dumbest jfet question yet

[pinkjimiphoton]

hi guys. been reading til my eyeballs melt but not grokking jfets at all.

i have a tester, i've sorted thru a whole bunch of these nasty little dead bugs for ids and vgss but i can't make my p90 circuit run.
really weird,
i can hear a bit of noise cycling, and am getting wiggle on pin 7 of my lfo so i'm assuming it works.

all the voltages reading on the sockets for the 4 trannys read identical voltages until i put jfets in.
then it's all over the freekin place.

frankly, at this point if i get this thing running i have a feeling i won't bother with this bull...oney again.

so my dumbest ultimate question is:

can i sort jfets of different types with close to identical pairs? i have tons of jfets of various types, some that seem to read very similar..
but the specc'd 5292's just ain't cutting it.

when i turn the trimmer, i can hear it phase. but i can't get it to kick in.

should i bite the bullet and just buy a "matched set"? after reading posts from the archives until 4 am last nite, i am more fu-nked up than i was before.

so like... if they match up can i stick a j201 in there if the ids and vgss or whatever ya call it match close enough?

i am assuming this is why i can't get this thing working.. i reflowed the board, checked for continuity and shorts, replaced the chips even as well with absolutely no difference in performance.

thanks for the potential responses. remember please i am but an egg, not an EE and technical jargon will likely be met with some nads scratching and head shaking.

can someone explain this shit in a way that makes sense?

[EBK]

Can I preemptively ask J.C. and R.G. to not fight in this thread?

[pinkjimiphoton]

i certainly hope so.
i for one don't need to see it.

i just need to know if its ok to mismatch jfets with similar enough characteristics to be useable.

i don't wanna have to buy 100 devices to get 4 useable ones.. i went thru over 60 of several types and they are literally all over the place.

i gotta little component testor which reads the jfets to permit matching, but even with proper matching... i actually found 5 that were 1.8 ids and .28 vgs i think it was that seemed to be a good match, but i can't get the damn thing to work.

i'm assuming despite matching up they just don't have the gain range required to make it work.

please, no fighting.

it doesn't help anyone, and there are many different roads to truth. every one of us walks our own.

i just need to know if this is possible before i go blow a lot of money i don't have on a bunch of parts i'll end up throwing away

[Kipper4]

If I remember correctly I used 2n5457 or 2n5458s. I didnt bother matching the last set i did.
I bought them all at the same time and just trusted that they all came off the same manufacturing run and I got lucky.

I cant get those 5292's at a decent price or I probably would.

So you adjusted the biasing trimmer and you cant find that little spot where the phasing is at its strongest?


It might be an idea to post up some volts please Jimi.
I know thats gonna hurt your eyes, but not as much as the head scratching will hurt.

Stay strong and Rock on.

schematic

http://www.electrosmash.com/images/tech/phase-90/mxr-phase-90-script-logo-schematic-parts.png

[pinkjimiphoton]

thanks rich.
my eyes are gonna need a break for a day or two i think. even with glasses and meds its hard to see today.
when i get a chance to get to my bench... soon... i will post voltages.

it's actually not a p90, it's a "doctor phibes" board barry at guitarpcb.com sent me a couple years ago. it's based on the p90 but not exact.
i did a couple mods to it... i "univibed" the cap values.
i bypassed the 3rd and 4th stages so i can have p90 or p45 on a switch
i added a switch for the feedback resistor, and added a 50k pot inline with a very small resistor to keep it from "shorting" (22 ohms flameproof if memory serves)

i've tried 2n everything i got here and can't seem to make it fly.
i DO get a weird oscillation when i touch the shaft of the feedback pot where i can hear a very asym wiggle modulation,... very high pitched and barely hearable.

tried 5534, 741 and finally the specc'd tl061 for the oa's, still no difference.

let me get some grub in me and i'll go hit my bench and post up some voltages.

thanks mate!!

ps.. got your email, dude, thank you!!


board wiring



board



schem


bom



the "mod"



just realized i subbed a 500k for the 250k trimmer. gonna put the proper value in (didn't have it at the time)

[Kipper4]

Your welcome Jimi. Have fun with those.
Speak later.

[PRR]

Agree on the bickering.

> all the voltages reading on the sockets for the 4 trannys read identical voltages until i put jfets in. then it's all over the freekin place.

The _DC_ voltages go off?? Something is very wrong with your wiring.

The JFETs do not give "gain" here. They are variable resistors. Same as if you made R5 R8 et al variable. This should not change the DC voltages a bit (not to measure). "Mismatch" may give sub-optimal flangeing/dips, but it should pass signal and wobble some.

[MrStab]

i might be remembering this wrong, but i think one or two people on DIYSB said J201s worked first-time without any matching at all in their circuit, due to some spec they couldn't figure out. i might've imagined it, but i'm pretty sure i read something along those lines.

[bluebunny]

I think that may have been Rob, and it may not have been the Grouse talking... 

[anotherjim]

Hey Jimi - what's the gate voltage compared to +VB? Those gates should never get a diode drop above +VB or it could pull the +in of the amps somewhere bad. Lucky to have the FET's socketed, you can try one (only one at a time) in each position in turn and see if the pin voltages are the same. If any are different, then you have a circuit error at that socket.

With no depth control, you might pull R23 to stop the LFO messing the test readings.

[pinkjimiphoton]

Agree on the bickering.

> all the voltages reading on the sockets for the 4 trannys read identical voltages until i put jfets in. then it's all over the freekin place.

The _DC_ voltages go off?? Something is very wrong with your wiring.

thanks paul, i figured they should be the same voltages on each socket... they are.. but as soon as i put the best "matches" together i can and pop 'em in, the top pin changes voltage. i got tied up yesterday with some family crap and then a gig, so will get some actual voltages this afternoon once i wake up a little bit.
i figured if they all read the same idle, and it changes, it's gotta be the q's. especially since the problem stays with the jfets if i switch them to different sockets.

The JFETs do not give "gain" here. They are variable resistors. Same as if you made R5 R8 et al variable. This should not change the DC voltages a bit (not to measure). "Mismatch" may give sub-optimal flangeing/dips, but it should pass signal and wobble some.

thanks for the simple answer on this. i will try and get some that are all around the same vgss or whatever ya call it. the problem i think right now is that either i subbed a 500k trim for the 250 specc'd (closest value i had at the time) so perhaps the "sweet spot" is too narrow to find on it or the q's i got are just very sub par... looking at rg's curve on his foolin with fets page and the jfet tester, it looks like most of them are around id1.x and a vgss around .28 or so... thats why i thought maybe the gain is too low with these particular devices.

[pinkjimiphoton]

i might be remembering this wrong, but i think one or two people on DIYSB said J201s worked first-time without any matching at all in their circuit, due to some spec they couldn't figure out. i might've imagined it, but i'm pretty sure i read something along those lines.

i did read that. i may have another matched set of j201's.


doing more research, it could be an open pot or switch too, so i will investigate when i get down there. thanks bro!

[pinkjimiphoton]

Hey Jimi - what's the gate voltage compared to +VB? Those gates should never get a diode drop above +VB or it could pull the +in of the amps somewhere bad. Lucky to have the FET's socketed, you can try one (only one at a time) in each position in turn and see if the pin voltages are the same. If any are different, then you have a circuit error at that socket.

With no depth control, you might pull R23 to stop the LFO messing the test readings.

hi jim,
i can't remember without looking at it to say which pin is which, but the top two holes in all 4 sockets read 5.02 the middle reads 5.03 and the bottom seems adjustable from around 3.5- 7 volts or so.
i will confirm when i get down there.
usually in the past probs like this mean a bad resistor or electro in the things i brought back from the shades.
i really wanna use this to replace my univibe with live. much lighter. i am old now .

it couldn't be from me "univibing" the cap values of the phaser, could it?

arrrgh

thanks bro

[anotherjim]

Bottom pins are the gates. Trimmer controls their bias. Top & middle will both be about 5V due to the zener. All that looks good.
Start off with the gates about 2 to 2.5 volt (might have to be lower). LFO should swing that up & down but not take it too close to 5v.

As already said, the FETs Q1 to 4 are being variable resistors to the AC signal and always the same DC at both source & drain ends -  you won't see if they are working in phasing range with a voltmeter. You can only surmise that they "should/could" work if the meter shows gate voltage swings lower than Source pins.

If, at each phasing stage, you lift the -input resistor (R4, R7 etc), and cut the dry path (lift R18), the effect should be like tremolo at the output.


Uni-Vibing? If that means changing the 47nF phase cap values by a lot, then the resistors across each FET (R5,R8 etc) might need changing too?
Have you heard this mod elsewhere? LFO wave on this may mean it dont have the watery warble of the sine + LDR 'vibe. Not said to stop you, I'm just wondering.

usually in the past probs like this mean a bad resistor or electro in the things i brought back from the shades.

If its a known good working scheme+pcb, then that's as likely as anything

[pinkjimiphoton]

 

i got it. i'm checking everything, trying this, trying that.....

went with the j201's reverse biased. got a very weak and subtle vibrato... wtf?

checking the bottom of the board for bridges... and sure enough
not a one.

thats when i noticed daylight peeping thru the board.
i forgot to solder ONE connection (damn dodgy eyeballs!!!!) as soon as i connected it,

viola. or was it voila. i tend to like girls named viola.
anyways.. i re-connected the 90-45 mod, and it works great.
i can't seem to use the feedback resistor deal... i tried subbing it with a 220r resistor to a 50k pot and a switch, that actually worked in p90 mode, but in p45 it kills the whole thing so gonna leave it out.
also, i tried going down to a 20k pot... i knew i bought it for some reason... and all it does is squeal like a pig and oscillate if i turn it even close to half way up.

so... gonna leave the feedback to my guitar on this one.
will probably make the leftover switch a dry kill. i likes vibrato.

i made r17 100k instead of 150k seeking more wet than dry,. seemed to do the trick nicely.


right now i got .047's in there as specc'd. i did go with the univibe values for caps 2-5 of the phaser. i imagine they will work better now.
barry at gtrpcb also has values he reccomended for the thing, i got a set of them picked out too and will see what sounds best.
i think it will be close enough for rocknroll for what i need it to do.

more later.

of course, i just bought a couple matched sets of jfets, right?

[MrStab]

glad you got it working, Jimi! doesn't the feedback path need to go back through an odd number of stages, which wouldn't be do-able in '45 mode? maybe you could switch it out, if you have a spare set of switch poles

[Ruptor]

The circuit would be better if it used a quad pack of FETs since the gate threshold would be the same on all the transistors due to them being fabricated on the same chip.

[aion]

The circuit would be better if it used a quad pack of FETs since the gate threshold would be the same on all the transistors due to them being fabricated on the same chip.

Are there any good options for these still in production? I remember that there were a couple of quad-FET ICs in the 70s and 80s that made it into pedals (the Pearl phaser used an AM79C11) but those are all long gone.

[Ruptor]

From memory the CPH6901 is a dual jfet but maybe there is a quad package in the ON Semiconductor range. This dual one is used for close matching JFets in a differential amplifier.

[pinkjimiphoton]

glad you got it working, Jimi! doesn't the feedback path need to go back through an odd number of stages, which wouldn't be do-able in '45 mode? maybe you could switch it out, if you have a spare set of switch poles

i cannot believe i missed one connection completely

it works great, but i gotta set the treadle up backwards, so the high rate in heel down. it actually kinda makes more sense that way to me.

before i do tho i'm gonna try and add a trimmer to the 500k speed pot. i bought a wah pot from joe gagan recently (he makes THE BEST hands down) and he thru in a 500k gagan pot for me to try.. i figured it would be a c taper, but it looks like it's a b taper. i've read RG's article on this and i think if i add a 100k trimmer to the center and top lug i may be able to spread out the taper so it's a little more useful wired traditionally.
i used a cheap ass pye pro wah shell... trying to adapt joe's pot to that was a bit of a challenge and required quite a bit of nard scratching. the gear was a different size from the rack so i had to change the rack out too. it's a little "stiff" but useable.

i ended up going with the stock univibe caps for stages 2-5  (1-4??) and it has a very good wobble. sounds a lot like my dunlop UV..even seems to take a split second to come up to speed sometimes.
i used j201's reverse biased. seems to work.
i made the 150k wet mixing resistor 100k to make it a little more wet. tried 82k but it didn't improve it at all, actually made it sound worse.
in the end i ditched the pot for the feedback... simply too unstable to work with.. and used a 47k fixed resistor and a switch to turn it off and on.
did the p90-p45 mod and it works great.
yes, the feedback doesn't work in p45 mode but i'm ok with that, frankly i kinda prefer the feedback resistor out of circuit anyways, especially since i'm trying to mimic a u-vibe "sound kinda like".
i stuck a rate led in the pot hole, connected it to pin 6 of the lfo opamp with a 4.7k resistor in line with the anode, and grounded the other side of the led. works great, no tickytickyticky and fills the hole that would have been in there. plus, it's kinda cool to look at... more for the peeps that come to shows than me.
with size 15EEE boots not much light gets around them
but anyways, when all was said and done, it works.
i'll try the matched q's i bought when they arrive, but i think i just wasted my money on them.
down the road when i am gigging a bit more again, i'll mount this in a crybaby shell. joe has a lot of them and has hooked me up with some great deals so i look forward to grabbing a few. it will make life easier... and the pot will no longer be "backwards"... joe's pots are designed for crybabys, with the gear to the left... but the pyle is backwards, with the gear to the right. oyyyyyyy. made my head hurt.

thanks for the advice and support guys... means the world to me.
now.... that said... NEVER AGAIN

lol

i gotta get back on the polyphase project now that i have a bit of a clue. but that will be down the road.
today i think i'll build my first RAT, i have a few original can type ic's i think may sound good in it.

i think that's it for the mods i did. surprised it worked.

incidently, the 500k trim pot was way overkill.. 250k has a very narrow "sweet spot " that it phases in, i couldn't quite hit it with the 500k.

again, thanks to all... peace!