J201 a signal switch?

[Toney]

Is there anything stopping the use of the J201 to as a switch ala the Boss bypass method?



Vgs off is very low anything else?
Will stop all this JFET switching stuff soon ( )

Perhaps this?
"Idss is an absolute limit on the amount of current that can flow through the channel, so you have to ensure that when the JFET is on, channel current never gets to Idss. Some small signal JFETs have quite small Idss. "

[amptramp]

This shows you how to get a gradual switchover from one input to another using a CD4007 or MC14007:

http://html.alldatasheet.com/html-pdf/11935/ONSEMI/MC14007UB/371/2/MC14007UB.html

Use the bottom diagram but instead of the inverter in the southwest corner, just connect the select lines to a D-type flip-flop via R-C delays and connect /Q to D to get a toggle action.  This can be connected through an R-C debounce to the SPST stompswitch.  The problem with individual FET's is that the switching pulse is capacitively coupled to the channel.  With a MOSFET switch of this sort, the effect of the N-channel and P-channel devices tends to cancel out.

[R.G.]

Is there anything stopping the use of the J201 to as a switch ala the Boss bypass method?
Vgs off is very low anything else?

Stopping? No. Any commonly available JFET (N.B.: That may be a contradiction in terms.) can be used as a voltage or current switch. The differences are in how well it works. JFETs intended for use as switches will have specified rdson, cgs, perhaps switching speeds, etc., and usually higher, and well specified Idss. Amplifier-specialized JFETs like the J201 will have specs on Yfs and other amplification items. I had to look pretty hard to find rdson in the Fairchild J201 datasheet at all. I only found it in one of the charts of typical characteristics. It's about 50-100 ohms at some set of circumstances.

Contrast what that tells you against one like the 2N4391.

Will stop all this JFET switching stuff soon ( )

I doubt it.    

Perhaps this?
"Idss is an absolute limit on the amount of current that can flow through the channel, so you have to ensure that when the JFET is on, channel current never gets to Idss. Some small signal JFETs have quite small Idss. "

Perhaps. Depends on whether you're doing voltage mode or current mode switching. Boss style is voltage mode. If you do it right, the load on the switch is such that not much current ever flows, and you're fine. the J201 can do half to one ma of Idss. That's plenty for switching voltage into a 10K or over load. A current mode switch is usually used for small currents too, but you do have to look at it.

Mostly the J201 is intended for amplification. It can work in the Boss style switch, but since that's not what the makers test it for, the results may be less than spectacularly predictable.

[Toney]

Thanks R.G. insightful, as always!

[R.G.]

One other thing I forgot to mention. When using a JFET as a switch, the "off" voltage on the gate needs to be big enough that the gate is never pulled out of Vgsoff territory by signal excursions. This would let the JFET start conducting again, and let signal leakage happen. This is a problem in some commercial pedals.

For a negative ground pedal, placing the JFET drain and source bias at the positive power supply would make sense for just the JFET bias. However, any noise on the power supply rides right through with the signal. The drain and source bias voltages need to be (a) the same, within millivolts to stop clicking and (b) quiet.

[Toney]

This shows you how to get a gradual switchover from one input to another using a CD4007 or MC14007:

http://html.alldatasheet.com/html-pdf/11935/ONSEMI/MC14007UB/371/2/MC14007UB.html

Use the bottom diagram but instead of the inverter in the southwest corner, just connect the select lines to a D-type flip-flop via R-C delays and connect /Q to D to get a toggle action.  This can be connected through an R-C debounce to the SPST stompswitch.  The problem with individual FET's is that the switching pulse is capacitively coupled to the channel.  With a MOSFET switch of this sort, the effect of the N-channel and P-channel devices tends to cancel out.

Forgot to say "thanks" amptramp for this.
Will be giving it a go and I'll post back a layout once I receive some CD4007.

Thanks again R.G. -
So basically, biasing should be at some mid point, which I assume can be stretched to within a couple of volts of supply for JFETs with a higher VGs off but not at supply. Also, the if VGs  off is too small, transients can cause partial switch on.
- That's how I read it, hope its right.

[R.G.]

So basically, biasing should be at some mid point, which I assume can be stretched to within a couple of volts of supply for JFETs with a higher VGs off but not at supply. Also, the if VGs  off is too small, transients can cause partial switch on.

That's right. Drain and source are generally tied to the signal reference voltage for reasons of economy.

There are a couple of variations. One is to use a P-channel JFET and choose the bias voltage to be 0V. In this case the "off" voltage for the gate is more positive, so switching it to a positive power supply makes it go off. The only reason this is better than N-channel at power supply is that "ground" is the common reference for the circuit, so the amplification stages see no difference from signal reference, and it's effectively quieter if wired right.

Another is one that just struck me as I typed. There is a gate driver for MOSFETs that accepts an LED current drive, but produces a true floating output voltage with internal photovoltaic cells. One of these would drive a truly floating JFET gate off as well.  The drain and source would still need to be biased somewhere and to the same voltage to prevent switching clicks, but the LED-PV isolator would let you put that bias voltage as much as a couple of thousand volts away from the drain/source bias voltage. Not that we're likely to need that in pedals, but it should work.

[Cliff Schecht]

but the LED-PV isolator would let you put that bias voltage as much as a couple of thousand volts away from the drain/source bias voltage. Not that we're likely to need that in pedals, but it should work.

Maybe not but last summer I was working on a (mostly) integrated 1000-3000V SMPS that was meant to go inside of a cell phone . Maybe it's not too long before we start putting micro-sized HV supplies in our pedals. You sure can't beat the headroom offered by a 1000V supply.. 

[R.G.]

!!??@!!!??


... and this was intended for... um? ah... training users not to do X by spark gap aversion therapy?

Um, possibly integrated brain X-rays?

CRT based cell phones?

Vacuum tube cell phones for radiation hardness? Field emission triode based?

Field emission display?

Integrated Tesla coil aps?

Combined cell phone/taser?

The ability to *really* reach out and touch someone, as the old phone company ad said??

Ah hah!! Countably accurate planned obsolescence as the cell phone zaps its controller and memories if you either don't pay the bill or keep it too long without upgrading!

Integrated Geiger-Mueller tube? Scintillation counter? Cosmic-ray detector (bubble chamber style) that also detects neutrinos and WIMPs? ( look it up...)

Wireless wide band Morse Code?

Help me here - I'm running out of plausibilities.   

[amptramp]

If you look at the schematic in the first post, you will see one issue that has been addressed in other designs: the voltage at the JFET gate is somewhat indeterminate, depending on the relative leakage of the gate and the diode connected to it when the gate is in the floating (ON) condition.  Other designs use a unity-gain buffer made of an op amp with output connected to inverting input and the non-inverting input connected to the JFET source.  The output drives a high-value resistor connected to the junction of the gate and the diode that drives it in the existing design.  This allows the gate to be referenced to the source with a finite (but high) impedance, like 2 megohms.  The gate drive can override this if it is significantly lower, but there is no problem with indeterminate drive in the ON state that can lead to dropouts caused by signal voltage spikes or interference.

[MetalGuy]

One is to use a P-channel JFET and choose the bias voltage to be 0V. In this case the "off" voltage for the gate is more positive, so switching it to a positive power supply makes it go off.

+1. Go for the good old J174/175.

[Cliff Schecht]

!!??@!!!??


... and this was intended for... um? ah... training users not to do X by spark gap aversion therapy?

Um, possibly integrated brain X-rays?

CRT based cell phones?

Vacuum tube cell phones for radiation hardness? Field emission triode based?

Field emission display?

Integrated Tesla coil aps?

Combined cell phone/taser?

The ability to *really* reach out and touch someone, as the old phone company ad said??

Ah hah!! Countably accurate planned obsolescence as the cell phone zaps its controller and memories if you either don't pay the bill or keep it too long without upgrading!

Integrated Geiger-Mueller tube? Scintillation counter? Cosmic-ray detector (bubble chamber style) that also detects neutrinos and WIMPs? ( look it up...)

Wireless wide band Morse Code?

Help me here - I'm running out of plausibilities.   

Artificial muscles was the original application, then they started talking about putting this stuff in a cell phone to give a more realistic haptic feedback (think Microsoft joysticks with the haptics stuff built in). It was a pretty cool project for an intern at a big semiconductor company. I think part of the reason I got the job was my experience with both HV tube amps and HV switching power supplies for use in tube amps (of which I've built a few but never shared, maybe someday guys ).

[R.G.]

Artificial muscles was the original application, then they started talking about putting this stuff in a cell phone to give a more realistic haptic feedback (think Microsoft joysticks with the haptics stuff built in).

Cool!

Reminds me of one of my patent applications. I used to see if I could confuse the patent attorneys that had to evaluate disclosures.  This was back when computer mice were new. I described a system where you'd add to the X-Y sense wheels on a mouse an X and a Y brake wheel. The idea was to do touch-sensitive feedback to the fingers when the mouse was moving by braking only. The brakes would be linked to mouse position on the screen, and could impart a "texture" to a window which would tell you whether you were in the window, and discriminate between windows. You could also make windows hard to get into, braking for a fraction as you crossed the edge, or "slippery" by making them harder to move toward the center but zero friction if you were going out.  Likewise, windows could "suck" or auto-center by no friction going in, braking going out.

I gave myself some extra points on this one, as it took the patent guys a month to get back with an evaluation.  They gave me a $50 dinner-for-two award (you really could get dinner for two for $50 then) for being so clever and then did not file for a patent. I think they just didn't want me to bother them any more. 

[FiveseveN]

Somewhat returning to the topic, would you guys say a CMOS solution like the 4053 would be more practical considering the typical applications in pedals/SS (pre)amps than discrete JFET switches like these (or god-forbid relays)?

[R.G.]

Purely a matter of design choice. Depends on what you're trying to achieve. This might be lowest cost, best layout, design with available parts, etc. Both are practical, and both are used in commercial pedals. I have used both professionally.