Asymetrical phasing....

[artsinbloodshed]

hey guys!
i've just finished another phase 90 clone.
The phasing is asymetrical!I mean the rising time of the phasing is way faster than the decreasing one...the sinusoide looks more like a triangle than a wave...
I used matched fets, tl 072 etc...the whole package is point to point correct.
i checked and this problem is somewhat common like that
:http://www.youtube.com/watch?v=r2AqCkT93Sg
could it be a defectuous TL072?

[Rob Strand]

Check the following:

- voltage across zener
- voltage across bias pot
- swing at output of LFO opamp (set speed to slow)
- 15uF cap orientation (maybe temporarily change the cap)

[Thomeeque]

hey guys!
i've just finished another phase 90 clone.
The phasing is asymetrical!I mean the rising time of the phasing is way faster than the decreasing one...the sinusoide looks more like a triangle than a wave...

Define "way faster" P90's LFO generates triangles (it gets rounded later on it's way to the FET gates, but base /on 15uF cap/ are triangles) and they are little asymmetrical - asymmetry is caused and grows with difference of V_REF to V_CC/2*.. T.

*
P90_LFO.asc

[artsinbloodshed]

Thanks guys!
Yeah weird!
I removed the zener diode and no change,the phasing was still asym ...then I resoldered the zener and no phasing at all...
Should be the zener. I think it's a common prob...

[Mark Hammer]

Smoothness and feel of the sweep are just about everythng when it comes to phasers.  Certainly FET-matchng plays a role in getting smoother-souding sweeps, but as a group, we need to spend a little more time talking about how to get optimum performance from LFOs.

[Rob Strand]

but as a group, we need to spend a little more time talking about how to get

I've spent some time on this in the DSP world but for the circuit world the Black Ross LFO is pretty good.

[Mark Hammer]

It may be but I find the sweep to be inconsistent from build to build.  Some are just "lumpy", with the upward sweep feeling like someone arriving late for a meeting, bursting in the door, and fiddling around with their papers.

[R.G.]

Mark is correct. The LFO in that circuit is inconsistent. It's working as designed; or, perhaps more to the point, working as implemented if there was no design spent on that part of the circuit.

The LFO is a classical single opamp Schmitt trigger oscillator. The input range of the opamp and how far its output moves when switched up versus down, and the exact value of the bias voltage change the symmetry of the pseudo-triangle the cap makes.

A Schmitt trigger of this kind has a reference voltage which sets the midpoint of its hysteresis band. That means the input will always switch the output symmetrically around the reference voltage. Those are the trigger points for switching. The hysteresis is set by the ratio of the positive feedback and input resistors to the + input.

The output of the Schmitt trigger will always be as close to the + supply and the - supply as the opamp can get. Note that this varies from opamp to opamp type. TL072s can only get their output to within about 1.5V of their + and - supplies when they're fully saturated. An LM324 style device can get its output to within 50mV of ground but only within 2V of the + supply - it's asymmetric.

The capacitor which acts like an integrator ramps up from the lower hysteresis point until it hits the upper hysteresis point, where the Schmitt trigger flips state and drives it down toward the lower hysteresis point again, and the cycle starts over.

The reference voltage for the hysteresis point is not exactly in the middle of the power supply, so the capacitor sees a bigger drive voltage for the up side (or down side) of its swing than for the down side (or vice versa, respectively).The bigger drive voltage one way than the other through the speed resistance makes the cap ramp faster one way than the other.

There are ways to force symmetry in similar circuits. I did some of them in the voltage controlled LFO for the P90+/P180+ at geofex. But they are complicating factors, and no attempt was made to put these things in the original P90 circuit. At the time it was thought up, it was considered revolutionary to simply have the circuit work inside a small stomp box and have a phaser at all.

We've moved on. I expect perhaps to see someone advertise a phaser with vintage asymmetry in the LFO wave as an advantage some day. Sigh.

[Thomeeque]

We've moved on. I expect perhaps to see someone advertise a phaser with vintage asymmetry in the LFO wave as an advantage some day. Sigh.

Well, 3V bias asymmetry is already pretty audible so if somebody is used to it, he may appreciate such a feature.. T.

[R.G.]

Well, 3V bias asymmetry is already pretty audible so if somebody is used to it, he may appreciate such a feature.. T.

What's not widely appreciated is that you can put in a symmetry control; and you can force symmetry by clamping the output of the opamp to ground or 2x the reference voltage. There's a lot of flexibility there.

[Rob Strand]

It may be but I find the sweep to be inconsistent from build to build.  Some are just "lumpy", with the upward sweep feeling like someone arriving late for a meeting, bursting in the door, and fiddling around with their papers.

Mark, I know what you mean!

I didn't want to get caught up in the thread but there's many distinct mechanisms at work.  Circuits variations and tolerances add to each mechanism making it had to see what is going on and therefore making it easy to blame (a non-specific aspect of) the LFO.

A brief list of the main are:

1) Sweep wave shape. This controls the general smoothness of the sound over the sweep.
    eg.  ross black phaser vs phase 90
2) Range of the sweep.  There's two aspects of this how low the sweep goes and how high.
3) Circuit variations

With modulation effects like the phaser, flanger and chorus there's two things going on simultaneously.  The first is frequency locations of the notch, and how far they sweep.  The second is the amount of pitch shift and the amount of shift at each frequency. Phasers produce different amounts of pitch shift at each frequency because the phase shift network doesn't act over the whole band width (changing the cap values and JFET variations will move this around).

Generally sweeping the notches in a log manner sound natural.

The ideal pitch shift isnt't so obvious.  One source of lumpyness is too much pitch shift at low frequencies.  A 10Hz pitch shift at 1kHz is 1% change, but at 100Hz it is 10% change. The amount of pitch shift is defined the derivative d (phase) / dt.   Fast and/or wider sweeps produce more pitch shift.   The "fastness" is also determined by the sweep wave shape.  If the waveshape is such that the LFO voltage doesn't hang around long in the low frequency region the pitch shift is high.   

With JFET phasers the behaviour of the sweep at low frequencies is highly variable.  It depends a lot on matching, bias set-up and also the size of the resistors in parallel to the JFET in relation to the JFET resistance.

When you consider JFET matching, LFO issues that RG outlined, and bias set-up you can see why there's quite a bit of unit to unit and design to design variation.

[Electron Tornado]

Using the schematic here (the second one, posted by tubelectron):     http://www.diystompboxes.com/smfforum/index.php?topic=89779.0     where would I put the probes in order to see the LFO wave on an oscilloscope? Pin 7 of U1 and Vref?

I repaired a Phase 90 by replacing the two chips. However, I just used what I had on hand which was a TA75558, previously removed from a TS-9, for U1 and an LM324 for U2. Would using either of those as substitutes affect the symmetry of the phase sweep?