Phase 45 (w/ Univibe mods) Troubleshoot

[Rob Strand]

I'm thinking it's something with the jfets. I actually completely replaced the IC's yesterday and have the same issue.

Thanks for all the help, I'll perform the trim pot test next and report back.

Ah OK, so yes, the JFETs are at the top of the list.

The weird thing is plenty of people have built Phase 90's with very low VP JFETs and they never complained about distortion issues.   The Phase 90 doesn't even have the linearization nework that the Phase 45 has.   The linearization networks are the 10k's and 10n caps around the JFETs, they reduce distortion when the signal level is high.   The downside of the Phase 45 is it boosts the signal at the front end which will promote distortion.  The Phase 90 doesn't do that as it actively mixes the clean and phased signals at the output.    I have a old Ibanez 4-stage phaser and it boost the signal at the front-end and passively mixes at the output like the Phase 45 and I can't say I've noticed excessive distortion from that unit.

We don't really know if you JFETs have a low VP, maybe they don't.  That would add more to the mystery.

This is why it would be good to check the JFET pinouts.

You might be able to make some small improvements by change the 2x10nF caps on the JFET gates to 50nF.   However, it's probably best to pursue all other angles before doing unnecessary mods.

I forgot to mention JFET phasers are known for not being absolutely clean.  To some that's the charm of them.  In your case we need to know if it's beyond the normal amount.

[DJPsychic]

Lifted long 10k and set trim pot to both extremes, clean on both sides.

Have another quad-matched set and (2) remaining from the quad I used for this pedal.

What is the best way to check V? What am I looking for?

could I use my tranny tester?

- as far as the amount of distortion, it may be normal. It's just not a very pretty sounding phase. When I play singular notes they get kind of muddied in the widest point of the modulation. (Maybe narrowest I don't know! Lol) just a certain point of the cycle it hits a thread hold. Again could be normal.

[Rob Strand]

Lifted long 10k and set trim pot to both extremes, clean on both sides.

So it's pretty clear now the objectionable distortion is due to the JFETs.

Have another quad-matched set and (2) remaining from the quad I used for this pedal.

What is the best way to check V? What am I looking for?

could I use my tranny tester?

You can measure the VP parameter for the JFET using a simple set-up like this.   Don't worry too much about the
corrections on the side panel.  It's really just documenting the small errors due to the test set-up.



If the VP value is low there is generally more distortion.  Typically you might get VP values around 1.5V to 2.5V for phasers but some of the ebay JFETs can have VP below 1V.

- as far as the amount of distortion, it may be normal. It's just not a very pretty sounding phase. When I play singular notes they get kind of muddied in the widest point of the modulation. (Maybe narrowest I don't know! Lol) just a certain point of the cycle it hits a thread hold. Again could be normal.

The distortion is definitely not a black and white.   It's always there to some extent.

For a given set of JFETs you can only go so far in reducing the distortion.

So what else can happen is if the JFETs have a low VPs, in relative terms the LFO circuit modulates the JFETs more.   If this is the case, when the modulation changes direction you can get very jerky turn arounds.     This can be fixed by changing a few resistor values.   A few years back a forum member built a phase 90 with very low VP JFETs.    With some tweaking the sweep was made to sound good.   I might add despite the low VP's he didn't complain of a lot of distortion.  Also, I don't remember a lot of distortion being present in the sound samples he posted.

[DJPsychic]

Thanks rob. I'll try another set of JFETS when I get to my bench again.

The cycle is a bit jerky, hard to dial in nice "round" modulation. It can be done, but mostly "square" sounding

[Rob Strand]

Thanks rob. I'll try another set of JFETS when I get to my bench again.

The cycle is a bit jerky, hard to dial in nice "round" modulation. It can be done, but mostly "square" sounding

OK, that might explain what's going on.

Checkout the mods on this post,
https://www.diystompboxes.com/smfforum/index.php?topic=122690.msg1159079#msg1159079
https://postlmg.cc/JshVHfzQ

The second circuit will work but third is likely to be better.

I don't recommend changing the resistor to 100k or adding the 150k.   That might be going too far the VP's in that post were quite low and the small values only apply to that case.   The large change from 1M to 100k meant we needed to change the 47nF (50nF) cap to 220nF.  Smaller changes to the resistor won't need to go as high as 220nF.

You could just add 1M in place of the 150k shown in the third circuit.  The 150k is an added component.  A 1M pot might do to find the right value.  If  1M outright is too low, ie.  too much loss of sweep, then perhaps a 1M resistor in series with a 1M pot in the place of the 150k.  Any change or adjustment to that resistor is likely to need a  bias trimpot adjustment.

Measuring the VP helps guide the choice of the resistor a bit better but at the end of the day the thing that will fix it will be tweaking that resistor - which you can do with measuring the VP's.  If you measure the VP's we know what we are dealing with.

[Rob Strand]

Something worth mentioning is when you test JFETs with large source resistors, noise can can affect the reading quite a bit.  The reading can seem stable but the value is wrong.    Placing a 1 uF film cap across the source resistor/meter leads can stabilize the readings.

[DJPsychic]

Something worth mentioning is when you test JFETs with large source resistors, noise can can affect the reading quite a bit.  The reading can seem stable but the value is wrong.    Placing a 1 uF film cap across the source resistor/meter leads can stabilize the readings.

Thanks Rob, good to know.

Waiting on a different set of JFETS from GGG.

I tested the ones I have and they all read 2.7-2.8

I may have been testing incorrectly though.

[Rob Strand]

I tested the ones I have and they all read 2.7-2.8

I may have been testing incorrectly though.

If the 2.8V is the real value it's going to be hard to reduce the distortion.     It's quite possible noise has stuffed-up the measurements.

If you doubt the measurements you can do a cross-check with different source resistors.

What you do is measure VGS in the test circuit with 10M (just the DMM), then add a 1M source to ground (ie. in parallel with the DMM), then repeat with 100k source to ground.   

Next, apply the correction factors in the table (for say rd_on  300 ohm) to convert the VGS measurement to VP.   100k isn't in the table but for 100k + 10M DMM the correction is about 0.925.

In theory the VP measurements should all agree but in practice they don't quite match exactly.   However what you will find is they are reasonably close.

If the VP values disagree then something might be wrong with the way you did the test.

It's not hard for the 10M test to be off if you are touching any of the wires  but it can be off even if you aren't touching them.  Sometimes powering the test circuit with a battery is more reliable than a DC power supply.   I have a 1uF film cap in my junk box I use for testing high impedance circuits.

Using lower valued source resistors tends to make the reading more reliable/stable.  The only down-side is the Vgs reading is a bit lower than the actual VP value (as can be seen from the small corrections in the table).

[DJPsychic]

Received another matched set today, they measured 2.5 & 2.48v

Same issue.

I think I'm going to put this project on the back burner for the time being.

I've done all I'm willing to do for now.


Thanks Rob for all the great tips.

[Rob Strand]

I can't see anything obvious.   The VPs are nice and high and the problem is consistent with different JFETs.  There's probably one silly thing not right.    Maybe if you give it a break and come back with fresh eyes you will see it straight away.

[fowl]

Pardon me if this has already been addressed, I didn't read the thread, but your trimmer appears to be in a different location and/or orientation than the layout.

[DJPsychic]

Pardon me if this has already been addressed, I didn't read the thread, but your trimmer appears to be in a different location and/or orientation than the layout.

I think it's just the photo but the trim pot looks to be in the right location 

[fowl]

I guess it's just a weird angle, but it just doesn't look right.

[DJPsychic]

That trim was a big big but was in the right spot. I've since changed and appears to be correct as well.

[fowl]

Right, sorry, I should have looked at your back side picture in your first post.

Carry on.

[DJPsychic]

Right, sorry, I should have looked at your back side picture in your first post.

Carry on.

All good, The more eyes on it the better. I was hoping you were right!

[Rob Strand]

Early on in the thread you had 2N5952's, which matched the layout.

I'm not sure what the new devices you got were.   For parts like the 2N5457's, the pinouts are flipped compared to the 2N5952's.

[DJPsychic]

Yes sir, I checked the pinout. They are 2n5485.

The pic I just posted is a bit older. New set are reversed correct.

The only other mod I've done is I raised the 150k to 220k (I think)  to match bypass unity volume.

But I was having the distortion issue before that so not the cause.

[Rob Strand]

Yes sir, I checked the pinout. They are 2n5485.

The pic I just posted is a bit older. New set are reversed correct.

Cool, I was just making sure.

The only other mod I've done is I raised the 150k to 220k (I think)  to match bypass unity volume.

No problem there.

But I was having the distortion issue before that so not the cause.

Yes, it was a long shot.  The thinking was it's possible a misconnected JFET can act like a diode.   You can actually get diodes to work in a phaser.  Normal they distort.   I saw one professional design which used 20 diodes in series to keep the distortion down  (!!!).