Phase 90 headroom

[Bucksears]

Is there any way to increase the headroom in a Phase 90-clone circuit? I have BYOC Phaser that distorts a bit when using humbuckers and I'd like to remedy that.
I've tried upping the voltage to 16v, but that just changed the 'shape' of the rotation (less 'round', more 'football').

I've read that there is some kind of noise reduction built in to the regular MXR Phase 90 that the MXR '74 reissue doesn't have. Any ideas? I really like the sound otherwise.

Thanks,
Buck

[frequencycentral]

I've found the same. I'm (eventually) going to try attenuating the input and boosting the output. Will play merry hell with the signal to noise ratio but what the heck.

[Mark Hammer]

This diagram illustrates what can be done to reduce distortion:  http://www5b.biglobe.ne.jp/~houshu/synth/PhaseFet0205.GIF

Apparently, while example 2 reduces distortion, essentially what it does is raise the threshold of clipping, rather than prohibit it entirely.  Several years ago, Mike Irwin informed me that while he had high hopes for it, he found that the distortion that did arise was ugly, and that in retrospect he would rather put up with the small amount of distortion introduced by the conventional means.  It is worth noting here, however, that Mike's principle interest is synths rather than guitar pedals, and these tend to have much hotter outputs than guitars do.  So, it may well be that the critical threshold is safely above what HB pickups will deliver.

Here is an example of implementing this scheme in a commercial pedal:  http://www.experimentalistsanonymous.com/diy/Schematics/Phasers/Ibanez%20PT-90.gif

[solderman]

Find and cut the feedback resistor with a switch , see the R22k at the www.Tonepad.com schematic, and read the build reports from there.
If you want you can check the sound samples at my site for "Captain America" with this mod.

[Bucksears]

Hey - just wanted to bump this one with an update.
Since I was a little skeptical about the matching done with the transistors that came with the BYOC kit, I ordered about 70 2N5952's from Mouser and started measuring with R.G.'s JFET matcher.
I numbered all of them with labels, noted their respective voltages and put that in an Excel spreadsheet, sorted by voltage. I was looking for the closest 2.00v or higher I could find. Got one perfect set, all four measuring 2.00v; another set, three measuring 2.04v, one at 2.02v. Then a third set with three at 1.99v, one at 1.98v.

Here's the kicker: I put the quad of 2.00v ones in my main phaser (right now), dialed in the sound I wanted on the trimpot and it did NOT distort. I tried several times slamming the strings on my humbucker neck, then the bridge and NADA! Don't know if this is a fluke or not, but I've got a second BYOC build that I'm working on that I'll try out another one of my matched sets on.

FWIW, the BYOC circuit is a clone of the 'script' version of the Phase 90, so it does not include the 22k feedback resistor.

[R.G.]

This one came through while I was at NAMM, so I didn't notice it until today.

Apparently, while example 2 reduces distortion, essentially what it does is raise the threshold of clipping, rather than prohibit it entirely. 

This is one example of using drain-voltage feedback to the gate to help reduce distortion. It makes sense that it would uglier when it distorts because feedback in general uses gain to hold off distortion longer - until it just can't any more and the underlying ugliness pops through suddenly.

Here is an example of implementing this scheme in a commercial pedal:  http://www.experimentalistsanonymous.com/diy/Schematics/Phasers/Ibanez%20PT-90.gif

Another one is the Phase 45. I have modified P90s with the additional resistors and cap to make them into phase 45+45s, and it's nice.

The underlying distortion is the fact that when used as a voltage controlled resistor, the JFET has only a very limited linear range. After that, it starts being a voltage-controlled current source. Feedback lets you use more of the variable resistance range before the curvature starts making distortion apparent.

[Mark Hammer]

Another one is the Phase 45. I have modified P90s with the additional resistors and cap to make them into phase 45+45s, and it's nice.

The underlying distortion is the fact that when used as a voltage controlled resistor, the JFET has only a very limited linear range. After that, it starts being a voltage-controlled current source. Feedback lets you use more of the variable resistance range before the curvature starts making distortion apparent.

Welcome back.

So let me ask you, would it be possible to use a pair of JFETs, with the drain-source paths in series, as a voltage-controled resistor with a larger resistance swing?  Or is this just begging, pleading on bended knee, grovelling, for trouble?

[Joe Hart]

Hey - just wanted to bump this one with an update.
Since I was a little skeptical about the matching done with the transistors that came with the BYOC kit, I ordered about 70 2N5952's from Mouser and started measuring with R.G.'s JFET matcher.
I numbered all of them with labels, noted their respective voltages and put that in an Excel spreadsheet, sorted by voltage. I was looking for the closest 2.00v or higher I could find. Got one perfect set, all four measuring 2.00v; another set, three measuring 2.04v, one at 2.02v. Then a third set with three at 1.99v, one at 1.98v.

Here's the kicker: I put the quad of 2.00v ones in my main phaser (right now), dialed in the sound I wanted on the trimpot and it did NOT distort. I tried several times slamming the strings on my humbucker neck, then the bridge and NADA! Don't know if this is a fluke or not, but I've got a second BYOC build that I'm working on that I'll try out another one of my matched sets on.

FWIW, the BYOC circuit is a clone of the 'script' version of the Phase 90, so it does not include the 22k feedback resistor.

Interesting. Let us know what happens with the other "close" sets.
-Joe Hart

[R.G.]

So let me ask you, would it be possible to use a pair of JFETs, with the drain-source paths in series, as a voltage-controled resistor with a larger resistance swing?  Or is this just begging, pleading on bended knee, grovelling, for trouble?

Oh, sure - ask the obvious question.  

Well, I don't think the resistance range would change, as any JFET can go from near-infinite resistance to it rds. Plenty of range there, but the voltage swing before distortion would double if you could stack them. Hmmm.

Just off hand I can't think of a good way to arrange the DC level on the top JFET's source so it's both at a fixed DC voltage for wiggling the gate voltage with reference to, and at the same time being a high impedance DC source so the source voltage does not shunt all the signal current away from the bottom JFET drain. I'll give it some more thought, but what might work is using an opamp to remove the voltage variation from the source-drain of a single JFET and then using the output of the opamp as either a linearizing voltage/current source or the "faked" resistance itself, much as gyrators and simulated inductors simulate the V-I characteristic of the simulated parts.

Of course, that probably collapses to be equal to Osama's #3 or #4 in the original list of solutions if you work at it.

Actually, if you could do that, a dual opamp amounts to two sets of three pins to do something with, plus two power pins. So each half of the package is the same number of pins as one additional JFET would be, plus a fixed cost of one power pin. With #3, you don't have any resistors purely for the opamp, so it's very much like adding one resistor and one JFET to the existing JFET, at least in terms of board layout space. It's probably less complex in terms of needing to match fewer JFETs. I just checked Mouser - the NJM4558D is the ... cheapest... dual opamp they stock! I didn't know that. It's 0.32 each in ones. N-channel JFETs similar to the 2N5485 cost about 0.11 each, so you're costing 0.22 for two JFETs against 0.32 for a dual opamp, and you get away from matching twice the number of JFETs.

I'll keep pondering the two-JFET solution - maybe I can come up with something. But I think the penalty for using #3 isn't that severe.

[Mark Hammer]

Of course, that probably collapses to be equal to Osama's #3 or #4 in the original list of solutions if you work at it.

Osamu.  Osamu is a respected electronic engineer for NEC who lives in one of the prefectures of Tokyo, works on digital reduction of ambient noise in cellphones, and has a lovely wfe and daughter.  Osama is an engineer of revenge who makes bootleg tapes, lives in a cave somewhere.......actually I dunno where, but not in Tokyo.

Osama seems to be mostly beard.  Osamu, on the other hand, as sweet a guy as he is, couldn't grow a beard on the beardiest day of his life if he had an electrified beard-growing machine.

Osamu is game for analog, digital, discrete or integrated, CMOS or linear, as long as it does the job.  Osama, on the other hand, thinks that the changeover from true bypass to JFET switching was an indication of how western decadence had compromised the morals of the youth.

Two completely different web-sites.   

[R.G.]

DOH!! 

If you think my spelling is bad in English, don't even let me get started in anything else.

My apologies to *both* Osamu and Osama!

[karis12]

Bumping this thread to get some confirmation, as I have just built a Phase 90 but am having problems with headroom as well.

Is it imperative to find matched sets of JFETs that are at least 2v or more in terms of Vgsoff? I currently only have access to sets that are a bit under that threshold, around 1.64-66v in range. I got it to bias and phase correctly, and unsurprisingly it was identical in sound to my Phase 95. The only issue now is the distorted output, which I hope could be remedied by simply sourcing better parts for the circuit.

Thank you in advance.

[PRR]

> at least 2v or more in terms of Vgsoff?

Common mix-up which has fueled years of pointless squabble.

The "2V" in Bucksears' report is NOT Vgs(off), but voltage from R.G.'s tester, which finds a non-OFF condition near 10K resistance (IIRC).

If you are not testing R.G. style, but measuring Vgs(off), all your FETs are low-voltage. If RG=2V is the criteria, we expect Rgs(off) to be nearer 3V not 1.6V.

[karis12]

Ah, yes of course, sorry about that. I am using R.G.'s tester circuit. So should I expect better headroom if I find sets that rate above the 2 volt mark, or should I look into other solutions?

[PRR]

1.6 is so close to 2, that I would ask you to play at 80% loudness, see if that gets it clean. If not, then a part-Volt more on the FETs is not likely your cure.

[karis12]

I still don't have new JFETs to test, but sorry if I'm being annoying making further inquiries regarding the subject. It's just frustrating to have a reference device on hand and hear what's supposedly an identical pedal circuit-wise perform differently.

If anything, I can hear significantly more distortion with a below-9 volt battery, compared to using a power supply that outputs 9.1v. Then again, Bucksears had already tried increasing the supply voltage and simply ended up with the pedal behaving differently from the intended design.