The tell tale heart is driving me insane! LFO ticking + decoupling!

[nelson]

Hi People, Gurus, novices, lurkers.


So, I built a working version of the polyphase/Deluxe Phaser. It sounds wonderful, envelope controlled phasing is seriously underrated.

Anyway, I didnt notice this at first. I boxed up the effect, which got rid of the usual RFI. Only to be left with some lovely ticking. Now I followed the schematic of the polyphase/dlx phaser. There is minimal decoupling.

I proceed to hack up my lovely PCB (the desktop I use to print pnp now has a blown up monitor) and decouple using 220uF capacitors and 22r resistors on the V+ lines of the CD4013, CA3140, LM1458 ahead of the 3 optocouplers in series, the input buffer, the 4558 in the last phase stage and the transistors in the main LFO. The opamps dont share a Vref line.

Now after all this I think what I am doing is overkill so I get my DMM and start probing around on the V+ and vref of the various stages and they are still fluctuating in voltage by a good 0.03v in time with the LFO. Argh!


Now I think it must be the optocouplers causing such a fluctuation.

Should I up the capacitor value to ground before the optocouplers?

Should I modify the "power supply section" to include a couple of huge capacitors? Currently its just a 22r resistor and 220uF cap just after a 7815 regulator.

All suggestions welcome, its driving me insane.
I have been working on this for months and it sounds so wonderful, everything I have been looking for in a phaser. The modifications possible are tantalising with the added bonus of the envelope control.

[Skreddy]

Can you adjust the side the optocouplers affect so they're more sensitve to less of a voltage change, and then limit the amount of current being consumed by the optocouplers?  Can you separate the ground traces of the current consumers from the audio path's grounds, up to a common point near the ground wire, or at least provide a fatter path for both ground and +V to the heavy hitters?

[bioroids]

Sometimes is good to put a 100nf ceramic in parallel with filter caps. That's because the LFO at some point is making a square wave, and the big filter caps take some time to provide current for the spike, so you need caps with a smaller time constant in parallel with the biggies. The ceramics should be very close to the component making the spikes.

I'm assuming the ticking is coupling through the ground/power traces, if the modulation is done with optocouplers.

Luck!

Miguel

PS: that's funny, the spell checker suggest "Episcopals" in place of "optocouplers"

[nelson]

Can you adjust the side the optocouplers affect so they're more sensitve to less of a voltage change, and then limit the amount of current being consumed by the optocouplers?  Can you separate the ground traces of the current consumers from the audio path's grounds, up to a common point near the ground wire, or at least provide a fatter path for both ground and +V to the heavy hitters?

I am very limited in terms of space. The layout only just fits in a 1590BB, there is no room to fatten up any traces. I may be pushing it with 9 IC's 2 LFO's an envelope follower, 4 Transistors + 3 dual optocouplers.

The LFO's, envelope section and optocouplers are all only connected to audio ground at two points.

I dont believe limiting current consumption and upping voltage sensitivity is an option, atleast not one I am willing to undertake. The optocouplers are rough in spec and being too close in voltage specification would likely result in a bad sweep. Plus I want to remain as faithful to the original design as possible.

I really am limiting my options, I really dont want to put it in a bigger enclosure.





Here is a pic.

Would better decoupling do it?

I will try the 0.1uf ceramics.

Thanks Boiroids!

[Skreddy]

Looks like an excellent candidate for a double-sided board.  You could eliminate most of the closely-spaced concentric/parallel traces (and most if not all jumpers), keep supplies for optos and audios separate, have fat supply and ground traces, etc., without sacrificing compact size.

[gez]

Anyway, I didnt notice this at first. I boxed up the effect, which got rid of the usual RFI. Only to be left with some lovely ticking.

If you didn't have ticking to begin with then it's just your wires being too close together or hovering over areas of the circuit board that house the LFO as a result of being crammed into the box (stray capacitance). 

This has happened to me a number of times in the past and the solution has always been pretty simple.  Usually rerouting the wires does it.  Failing that use screened wire.  I now design layouts so that wiring will go nowhere near the LFO.  Watch out for wires to pots as well, keep them short and screen them if necessary.

[nelson]

Anyway, I didnt notice this at first. I boxed up the effect, which got rid of the usual RFI. Only to be left with some lovely ticking.

If you didn't have ticking to begin with then it's just your wires being too close together or hovering over areas of the circuit board that house the LFO as a result of being crammed into the box (stray capacitance). 

This has happened to me a number of times in the past and the solution has always been pretty simple.  Usually rerouting the wires does it.  Failing that use screened wire.  I now design layouts so that wiring will go nowhere near the LFO.  Watch out for wires to pots as well, keep them short and screen them if necessary.

Thanks Gez, I am using PCB mount pots. Its kind of hard to keep wires + traces completely away from the LFO seeing as the whole right hand side of the board is an LFO/envelope follower, I will try shielded wires for input and output. Plus there is a switch is connected to the 2 LFO's. However it was wired in the same configuration before I put it in the box. I think the RFI was just masking the ticking.

I appreciate everyones suggestions, I will be publishing the layout when I get this minor bug sorted.

Thanks

[bioroids]

I once did a little board with an LFO (standard doual opamp square/triangle) to enhace a delay pedal with a little modulation. It worked great outside of the box, but once I mounted everything inside, the ticking appeared.

Only solution I found: isolating the LFO board with aluminum foil (connected to ground). Basically I had to shield between the two boards to get rid of the ticking. It was not emmited nor received by the wires, it was the board itself emmiting radiations.

If you can't get rid of it with decoupling caps, I would suggest making this a two board project, mounted one on top of the other to save space, with a good shield (can be a PCB board) between the two. That would separate   the stuff a lot and the boards can be smaller/less tight

This sounds like a very interesting effect

Luck!

Miguel