Phase 90 mod - Univibe mod?

[Mark Hammer]

In my own experience, there is little value to be gained by going below, say, 350k or above 1.5k.  If you want a variable offset/range control, a 390k resistor in series with a 1M pot should be sufficient. 

Remember, there isn't much musical value in any notches produced above or below the range of the guitar signal, or above/below the range of the speakers.  They have to be where you can actually hear them.  That's one of the reasons why I don't think there is much to be gained by having more than 8 phase-shift stages for guitar.  Bear in mind that having more stages does not space the notches any closer together; it moves the location of the notches down a bit, but mostly adds more notches above where the existing ones are.  For ultra-wide-bandwidth signals like synth, 10 and 12 stages can sound spectacular (I've heard 24 stages with white noise and it was amazing), because there is signal content "up there" to insert notches into.  If those synths are played through amps using 4x12 cabs with a 6khz rolloff, though, it may still not be particularly spectacular.

So, even with only 4 stages and 2 notches, being able to bias the FETs so that the notches are created way up high, is not going to be audible or musically useful.

[ItsGiusto]

I ended up using my extra drilled-hole to put in a 1M log pot (log was all I had) wired to be a variable resistor. The pot actually measured at a little over 1.1M, which worked out well. I connected it to a couple of resistors summing to 2.75M. That way, with the pot completely rolled off, it'd be 2.75M for the widest sweep, and with the pot dimed, it will come to 3.85M for the smallest sweep. I'll calibrate the trimpot again and try it out tomorrow.

When I said this, I'm referring to that I put in this pot-resistor combo in place of the 3M9 resistor, for a sweep-width control.

[Mark Hammer]

That's what I get for quickly reading posts while at work. 

Note that the range of adjustment that the FETs permit can limit how much the sweep width and bias can jointly contribute.  Keep in mind they sum at the FET gates.  The FET can only change its drain-source resistance so much in either direction before it craps out.  That is precisely why folks are encouraged to match FETs in this, and similar, circuits, so that no FET craps out at one or the other extreme before the other FETs do.  It's also why many folks like LDR-based phasers since LDRs are rarely pushed to their very extremes, with no more resistance-change possible.

The bottom line is that, while it is possible to play with the current coming from the LFO and the current coming from the bias voltage, one can't have their total adding up to too much.  That can be tackled by either abandoning one so that all effort can be put into the other, or by simply being moderate in expectations for what each current source is allowed to provide.  My own preference - which doesn't have to be yours - is to pursue the second option.

At the same time, as long as it doesn't damage anything, there is no harm in having control ranges that exceed what is possible, but are reasonable when some other parameter is tamed.  In other words, width-plus-bias-current may cause problems if both are maxed, but NOT if one is restrained to allow the other to be maxed.

Manufacturers generally like to design circuits such that there are no "trouble zones" with any of the controls (i.e., everything seems to behave well unless I turn knob X up past 7), since it would require fending off complaints and providing more documentation to consumers than the consumer is perhaps prepared to read or make sense of.

[ItsGiusto]

Cool. My plan is to calibrate it for all Phase 90 stock settings (width set at 3M9, depth set at 1M) so it sounds exactly like a P90. Then I'll see what I get by messing with the dials. If I find ranges that are unusable, I might leave them for now, or try to make a quick adjustment under the hood. But probably the former, since all my autowah parts came and I want to get finished with this one and get started on that!

[ItsGiusto]

By the way, getting back to the main subject of this thread, I do happen to have a spare set of matched Phase 90 transistors (2N5952? I forget) lying around (bought originally because I thought maybe the first ones were defective). I could maybe make an entirely different pedal to clone the MXR M68 (phase 90 that sounds like a univibe), calibrated to actually sound univibe-like. If I were to do this, do you have recommendations of what specific adjustments I should make to the P90 schematic to make that happen?

1) You said that I'd need a less-wide sweep for a univibe sound. Or was that a wider sweep? I think maybe you said wider for vibrato, and less-wide for vibe? So what might the value of the 3M9 be changed to? 4M7, maybe?
2) I'd of course need to replace the 4 phase caps for the univibe-valued ones
3) You said that the swirling would have to take place around the lower region instead of the spectrum, so maybe I'd change the 1M depth/bias resistor to a 470k? Maybe less?
4) You said that I should cut the speed pot from 470k to around 200k
5) Feedback: you said that vibes have no feedback. Does that basically mean to keep the infamous 22k "block logo" resistor out of the phase 90/vibe circuit to make it sound more like a vibe?
6) Any other adjustments?

If you have recommendation for values to start at to try this out to try to find a good univibe-phaser sound, I'm all ears. I would have thought this info would be more readily available online in searching, since I do see a lot of phase 90s modded to sound like univibes for sale, not to mention the MXR M68.

[Mark Hammer]

How far the sweep goes, and where it happens, is a function of more than what you list.

If you look at each phase-shift stage you'll see that the inverting and non-inverting inputs each receive a different signal.  The inverting in put gets a full bandwidth version, via a simple 10k resistor.  The noninverting input gets a highpass-filtered version of that same signal.  The rolloff point of that highpass version is given by 1 / (2 * pi * C * R), like any other one-pole filter.  Where it is different is that the 'R' is a joint function of the fixed resistor and the drain-source resistance of the FET in parallel with it.  Make the C part of the equation lower in value (e.g., .039) and the rolloff point moves upward.  Similarly, make the combined parallel resistance smaller, and the rolloff also moves upwards.  Conversely, make either the C or the R, or both, larger in value, and the rolloff point moves downward.

So, while, yes, dickering with the bias can move the range around, so can dickering with the cap values and the parallel resistors.  The trouble is that "classic" vibes have such a different structure, I find it hard to do the necessary calculations to be able to replicate where the dips occur, within a P90 structure.  If somebody else knows how to calculate them, or has ever measured them, that would be wonderful.  But in the history of this place, all folks seem to have done is simply moved the stock cap values of a Uni-Vibe to a P90, and I doubt that is sufficient to duplicate the sound.  It is certainly more Vibe-like than the stock P90 values, but it is unclear whether nailing the sweep range and width of a Vibe is captured by just porting over the cap values.

[ItsGiusto]

Oh, that's too bad, I'd like to know how its done, because, like I said earlier, these modded phase 90/univibe hybrids do seem to exist and be somewhat abundantly available on Reverb. And the MXR M68 Univibe has been in production for a couple of years now. I would have thought someone would have dissected it to figure out what makes it tick, or dissected one of the aforementioned available mods. I've been told that the M68 is definitely a modded P90, and that there's no way it's actually an LDR-based univibe, like they'd have customers believe. For what its worth, I think the M68 sounds very much like the univibe-modded P90s I've heard demoed on Youtube.

[PRR]

> I thought that FETs were

You are describing MOSFETs.

MOSFETs can be build several ways. But by far the most common is Vgs=0V means "off" and Vgs= several Volts means "ON". This works dandy with digital chips, and can be used for some audio work.

In an N-channel Junction FET, we can't take the Gate positive because it will conduct like mad. With Vgs=0V the gate is not conducty and the channel conducts near maximum. We take the gate negative to reduce channel conduction. Near Vgs(off) the channel is nearly off.

[ItsGiusto]

I completed the build last night - it sounds really great, just like a p90, at stock settings! Also, I really like the depth knob all the way down, it starts to get some vibey wobble on the fundamentals of the notes, like Mark was saying.

I may make a few more changes in the future, like increasing the depth knob to be 1M instead of 470k, changing the width control to either be linear taper, or to just be a DPDT ON-OFF-ON with some resistors in parallel - there seem to be just a few useful settings of the knob right now, so I could simplify it. But for now, I think I want to finally start my autowah.

[Mark Hammer]

As you've discovered in a profound way, where something happens in the spectrum is often just as important, if not moreso, than what is happening.

And that's a big part of satisfying pedal design/modding: where in the spectrum will doing X deliver what I want?

Congrats.  Now get cracking on that auto-wah - the pedal named for my hometown.

[pinkjimiphoton]

so far, the best sounding "vibe" project i've made is the eqd the depths. its like a simplified univibe using an led, you can use the uv values and it sounds great. better to my ear than the dunlop by far, way more transparent... almost as good as the black cat vibe, which to my ear is the best vibe on the market.
i tried making a vibe-a-like with a p90 circuit... i could make a nice vibe with it but never even close to a univibe no matter what i tried.
my 2 cents

[ItsGiusto]

so far, the best sounding "vibe" project i've made is the eqd the depths. its like a simplified univibe using an led, you can use the uv values and it sounds great. better to my ear than the dunlop by far, way more transparent... almost as good as the black cat vibe, which to my ear is the best vibe on the market.
i tried making a vibe-a-like with a p90 circuit... i could make a nice vibe with it but never even close to a univibe no matter what i tried.
my 2 cents

Ah, that's good to know! I didn't know that the Depths used an LED, though that makes sense - I always wondered how they managed to make a vibe circuit that cheaply and compact. But I had been told years ago that it was impossible to make a vibe circuit that actually worked with an LED, and that you would actually have to have an incandescent bulb to make it sound similar.

[pinkjimiphoton]

well, it sounds so much like the black cat is blows my mind. the eqd thing i built (thanks to Aishabhag for rolling the board for me) uses a green led and sounds way better than the dunlop i have on my board... the old-school giant one with the external crybaby kinda speed controller.
you could convert it to an incandescent or neon lamp i'd imagine. but really no need to. the depths is the best one of the many i;ve tried/built. and it fits nice in a crybaby shell. little hard to find a suitable dual gang 100k d shaft pot, but they are out there.

also, FWIW, i added a rate indicator in the heel of the crybaby shell, simply run in parallel with the led lamp. works great.

the most transparent vibe i've encountered is definitely the black cat... that has the least coloration with the best actual vibe to my ear. but the eqd depths is really really close. very quiet, and can be very transparent... or very colored.

fwiw, i gave analog tom the p90 in a crybaby shell i'd built. once i got the depths going, the phase 90 just wouldn't cut it anymore. it sounded great, but it didn't sound like a univibe at all.

ymmv of course

i did the "univibe values" caps to a behringer small stone clone.. and it sounded a litle more leslie like, but not univibey at all. at least not to my damaged old ears

[Mark Hammer]

1) Like I say, it's not the cap values per se, but the location of the dips in the spectrum.  The cap values certainly play a role in that, but they aren't the whole story.

2)  The difficulties people have in producing a satisfying Vibe using one and two-op-amp LFOs, and something other than an incandescent bulb only underscores how much the "classic" sound is a function of the idiosyncrasies of the LFO circuit and the sort of waveform that a lamp/LDR combo yields.  Well, maybe waveform is inaccurate.  The resistance-change function, since whatever waveform the LFO produces only contributes to the manner in which the LDRs shange.  The LDRs have their own idiosyncratic rise and fall and responsiveness to light.

[ItsGiusto]

The 1M resistor from the trimmer  sets where the sweep begins.  It limits the current from a stable bias voltage applied to the FET gates.  The FETs are being used as voltage/current-controlled resistors, each of which is in parallel with a 22k (24k on some issues) fixed resistor.  When turned "off" the FET resistance is much much higher than 22k, so we'll consider the parallel resistance of the 22k+FET to be = 22k when the bias voltage is minimal and the LFO is not adding to that.  22k and .05uf at the non-inverting pin provides a steady 90-degree phase shift at 144hz.  As the FET resistance is pushed lower, its combined parallel resistance with the 22k resistor starts to move that phase shift point upwards.  So when the FET+resistor = 18k, maximum phase shift is at 176hz.  When  FET+resistor = 15k, max phase shift is at 212hz, 318hz when it drops to 10k, and 636hz when it reaches 5k.

Hey Mark, I have another question. I think I understand phase shifting and thanks to your explanation, how it works in something like the Phase 90. And I understand why having a phase-shifted signal in combination with a dry signal would cause phase-cancellation notches.

What I don't understand, is why is it that when I take away the dry-signal (say like the chorus/vibe switch in the Univibe, or the switch I put on my Phase 90 clone to lift the resistor to mix in the dry signal), I still hear a vibrato sound? Theoretically, I should just be left with a phase-shifted sound, and I shouldn't be able to hear the difference at all, since the human ear can't hear phase, just the effects of phase-cancellation. But for some reason, I actually hear a vibrato sound. Maybe it's different from a chorus's wet-signal vibrato sound, which is produced by using a modulated delay on the signal, but the phaser/vibe's wet sound is definitely modulating the pitch in some way. So why is that happening? Like I said above, shouldn't I just hear nothing, instead of modulated pitch?

[pinkjimiphoton]

hey justin,
mark is THE guy for this stuff,
but i believe what you're hearing is what you should be, as the phasing is created by MODULATING the phase shift stages with the lfo, so each stage adds a bit more to the shift and makes that wobble sound like a chorus. if you take the dry signal away, all you hear is the modulated and phased signal, and without the dry signal the actual pitch modulation doppler effect becomes much more apparent than when its being swept against the dry signal. without the dry signal, there's not so much a phase shift effect against the actual input signal, so all ya get is the modulated one.

i hope i didn't make stuff more confusing.

but without the lfo modulating the phase shift stages, the whole thing would sound more like a comb filter i think.
i was just restoring an old polyphase recently with the help of scruffie, and when i got close, i could definitely hear the phase change to the signal.. even tho it wasn't sweeping.

i believe its all about sweeping the filters. the whole idea originally was to mimic the doppler effect of a leslie speaker.
no sweep? no phase.
no dry signal? no phased signal to modulate it, so all you hear is the slight bit of pitch modulation. i dunno if its really PITCH, or just gives one the perception of it.

gonna bow out now cuz i may have talked myself into a corner, and mark can definitely answer this better than me. peace!

[PRR]

> I still hear a vibrato sound?

Variable phase shifters are never perfect. There will be a small amplitude change. I assume this is what you are hearing.