Choosing JFET for bypass switching

[mark2]

Using a configuration similar to the TS (except replacing the toggle circuit with a uC), and trying to choose an appropriate Q2 and Q4, am I simply looking for something with a VGS(OFF) that covers Vf of D3 and D4? i.e. -0.7V

e.g. is the following true?
5457 has VGS(off): -0.5 to -6, so it would work?
J202 has VGS(off)  -0.8V to -4.0V so it may be problematic?

[mark2]

In writing this out, I guess I've come to realize I don't fully understand how it turns on.  When 9V is applied to the diode's cathode, do you have 0V across the diode? What happens with the JFET?

[PRR]

The diode+gate string does not make sense, until you learn that the gate leakage is much less than diode leakage.

[mark2]

This probably also explains why I was having trouble modeling it. I'll study it some more, thank you.

[PRR]

> trouble modeling it

I dunno if all models "know" that JFET buyers pay more for less leakage, so that typical JFET gate leakage is less than typical GP-diode leakage.

My antique sim does a right thing. Bias 4.5V. Control voltage swept zero to 9V (more than real life). A 0.1V test probe to tell if the JFET is ON or OFF. Vto for this simmed JFET is 3.0V. So below 4.5V-3.0V=1.5V the FET is off. Much above that, on.

Note that the 'opposing diodes' do an iteresting thing. The mid-point follows control voltage up to bias+0.6V, and then the forward conduction of the Gate completely swallows the diode's reverse current. There is no large DC flow through gate into signal path.

I've made some values ridiculous to isolate side-effects.


Your flip-flop stands on a Volt or so of common emitter impedance. So a Vto as large as 3.0V gets marginally "off", especially as battery sags.

[mark2]

Thank you, this is proving very helpful. I'm still working through it to make complete sense of how this works.

I'm also experimenting and noticed that an audible amount of signal is "leaking" when off.  I have just the switching part of the circuit set up, i.e. this part (minus the LED+4.7k, lifted from Equinox II) and no buffers; Just a signal biased to 4.5V. Could the impedance issues be related? It seems the same whether or not I load an amp on the output.


It's approximately 500mV peak-to-peak when on, and 80mV when off. Edit: this is a J112

I also tried biasing the input to 7V and 8V but it didn't turn the jfet "more off". Going down to under 3.3V (very roughly) seemed to be the tipping point to turn it "more on."

[PRR]

An imperfect switch is always a voltage divider.

What is your "R2", the lower leg of the voltage divider?

In most BOSS-like switches, there are two switches. When one is off, the other side loads it with a few-KΩ.

[mark2]

For the bias?  I was using 10k + a varying R2. e.g. for 8V it was 82k or something in that ballpark.

[PRR]

[mark2]

Ohh, I clearly misunderstood. Thank you.

I will take some time to digest this and cross-reference with real world examples, then add another JFET (edit: or alternate source of resistance) and see how it sounds (or hopefully doesn't!).

[mark2]

The quiet input tone persists even when I remove the jfet.  I stripped this breadboard down the absolute bare essentials, even removing my test box, and it's still there.

I suspect the "Dollar Tree" cable I'm using to transfer the tone from my phone may be contributing. I'm going to rig up a tone generator with individual wires later to try again. Of course, this is in addition to testing it out with input/output buffers and a second jfet.

On the original topic's note, any thing in particular to look for in a JFET?
* Low Vgs(off) ~ -0.5V
* Low Rds ~ 100ohm or less

They're fewer and farther between now, so I was hoping to just pick whatever was cheap and readily available on Mouser.

[PRR]

> input tone persists

OK, now put 100Ω from switch output to ground. Is the leakage less?

[mark2]

Yeah, that seems to get rid of it. Of course it also 5x reduces the "on" signal too. But it takes the "off" signal from a distinct sine wave (a small version of the input) to indistinguishable noise.

I'm still working on my buffered 2-switch version, too.

[PRR]

So try an in-between load. Like 5k.

[mark2]

I'm doing this in parallel, and wondering if anyone sees any obvious problems?



I'm having weird issues (e.g. Vbias too high, loss of signal) that could very well be shaky breadboard connections but I figured I'd check in while I'm at it.

Thanks!

[ElectricDruid]

The first thing I notice is that there's no bias on the source side of the JFET. I thought it was generally important to have both sides biased to the same level.

http://www.geofex.com/Article_Folders/bosstech.pdf

[mark2]

Thanks, I'll add that on both jfets. I had it on earlier experiments, then removed one side and it still switched. I suppose it may be more error prone though?

I don't need to save 2 resistors so if it doesn't hurt I'll just put it back.  I tried removing it out of curiosity since I saw neither the TS nor the Equinox II seem to have it.

[m4268588]

http://www.geofex.com/article_folders/tstech/tsxtech3.gif

[ElectricDruid]

Thanks, I'll add that on both jfets. I had it on earlier experiments, then removed one side and it still switched. I suppose it may be more error prone though?

I don't need to save 2 resistors so if it doesn't hurt I'll just put it back.  I tried removing it out of curiosity since I saw neither the TS nor the Equinox II seem to have it.

Oh, it'll still switch. That's not the problem. The problem would be DC level shifts causing clicks or bumps.

It's true the Equinox3 seems to only have one side of it's JFET biased, but I think the tube screamer has both sides tied to 4.5V if you check (although there's an error on electrosmash's schematic like m4268588 said).

https://www.electrosmash.com/tube-screamer-analysis

compare with:

https://schematicheaven.net/effects/ibanez_ts10_tubescreamer.pdf

Sorry, couldn't find a direct TS808 full schematic (everyone takes the JFETs out and does it true bypass!).

[Rob Strand]

Check out,
https://www.diystompboxes.com/smfforum/index.php?topic=126415.0

Actually it's not this thead.  I can't find the one I'm looking of for.
Also reply 28 in this thread,
https://www.diystompboxes.com/smfforum/index.php?topic=127813.20


Reply #9 on this one,
https://www.diystompboxes.com/smfforum/index.php?topic=103927.msg1227335