Distortion on Mosfet based phasers experiment

[bioroids]

Hi!

I posted the results of my latest experiments. I breadboarded various stages from this page: http://www5b.biglobe.ne.jp/~houshu/synth/PhaseFet0205.GIF
and tested the harmonic distortion on my computer spectrum analyzer.

I tried the original circuit #4 but it distorted a lot more than the others, anyone has any experience with it?
It was kinda dissapointing, because the other linearization circuits have some disaventages, and this was a new hope for finally making a decent phaser with the CMOS mosfet trick.

Anyhow, if anyone wants to see the graphs here's the link www.bioroids.com.ar (click on Phase stage distortion)

Any comments or corrections are welcomed

Miguel

[George Giblet]

There's two possibilities:

Haven't you got the biasing correct.  The original ckts are for dual supplies with a ground.  If you have converted to single supple you have to do it right.

While the linearization may work MOSFETs have an additional cause of distortion ie. the built-in diodes from D to S.  Linearization doesn't help with the diode distortion they are basically outright clippers and when they clip the DC conditions at the source could change completely shafting the behaviour.   To remove the diode effects you need to drive at less than 10mVp-p.

[bioroids]

You are correct, I transferred the circuits to single supply. Basicaly the mosfet source connects to Vref instead of gnd, as in most phasers.

So, you mean circuit #4 will only work with dual supplies. Can it be adapted to single supply systems?
I'm not really sure how this linearization works...

About the body diodes, I think 10mVp-p is really a low number, that would make it completly useless for guitar!
I managed to get a good sounding phaser, with not too much distortion or noise (far from perfect though), and feeding it signals louder than 10mVp-p, but this judging only by ear.

I have yet to try this at different bias voltages for the mosfet, corresponding to different parts of the sweep.

Thanks and Luck!

Miguel

[Mark Hammer]

I can't see your page at the moment, so this might not apply.  However, some time back, Mike Irwin told me that while he had been a convert to the #2 circuit in the linked document from Osamu Hoshuyama, he had come to the conclusion over time that he did not like it.  His rationale was that while the drain/gate RC network DID keep distortion at bay for a bit, once the signal level rose to the point of clipping, the clipping was much uglier than what he was experiencing without the network.  I guess, putting it another way, if clipping ugliness were on a scale of 1-10 (10 being ugliest), circuit #1 got you clipping with a 2-4 rating up to the critical threshold, and maybe 6 over that threshold, where circuit #2 got you clipping with a 1-2 rating up to the critical threshold and a 7-8 above that.  Mike told me he felt the costs associated with circuit #2 did not outweight the benefits.

BUT....he's thinking in terms of synths and synth signal levels, where the risk of exceeding that threshold are much greater than with guitars (he doesn't have a guitar on his premises for testing circuits, to the best of my knowledge).

I mention all of this to raise the possibility that the distortion from Sam's #4 circuit may well follow the same general pattern Mike observed with #2: namely that distortion may be negligible up to a certain point and disturbingly nasty above that.  Sam would have tested it out with higher signal levels, I imagine, so when he says "low distortion" I am pretty sure he is not really saying "low distortion with unboosted single coil pickups".  Still, the point is that the degree of clipping in these various circuits can sometimes show nonlinear relationships to input level; i.e., great performance...until the dam bursts.

[bioroids]

There seems to be a need for more experiments with this stuff.

The results I got led me to believe this kind of circuits wouldn't perform very well with synth levels, unless you are willing to tolerate big amounts of distortion or hiss. Or I am doing something wrong (that could be posible  ). In either case I'd like to get to the bottom of this, as the cmos ic as variable resistor is an often mentioned trick, but is hard to find info on using it in a reliable way.

Luck!

Miguel

[amz-fx]

Good work, Miguel.

I'd like to see more sites with real-world research like that!

regards, Jack

[bioroids]

His rationale was that while the drain/gate RC network DID keep distortion at bay for a bit, once the signal level rose to the point of clipping, the clipping was much uglier than what he was experiencing without the network.

I'm at the bench rigth now and I found "experimental" evidence of this happening! I'll post it soon

I'd like to see more sites with real-world research like that!

Thanks Jack!

Luck

Miguel

[lovekraft0]

Just a thought - could the opamp follower in figure #3 be replaced with a bipolar emitter follower?

[bioroids]

Just a thought - could the opamp follower in figure #3 be replaced with a bipolar emitter follower?

I guess you could. You just have to be carefull and readjust the bias going to the mosfet.

Luck

Miguel