Auto kill / rage

[Gurner]

I got poetic with the subject heading - this is an autokill that's presenting an issue (hence the rage - like I say, poetic license!) Ok, simplest of circuits but an unsatisfactory result at present. Here it is, basically a kill switch under Uc control...



It all works...*but* with no guitar signal being present (ie quiescent conditions)...the audible clicking in sympathy with the Uc gate signal is a bit too much.

So stripping this back to basics...clicking in the main comes from abrupt changes in DC level.

when I roll the guitar pot right down, there's no clicking - so I'm figuring this clicking must come from a small  amount of DC on the MOSFET drain when it's 'off'? (I've not this one on breadboard yet, so just floating ideas!). If it is down to a small amount of DC there (which gets cranke dup considerably by the high gain amp sim I'm using)....where do you suppose it's coming from? The guitar coil is essentially a 7.5k pull down resistor to ground ...ok, so the leads I'm using to interconnect at present are about 4ft long, and this may cause small potentiial differences....but I tried a high value resitor across the drain source, but this made squat difference to the outcome (and if we're talkiing a very small amount of DC that's present, then a high value resistor isn't like to help here anyway?)

[CynicalMan]

Grid leak from the amp?

Try putting capacitors on either side of the MOSFET, in series with the signal, and a resistor to ground parallel to the MOSFET. This would filter out any DC bias from the guitar or amp.

[Gurner]

No, there's no amp involved here .....signal path is guitar-> soundcard (an RME Fireface 400).

Well, this issue not down to stray DC...my multimeter, goes to 4 decimal places & just measured the mosfet drain - zilch DC present.

here's a scope trace (we're right down in amongst the noise floor here)...



....that was taken with the soundcard disconnected (to eliminate possibilities)

So the path is now.....guitar -> auto kill MOSFET. How one earth can I be getting those spikes? (current bleed into the signal ground at the mosfet?)

Wondering if a small inductor might help on the mosfet source pin?

[R.G.]

It's capacitive coupling from the control signal.MOSFETs have large capacitances from gate to the channel. You have to either get another device or make the control signal so slow it's below the pass band of the rest of the signal.

Get a better (not bigger!) horse. The H11F1 through H11F3 LED-photoFET devices have optical coupling to the FET. They have essentially zero feedthrough.

[Gurner]

Thanks for the input - that saved me a lot of hair tearing - I've only ever used mosfets on high level signals, where I've not really had to even ponder their capacitive vibe.

[Gurner]

Ok, so a bit more googling reveals the term to be in play here is "charge injection" ....and over and above what RG said, it is directly proportional to the magnitude of the gate switching waveform on the gate (which figures).

Now as luck would have it, the PCB I knocked up to switch the MOSFET gate on/off  happens to the PIC's DACOUT pin too. The PIC has 32 voltage levels the DAC can output, so I'm figuring if I can establish the actual switching theshold/point for the MOSFET, then I can simply switch the DAC pin *just* above & below to switch the MOSFET on/off (rather than the full 3.3V the MOSFET GATE gets at present) ...the gate voltage excursions will then be much less (& by default the clicks that are permeating through the MOSFET's capacitance into the small signal will be less).

I've ordered some H11F1SMs up anyway, but they're quite chunky at about 9mm x 7mm (ok, that's still small but it'll dwarf everything else on my board!)

[Gurner]

Just by way of closure, the opt isolator arrived...slapped one in situ but still got clicking

Transpires there was actually a bit of DC on the signal line (so it wasn't a pointless procedure - ie it really helped me establish the root problem)

I therefore went back to the mosfet method (having tackled the stray DC) ....sure enough it clicks when the gate is switched with a full 3.3V ...by creative use of a dac to switch the gate, I could eliminate the click by switching the gate in and around it's threshold level.