fix. You might well have enough headroom doing that since the LM358 can swing to 0v, 5v p-p signal before clipping ought to be enough?

[BluffChill]

Thanks, I'll try that. What's the difference when using a TL072 or 4558 though? I get 'slightly different noises' with each one!

[anotherjim]

All opamps have a voltage swing limit in and out and that varies between types. Also, how it behaves when it reaches a limit can be different. In most cases, if a DC voltage forces the output too close to a supply voltage, any signal will, at best, be asymmetrically clipped - at worst, it performs a latchup stutter (dependant on slew rate) or momentarily reverses phase. The bi-fet opamps like the TL072 are the worst behaved and the bipolar like the 4558 (and its 1458 grandfather) are much better.
 

[BluffChill]

Thanks for the explanation, I've always been a little unclear on what happens when they reach their limit. That explains why a TL072 would produce sound for a few seconds, then stop entirely.

The bias with 2.5V from pin 2 worked a treat by the way - thank you very much. When these boards are ready, you're welcome to one!