dumb univibe question?

[pinkjimiphoton]

hey guys,

if you were to change the order of the magnitude of the caps in a phaser doing a "univibe" kinda mod for the values,

say,  from

47n
470n
470p
4.7n

to

47n
470p
470n
4n7 or so (just picking some off the top of my head) would you get TWO peaks in the sweep instead of just one? would it sound more wobbly, more lopsided, or just the two peaks i'm thinking it might have?

i've been messing around with a behringer vp1 phaser, and did a couple mods on it last nite. one was changing the cap values in the phase shift oscillator to the following values:
22n
330n
470p
4.7n  as suggested for a "leslie" sound on the brazillian handmades forum. it sounds good (a little weird at some settings with the vibe mod done) but i'm not sold on the way it sweeps. i'd like to make it sound more unbalanced and deranged. i figured this may work. i have some 3pdt center off toggles and figured that could give me 3 sets of caps to choose from per switch.
may be overkill. i'd be indebted for guidance and suggestions.

[DougH]

This is just off the top of my head so YMMV, but I think to affect the "wobble" you need to look at the LFO instead of the filter section. But I could be wrong.

[Mark Hammer]

Well, Sir James, you have the Takeda phase-shift calculator at your disposal.  That will let you calculate how much cumulative phase-shift there is across the 4 stages at any given R-value.  Keep in mind that the effect relies on the sum cumulative total of phase shift across the spectrum.  One can have less than 90 degrees added at any given frequency by any given stage, depending on the cap value.  So, a larger cap value at an earlier stage may contribute, say, 60 degrees of shift, which is not added to be a subsequent stage with a much smaller cap, but is supplemented by another 30 degrees added down the line by a stage with a cap value a bit smaller than the earlier one.  And so on.

The net effect of the manner in which the cap values are distributed is that there are no steep pronounced notches, or resonant peaks, created.  Instead, what you get is broad and shallow dips.  And the net effect of that is that there is none of the now-it's-here-and-now-it's-over-there that you get with a phaser, due to the focussed contrast that directs your attention.  And I suppose that's also partly why some folks can often prefer an older model phaser that does not incorporate any feedback...since that would only increase the apparent resonance.

[Digital Larry]

Changing the caps in the phase shift stages themselves changes the locations of the notches for a given value of "R", you know, whether it's a FET or LDR or OTA arrangement.  I'd expect it to scale linearly if you changed all the caps by the same amount, e.g. make them all ten times bigger, then the notches are at 1/10 the freq as before.

Last summer I spent a lot of time thinking about this in the digital realm, and got a headache.

I haven't experimented much with analog phasers to know whether the tuning also changes the sweep range.

[pinkjimiphoton]

WOW! thanks for the great replies guys. i'm still, 10 years in, enough of  a newb that i think i "got" a little of that.

so i guess then the cap values shouldn't matter as much "tonally" as i had thought. when i changed the values of the phaser's stock caps it seemed like it gained wobble and sweep and lost some of the more phaser-esque sound it had stock... or rather, seems like it's there, but at different settings. this one is way too small i think to do much more with (even with my jeweler's loupes SMD stuff is tiny).

so i take it the phaser calc thing i found was useful. cool! maybe i can scratch something in abject wonder and figure some of it out. so i'm taking it that say i want 180 degrees, each cap contributes 1/4th to the phase shift, but each "quarter" can affect different amounts of time kinda. i don't think i explained that right, but i'm guessing you divide the total sweep between the 4 caps in a case like this.

i was wondering why so many variants seemed to have different values. i still tend to think of components like little individual effects kinda... i know i shouldn't.... but i mean like, this value cap will affect this range of tone, this kinda transistor is "softer" than the next. etc/. i saw on the forum i think yesterday one of our peeps here is working on a modular system held together by magnets. cool idea, saw it in a late 50's/early 60's pop electronics (or maybe electronics illustrated?). in that case they thought of them as dominoes, where you could assemble anything with the basic components... held together magnetically. seemed like a great idea, but it never took off... i don't think the tech was up to snuff then to pull it off correctly.

anyways, in essence, the size of those caps has zero effect on the actual guitar signal or tone other than creating the time constraints for the sweep of each individual phaser stage.... did i get it finally?

thanks for the patience, i for the life of me don't understand how you guys can put up with my dumb hippy ass sometimes. but i am grateful you do.

[Mark Hammer]

so i take it the phaser calc thing i found was useful. cool! maybe i can scratch something in abject wonder and figure some of it out. so i'm taking it that say i want 180 degrees, each cap contributes 1/4th to the phase shift, but each "quarter" can affect different amounts of time kinda. i don't think i explained that right, but i'm guessing you divide the total sweep between the 4 caps in a case like this.

The phase-shift calculator shows you where the peaks and notches will be, based on the cap and resistance value.  But you will note that the R and C in that formula actually form a lowpass filter going to the "other" input on the op-amp.  As a lowpass filter, it is not a "brick wall", but has a 6db/oct slope.  So, at the frequency designated by Motohiko's calculator, there is 90 degrees of phase shift, and with 2 identical stages there is 180 degrees at the designated frequency, and a notch produced (this is why if you do the caolculation for one stage you won't see the down and up-pointing triangles).  But at frequencies below the "calcurated" cutoff, there is still phase-shift, just not as much.  That is, the cutoff shown is where the total phase-shift is maximized.  But in the same way that a 6db/oct lowpass also attenuates content below the calculated corner frequency - just not as much - a phase-shift stage also contributes less than 90 degrees for content below that calculated frequency.

Incidentally, I learned something from that calculator: when you add stages, not only do you add notches/peaks, but they are shifted lower in the spectrum, NOT added on top of what you previously had.  The implication is that if you have a phaser that can select between 4 and 8 stages, setting the resonance/feedback high, and then switching from 4 to 8, will get you annoying howling.  Why?  Because there is more signal at the low end, and emphasizing a peak down in the basement will increase the risk of oscillation.

So, if you have a phaser that offers a 2/4/6/8/+ option, you probably want to either have a means of adjusting the offset or bias, to move the overall range upwards when adding more stages, or else some means of automatically cutting the bass when you switch from less to more stages.

And, to tie this all together, this is precisely why adding stages provides more notches at the lower end.  Think of it this way.  If I have 2 stages that each provide 90 degrees of phase shift at frequency F, together they will produce a notch.  BUT, they also produce 45 degrees of phase shift at F/2.  With only 2 stages, I won't hear that.  But if I have 4 identical stages, 4 x 45 DOES end up producing a notch at F/2.  Again, one has to think not just in terms of what each stage is doing, but in terms of what it all adds up to.

I encourage people to use the calculator ( http://www.aleph.co.jp/~takeda/radio/psDesigner.swf ), pick a pair of R-C values, and play with the number of stages, keeping track of where the peaks and notches are.  You'll find it interesting.

Finally, we are most familiar with one-pole phase-shift stages.  But such allpass stages can be 2 and 3-pole as well, just like any filter.  Mike Irwin demonstrated a triple 2-pole phaser to me some years back.  That is, instead of 6 stages of single pole, there were 3 stages that were each 2-pole allpass.  What was interesting about it was that the notches and peaks were closer together than you get with the standard 6-stage single-pole design.  Very interesting sound.

[matt239]

Ok, How do we "calcurate" for an OTA based phaser?

The circuit doesn't look the same. which resistor are looking at?
Is it more complicated than that?

[matt239]

- That is, looking at say, a Small Stone, I see the cap, but I don't know which resistor forms the filter with it, so what to put in the calculator.

Also, I understand that having 4 different caps for 4 stages, (as in univibe mods) doesn't create 4 notches,

- but does having different caps for different PAIRS of stages create more notches? Or just change the depth?  - It should, right? :

The first pair of stages creates a notch at "F1." If the second pair has different values, you should get a notch at a new frequency, plus there will still be some frequency at which the cumulative phase shift produces a notch. - So 3 notches instead of the 2 you get with 4 identical stages, right?

- The notch at F1 just won't get deepened by the next stage, and you won't form much peaks.

- As far as the frequency of the notches being distributed lower as you add stages, it seems like you could make additional stages have smaller caps, so the distribution would be both higher and lower?

[matt239]

 It's my understanding, that having 4 different caps for 4 stages, (as in univibe mods) doesn't create 4 notches,

- but does having different caps for different PAIRS of stages create more notches? Or just change the depth?  - It should, right? :

The first pair of stages creates a notch at "F1." If the second pair has different values, you should get a notch at a new frequency, plus there will still be some frequency at which the cumulative phase shift produces a notch. - So 3 notches instead of the 2 you get with 4 identical stages, right?

- The notch at F1 just won't get deepened by the next stage, and you won't form much peaks.

Right??

[Kevin Mitchell]

Not to jack and steer this thread but I've been on the same thing pretty much. Working on a 12 stage vibe I have no clue on ideal caps to use for the phase stages. I get ideas from tinkering alone but knowing some more science behind it would be great. Of course there's great info at Geofex (The Technology of the Univibe) but since we're on the subject of I assume opamp phasers? Not aware of major discrepancies between the information provided.

But what I'd really like to know is what you fine fellows would recommend for such vibe concepts?  Does it come down to luck and preference or is there an ideal way of going about phase cap values?

Great info in this thread already! I'm open to some opinions and further clarity. Thanks!

[Mark Hammer]

Mike Irwin (my personal phaser guru) showed me a frequency plot for a 4-stage phaser, and a 4-stage Uni-Vibe.  The two devices yield 2 "impositions" on the frequency response.  In the case of the phaser, it produces two very apparent notches, and introducing feedback enhances the depth of the notches, in part by increasing the amplitude of the signal just above and below where each notch is.

The Vibe, however, because of the staggered cap values, does not yield sufficient phase shift at any single point to result in obvious notches.  Rather, it yields broad and shallow "dips" in the frequency response.  This is why people like them.  Unlike the more resonant phaser where one's attention is drawn to the more pronounced peaks and dips, the vibe simply provides some "animation" to the sound, such that one's attention can be focussed on the melody or harmonies.  This is also one of the major differences between chorus and flanging: the extent to which they direct the listener's attention to this aspect or that.

Because the vibe effect is less obvious, the speed range is more restricted.  A vibe sweep that takes 10 seconds from end to end is simply inaudible, whereas long slow resonant phaser sweeps are easily noticeable  So if one is adapting a phaser design to functioning like a vibe, drop the value of the Speed pot because well over half the range of speeds will be all but useless.

Similarly, feedback/resonance/emphasis in vibes is useless, which is why none of them include such a control, or even incorporate any fixed feedback (the way an MXR P90 does).

[matt239]

Not to jack and steer this thread but I've been on the same thing pretty much. Working on a 12 stage vibe I have no clue on ideal caps to use for the phase stages. I get ideas from tinkering alone but knowing some more science behind it would be great. Of course there's great info at Geofex (The Technology of the Univibe) but since we're on the subject of I assume opamp phasers? Not aware of major discrepancies between the information provided.
But what I'd really like to know is what you fine fellows would recommend for such vibe concepts?  Does it come down to luck and preference or is there an ideal way of going about phase cap values?
Great info in this thread already! I'm open to some opinions and further clarity. Thanks!

It seems like we SHOULD be able to calculate the filters on our own (without the calculator.) and figure out the notch frequencies.
- but actually doing this is over my head.
- I can calculate filters if they are of one of the common basic types, and I have the explanation right in front of me, but I don't quite get how to do it with phasers.

The "phase shift calculator" is only for identical stages. It's very interesting, but wont help you calculate vibe values.

A search will reveal values folks have chosen to "univibe mod" Small Stones, Phase-90s, & Ross.. (& values for the original Univibe & various clones..)

If you're doing more stages, I suppose you'd make the caps all a bit smaller, but IDK by how much.

How are you implementing your univibe phase stages? Lightbulbs & LDRs, or some other way?

It would be fairly easy to use 2x Tonepad-ROPEZ-Ross boards, (4-stage OTA) each with a Tonepad extra-stages-board (2 more stages) added on, for a total of 12 stages.
- You could run them all off one of the LFOs, or go ahead and use 2 LFOs. (one comes on each main board)
- Setting them at different speeds would be a cool way to randomize the modulation a bit.

I still can't say exactly what values to use..
Maybe you could make your first 4 stages whatever your univibe spread is, make the additional 2 stages a bit higher than that, (smaller caps, following the same formula as the univibe stagger..) - then make your next 6 stages all 25% smaller caps, respectively, than your first 6 stages?
- Don't want to make any of them TOO small, or you'll start having notches higher than you can hear.
- OR, make each PAIR of stages to the four univibe values, (8 total stages) - the next pair to Ross-ish values, & the next pair slightly smaller than that..?

It will be nice to hear more on the science of it from our resident experts!

[matt239]

I still want to know if this is correct; that you could create 3 notches, with only 4 stages:

That, as mark says, (If I understand correctly) in a normal 4-stage phaser, (identical stages) - the first stage makes a notch "F1", the second stage deepens the F1 notch, and together they create a shallower notch @ "F2"

But IF: " The first pair of stages creates a notch at "F1." If the SECOND PAIR HAS A DIFFERENT VALUE, you should get a notch at a new frequency, "F2'" plus there will still be some frequency at which the cumulative phase shift produces a notch, "F3". - So 3 notches instead of the 2 you get with 4 identical stages, right?

- The notch at F1 just won't get deepened by the next stage, and you won't form much peaks." 

So 3 notches, but all at "Phase-45-style" depth.  Right?