Fulltone AB box plops :(

[manson]

I just made the Fulltone AB box (http://www.fulltone.com/PDfFiles/AB_switcher.pdf). I changed / added a few things:

- Made a connection between switch lugs 2 and 3, middle row. To connect 'audio' ground to 'power' ground. It hummed before doing that.
- Put a 2M2 pulldown resistor on the input jack.
- Placed a fet buffer between the input jack and the switch. Signal always goes though it. I used 2x 4M7 resistors instead of 1M. Together with the 2M2 pulldown that would result in a 1M1 input impedance.



Now first of all, it works. Channel switching is ok, led's light up, the buffers works. But it plops when switching channels!
What can I do about this? Have I forgotten something? Do I perhaps need a pulldown on the outputs, or after the buffers' 10uF output cap?

Thanks!

[Toney]

- Made a connection between switch lugs 2 and 3, middle row. To connect 'audio' ground to 'power' ground. It hummed before doing that.

The quote above may well be the issue.
I would break it into it's simple components and isolate the problem.
The in-rush from the led ramping up is enough to cause a pop in a switching box
Fuller would have purposely isolated the audio and power grounds so that means no 'jack trick' to switch the 9v. To do this requires either using a toggle to switch on  the battery, having an always on 9v socket or using switching jack. 
I would undo that modification and check carefully my wiring.
Having sorted that, yes the 10uf left hanging can offer plenty of discharge and cause a pop.
Your pull down resistor should go after that.

[manson]

Thanks, I'll have a look at the grounding. I'm not using a battery, only a 9v adapter. Here's another buffer schematic from Jack Orman's site:



Here he's placed a 100k at the output. Would that help?

[R.G.]

Yes.

EVERY electrolytic when left to its own devices will slowly leak down because they don't have all that big a series resistance.

[manson]

Thanks R.G., I'll add a resistor a see if it helps. Would changing this electrolytic cap to a film type cap solve the problem too?

[R.G.]

Maybe, maybe not. It would make it better.

What's at play here is the relative resistance of the insulation in the capacitors. Plastic films have a much higher resistance to leakage than does the formed-oxide insulation inside an electro cap. An electro cap might only have a 10M-100M (or so!) resistance, depending on how old it is, what kind of treatment it's had, etc. A film cap might have a resistance from, say, 500M to 100G (yep, giga-ohms).

When you put a voltage across a resistor, even a very big resistor, current starts flowing. If you charge up any cap and simply leave it alone, it will slowly leak down to zero volts, just from the current running through its internal leakage resistance. How much it leaks per time period is dependent on the capacitance and the resistance, just like if the resistor was a 10K.

On top of that, there is an effect called dielectric absorption where the voltage across the cap actually mechanically pulls the atoms in the insulation around a bit, and this can relax back, giving up some more voltage.

So how much of a plop you get from a flying-lead output capacitor depends on the voltage, the capacitance, the type of insulation in the cap, the age of the cap, the humidity in the air, whether there is dust on the body of the cap, the phase of the moon, and whether your brother in law is coming over this weekend.

I have this philosophy that bullying electrons is OK. Electrons are there to do what ...I... want them to, and I don't care about their feelings. So I tend to design things that force them to do what I want, and not presenting a situation and seeing if they do the right thing. Electrons have short memories and will go off on their own if you let them. So I put in stuff just to be sure they stay where I want them to. So I would put a pulldown resistor on every single capacitor that is switched so it sometimes has an open lead. That way I can be sure that it is sitting at the correct DC voltage the next time it's put into use. The electrons go where I want them to, no questions, and no counting little electron heads to see who showed up and who didn't.

Put in the pulldowns unless there is a positive, clear, unabiguous, rock-solid reason not to. Messing around with "I think I can get away with it again" reminds me of my old friend Ivan, who was the world's greatest champion at Russian Roulette. His lifetime record was 359 and 1.

[manson]

I understand, make a robust design that does what you expect it to do.
What value pulldown resistor would be a good choice in this case? And what does it do with the output impedance (which I guess should be as low as possible)?

Thanks for your comments, I always enjoy reading them

[R.G.]

What value pulldown resistor would be a good choice in this case? And what does it do with the output impedance (which I guess should be as low as possible)?

You're welcome, glad to help.

The exact value of a pulldown is somewhat arbitrary. On inputs you want to keep them high. On outputs, it doesn't matter much. You could use anything between 10K and 10M. The lower the value, the lower hiss, but there isn't likely to be enough hiss to matter anyway. 1M is good, and safe. Putting a resistor across the output can only lower the output impedance. In fact, using a piece of wire of maybe 10milliohms will make the output impedance be 10milliohms. It will also prevent any signal from getting out because it's much lower than the impedance of the circuit. Make the resistor higher than the circuit by at least a factor of ten to avoid loading down the source follower. I would use from 100K on up, and again, 1M is a good choice.