SIx Stage Phase B. Report

[petemoore]

I recieved my 100 x 2n5952 jfets int he mail yesday!!!
 I began to test them out in the Improved Jfet matcher.
 To hold the tested jfets next to a marking of the measurment, I used a piece of perf, with a paper taped to it, poked holes down a column three holes wide [using a 1/2w resistor lead]...drawing the Jfet at an angle across the three holes, the leads would find the holes pretty easy, and the wedge of the three holes/leads firmly holds them...then I used a fine tip pen to mark the readings next to the Jfets. Anything that holds them and you can mark would do. You need something to hold them...this was handy.
 I tested them all, and figure about 1 in 10...maybe one in eight, could form a tightly clustered group. I tested about 75 Jfets and found a very tightly grouped Vgsoff in 6 individuals for the phaser.
 PHasers are scary...first times through testing...takes some knob twiddling before even a little phase is detected, I thought I had a 'real ' problem..
 I checked the OA voltages and did other stuff then went back to tuning/twiddling when I noticed a littel phase sound as I adjusted the bias pot.
 NExte theing I know I had it phazing PDGood, and swapped the TL082's for the NE5532's...I started using the 7 capacitor sockets, 1 for each phaze stage, and the input cap.
 I tired "Uni' type 'staggered' values, with 10x difference, kind of wobbly sounding, increased 'tremolo' pulse type sound.
 I also tried all six stages using .05uf's [473's] like as per MXR schematics, which sounded pretty good.
 Right now it has a small 'hilliness' to the caps choice, something like .047uf, .022uf, .15uf, .047uf, .068uf, .047uf, .022uf.
 I think I've made an excellent sounding phazer !!!
 I also thinking about the truth of LDR phazers...maybe need to try more of that to induce a deeper phase from each stage, I was reading how Mike Irwin somehow found Jfets 'napping' on the job, and how he likes LDR's cause they go all the way up and down...don't go to sleep mid sweep or something to that effect.
 ANyway I'm liking this sweet six stage phazer   , and wondering how close I really am to boxing it up...if there's somthing else I should look at before 'closing the lid' on this one.
 Thanks to RG and everyone for helping me out on this !!!
 The improved Jfet matcher works great...I'll have to pick a second cluster from the 5952's for my Phaze 90 also...I got that one sweeping with the 'ol swap till you drop or it starts sweeping method !!..The IJ Matcher is much more effecient...once it's wired right !!!

[RedHouse]

Can you explain more about what you went through getting the matching thing going using the matching jig? as I recall you were having some issues and I'm interested to know whay you did to get things working if you don't mind typing a bit about it.
(I'm going to be getting a load of FETs and match-up some sets in the near future)

Did you use the advanced FET matcher from RG's site?
(the one with the PCB)

[petemoore]

Can you explain more about what you went through getting the matching thing going using the matching jig? as I recall you were having some issues and I'm interested to know whay you did to get things working if you don't mind typing a bit about it.
(I'm going to be getting a load of FETs and match-up some sets in the near future)
 
Did you use the advanced FET matcher from RG's site?
(the one with the PCB)

I used the Improved Jfet Matcher GEO.
 What you need to know is: when the matcher is wired right  it works just great !!!
 I used a solid core lugged socket...wouldn't recommend the ribbon type [too flimsy, you want a very strong physical mounting] and used enough solder to get a nice 'hershey kiss' on the pads, also used 5 pins adding strength to the 'sides', the 2 extra pins not needed for connections.
 Also helps to make a board that can be solidly mounted to something so you don't have to hold it while plugging Q's in.
 I got a piece of perf, taped a paper to it same size, then poked holes in three adjacent columns, all the way down...this holds the tested Jfets, and leaves a place to mark the readings on the side...If I had that stuff Opamps come in from mouser...it's like foam...that'd be nice for holding Jfets. I like having them through the paper so I keep the transistor at it's marking on the paper...EZ to mix 'em up and have to do re-testing...

[petemoore]

Can you explain more about what you went through getting the matching thing going using the matching jig? as I recall you were having some issues and I'm interested to know whay you did to get things working if you don't mind typing a bit about it.
(I'm going to be getting a load of FETs and match-up some sets in the near future)
 
Did you use the advanced FET matcher from RG's site?
(the one with the PCB)

I used the Improved Jfet Matcher GEO.
 What you need to know is: when the matcher is wired right  it works just great !!!
 I used a solid core lugged socket...wouldn't recommend the ribbon type [too flimsy, you want a very strong physical mounting] and used enough solder to get a nice 'hershey kiss' on the pads, also used 5 pins adding strength to the 'sides', the 2 extra pins not needed for connections.
 Also helps to make a board that can be solidly mounted to something so you don't have to hold it while plugging Q's in.
 I got a piece of perf, taped a paper to it same size, then poked holes in three adjacent columns, all the way down...this holds the tested Jfets, and leaves a place to mark the readings on the side...If I had that stuff Opamps come in from mouser...it's like foam...that'd be nice for holding Jfets. I like having them through the paper so I keep the transistor at it's marking on the paper...EZ to mix 'em up and have to do re-testing...

I laid the board with the paper down on a thick cloth, then poking the holes didn't rip it, and I could go from the back where I can see the hole, I used a 1/2w resistor as the punch.
 Pulling the resistor at an angle [and slight pressure] to the paperpunched over the board holes, the leads 'find' the holes easily. Poking from the top they'd find the sides of the holes [top of the board] alot, and took too much time to insert.

[RDV]

Did you just add two more stages to a Phase90?

RDV

[stm]

Don't forget the National Semiconductor 6-stage phase shifter; it has a nice capacitor staggering.  This could be thought maybe as the sixtuple version of the Univibe.



Please let me know if you try these values and found them to be better than six identical caps.

Regards,

Sebastian.

[petemoore]

Did you just add two more stages to a Phase90?

RDV

>>>Pretty much just added a pair of stages to one yes. Since I don't know how many stages these other 100's and 180's have, I'll just say Mine has six stages.
 I used a couple aa batteries, but installed the right zener in it's socket this afternoon, all power comes from the battery clip now, and the phasing deepened when the PS god sorted out.
 I just Boxxed the bastard...WHEW alot of cut in's and out's...I had a little 'tangle' on the bottom board that was giving me fits for a time...
 It has a 3PDT, indicator and Jfet Gain recovery stage added to the output, very close or slightly over unity.
 It's in a Homey, but nice looking Large Enclosure I pounded out of old amp frame, all shielded.
 Whew am I glad that's in the box now...what a bear !!! I socketted all the coupling and input caps, and the OA's, tested with TL082's then installed the NE5532's...hafta say, one whompin' phazer. Nice deep lush phasing, all knobs work but are 'twiidly in the middle' [a small area of control], I put the bias knob on top because I know what it is and it definitely alters the phasing sounds, only makes it work in about 1/8th of the travel,,,cool...pretty much a set and forget knob, has enough coarse control that with a slow hand I can find the setting I want. Fortunately that knob works smooth and turns stiff, the shaft is very short.
 From what I read LDR phasers get full sweep from every stage, with the very closely matched set of 6 Jfets from a 100 batch[I don't know, I just picked the best set I saw, all within 1.5% of another], I had better results using the specified transistors, and choosing from a large batch..of course.
 The phase sound coming from this box is excellent. I should build one about the same but with LDR's just to see what the actual differences are.

[Mark Hammer]

Try adding 2 more stages of fixed phase shift as in the MXR Phase 100.

Note that the NatSem unit has no feedback/regen path.  That may well be because it is intended to yield a Univibe-like tone fedback doesn't improve Univibes).

With so many allpass stages, you have more options for tapping feedback.  Exploit them happily, my friend.

[petemoore]

Try adding 2 more stages of fixed phase shift as in the MXR Phase 100.
 >>>for a total of [    ] stages

Note that the NatSem unit has no feedback/regen path.  That may well be because it is intended to yield a Univibe-like tone fedback doesn't improve Univibes).

With so many allpass stages, you have more options for tapping feedback.  Exploit them happily, my friend.

>>>I'm not following the concept of feedback,...

[petemoore]

STM those cap values aren't far off from what I have in there...I'll try to recant.
 .22uf .068uf .047uf .022uf .1uf .047uf
 Something like this, not necessarily in that order, nice haveing those caps socketted, I can get from very tremolo like with a fat/skinny cap combo very wobbly to more like what I have in there...lush and more even sounding, no big 'bumps'.
 Before I was having .15uf's mixed in with .1uf's and .047's but it got bumpy or too much bass overall trying much over .1uf's. It's nice to have some stagger, but too thin or too fat can attenuate or boost a portion of the sweep cycle.
 The thing took some tuning and knob twiddling to get where it is now, which is a very nice place. I think it's set up to bees knees, sounds just superb, has alot of headroom.
 gets that stoned wobble through chordings, nice clarity for lead wailing, Thick smoth bass response...I think I upped the input cap value a bit to .1uf  ...really I couldn't be happier, does an excellent job...definitely nice having a Mix control.

[stm]

Thanks for the feedback.  I've always had the feeling the NatSemi phaser should be a good one, even if it is a cookbook approach.  I like the function of the depth control, which allows you to travel continuously from very soft to chorus to vibrato.

Regards and congrats for your work!

STM

[stm]

Slightly OT:  Many years ago I tried very hard to get those bloody AM9709 quad FETs.  No luck :-(  Has anyone ever seen one of those?  I even tried with obsolete parts suppliers, and also with Part Miner, but no luck at all.  OK, I'll use photocells and forget about them.

[StephenGiles]

Hi STM I got a couple of AM9709s many years ago with a view to building that phaser and never did. They are no longer in my IC box so probably lost for ever.

Do have a look at my EH Microsynth thread, I have posted original schematics.
Stephen

[Mark Hammer]

Try adding 2 more stages of fixed phase shift as in the MXR Phase 100.
 >>>for a total of [    ] stages

Note that the NatSem unit has no feedback/regen path.  That may well be because it is intended to yield a Univibe-like tone fedback doesn't improve Univibes).

With so many allpass stages, you have more options for tapping feedback.  Exploit them happily, my friend.

>>>I'm not following the concept of feedback,...

Many, though not all, phasers have a path from the last phase shift stage back to an earlier point to provide some regeneration.  Normally, this path goes back to the 2nd phase shift stage or whatever stage permits the signal to be fed back an odd number of stages.  For a 4-stage unit, the feedback goes from the 4th stage to the 2nd one such that the regen loop is around 3 stages.

The NatSem design has no such path.  If you look at Univibe and similar pedal designs (e.g., Shin-Ei Reslie-Tone), there is no feedback path there either.  One of the reasons why I suspect it yields a more Univibe-like than phaser-like tone.  

Earlier this year, I built a Ross Phaser clone for one of my son's friends, adapted to function like a Univibe (does a nice job, too).  The Ross normally has a feedback/regen control, but I found that increasing regen in Univibe mode did not improve or even vary the sound in the way it did with phaser cap-values, so I just removed that part of the circuit and it worked just fine.  

Trivia tidbit 1:  Part of the difference between the original "script" MXR Phase 90 and later issues is the value of the feedback resistor (the P90 uses a fixed regen level component rather than a variable control).  

Trivia tidbit 2: In one of the Ross phasers (the multiple-FETs-on-a-chip one; schem available from ROG), a clever arrangement is used in which there are 5 phase-shift stages.  Four are swept by FETs, and the 5th is unswept.and placed in the feedback path.  Where most other phasers would use an even number of stages and feed the signal back to one less than the total (e.g., from stage 6 to stage 2 if 6 stages, from stage 10 to stage 2 if 10 stages), this one feeds the signal from the 4th stage to the output AND back to the first stage, but passes it through the additional fixed stage on the way back such that the feedback still passes through an odd number of stages.

[The Tone God]

I've built that circuit a long time ago with that FET array part. I wasn't impressed. You could try to use a bunch of FET instead. I actually liked using optocouplers with this circuit.

I found that circuit a LONG time ago and use it as a base to redesign the univibe since I didn't like the originals that I could fine. I ended up throwing away the LFO, FETs, buffer, and mixing section only keeping the phaser sections, which are pretty standard to begin with, then redesigned the rest from there. This became my first experiment into vibe land.

Andrew

[snap]

how about a 4007 solution?

[petemoore]

Even if the Jfet's are 'sleeping' on the job once in a while, they're still pumping out what I think is a very cool sounding phaze.
 i've kind of opted for a less Uni-Ish sound and cap value stagger, opting for an 'inbetween' stagger, smallest cap is .01uf and the largest is .15uf [something like that]..I sockettted them so I just kinda played around with values...going for 'rolling hills' as opposed to 'cliff's and mountains'.
 It can get wobbly, but the linear sweepeing becomes quite apparent, [weeshaw weeshaw] and beging to sound 'lumpy'. My guess is that Uni's have some 'play' in the way the up down moves, as opposed to being 'tightly bound to the wave'...I'm guessig there's an 'inconsistancy' with the up down' sweeping of the Uni that makes the 'cyclic-ness' less apparent.
 So whatever the case on 4007's and LDR's may be, I like the sweep of this one, and at any rate it's already in the box, I'm not ripping it apart to try stuff on it...it's working PDGoode IMO, fine enough.,
 ....
 ...Next PHazer !!! ;lol;...Next one uses 'choice' LDR's...
 So Far I have:
 EZ Vibe...I call it heaven
 Phaze 90...Nice Phazer, sounds much less 'organ-ey' than the MXR unit that "Bill has.
 Phaze X6 [pretty good name, I think you know which one this is], similar but stronger and more 'variant' than the 90. My personal opinion is that it works just great, does what it does very well, didin't give me fits going together, and I do like the sound of it. Has more knobs that work than the phaze 90, and provides much more adjustability. Doesn't nail the Uni-thing, has a touch of it. IMO Great sounding phazer...Napping Jfets or not !!!

[stm]

...My guess is that Uni's have some 'play' in the way the up down moves, as opposed to being 'tightly bound to the wave'...I'm guessig there's an 'inconsistancy' with the up down' sweeping of the Uni that makes the 'cyclic-ness' less apparent...

Pete:

Very interesting what you point here.  In a phaser with equal capacitors, the phase curve is monotonic and has only one inflexion point in the middle.  This gives what you describe as a "linear" sound.

In some simulations I made some time ago I noticed when you use the 'vibe caps, the phase curve has several inflexions--three in the case of four stages.  This traduces to something like additional frequency changes on top of the main linear change (difficult to express in words, I think you stated this quite clear on the text I quoted), thus making the linearity of the sweep less evident. I think this explains the sonic differences between the univibe and the equal capacitor values.

Starting from the above, it is possible to optimize the capacitor values to maximize the effect of these extra inflexions to maximize this "multiple swirls" effect.  At that time I found one way to go was to have two different capacitor values applied to half of the stages each.  In a six-stage phaser this would be like:  three stages with 100n and three stages with 10n.  I would really appreciate if you could try those values (or some other closer ones) to see if the sound is really more vibe-ish or swirly as the theory indicates.

Best regards,

STM

P.S.  By the way, the values you mention are very close to the NatSemi:  you said caps between 150n and 10n, or a 15:1 ratio.  On the vibe the caps vary from 220n and 470p, or 468:1 ratio.  Finally, on the NatSemi the caps vary from 1u to 50n, being a 20:1 ratio, which is pretty close to your values.

[stm]

I correct my last post about cap values for extra swirl:

You should use 3 stages with 100nF + 3 stages with 2.2nF.  This gives nice inflexions on the phase curve, even higher than the 'vibe values.

I checked the above in the simulation.  If you use the caps with the original values I suggested (47n and 4.7n) you should get perhaps a less pronounced phasing effect.

Regards,

STM

[petemoore]

I'd like to try that, but I'm getting lost with all the many caps values discussed for the particular 'enhanced depth' mod.
 I follow the basics of the texts, as applied to the phazer, just came back up from some retunings...
 If you state just the values you're suggesting for stages 1-6, I'd be glad to check stox and plug what I have that on or close to the values.
 Are you saying
 3 Stages 'in a row?' with 100n's...or [code '103'] or .01uf ?
 and the last 3 stages with 2n2's ?