Debugging Valve Wizard Small Time switch

[PBE6]

I finally finished building Valve Wizard's Small Time PT2399 Echo, and it sounds great! Only trouble I'm having is the switching mechanism. In this case when the unit is "off", it's still engaged

For the test I'm using a 8.60 V battery. The "9V" connection is 6.20 V after the power filtering section.

When the circuit is ON, I get:

D: 3.11 V
S: 3.10 V
G: 0.07 V

When the circuit is OFF, I get:

D: 3.11 V
S: 3.03 V
G: 0.00 V (sometimes -0.01 V briefly)

Do these seem right? Everything else is working fine, including the LED. The darn thing just won't shut off!

[Groovenut]

In one state the gate voltage should be greater than about 3V-6V, in the other state it should be 0V. It looks as though you may have a bad connection at the gate. Or perhaps the diode is the wrong way round?

[PBE6]

I checked and the diode is installed as per the diagram, but the diagram has the cathode facing the power supply/ground. Is that correct? How does that work? I figure pushing current uphill against the diode would limit the voltage at the gate to the small voltage associated with the reverse leakage current.

Here's the diagram:
http://www.valvewizard.co.uk/smalltime.html

[merlinb]

For the test I'm using a 8.60 V battery. The "9V" connection is 6.20 V after the power filtering section.

That's a lot of voltage drop... doesn't sound right.

When the circuit is ON, I get:
D: 3.11 V
S: 3.10 V
G: 0.07 V

These look OK. The gate voltage will be pulled down by the multimeter, making it look like zero volts when it isn't normally. It it pulled up by the FET's own leakage.

When the circuit is OFF, I get:
D: 3.11 V
S: 3.03 V
G: 0.00 V (sometimes -0.01 V briefly)

Correct.

What FET are you using?

[PBE6]

I used a J112 for the JFET. I made one substitution in the power section, I used a 1000uF cap instead of a 100uF cap, but I don't think that should make any difference should it?

Ok well at least the JFET voltages seem about right. I probably made a wiring error somewhere else. Just to be clear for myself, the only elements seeing the switch in power are the LED/resistor and the JFET gate, correct?

[PBE6]

BTW, the delay sounds great!! Thanks for posting this project.

[PBE6]

Hmm..just realized that 6.20 V is in the ballpark of (V+ + Vref)/2 = (8.20 + 8.20/2)/2 = 6.45 V. I wonder if I made a goof somewhere and V+ and Vref are fighting it out?

[Groovenut]

You can try checking the voltage in both modes at the cathode (striped) end of the diode. It should be either V+ or ground.

[blackieNYC]

Forgive this stupid answer but I won't sleep until someone suggests-
   This isn't true bypass (for the tails) , and I could see myself thinking it's always on.  The feedback ringing on after switching it off, or putting it in Off yet hearing it go silent when the power plug is pulled.  I only offer this because Merlin finds your FET voltages to be ok.
(99% chance) you may now return to your loftier discourse.

[PRR]

> Is that correct?

Yes.

And it does look odd. But the diode leaks more than the JFET gate. (The diode is processed for extra conductivity, the JFET Gate junction is not, so the diode has more back-leakage.) This is reliable: this idea is used hundreds of times in many consoles.

"Assume" the diode and JFET are good. Measure the voltage at the 1K+10ufd junction. This must go up and down full swing. If the cap is very bad (backward or solder-blobbed), it won't. If the 1K is very bad (poor joint or wildly huge value), it won't. Also go back to the switch, because they are at least as dubious as Rs and Cs (and far more dubious than a D and a JFET).

[PBE6]

Aha! I found the problem. It seems my dinky little run down battery wasn't providing enough juice to turn off the JFET. I tried it with an adapter that was putting out something like 9.7 V and now it switches on and off with the best of them! The gate reading is now something like -0.70 V when OFF, I guess the -0.01 I had previously just wasn't enough.

When checking over the circuit I did notice that I made another substitution (!!) in the power section. I added a 100 ohm resistor after the diode to ensure the cap would provide at least a 30 dB reduction in 60 cycle hum, but that drops the voltage significantly. Is this resistor superfluous? Will the resistance of the diode itself form a low pass filter with the cap? If so, I can just bypass the resistor and get some headroom back. Thoughts?

Thanks for all the helpful replies BTW!

[Cozybuilder]

Instead of the 1N400x, for reverse polarity protection why not use a 1n5817 and get a few tenths back, or a BS250 P-channel MOSFET (D to 9V, G to ground, S to V+) and get almost all the voltage (my builds usually drop 1's to 10's of mV)? And you can join the magic smoke club if you put the mossy in backwards 

[Groovenut]

Great to hear it's working for you!

This circuit draws about 25mA or so. The diode is going to give a .7 volt drop leaving 8.3 volts. If you use ohms law (V=I*R), you get a 2.5 volt drop across the 100R resistor, leaving you only 6.1 volts to work with for your supply. The 100u//100n caps should provide enough filtering for the average wall wart supply. You have extra filtering built in to the 5 volt regulated supply for the PT2399 and opamps have a pretty decent power supply rejection ratio (PSRR), enough to handle whats left over from the input filter pair. IMO the added 100R resistor is not needed with this circuit.

[PRR]

> I added a 100 ohm resistor after the diode to ensure the cap would provide at least a 30 dB reduction in 60 cycle hum

Why?

Only the TL072 and the FET switch eat the 9V direct. The TL0 has plenty of power supply rejection. The FET switch comes out at 1K+10uFd which is non-negligible buzz rejection. (If you can truly prove buzz modulation of FET conductivity, the 1K could be 10K or so.)

Vref is 10K+10K||10uFd, well cleaned.

Everything else is on the 78L05 which will usually clean up any supply that isn't utter crap.

Unlike many simple pedals, this PT2399 eats LOTS of current. So 100 Ohms is lots of drop.

And the computed ~~6V left is well below the TL072's happy-zone. (Also doesn't leave the 78L05 enough headroom to deliver a full 5V.)

You probably also have a high-limit Vto JFET. A low-limit 1V Vto FET would turn off with -1V gate voltage. High-limit JFETs may need -3V or -4V to really turn-off. Sitting at half of 6.1V it is pretty marginal.

[PBE6]

I just realized that the only reason I put the 100 ohm resistor in was because I had hum issues in some high gain pedal builds, of which this is not one! I have bypassed it now and it's working fine.

I also decided to put a 130k resistor across the DELAY time pot as I wasn't a fan of the degraded sound on longer delays. Now the pot only goes up to about 36k instead of 50k and I find the whole range useable.