distorted phaser..

[vdm]

hey guys..

i built a phase 45 a while ago, and i can't seem to get rid of this very slight distorted tone that is mixed in when i hit the strings at anything louder than a very gentle touch.

i was first wondering if anyone had any solutions to that... i was thinking maybe run it off a 12V wallwart to get more headroom...

secondly i was thinking of where the original signal (from after the buffer/preamp) is mixed back into the 'phased' signal, that i could replace the clean section and incorporate some kind of fuzz or distortion.

would this create an interesting tone when mixed in with the phased signal, or has someone tried it and it's nothing special? I was thinking of putting it on a dpdt switch, so i could go between clean phase and dirty phase, thinking that the distortion might create some strange or interesting interaction as it is being run in parralel with the phasing section...

i built the P45 on perfboard, and it works great with the univibe cap mods, the flashing led and the mix pot, although i would suggest to most people to only try the mix pot and see where you find it useful, because the subtle (compared to other phasers) effect is not heard *that* much when turning the mix pot.. though you can go from clean signal right through to warbly detuning..

thanks everyone, and sorry again for the long post

trent

[niftydog]

have you checked the op amp biasing? (inverting pins at roughly half supply, except the last op amp, which is biased on the non-inverting pin) FET biasing? (looks like roughly half supply again at the gates)

It's unlikely that you're overloading an op amp with your guitar, but if there's some DC shift in the signal it could be clipping.

[Kleber AG]

:?  I built the Phase90(pretty similar) and had this distortion problem as well. So after reading RG's sugestion about changing the 4 Phase90 Jfets sections to the Phase45 configuration(search the archives for all details), it's better now, I can hear the difference.
But I think I remeber he said these Fets would not take more than 0,5V on it's inputs without distorting, and now (after these mods) I think It is doubled "1,0V" but my guitar (hambuckers) may be puting way more than 1,0V at the peaks :?

Resulting... it still distorts a lot, it's impossible to have the "R28" in place (wich I like the tone with it), even without R28 still there's some distortion.

All of these distortion only goes alway with my single coils strato guitar :x

I tried many thing without a good succes, the circuit is very sensitive to it's input if you pad down enough volume at it's input the output will a lot lower, then you try to bring output higher again and you'll have noise (bad Signal/Noise trade)

I do prefer the way Phase 90 phases to the SmallStone, but my only problem is this freak distortion :x

Kleber AG

[gez]

Yeah, it's probably FET distortion.  What I used to do is divide down the input then use an active mixer with a little gain at the output to restore level (you sometimes see this in schematics).

[Kleber AG]

Hi Gez

How would be the best way to divide the input? Maybe without loading down the input/guitar? Values?
And then we need to compensate at the phaser/mixer output, but that configuration of the PNP trans at the end of the mixer IMHO doesn't looks to be the best way to do it, it's easy to get it distorting as well :? does it needs to be PNP??? :?

Would a TL062 op-amp make "any" difference for the Jfets? Less distortion for some reason? too bad I have no tl062s here(i'm using 072), but I would by and try if it could make any difference.
Or using some different Jfets? I have 2N5952s matched there...

And the last question, would it have difference changing the PNP trans for one with more gain to compensate any volume???

I wish I could have the answers for all this questions, but I need your help...

Thanks
Kleber AG
PS: I hope I'm not changing the original topic, as the solutions for the 90 would apply to the 45 as well  

[gez]

This is how I'd do it.  You can probably get away with loosing that 4u7 NP cap at the input of the mixer.  I can't remember why I put it there, probably the amps I was using created an offset...lost in the mists of time.

No guarantee that you won't get distortion if your pickups are hot, in which case you'd have to divide down even more.  This creates noise though (that's why the 470n cap in feedback loop is there).  As shown it's pretty quiet though.

[RDV]

I changed the opamps in mine to Burr/Brown OPA2134 and it helped a bunch, though not a 100% solution.

RDV

[Kleber AG]

errr... Thank you very much Gez!

Yeah I can see how you divided the input, but hey you got a lot more things changed there  

Please could you explain the changes? And maybe describe the tone diferences? It's not only some Rs or caps :shock:  Seems to be redesigned
Looks very cool!

I'm trying to figure it out.

Thanks
Kleber AG

[gez]

kleber, for the moment just concentrate on the buffer and mixer.  The output of a phase 45’s buffer could be divided down in exactly the way I’ve shown in the above clone, only Vref would be the junction of the cathode of the 1N5230 with the 20k and 470k resistors (as shown in the phase 45 schematic linked from Aron’s schematics page).  If you add another op-amp at the output and do the mixing actively instead of passively, this will allow you to add gain to the wet signal, which is necessary as it’s been divided down after the buffer to keep the signal across the FETs small.  For quieter operation the dry signal isn’t divided down and gets unity gain at the mixer stage, whereas the wet signal gets a gain of three.

You might have to divide down the buffer’s output even more and then boost gain in the mixer stage some more, all depends on your pickups.  This can get noisy though.  If you do go down this road then it’s better to increase the value of the two 30k resistors rather than make the 10k resistor smaller (you’ll have to make the cap in the feedback loop smaller too).  That’s all you need to do, or something like it.  

If you’re interested in the rest of the circuit I posted then here’s a little info.  The MOSFETs are from a 4007 chip, which means you don’t have to bother doing any matching.  The downside to using them is that it doesn’t take much to make them distort (chips tend to be worse that discrete devices in my experience) which is why I divided down the signal going into the phase stages.

All the amps are biased from the divider coming off the 5V regulator.  It keeps the DC bias for the MOSFETs stable regardless of what your battery/supply is doing so phase is consistent as your battery flattens.  The trimpot connected to the Depth pot sets the DC bias on the gates of the MOSFETs.

The LFO is a cut-down variation of the ‘easily tuned sine wave oscillator’ from a National application document (it’s online somewhere).  You sometimes see it used in commercial effects.  Usually the comparator has a huge value resistor in parallel with a cap in the feedback loop (the cap acting as a crude filter to reduce distortion) and its output is clamped to produce a squarewave.  This square is fed back to the negative input of the first op-amp which has a bridged-T network in its feedback loop.  Gain isn’t high enough for self-oscillation so the output of the comparator is used to ‘energise’ the circuit into oscillation.  With this signal present at the amp’s -ve terminal the output of the first amp will do whatever it takes to equalise its inputs  but because of the 180 degree phase shift introduced by the T-network feedback is positive, so the output goes high when the comparator’s output is high and low when it’s low.  As the sine output goes positive it ‘trips’ the comparator which will then go low, in turn causing the sine output to go low which then trips the comparator again…and so on.

The ICL7621 is a CMOS amp and because of its symmetrical output distortion is low, so low in fact that the comparator can be simplified somewhat by ditching the components in the feedback loop.  Can’t remember what the peak-to-peak voltage of the sine wave is but I think it was 1V.  The LFO can be run at higher voltages but you need to divide down the comparator’s output and it helps if the tail of the T-network is tied to half the supply rather than ground (you’ll most likely get some distortion at lower frequencies otherwise).  Keeping the amplitude small also keeps distortion low.  The LFO is run off 3V from a crude discrete regulator (it works, that’s all that matters) so there’s no need to divide down the comparators output, instead a large value resistor limits the signal fed back to the first amp .  Also, current consumption of the amp is a lot lower at this voltage (and this amp is micro power so it’s low to begin with) so current spices are miniscule, which almost guarantees click free operation (one of the downsides to this oscillator).  The LFO output is coupled through to the gates of the MOSFETs and modulates them.  

The ICL7621 amp is reasonably easy to get hold of here in the UK (though Rapid no longer stock it) as is the regulator, which is also micro power (can’t remember what the total current consumption for this circuit was but if you use micro power amps for the audio part too it’s miniscule).

[Kleber AG]

Thanks a lot Gez, very well explained!  

Looks really interesting, altought I do not would like to put a 4047 there
Cool, the mixer looks perfect.

Why have you stopped with two stages(phase 45) instead of the 4 stages (phase 90)?

How does it sounds in the end? Compared with the originals?

thanks for your time Gez!
Kleber AG

[gez]

Why have you stopped with two stages(phase 45) instead of the 4 stages (phase 90)?

How does it sounds in the end? Compared with the originals?

I've only ever breadboarded the above circuit, I was testing this variation of the LFO out to see how well it performed at lower voltages and a simple two stage phaser was quick and easy to breadboard.  You could add more stages and use the MOSFETs from a 4049 or a similar chip.

Sound wise, the sinewave gives a swirly sound more reminiscent of a Univibe.  I've never built the proper version of a phase 45 though, I'm only going by soundclips I've heard.

[mikeb]

Gez, thanks for the info and further education - you are a true British gentleman (although that Rasputin-like photo I saw before still haunts me  :shock: ).

Mike

[gez]

although that Rasputin-like photo I saw before still haunts me  :shock:

Good! :twisted: ( )

[gez]

I should mention that the LFO can be sensitive to component changes.  It's consistent from circuit to circuit, but only if you stick to the same parts - change the caps to a different type/tolerance and you can get changes in amplitude.

The LFO is worth playing with.  At higher supply voltages the amplitude is a lot bigger and is useful in OTA circuits.  Another way to vary amplitude is to put a pot in series with the large resistor coming off the comparator's output, but if amplitude gets too low oscillation stops, which is why I did the depth arrangement as shown.

The range isn't great with these LFOs either.  Increase it too much and you get a very non-linear feel to the 'speed' pot (everything bunches up at one end).  It's possible to compensate for this by using a non-linear active as the variable resistor but then things get a little complicated.  With a large range, distortion of the waveform also tends to occur at lower frequencies.

Still a good sinewave oscillator mind, nice and simple and you can vary amplitude (check out the National app note for frequency calculations and more info).  Works best with a CMOS amp if you go for the simplified version I've shown.

[Kleber AG]

Hello
I just found this:
http://www.fortunecity.com/tinpan/humperdinck/223/getmsg.gif
You may have to enter here:
http://www.fortunecity.com/tinpan/humperdinck/223/effect.htm
And scroll down until "Delay Effects" and click on "MXR PHASOR"
That's because of Fortunecity :?

But this circuit is somewhat using the same aproach as yours gez, and it's a MXR phaser, and I know nothing about it, could it had been produced as a Phase90 in the past??? :shock:

So it solves the JFets(no fets there) matches issues(?), and what else?
Oh, and theres no pad-down (divided input) here...actually there's some gain at input :shock:  a gain factor of 10!!! :shock:

Kleber AG

[gez]

But this circuit is somewhat using the same aproach as yours gez, and it's a MXR phaser, and I know nothing about it, could it had been produced as a Phase90 in the past??? :shock:

So it solves the JFets(no fets there) matches issues(?), and what else?
Oh, and theres no pad-down (divided input) here...actually there's some gain at input :shock:  a gain factor of 10!!! :shock:

There's only so many ways you can do phasing so it's taking the p*ss a bit when people claim that phase circuits are 'their own design'.  Most phasers are just variations on a theme and you rarely see anything new (though I'm working on it!  :twisted: ).

The above schematic you linked to uses a 4016 chip so it's pretty safe to assume that PWM at high frequency is being used to switch the individual devices in the 4016 to create variable resistance.  The switches in a 4016 can take quite large signals without distorting so there's no need to divide the input down.

There's a schematic on line of the eti 447 phaser which shows the input divided down and boosted at the mixer.  Can't find the link though...anyone?