Boss-style JFET bypass switching - are buffers necessary?

[matt_garman]

I've created a couple relay-based bypass circuits (one using a CMOS inverter, the other using a microcontroller).

But now I'm looking at replacing the mechanical relay with electrical switches, such as the JFET switches used in Boss pedals.  Here's a nice primer on JFET switching by RG Keen.

My question: is it really necessary to have the input and output buffers when using this electronic switching scheme?  In other words, is it possible to do true bypass with JFET switching?  Should they be included in a "plug and play" bypass PCB?

I'm not coming at this from a buffered bypass vs true bypass debate, but rather wanting to create a generic yet simple-as-possible electronically-switched bypass circuit that I (and others!) can use in DIY pedal builds.

I'm using the Boss OD-3 schematic by kanengomibako as a reference (alternate link).  I count three buffers outside the actual effect circuit: Q8 input JFET buffer, Q1 output BJT buffer, Q5 bypass BJT buffer.  Are all these truly necessary?

Beyond the direct question, I'm open to general considerations about electronic switching.

[R.G.]

My question: is it really necessary to have the input and output buffers when using this electronic switching scheme?  In other words, is it possible to do true bypass with JFET switching?  Should they be included in a "plug and play" bypass PCB?

That depends on what you mean by "true bypass". The conventional wisdom is that the pedal circuit is completely disconnected from the signal path when bypassed, with (for instance) air gaps in switches or relays. JFETs and CMOS switches can provide very good isolation of AC signals, but the signal circuits are necessarily still connected in bypass. Buffers are a different issue. Buffers can prevent loading the signal path down and "tone sucking" treble loss when non-true-bypass is used. Buffers are a means to an end.

I'm not coming at this from a buffered bypass vs true bypass debate, but rather wanting to create a generic yet simple-as-possible electronically-switched bypass circuit that I (and others!) can use in DIY pedal builds.

You will probably run into people who complain that without a switch or a relay, it cannot be true bypass. That's the underlying item with claiming "true bypass" with FETs in any form.

[FiveseveN]

air gaps in switches or relays

For me it's sulfur hexafluoride or bust
The question "is it possible to do true bypass with JFET switching" is more philosophical than anything; not surprising when it comes down to what is "true". Air gaps are not the only way to achieve "practically infinite" impedance between two nodes but different people will have varying standards for "practically".

[matt_garman]

I guess I was thinking of "true bypass" as simply "not buffered bypass".  As you say, mechanical bypass (e.g. relay or 3pdt), truly switches between the effect and a "wire".  I guess electrical JFET switches aren't "true" in the sense that, like you said, the effect circuit still has some connection to the bypassed circuit.

Either way, I'm not necessarily trying for "true bypass", no matter how you define it.  My goal is circuit simplicity: do I need one or more buffers in a JFET switching scheme?  How minimal of a circuit can I get away with and still have a robust bypass scheme?

My perspective is that I've never heard of or seen a JFET switch-based bypass without buffers.  Is that simply because I haven't seen enough bypass schemes?  Or because it's a bad design?

[Rob Strand]

One aim of JFET switching which is not really possible with a mechanical "true bypass" switch is to de-glitch the audio when switching from one source to the other.   JFET switching have gate time constants which gradually transition between the effect and dry sources.   The smooth blend crossover prevents glitches caused by instantly joining two different waveforms - you get a discontinuity in the level or the derivatives (slope).


You can see in these pedals there is no output buffer.

This one uses MOSFET switches:
https://schematicheaven.net/effects/pearl_overdrive.pdf

This one JFET:
https://www.schematicsunlimited.com/d/dod/dod-fx25-envelope-filter-fx-pedal-schematic

Switching the unbuffered guitar signal directly has a slight risk of glitching due to the JFET switch capacitance.

As far as TRUE bypass goes, if you define it as not loading the source at all then you might struggle to do that with JFETs.   You need to bias the drain and source of the JFETs so a potential can existing on the gate.   That means some sort of resistor and hence not true bypass.   You could use large resistors to have no loading in a practical sense.    Unless you can bias to 0V you will also need coupling caps and that will modify the frequency response - again low cutoffs would approach a true bypass.   Lastly there are headroom limitations.

JFET switching has a lot of subtle quirks and no doubt these quirks get more pronounced when you start switching with high impedance sources and high impedance loads.
I rememeber this thread which did not work when some of the resistors were changed from 1M to 10M

https://www.diystompboxes.com/smfforum/index.php?topic=121972.0

Unfortunately a lot of the simulations waveforms are not displaying because the postimage image server changed names.   The links are there but hidden.

The thread  might have showed the difficulty getting the waveforms to switch seamlessly with the wrong JFET VGS_off.  (I could be mistaken it might have been a Boss Tremolo.)

[R.G.]

For me it's sulfur hexafluoride or bust
The question "is it possible to do true bypass with JFET switching" is more philosophical than anything; not surprising when it comes down to what is "true". Air gaps are not the only way to achieve "practically infinite" impedance between two nodes but different people will have varying standards for "practically".

I came up with the Clinton Bypass. For those of us old enough to remember William Jefferson Clinton's sworn testimony, it will strike a chord of memory to think that "I don't think you can find any evidence that it's not true..."

[R.G.]

I guess I was thinking of "true bypass" as simply "not buffered bypass".  As you say, mechanical bypass (e.g. relay or 3pdt), truly switches between the effect and a "wire".  I guess electrical JFET switches aren't "true" in the sense that, like you said, the effect circuit still has some connection to the bypassed circuit.

I brought that up because I got a LOT of flak from proposing non-hard-metal switching in early bypassing schemes in the early 2000s. There are diehard One True Bypass folks around.  But no, hard metal switching and disconnect are not needed for good bypassing.

Either way, I'm not necessarily trying for "true bypass", no matter how you define it.  My goal is circuit simplicity: do I need one or more buffers in a JFET switching scheme?  How minimal of a circuit can I get away with and still have a robust bypass scheme?

You're not going to like this.
It depends on the circuit you're bypassing and your JFETs. As Rob says, designing JFET bypasses includes some subtleties. 
JFETs with no particular voltage impressed between the gate and channel have their normal channel resistance, rdson. This is a characteristic of that particular JFET type number and that particular JFET. The small-signal resistance of the channel acts like a resistor for signals smaller than the channel pinch-off voltage - which varies. To get good switching you need the drive signal for the gate to be big enough to never let the signal being switched lower the available gate drive below the Vgsoff and into the pinchoff region.  This is a problem with 1V or bigger guitar/effect signals and 9V limited power supplies. The sweet-spot switching JFETs for 9V pedals is a JFET with about 4-5V of Vgsoff as a result.
And the circuit being switched needs an input impedance significantly larger than the JFET channel resistance. If it doesn't, you get signal division and loss, as well as tone loss effects.
Taken together, this is a very short synopsis of why there are nearly always buffers or quite-high input impedances being switched.

My perspective is that I've never heard of or seen a JFET switch-based bypass without buffers.  Is that simply because I haven't seen enough bypass schemes?  Or because it's a bad design?

It's a tricky design. Making a JFET switch without buffers can be done, but there are very often special cases.