jFet not opening completely

[Lithium_Grim]

I want to make a tails switch for a DBD. There are multiple examples on the internet, but I just can not seem to get it to work correctly. Must be my lacking knowledge of transistors, so I am here to be taught

Attached a picture by R.G. Keen with a basic jFet switch as in boss/ibanez. If I am not mistaken when Vgs > V cuttoff the switch will be closed and if Vgs < V cuttoff it should be open. As an example
a J112 with V cuttoff between -1v en -5V

So if:
V control 9V, Vref 4,5V makes Vgs 9V-4,5V = 4,5V which is higher than V cuttoff -> Closed
V control = ground (0V), Vref 4,5V makes Vgs 0-4,5 = -4,5V which is in the Vcuttoff range -> Open

However... which ever jFet I try, when it is open it still bleeds signal between D/S. The louder the signal is (say humbucker with a full open OD) the more audible the (bleed) signal gets even at lower amp volumes. Why? How do I get to open the jfet/switch all the way and kill any S/D bleeding?

[Lithium_Grim]

Ok tested it in this setup (see attached pic). RDS off = 974k and RDS on = 27R. Still bleeds through a lot of sound with high gain input on the opamp.

[lcv]

Try connecting 10K from the output to the amp and ground.
Without that you will have as output signal the partition between 974K and amp input impedance.

[Lithium_Grim]

That makes it better, but still does not open it enough. I know there is no perfect jFET switch, but when hit with a hard signal (like a dimed out OD) is still sputters a lot.

[Manfred]

This might be an alternative circuit with better switching characteristics.
https://g4tgj.github.io/Audio-Switch/

[stolen]

Hey! Generally, if one transistor doesn't close enough for your application, you can just use two! Put another in parallel with R6 and drive it with the opposite of V_control.

But before going down this brute force path, there's a couple of things you could optimize. How is VB stabilized? Maybe the electrolyte is a little old and lets a bunch of signal pass? lcv's proposition of lowering shunt resistance is important, keep that 10k resistor.

One extra thing to try would be increasing VB in the switch section; the j112 has a rather big variance in Vgs(off), -1 to -5V, so if you got one from the lower end your control voltage is not sufficiently negative! The j113 is often the better alternative. But if you make 2 seperate VB rails, one for the OPA at 4.5V and one for the FET biases around 7V-ish, maybe with a trimmer, you might find that the FET can close more than it does at the moment. Or, if you're lazy, just try a control voltage of -5V and see if it closes more :3

Cheers,
stolen

[Lithium_Grim]

This might be an alternative circuit with better switching characteristics.
https://g4tgj.github.io/Audio-Switch/

Ok tried it with a control voltage of +9V and ground (using a 4u7 and 150k making it around 177ms switch time).

It still bleeds on high gain signals. It is better than my previous design, but it will just not open up totally. I also noticed that the first not I play after opening it gives a small "thumb" even if I wait a few seconds before I play a note.

[Manfred]

Try connecting 10K from the output to the amp and ground.
Without that you will have as output signal the partition between 974K and amp input impedance.

You are right, that helps.
I omitted the capacitors to calculate the frequency-independent damping.
To do this, I drew out the equivalent AC circuit of R5, R6 and R_channel of the FET, and then did a
delta-star-transformation, so you can see the ratios better.
If one connects a resistor of 1 megohm to the output,
which is a common input resistor of an amplifier, still 25% of the signal in the off-state arrives at the output.
With a 10k resistor only 1% in the off-state and 99% in the on-state.