What is a good Vgs-off value for JFET matching

[cjlectronics]

I'm working on a Univox Micro Fazer and have determined that the weak phasing is a result of weak JFETs. Everything else looks and works without a problem and I pulled the original 2SK34's and they measured between -0.74v to -0.89v on R.G.'s improved JFET Matcher (IJM). Which I think is not good?

So, I bought 50 BF244A JFETs, measured them on the IJM, setup my bench to record values and categorized 3 sets of 4. One particular set has a value of -1.06v. I put these in and the phasing is still weak.

So here are my questions
1. What is considered an optimal Vgs-off value (using the IJM) for the standard Phase 90 circuit?
2. Vgs-off means the voltage at which the JFET turns off. If the JFET is biased at 4.5vdc, is the LFO voltage (at the Gate) the control that turns the JFET on and off... I would think the answer is yes! Vgs-off must be the voltage difference between the Gate and the source...correct?
3. If question 2 is correct, then the optimal Vgs-off is dependent upon the sweep range of the LFO... correct?
4. If this is true then the LFO voltage determines the optimal voltage difference between the gate and the source and I should be able to reduce the sweep range to match my Vgs-off values of the JFETS ...correct?

I better stop there before I either answer my own questions or drive myself into a JFET induced coma.

Thanks in advance,
CJ

[R.G.]

I'm working on a Univox Micro Fazer and have determined that the weak phasing is a result of weak JFETs. Everything else looks and works without a problem and I pulled the original 2SK34's and they measured between -0.74v to -0.89v on R.G.'s improved JFET Matcher (IJM). Which I think is not good?

It's not good that they vary so much from one to the other.

So, I bought 50 BF244A JFETs, measured them on the IJM, setup my bench to record values and categorized 3 sets of 4. One particular set has a value of -1.06v. I put these in and the phasing is still weak.

This likely indicates that the issue is not only mismatched JFETs. Never assume there's only one issue.

1. What is considered an optimal Vgs-off value (using the IJM) for the standard Phase 90 circuit?

Not possible to answer. Ideally it is as big as possible (for widest range on the LFO) but less than the nominal 3V zener bias. JFETs all have about the same Rds range - that is, a few hundred ohms or less up to multiply many megohms. Only about the few-K to 1-2Mohm range is useful in phasers. Essentially all JFETs have that range, varying only in what Vgs makes them be in that range. The J201, for instance, has a Vgsoff of as little as 0.1V. That seems good, but actually is not because signals bigger than about half the Vgsoff modulate the conductivity all by themselves. A better answer is probably that they should be something like the 2N5292 of the orginal P90 or the 2SK30A of a few other phasers. These are in the 0.5 to 4V range, with most of the selected groups about 1 to 3.

2. Vgs-off means the voltage at which the JFET turns off.

Yes. As such, it's not the voltage you want. You want Vgs at a drain resistance of maybe 10K. That's one reason I designed the JFET matcher that way. Vgsoff is not all that useful except for selecting likely part numbers. The actual groups get selected by individual test. But you knew that already.

If the JFET is biased at 4.5vdc,

That is, the source and drain are held at +4.5Vdc.

is the LFO voltage (at the Gate) the control that turns the JFET on and off... I would think the answer is yes!

The difference between the source voltage and the gate voltage is what modulates the JFET channel resistance.

Vgs-off must be the voltage difference between the Gate and the source...correct?

No. Vgs-off is that value of voltage from gate to source that just barely reduces the channel into non-conduction, usually taken as a leakage current in the nano-amps range. Vgs-off is a datasheet number. Vgs is an actual operating number. When Vgs exceeds Vgs off, the transistor is non conducting.

3. If question 2 is correct, then the optimal Vgs-off is dependent upon the sweep range of the LFO... correct?

To a certain extent. It is true that the LFO voltage, including any DC offset, must be in the range of 0V to Vsource. However, the change in resistance with voltage of the JFET channel is nonlinear within that range. So the LFO sweep voltage must be smaller than the Vsource bias, probably significantly smaller when measured at the gate.

4. If this is true then the LFO voltage determines the optimal voltage difference between the gate and the source and I should be able to reduce the sweep range to match my Vgs-off values of the JFETS ...correct?

You're way out on a banana peel. Yes, if everything else is lined up.

I would say it the other way round. For a given JFET (because matched JFETs are harder to get than a certain LFO voltage swing), the bias voltage of the source and drain in a phaser circuit should be comparable to or greater than the Vgsoff of the JFET. This ensures that an LFO voltage which is equal to the bias voltage from peak to peak can run the JFET from full on to fully off. As a second condition, the sweet spot for Rds variation is smaller than zero to infinity, so the actual desired LFO voltage swing from peak to peak is within the 0-to-Vsource range and centered over the 10k-to-1M range of the JFETs at hand - which varies.

I think you're overthinking it. On the microphazer, replace the electros and remelt all the solder joints. Check the Vsource/Vdrain bias voltage, and if it's between 3V and 5V, it's probably OK. Then see the voltages at the peaks and valleys of the LFO. Should be maybe 1-2V wide, and trimmer adjustable from near 0V to nearly the bias voltage.

If that all checks out check for drift in the mixing resistors that mix the dry and phased signals. It's not widely known but the cancellation with those resistors has to be really good to get a notch of maybe 20db.

[Enzo]

Great post RG, I enjoyed that.

[cjlectronics]

Thanks for the explanation RG. You cleared up a lot of questions. However, the original question, What is a good Vgs is still not clear but you answered that by stating it not possible to answer.

I'll go with hands-on learning for now. I got the fazer to work and I have to say it sounds superb. Using the BF244A's (with the -1.06v Vgs) I compared the results with the original 2SK34's and the exact matched set (BF244A's) sounded much better. You were right, there were multiple problems and I suspected that but didn't realize the stomp switch was one of them. The stomp switch would work intermittently and once I replaced it the pedal worked a lot better. It was probably adding resistance to mixing resistors like you stated?

When I compared the original 2SK34's to the matched BF244A's (and the circuit was operational), the 2SK34's weren't as expressive or didn't have as much of rotational pulse as the BF244's (at least to my ear they didn't). I probaly could have gotten away with the phase shifting effect they produced, but when the BF244A's were used, the phase shifting was much better...more of a deep pulsing and thumping sound. I also replaced all of the caps (I did this early in the repair) and re-flowed the solder on many joints.

So, what have I learned:
1. BF244A's will work in a phase shifter (when matched using the IJM). I will assume that BF244B's and BF244C's will work as well.
2. A Vgs value of -1.06v (in and around this value) is OK to use in phase shifters.
3. It is best if all JFETs are closely matched (but RG's "JFET Matching for Effects" paper states that fact)
4. I now look at the JFET (in phase shifting circuits) as a water valve with the flow controlled by the LFO. The JFET is turned on to allow current flow then as the LFO sweeps the valve is shut off and this happens in unison with all JFETs in the circuit.

Thanks again RG!

CJ

[Gokarna]

Also problems with my Univox Mini Fazer

Hello,

I've just bought one of these and it doesn't work correct.

First testing it, the sound was only kind of tremolo-effect and no phaser sound.

So I played around with the trimpot and checked the soldering, 2nd test was no effect at all, but signal comes through

Chicking the pcb again, I found one of the op-amps (TI sn72558P) has a nearly broken pin.

A few questions:

1. What is a equal substitute for the Op-amps ?
I found a seller offering TI MC1458P (LM 1458) DUAL COMPENSATED for a reasonable price, can I use these ?
Can I also use RC4558 op-amps ?
2. I found also comments, that it will bw possible to use SK30A or 2n5484 J-Fets. Is that correct ?

3. I want to change also all the caps. But I have found two different schematic with different values (of the caps)
My pcb is nearly the one, I attached, but has a 4.7uF /35V instead the second 10uF / 16V in the schematic.
So what to do here ?

I'm happy for any help,

Best reagards

Till

[Mark Hammer]

Take a gander here: The inner life, and rehabilitation of, stompswitches - YouTube

Could come in useful at some point in the future.

RE: JFETs and matching. Apparently what is critical is that the JFETs all "move" in response to the LFO voltage. The matching is simply to assure that they are all able to change resistance in response to the specific LFO voltage swing. What you want to avoid is any one or more of them reaching their individual "I ain't a-movin' and you cain't make me" point anywhere in the sweep cycle, or else the turnaround at the peak and valley of the sweep cycle will suddenly get wimpy.

This is apparently one of the strong suits (among many) of LDR-based phasers: the LDRs may not be matched, but their resistance will continue to change in response to changes in the light source.

[Mark Hammer]

I can't see anything about the circuit that would make a 4558 or 1458 of any type inappropriate.

As far as I can see, 2SK30A and 2N5484 JFETs should be okay. BF245 is also good, and 2N5952 will work well. You will need to bias them properly, whichever kind you use. Note that the zener diode shown in the schematic can beof different voltages. The zener simply provides a stable and reliable supply-voltage that can be divided down. You can just divide the battery voltage down for the bias, too, but when the battery drops from 9.4V fresh to 8.7v after some use, you would need to open the unit up and rebias again. Using a zener that is less than the battery voltage means you can bias once, and be content.

I do not know what the on/off resistance range is for the K34, compared to other JFETs. The 100k parallel resistor might be too high or even too low. Remember that the location of the notches will be predicted by F = 1 / [2 * pi * R * C]. With C = 50nf, if the combined parallel resistance of 100K and the JFET got as low as 25k, that would put the notches in the range of around 32hz (a little on the low side) to 127kz. If the JFET+resistor parallel resistance swept back and forth between 60k and 10k, the notches would sweep between 53hz and 318hz. And so on.