GGG Phase 90 Problem (PCB version)

[rrichard]

Greeting all,
First post, been lurking and researching here quite a bit for while now. It's pretty awesome what you can learn here just by searching for whatever you need. I don't play guitar but I've enjoyed building a few FX's for my #2 son as he really is into music and tone stuff !

Having problems getting Phase 90 going that is built from kit with PCB from GGG. Have found several posts and although very similar and very informative answers,none seem to have the solution for my particular problem.

Most notably are "GGG Phase 90 Problem" and
"MXR 90 not phasing".

Original "GGG Phase 90 Problem" thread turned out to be "solder bridge" on the PCB trace between Source and Gate.

The "MXR Phase 90 not phasing" was apparently caused by lower than "Normal" V-ref and "Bias" voltage, cured by subbing in slightly higher voltage zener.

I've been in touch with JD about this and he said it was either "solder bridge" , improper FET orientation, or highly unlikely but possible the FET's themselves. He has graciously offered to send me another set of FET's @ No Charge and I really like that, hard to find Good Customer Service these days you know!

Wanted to run all the stuff I've checked and done past you guys and try to make sure it's not something stupid (very likely in my part) or something really obvious I'm just not seeing!

Name of the circuit = General Guitar Gadgets P90 (MXRTM Phase 90TM Replica)
Instructions and GGG Voltages, http://generalguitargadgets.com/pdf/ggg_p90_instruct.pdf
Schematic, http://generalguitargadgets.com/pdf/ggg_p90_sc.pdf
BOM. http://generalguitargadgets.com/pdf/ggg_p90_bom.pdf

Built with parts supplied in kit as well as ready to solder PCB supplied with kit, all resistors and caps were sorted and tested prior to installation, did use IC sockets, ended up installing sockets for zener, however NO sockets for FET's. Tantalum caps are proper polarity (little red +++++ signs are on square pads), All IC's are Pin1 on square pad, Diode D2 Zener is black band on square pad, FETs and PNP are properly oriented on board (have checked and rechecked pin out of these and all are correct as shown on board and in circuit)

Only "mods" done is using 25K pot in place of "22k" Feedback resistor, figured that would be more user tweak-able than putting it on switch (I realize it will go into "self oscillation" if cranked full CW), LED resistor changed to 4.7 K.

Here is what I'm seeing happening,
Bypass signal good, Engaged sounds mostly like Bypassed, with only very, very "faint" phasing, so "faint" it's kind hard to tell but seems to get faster and slower with "Speed" knob. To get "faint" phasing 250K trim-pot has to be just so, any more, any less then NO phasing.

Pin and Component Voltages,
DC @ + of board 9.02 positive voltage (Using know good power supply, we don't use batteries)
         - of board 0 volts
D1 (1N914)                          D2 (swapped original 5.1 F231B with what I believe is 5.2 out of spares, tested @ 4.93 w 10K)
A= 0                                  A= 0
C= 9.02                              C= 4.89

IC1 
pin1  0               pin5  0
pin2  4.87           pin6  4.87
pin3  4.65           pin7  9.01
pin4  0               pin8  0

IC 2,3,4,5
pin1  0               pin5  0
pin2  4.89           pin6  4.91
pin3  4.89           pin7  9.01
pin4  0               pin8  0

IC 6
pin1  0                               pin5  0
pin2  4.14 vary to 5.50         pin6  1.25 to 8.34 (cycles faster w/ speed knob more CW)
pin3  3.95 vary to 5.65         pin7  9.01
pin4  0                               pin8  0

Q1,Q2,Q3,Q4 Fet's

Gate      4.84 vary 4.85        with trimmer adjust all gate voltages range from
Source   4.89                      MIN 4.81      Max 4.88  (approx .07 volt range total adjustment and only varying .01 within that range)
Drain      4.89                                 (Checking Gate voltage with respect to ground, not across Gate and Source) 

Q5 PNP

Collector  3.11
Base       4.31
Emitter    4.89
                                                                                                           Looking @ component side
R-19 (150K) side one 4.04 vary to 5.68      side two  4.12 vary to 5.60       side one is toward center of board
                                                                                                      side two is toward outer edge of board
R-18 (3.9M) side one 4.85 vary to 4.84      side two  4.08 vary to 5.65

R-17 (1M)    side one 4.85 vary to 4.84      side two   2.68   moves with trimmer MIN 0  MAX 4.89

Misc testing notes,
Have about gone bug-eyed looking for possible "short" between Gate and Source, see nothing visually,have pulled all IC's and continuity tested the traces related to source and gates with meter not finding any blatant shorts of continuity of any sort

Seems to me I should be getting good swing of oscillation to the gates to let the fets work but just ain't letting gate voltage swing like it should. Could it be the one of the gate legs on one of the fets is somehow just keeping the gate voltage up?

This thing is kicking my butt big time, any thoughts would be most appreciated !!
Apologize for lengthy post, hope it makes sense to someone !!

Rick     

                         
         

[PRR]

Welcome.

Power off, what is the resistance across C10?

Should be half-Meg or so.

Readings suggest it "must" be a few K or less (or my guess is wrong).

Shorted/sick C10. R17 a few K instead of 1Meg. R18 bad joint or bad resistor. (You can tack 1Meg at R18 just to see if that makes it wobble, though it will be too much).

[rrichard]

Hey Paul,
Thanks for checking it out for me, sure enough C-10 tests out @ 13.9K ohms (all powered down)
Gonna desolder that and stick a new one in there
Suppose we just go from there on the other resistor checks.
FWIW, R-17 was tested @ 1M ohm prior to installation
I'll try the C-10 replacement, and then the R18 and R17.

I'll post an update as soon as I'm finished.

Once again, I really appreciate you talking a look at this.
Rick

[rrichard]

I'm back with the promised update.

C-10 was replaced with a tested 47n cap, all I had was a 250v 47n (just wanted to explain why that cap looks so big in the pictures below.)
After replacement, resistance across C-10 is @ 13.7k

After frenching in the 1Meg on R18, the gate at each of the JFETs is 4.21-4.22v.

I'm not sure if it is helpful:
In circuit resistance of R-17 (1Meg) is 14k.
In circuit resistance of R-18 (3M9) is 614k.
(Readings Taken w/ Circuit Unpowered)

Components were checked before installation.
Here are some pictures of the board after C-10 replacement.

Thanks Again,
Rick

Component Side:

Full Size: http://i.imgur.com/OjN0GZR.jpg

Bottom:

Full Size: http://i.imgur.com/0aazNiY.jpg


Full Size: http://i.imgur.com/Kx3AiJu.jpg


Full Size: http://i.imgur.com/XROaXhZ.jpg


Full Size: http://i.imgur.com/kjeF1JD.jpg

[rrichard]

Hey again all,

Just a little more to update,
Went ahead and pulled R-18 and R-17 off board, R-17 measured .999M      R-18 measured 3.79M
reinstalled them both, with as good a soldering as I can do. Problem still remains with very low "swing" @ gates 4.86 to 4.87 now!

Dang it, at least getting better @ desoldering !

I guess my thing is, I grasp the concept of the FET gates needing a voltage more negative with respect to source voltage level to function, I can even understand mostly what the LFO is doing and it's intended purpose. When its all added together and you are feeding (or supposed to be feeding) this "control" voltage to the actual gates, what the heck could be causing it not to function?

Reckon that maybe needs to go in folder #2, things I may one day fully understand !

Thanks again,
Rick
                                                                 

[rrichard]

Got it going !!

Replaced the fets (2N5952) and that got the gate voltage to respond properly ! Short story, Longer story below .........

Would like to thank PRR for his input and taking a swing @ this one for me, his response ultimately helped me figure it out !

Really must say this one gave me more trouble than any other pedal I've attempted so far, however it forces you to get a real understanding of the circuit and that,s really what it's all about for me.

Lessons learned for me on this one,

#1 Even if you have a "Matched Set" of Fets it is in your best interest to build the tester and get an idea of what voltage level they are actually "matched" at !

#2 The 25K pot I had intended to use for the "feedback" control did not work because my thinking was backwards on it, it needs to be
minimum resistance going to LARGER resistance not 25K going to SMALLER resistance, DUH!
One is supposed to use like a 50K pot with whatever min resistance you like in series with pot.

#3 You can bench test your zeners to get better idea of what the voltage, be will be in circuit.

Thanks again to everybody !!
Rick

[Neivid]

Hi Rick, congrats for making your circuit work. Bro, I have a question, which is the (average) measured voltage of the new matched JFet set that you put and make your pedal work?? Please, that info will be so much appreciated for anybody that tries to make the same project of GGG. Thanks in advance mate.

Dave

[rrichard]

In respect to this answer in general...
A strong-basic understanding of the circuit is critical to the interpretation of any answer you will receive. (this is true of basically every project)

In respect to your question:
(TL;DR: Get a matched set of fets and adjust the bias)

For a more complete understanding:
DIAGRAM A:

What I found out was that the actual Vg is immaterial (within reason).
The important part in DIAGRAM A is Vgs, the relationship between gate and source voltage.

Understand the following...
Any functional matched set of 2n5952 n-channel jfets, when biased properly in a working circuit, will operate in this project.
Certain Vgs values will operate more optimally than others. (The better the bias, the lower the risk of op-amp clipping)

Here's some great resources for this project:
R.G. Explains: http://www.diystompboxes.com/smfforum/index.php?topic=79931.0 see reply #2 by R.G.
R.G. Explained: http://www.diystompboxes.com/smfforum/index.php?topic=111358.msg1023928#msg1023928
JFET Matcher Circuit: http://www.geofex.com/article_folders/fetmatch/fetmatch.htm

[Mark Hammer]

All FETs here need to be biased appropriately to a) be able to respond to the LFO without hitting the ceiling or floor (where they will respond no further) and b) yield notches in the human hearing range.

The trimmer sets that bias but what is it trimming?  A: The trimmer is trimming back on a stable bias voltage provided by that zener.  So, it can provide a bias voltage of anything up to 5.1v, but no more.  If you have FETs that require more than that, they are probably not well-suited to being able to swing fully up and down in response to the LFO (which is "added" to the basic bias voltage/current).  But equally important, they would probably not be suitably biased by a stock board (i.e., where the bias is derived from 5.1V).  I'm guessing this is likely why you could get weak faint phasing, but not much more.

But welcome on board, and thanks for a terrific description of the pre-solution board and its functioning.  Led to a speedy resolution.

[Phend]

Reviving this post since it relates to my latest Phase 90 attempt.
It is not a PCB version, it is my own Bread Board build.
Circuit is from the GGG site, which I have added 2 pots.
One for Drive and the other for Feedback with a switch.
Inspected thru the whole process, magnifier, measured components etc.
Q1-Q4 are purchased as matched, they are as I measure them.
Plugged it in and adjusted the trim to get phasing, yes it does work but probably could be better. Drive and Feedback work also.

Here is my question.
The Zener produces only 4.25 volts instead of 5.1 . I have some others on order.
Will increasing the Zener to say 5.6 help ?

Readings and diagram :

Power 9.5

Q1 - Q4 2N5952
G 1.3
S 4.25
D 4.25

Q5 2N3906
C 2.6
B 3.7
E 4.25

IC1
1 0
2 4.23
3 4.05
4 0
5 0
6 4.23
7 9.5
8 0

IC2 - IC5
1 0
2 4.25
3 4.25
4 0
5 0
6 4.23 - 4.27
7 9.5
8 0

IC6
1 0
2 4.2~4.6 Varies Wildly
3 3.9~4.5 Varies
4 0
5 0
6 3.0~5.9 Varies
7 9.5
8 0

C10 .5 to .9 Meg depending on drive pot
R18 3.9 Meg
R17 with 510K lifted is .9 Meg




 

[duck_arse]

Here is my question.
The Zener produces only 4.25 volts instead of 5.1 . I have some others on order.
Will increasing the Zener to say 5.6 help ?

if your trimmer is not against the stop at one or other end, then no, more zener volts won't probably help. if the fets will bias happy, the zener voltage is good enough.

[Kevin Mitchell]

Don't worry about the zener. You have to consider a few things;
-The R16 10K current limiting resistor
-The tolerance of the zener (typically 5%, 10% is common too)
-The summing of the input signal through R2

"Probably could be better" doesn't suggest anything is wrong. But, I will say that in my experience the phasing doesn't quite cut through much without beefing up the input signal.

[Phend]

Thanks Stephen and Kevin for the guidance.
The trim is not near the stops, plenty of adjustment back and forth.
R10 does measure at 9.9K, so that is good.
Actually it does sound good, swapped 3M9 for 3M3 and went back to 3M9.
Changing the Feedback resistor (a little) to hear what that does.
Thanks again, it works, now for the box face plate art work, the other fun part !

[Phend]

Changing the Feedback resistor to 12K made a Huge difference. Not 10K.
 That as shown in the pic I posted earlier, R15.
(I have seen this mentioned somewhere in the deep abyss of Phase 90 on the Forum)

[jorg777]

Hi Rick, congrats for making your circuit work. Bro, I have a question, which is the (average) measured voltage of the new matched JFet set that you put and make your pedal work?? Please, that info will be so much appreciated for anybody that tries to make the same project of GGG. Thanks in advance mate.

Dave

It's important to note that the Vgs threshold voltage for most JFETs varies WIDELY from unit-to-unit; generally by about 3:1 (1V on one piece, next one out of the bag might be 3V).  The bias that works for one person is meaningless for the next person with a different pair of JFETs.

[amptramp]

The zener diode is being biased at less than 0.5 mA, so it may be that it is not conducting at all.  I have modified circuitry before that had the same problem - not enough bias current to actually turn the diode on and in this circuit, the current is also diverted through C6 and R22.  Old electrolytics can be a problem due to electrical leakage.

[Mark Hammer]

I contend that the role of the zener is to provide a stable voltage to the bias trimmer, such that the trimmer setting remains valid until the battery is too weak to power the audio path.  That's why the voltage of the zener is chosen as it is.  Like so many older pedals and many of their critical components, they are based on the assumption of battery-power, and compensating for how they change with extended use.  In theory, given that most of us operate our pedals from an external power brick, rather than batteries, that zener is not critical anymore, although eliminating it would likely require a slightly different biasing arrangement.