CD4007 bypass: how does this work

[Fancy Lime]

Hi there,

I've been experimenting with CD4053 and CD4066 based switches (thanks to R.G. and the Tone God), which work nice and all. I have, on occasion, also seen the CD4007 mentioned for bypass switching but have been unable to find a schematic until Ron R. kindly made me aware of this thread:
https://www.diystompboxes.com/smfforum/index.php?topic=120006.msg1122270#msg1122270

Looking at the schematic (end of the tread, first of the images) I wonder: How does that even work? If the same thing were implemented with discrete MOSFETs, the body diodes would open if the signal went over 0.7V (or whatever the breakdown voltage is). Am I right in assuming that that is not a problem in this case because all P-MOSes share a body (tied to pin 14) and all N-MOSes share one (pin 7)? Why do I need one N-MOS and one P-MOS where otherwise one JFET would sit?

Do I furthermore interpret the data sheet correctly in assuming that the individual transistors of the complementary pair with pins 1-5 are symmetrical (meaning source and drain can be flipped like 1<->2 and 4<->5) because none of their pins is connected to the bodies (the same would apply to the inverter pair if they were not tied together drain to source)? Does that mean I could use the 1-5 pin pair as symmetrical variable resistors? what would the difference to JFETs be?

The .01 cap sets the switching speed, together with the two 100k resistors, right?

And lastly: Is there a reason why the status LED in the schematic in the thread linked above is driven by an additional discrete MOSFET instead of one of the uncommited ones? Could I not use the uncommited P-MOS to drive the LED and the N-MOS to short the effect input to ground when the effect is off?

Thanks,
Andy

[anotherjim]

The body diodes are indeed out of the way in the 4007.

I'm not sure if S & D can be swapped, if so, it looks like the middle pair in the above drawing have the most freedom.

If you use a high-efficiency status LED so that the CLR can be higher than 6k, The unused N channel could indeed be used to switch it. Depending on what drives the eff/-eff control, that might be able to do it directly anyway.
That said, I've used standard 3mm LED's with 6k8 CLR and get plenty of light.

[Fancy Lime]

I'm actually pondering building an octave fuzz using 4 out of 6 inverters from a CD4069. So I could use the 2 remaining inverters plus the leftover pair from the 4007 for the switching logic. In that case I'd either have to try and run the LED directly from the flip flop without an extra buffer, or I could just stick a BJT in there as a current source, to be on the safe side.

[amz-fx]

Look up schematics for some of the Pearl pedals to see the 4007 used for bypass.

regards, Jack

[duck_arse]

DOD circuits use the 4007 bypass as well.

[Fancy Lime]

Look up schematics for some of the Pearl pedals to see the 4007 used for bypass.

regards, Jack

Thanks! Did not know "Pearl pedals" were a thing, shame on me. I had a look at the OC-7 schematic here:
https://www.zikinf.com/materiel/pearl-oc-07-octaver,25606
Very nifty. One 4007 to do the flippedy floppedy and the actual switching. This is so compact, simple, and low-part-count that I have to wonder what the catch is. And if there is non, why is anyone still using discreet JFETs for the switches?

One thing I do not understand, though, is what the arrangement arround Q2 (lower right corner of schematic) does. Can someone explain that? Looks like some awfully elaborate stabilization for the status LED voltage to me.

DOD circuits use the 4007 bypass as well.

Do they though? All DOD schematics I am aware of (that is, the few of those that also show the switching circuit) use two inverters from the 4007 for the flip flop but then discreet JFETs (J113) as switches. Some even use an additional JFET to trigger the flip flop, which seems a bit redundant to me.

Thanks,
Andy

[anotherjim]

I've used a 4007 fet individually as a series cut input to allow reverb tails from a PT2399. In that case though, I know the signal amplitude is limited for the 5v supply for the delay and the 4007 substrate can be wired to the 9v. The possible flaw is if the signal gets large enough to affect Vgs, however, the body diode isn't a concern. The rest of the 4007 made a 5v limited soft clip input and the 9v inverter drove the status LED.

[amptramp]

Look up schematics for some of the Pearl pedals to see the 4007 used for bypass.

regards, Jack

Thanks! Did not know "Pearl pedals" were a thing, shame on me. I had a look at the OC-7 schematic here:
https://www.zikinf.com/materiel/pearl-oc-07-octaver,25606
Very nifty. One 4007 to do the flippedy floppedy and the actual switching. This is so compact, simple, and low-part-count that I have to wonder what the catch is. And if there is non, why is anyone still using discreet JFETs for the switches?

One thing I do not understand, though, is what the arrangement arround Q2 (lower right corner of schematic) does. Can someone explain that? Looks like some awfully elaborate stabilization for the status LED voltage to me.

DOD circuits use the 4007 bypass as well.

Do they though? All DOD schematics I am aware of (that is, the few of those that also show the switching circuit) use two inverters from the 4007 for the flip flop but then discreet JFETs (J113) as switches. Some even use an additional JFET to trigger the flip flop, which seems a bit redundant to me.

Thanks,
Andy

Thanks for posting this, I was not aware that anyone was actually using the 4007 as a switch in a production pedal and the circuit is a bit different from what I had suggested (which was copied from the application note in the Motorola MC14007 datasheet).  I had shown a separate LED driver to avoid effects from loading CMOS logic but maybe it is not necessary with high-efficiency LED's.  I like using the complementary pair switching since this gives the most symmetrical switching and decently low resistance for the connected switch over a reasonable input voltage range.

To answer another question, the R-C network going to the gates of the switches were to slow the switching to the point where there would be a short duration of no connection from either switch during the switching interval.  This avoids switch pops by slowly raising the resistance from one input then slowly reducing it for the other input.

[Slowpoke101]

One thing I do not understand, though, is what the arrangement arround Q2 (lower right corner of schematic) does. Can someone explain that? Looks like some awfully elaborate stabilization for the status LED voltage to me.

Looking at that part of the circuit I think that you will find that Q2 forms a low voltage indicator. The associated diode is a Lamba type diode which do behave oddly. The diode will conduct at a certain voltage level but will not conduct at voltages above or below that point. The trim-pot would allow you to set the battery voltage level that would be considered too low. I could be wrong but that seems to be what that part of the circuit does.

[duck_arse]

you had me doubting myself, so I went and looked at a few circuits, but when you put it like this:

DOD circuits use the 4007 bypass as well.

Do they though? All DOD schematics I am aware of (that is, the few of those that also show the switching circuit) use two inverters from the 4007 for the flip flop but then discreet JFETs (J113) as switches.

.... quite right. no, I suppose they don't.

[edvard]

Look up schematics for some of the Pearl pedals to see the 4007 used for bypass.

regards, Jack

Thanks! Did not know "Pearl pedals" were a thing, shame on me. I had a look at the OC-7 schematic here:
https://www.zikinf.com/materiel/pearl-oc-07-octaver,25606
Very nifty. One 4007 to do the flippedy floppedy and the actual switching. This is so compact, simple, and low-part-count that I have to wonder what the catch is. And if there is non, why is anyone still using discreet JFETs for the switches?
...

DOD circuits use the 4007 bypass as well.

Do they though? All DOD schematics I am aware of (that is, the few of those that also show the switching circuit) use two inverters from the 4007 for the flip flop but then discreet JFETs (J113) as switches. Some even use an additional JFET to trigger the flip flop, which seems a bit redundant to me.

Thanks,
Andy

The original idea for this circuit was published in the June 1983 issue of Polyphony magazine, by none other than David DiFrancesco, the founder of DOD.
From Mark Hammer's pages at ampage.org:
http://hammer.ampage.org/files/dodswitch.gif

I'm sure that improvements and mods were made over the years, but this is the original.  Yeah, using one of the 4007 MOSFETs as a trigger is kinda redundant. If you replace that with a momentary switch, the results are the same (at least, according to simulations)
The reason for external JFETs is probably two-fold;
One, because in many cases you need more than two switches (FETs), especially if you're going to do anything resembling true bypass (which would require at least three), and the arrangement in that Pearl schematic you linked has the two MOSFETs arranged to function basically as a SPDT switch on the output.  Common in the early days of pedals, but we're so much more sophisticated these days... 
Two, sometimes it's easier to have the switch FETs placed elsewhere on the board just for convenience, such as closer to the in/out points on the board.  I've always felt that some high-gain circuits especially should have the input and output separated as much as possible to eliminate inadvertent feedback paths.  Though I must admit, stringing in/outs to within 1/4" of each other on the solder lugs of a DPDT kinda takes some wind out of that argument...

Other reasons to consider: The switch MOSFETs are different, and so need Sink and Drain to be biased differently than the other; from the Pearl schematic, it looks like Sink and Drain on the P-channel get biased to Ground, and on the N-channel, Sink and Drain get about +3V. That's not such a big deal, really, but should be kept in mind.  Also, the 'on' state resistance for a J113 is on the order of 100 ohms or so, which is good; the 4007 MOSFETs I'm not sure, the spec sheets don't directly say, and I'm not enough of a math whiz to take on extrapolating that.

[Mark Hammer]

You said exactly what I was about to say: that the J113s may have been used, instead of just the CD4007, because their "on" drain-source resistance is pleasingly low.

[Fancy Lime]

@ Eddy and Mark,

thanks, that's the info I was after. I hope I'll get around to drawing a few schematics of the different variants one of these days for different scenarios: with and without input switching, etc. to facilitate comparison. Have a feeling that it should be possible to develop a layout that allows several variants for different purposes that can be used as a secondary board replacing existing true bypass switches. Would also facilitate remote switching, which IIRC Mark expressed interest in standardizing, right?

Andy

[anotherjim]

Not only do the CMOS transistors channels have a fairly high Ron, but it seems to vary between the different chip manufacturers.

[Mark Hammer]

@ Eddy and Mark,

thanks, that's the info I was after. I hope I'll get around to drawing a few schematics of the different variants one of these days for different scenarios: with and without input switching, etc. to facilitate comparison. Have a feeling that it should be possible to develop a layout that allows several variants for different purposes that can be used as a secondary board replacing existing true bypass switches. Would also facilitate remote switching, which IIRC Mark expressed interest in standardizing, right?

Andy

Yep.  Preferably using a momentary connection to ground.

[Fancy Lime]

@ Eddy and Mark,

thanks, that's the info I was after. I hope I'll get around to drawing a few schematics of the different variants one of these days for different scenarios: with and without input switching, etc. to facilitate comparison. Have a feeling that it should be possible to develop a layout that allows several variants for different purposes that can be used as a secondary board replacing existing true bypass switches. Would also facilitate remote switching, which IIRC Mark expressed interest in standardizing, right?

Andy

Yep.  Preferably using a momentary connection to ground.

Aha! And that is where the seemingly redundant JFET in the DOD circuit comes in, isn't is?

Andy

[edvard]

Ah, I hadn't thought of remote triggering... yep, a FET would be just the thing for that.  Would that also help with de-bouncing the switch?