ADHD jfet voltages

[DGWVI]

Have one of these worked up. The distortion and true bypass work, but the buffered bypass is faint and fizzy sounding. Anyone have voltages?

 

[Chuck D. Bones]

I suspect you got one or two hot 2N5457s. Their Vp can be anywhere between 0.5V and 6.0V. Anything much higher than 3V won't work for Q6 in this circuit. The gates (pin 3) of Q5 & Q6 are at 0VDC. Measure the source lead (pin 2) of Q6. if it's close to 4.5V then we know Q6 is too hot. For comparison, measure the source lead on Q5. I'm guessing it's over 5V.

I'm going to assume that Mr. PedalPCB replicated the OCD v2 faithfully and Mark Fuller wired Q5 upside-down on purpose. JFETs will work either way, but the biasing and gain will be different.

If Q6 turns out to be too hot, we have at least four choices:
1) Cherry-pick a 2N5457 with a Vp less than or equal to 3V. This presumes you have a few spares.
2) Use a different part # JFET with a more favorable Vp.
3) Wire Q6 "backwards" like Q5 by swapping leads 1 & 2. In theory, this should work, but I haven't tried it.
4) Don't use buffered bypass.

This will take some time to sort out. If we're lucky, someone with a working ADHD / OCD v2 will chime in.

 

[DGWVI]

Swapped out the 5457s for 2sk30s (didn't have any 5457 or 5458 left, and saving my j201s), and it works splendidly.

 

[Chuck D. Bones]

Did you measure the voltages after the change? Might be helpful for others having similar problems.

 

[DGWVI]

Q5
D- 1.16
S- 7.65
G- 0.01

Q6
D- 7.76
S- 1.17
G- 0.01

 

[Chuck D. Bones]

Q5's source and drain are effectively swapped. JFETs can do that. Drain current on both is around 110uA, pretty low, but if it sounds ok, then that's all that matters. In this circuit, maximum headroom occurs with drain current at 225uA for both FETs.

 

[DGWVI]

Q5's source and drain are effectively swapped. JFETs can do that. Drain current on both is around 110uA, pretty low, but if it sounds ok, then that's all that matters. In this circuit, maximum headroom occurs with drain current at 225uA for both FETs.

The buffered option doesn't sound appreciably different for the hardwire bypass, so I'm assuming it'll be fine

 

[Bravin Neff]

I'm following this conversation intently, as I'm getting ready to build my own ADHD. What I'm missing is... my board doesn't have any of the transistors labelled as normal (Q1, Q2, etc.), but rather just by their type (2N7000, etc.). It's not hard to zero in on what you guys are talking about, except for stuff like this:

I'm going to assume that Mr. PedalPCB replicated the OCD v2 faithfully and Mark Fuller wired Q5 upside-down on purpose.

and

3) Wire Q6 "backwards" like Q5 by swapping leads 1 & 2.

How does one tell what is backards and not, since graphically there is nothing to go by one way or the other?

 

[Chuck D. Bones]

As long as your pickups aren't too hot or you don't have a boost in front of the ADHD, it should be good to go.

 

[Chuck D. Bones]

I'm following this conversation intently, as I'm getting ready to build my own ADHD. What I'm missing is... my board doesn't have any of the transistors labelled as normal (Q1, Q2, etc.), but rather just by their type (2N7000, etc.). It's not hard to zero in on what you guys are talking about, except for stuff like this:



and



How does one tell what is backwards and not, since graphically there is nothing to go by one way or the other?

There is enough information, it just takes some effort to sort it all out. The older build docs used to label the ref designators. Made life real easy. Now, if we want to figure out which device is which, we have to do it the hard way. It's kind of a ♪ Leg bone's connected to the hip bone... ♫ sort of thing. Compare the schematic to the board (if you have a board) or the board artwork on the website (if you don't). Look for traces connecting 2 or 3 parts together and then find those things on the schematic. One of the JFETs has pin 2 connected to two 22nF caps, that has to be Q5. One of the JFETs has pin 1 connected to a 100R resistor, that has to be Q6.

Now about upside-down JFETs...
PedalPCB uses JFET schematic symbols that are unambiguous. In the case of the ADHD schematic, notice that on Q6, the gate arrow (pin 3) points right at pin 2. That identifies pin 2 as the source. Also, you can download the 2N5457 datasheet from the web and it says that pin 2 is the source. It's the same way with Q5, except for whatever reason the pin numbers are hidden on the schematic. The gate arrow points at the source and if you trace the board like I did, you can see that pin 2 of Q5 connects to C1 and C13.
Now here's where is gets a little weird...
JFETs don't really care if you swap source & drain, at least not at audio frequencies. With N-channel JFETs, whichever pin is at the higher DC voltage, that's the drain. So in the case of Q5, even though the board artwork and the schematic identify the lead that connects to C1 & C13 as pin 2, it behaves like the drain because it's at a higher DC voltage.

Clear as mud, right?

 

[Bravin Neff]

There is enough information...

I'm gonna have to read that a couple more times, LOL.

 

[Paul.Ruby]

Building mine now and definitely must cherry-pick the jfets. I have 5 good and 5 bad out of 10 bought. I strongly suggest builders breadboard this and check DC bias levels using this circuit. Shoot for ~2V on the source and ~7V on the drain, but there is a wide range suitable, given guitar signals are sub 1VAC. Reversing the drain and source had no effect on any given device to be suitable or not. I would actually bet the "upside-down jfet" was a layout choice, not circuit function choice, of whoever did this originally. It's the kind of thing I would do when building a prototype or laying out a PCB by hand. FYI, I tested mine to be suitable up to 1Vpp at 1khz with the full circuit implementation. This simplified DC bias test is plenty good enough for cherry picking with confidence. My "good" jfets all landed at ~2.5V on the source, ~6.5V drain, which is ideal. If you only have "bad" jfets, then the 10k resistors can be tweaked up or down BUT must both remain the same value. Use the breadboard method to pick resistor value instead. The 27k resistor in the full circuit to get unity gain does not need to be tweaked.

 

[Paul.Ruby]

Also, the circuit could be made immune to jfet variation by replacing the two consecutive jfet stages with one source follower, like this. Can still be a buffered bypass. The only downside is it's not quite unity gain. Unity gain doesn't matter at all feeding to the OD circuit. If using a true-bypass instead of buffered, then no downside, simpler and immune to jfet variability. The original circuit is also limited by the zener diode to ~500mv negative swing, so I'm not installing that in my build and not shown here. This alternate circuit is also not limited by the zener if it is desired but I would still leave it out so that the circuit works fine up to 18V supply. (I imagine this might be rehashing old news, but I didn't see this discussed elsewhere.)

 

[micmac]

Hi Paul, hi all,

Sorry for the necro post Last entry in here happened even before Corona I reckon.

I tried the measurement circuit for the JFET you posted (2 x 10k, 1 x 1M, JFET, DC@9.18V). I tested the two JFETs I used in my ADHD breadboard circuit. All Vp and Idss measurements done on my T7 tester. It's cheap, but I did some measurements on a breadboard earlier, using the Fetzer Valve circuit for measurement. The T7 is not exact, but in the ballbark.

Q5 2N5458
Vd 7,60
Vs 1,56
Vp 1,86V
Idss 4,8mA

Q6 2N5458
Vd 7,50
Vs 1,66
Vp 2,04V
Idss 4.1mA

You wrote: "~2.5V on the source, ~6.5V drain, which is ideal". My two JFETs are a bit away from ideal then, it seems. But these two fets in the ADHD can handle 2.5V peak-to-peak on the input with _very_ little clipping. The output is around 2.7V ptp (with the tiniest bit of clipping).

I have some 2n5858s with Vp around 3V. So I put them into the small test circuit as well:

a) 2N5458 #1
Vd 6,70
Vs 2,47
Vp 2,99V
Idss 5,5mA

b) 2N5458 #2
Vd 6,85
Vs 2,32
Vp 2,82V
Idss 5,3mA

These measure closer to the voltages you wrote were ideal. But when I tested these in the ADHD circuit I saw them actually clipping the 2.5V ptp input pretty hard. I also have some 2n5457 with Vp of ~1.5V. They clip this signal as well.

Chuck wrote above that drain current at 225uA has most headroom. Not sure if below formula applies for the ADHD, but here it goes:

Vp/Rs=Id
Vp=Id*Rs=0,000225A*10000Ohm=2.25V

So for drain current of 225uA you'd need a Vp of 2.25V. If that's optimal headroom then the two 2n5858s I have with Vp~2V seem to be my best shot. And my scope seems to agree, as these clip the signal the least.

But down the rabbit hole I go and try, like suggested, different drain and source resistors (always pairs!). I try this with a 2n5857 to see how the changed resistors influence the voltages. I also measure drain current with the same multimeter.

Vp 1,50V
Idss 3,8mA

@18k
Vd 7,86
Vs 1,21
Id 72,6uA

@10k
Vd 7,92
Vs 1,31
Id 125,1uA

@7k5
Vd 7,96
Vs 1,21
Id 162,1uA

@5k6
Vd 8,03
Vs 1,17
Id 209,2uA

@4k7
Id 243,4uA

@3k3
Id 329uA

So, varying the resistors doesn't really change the voltages in any big way.

But anyway, at least the drain current changes. The resistors at 5k6 get me pretty close to Chuck's 225uA. So I put them into my ADHD circuit and change all four drain/source resistors that go to VCC or ground to 5k6. First thing to note is that the drain and source voltages are different to the ones in the small measurement circuit. Then I measured drain current, alas far away from 225uA, more like 150uA.

Then I connect the scope. I see no clipping. Yay! But I see the signal has been attenuated from 2.5V at the input down to 2.1V at the output of the ADHD's buffered bypass. Not good.

Sorry for the long post. But maybe it saves somebody else some time.