Switch glitch

[gez]

Why is it that that in the following schematics, A produces noiseless switching but all the others produce an audible click (to a greater or lesser extent):



Subbing in a BS170 MOSFET, as shown in B, creates an audible click.  C is the worse offender and D is about the same as B.

The outputs of the flip-flop control FET switches (all works fine).  There's a Schmidt buffer in front of the clock input so how on earth is noise being coupled through in schematics B,C & D?!  Is it that the 4093 is acting as an amp during the brief moment that current flows during switching?  I thought the whole point of the Schmidt input was to prevent this sort of thing!

Why the 4093 in the first place I hear you ask?  I have a batch of 4013s made by ST and the clock inputs are incredibly picky (not the only ST chip I've had problems with).  In many applications where the 4013 should work, the ST ones don't, unless buffered by a Schmidt input logic device (as shown above).  I'm reworking an old circuit for momentary switching and I need the rest of the 4093 for logic purposes so there's no problem with using one of the remaining devices as a buffer, thus saving me the hunt for 4013s made by another manufacturer.

Anyway, why the noise in all but the first schematic?  

Baffled of London…

[gez]

Whoops, drew the last one up slightly wrong.  It's now been edited.

[puretube]

A) has a conducting low impedance to ground (B-E)...
(you`re only switching 0.7V to =0V; the others switch (almost)V+ potential to ground)... ?

[gez]

Hmmm, I think you might be onto something there Ton.  Of course, it doesn't help that I used the same ground path for the switching trannie (+ supporting circuitry) and the 4093!  :roll:  It's only on my breadboard though, so no big deal.

I'd like to get D working as it's battery friendly.  I'll try connecting a resistor from the lower pin of the switch to ground.  If I can set it up so that the base is only pulled down enough to turn the trannie on then this might reduce the click.  Will let you know.

Thanks for your input.  

[puretube]

how about a large pulldown (4.7 to 22 M) across the cap?

[gez]

how about a large pulldown (4.7 to 22 M) across the cap?

You mean permanently, ie with switch open too, and the switch shorts it?  :?

I was thinking of switching the resistor in when the switch is closed, just bias things so there's a slight pulldown on the base (enough to turn the trannie on).

[puretube]

btw: you do have an electrolytic cap across the (logical) V+ , I presume,
and have this separated from the audio B+ by some R (e.g. 100 Ohm)?

[puretube]

how about version C,
with R=470k; C=100n; and adding 470 Ohm series-R on the ungrounded side of the switch (i.e.: between the "northern" side of the switch, and the R/C/4093-input node), to "slow down" the cap-discharge?

[gez]

how about version C,
with R=470k; C=100n; and adding 470 Ohm series-R on the ungrounded side of the switch (i.e.: between the "northern" side of the switch, and the R/C/4093-input node), to "slow down" the cap-discharge?

Ahhh, good idea Ton! I've seen something similar in some (Boss?) schematics.  I'll mess about a little more tomorrow (or today if I get the chance) and let you know how it works out.

[gez]

how about version C,
with R=470k; C=100n; and adding 470 Ohm series-R on the ungrounded side of the switch (i.e.: between the "northern" side of the switch, and the R/C/4093-input node), to "slow down" the cap-discharge?

[R.G.]

It's the combination of the fast capacitor discharge and the grounding you have set up for the unit.

If you isolate the power and ground of the switching circuit from the switched circuit and don't have your wiring trailing about so that it picks up the transition capacitively, all versions work equally. To the extent that the switch section can couple transients into power or ground lines or signal wiring, you'll get a tick.

Note that you don't need logic to make this happen. 3PDT switches wired by inexperienced people to share power and ground lines with high gain pedals usually pop too.

[gez]

RG, yes I had to change the grounding.  The breadboard with the main circuit on is absolutely full so I didn't have much of an option but to share.  I ended up clearing a mini-board and putting all the switching circuity on that.  C is the obvious choice and it's also the quietest now, so I'm a happy bunny.  

Thank you Ton!

[puretube]

welcome - next time we`ll try without a 4093...

[R.G.]

welcome - next time we`ll try without a 4093...

The reason the 4013 doesn't work well without a 4093 and that it varies from maker to maker is that the 4013 is an edge triggered flip flop. So the quality of the edge is what makes it work. The datasheets specify a maximum rise and fall time on the clock edge to trigger it. Too slow an edge, no trigger. Worse, intermittent triggering, which is worse.

The very process of slowing down the capacitor discharge to get less pop is one thing that poisons the trigger.

Why not try without the cap? Because the switch is mechanical and bounces when the contacts hit. If it happens to bounce an odd number of times, the 4013 triggers an odd number of times and it appears to work. If it bounces an even number of times, the 4013 appears not to have worked. The cap is for debouncing the switch.

I prefer the hex schmitt trigger gates like the CD40106, 74C14, and 4584 for cleaning up a debounced switch because there are six in one package and there's no second input to each one of them to have to figure out what to do with.

[gez]

The reason the 4013 doesn't work well without a 4093 and that it varies from maker to maker is that the 4013 is an edge triggered flip flop. So the quality of the edge is what makes it work. The datasheets specify a maximum rise and fall time on the clock edge to trigger it. Too slow an edge, no trigger. Worse, intermittent triggering, which is worse.

Bingo, and the clock inputs of ST's 4013s are really picky!  One manufacturer used to do their 4013s with a Schmidt clock input, which wouldn't need the extra buffering, but I've yet to lay my hands on any.  Could have been Fairchild, I'll have to check the data sheet.

I prefer the hex schmitt trigger gates like the CD40106, 74C14, and 4584 for cleaning up a debounced switch because there are six in one package and there's no second input to each one of them to have to figure out what to do with.

Unless I need the rest of a 40106/whatever, I usually use a 555 wired as a Schmidt as it takes up less board space.  The reason I used the 4093 is because I needed it for a simple logic problem (one switch toggles between two different LFO waveforms and a bi-colour LED shows which is selected, another switch 'bypasses' the effect and turns the LED off - works beautifully!)

[gez]

One manufacturer used to do their 4013s with a Schmidt clock input, which wouldn't need the extra buffering, but I've yet to lay my hands on any.  Could have been Fairchild, I'll have to check the data sheet.

Hmmm, might be getting confused with one of the Phillips 47 series devices.  Could have sworn there was a 4013 with Schmidt input, but can't find anything...

[puretube]

:lol:  I`ve been searching half of the afternoon for that one, too
(actually I intended this morning, to reply that there must be a ST-input-FF...).

OTOH, I`m also sure, I`ve seen some differentiating-de-bounce-edge-trigger-button-control somewhere, with a diode and a 2nd small cap... :roll:

[gez]

Phillips have quite a few D types with ST-input, but none (that I could see) that opperate from a 9V supply.  6V was the best I could find...