Fet Phaser Idea

[mac]

I was reading the National Fet Circuit Application, note 32, Feb 1970, page 9 and decided to try the idea.



As long as the LDRs are from the same batch there are no critical parts to match.
I'l try to post a sound clip soon... broken wrist


mac

[gaussmarkov]

cool.  seems like it should be robust to expected variation in JFET characteristics.  it's tempting to use JFETs for the voltage controlled restistors (VCRs) as well (a la MXR Phase 90) so that the board would be a forest of fets.    but then you'd lose your no-matching-required feature.

[bioroids]

I like it, and the good thing about the LDRs is that you can replace the LFO with anything you like without having to mess with the signal path.

Have you tried using 2 leds instead of 4, each led driving two LDRs placed side by side? That would be cool to reduce consumption and space requeriments.

Regards!

Miguel

[Gus]

It is good to see something not a ff or ts.

  I like the use of the jfets reminds me of the univibe bootstrapped darlington stages but you are  getting high input R with one active device and fewer parts.

IMO LEDS are kind of hard to dim in a controlled way.  If I was to build this I think I would look into making a voltage controlled PWM drive led drive circuit to mimic the univibe lamp time constant and maybe other leading and trailing edge ramping up and/or down time adjusts contol

[slideman82]

Nice nice nice! I love FET's, and this seems to work really nice! Great work, Marcelo! Miguel's mods seem to work, and probably sounds the same... am I wrong? You have just gave me some ideas, I'll start to read some datasheets and looking for some circuits.
What about using a secuencer with those LEDs? If it is mixed with the LFO, it probably could be done! Just some ideas...

[Isaiah]

That looks cool, Mac
Funnily enough, I've been working on something very similar.

[mike_a]

This looks like a great project!

A few suggestions:

I would like to suggest that you put a 150 ohm resistor in series with each LED.
They don't always have EXACTLY the same forward voltage and this way you have the resistor to handle the differences.

Another thing which I am concerned about is the ability of the opamp to drive the the amount of current needed by the LED's.
Maybe you should use low-current LED's.  However, I'm not certain it will pose a problem - this needs to be checked.
edit: another thing that might be better for current consumption is to connect the LED's between the opamp and V+ rather than to the ground.

For the LFO, I would recommend TL062.  They are lower current and this might help in case of ticking.

Final suggestion - maybe a univibe mod?  you can turn the feedback off and put a cap switch.  then you have both a phaser AND a vibe 

I'm really looking forward to hear how it sounds. 
Many thanks, Marcello!
Mike.

[RickL]

Am I missing something? This looks like a vibratro with feedback not a phaser. Most phasers achieve their sound by mixing the dry signal with the phase shifted signal which this design doesn't appear to do. If the feedback loop was completely disconnected I would expect it to sound like a vibrato, as more feedback is introduced it probably sounds more like a phase shifter.

I would say your mix pot (R21) is actually controlling the amount of feedback, not the wet/dry mix.

It might be worth trying connecting the output of the phase shift sections to Q5 via a resistor (10k maybe - I'm guessing) and also connecting the input (dry signal) to the same place via another resistor (start with the same value, you might have to alter it up or down to get the most intense phase sound).


Or someone with better knowledge of phasers can tell me how badly I've misunderstood the circuit 

[mac]

Sound sample at:
copy to a new browser window
or go to
http://guitarfxs.tripod.com

Guitar to breadboard to Laney LC30II to mic to sound card.

Last two takes were recorded with a mod I have to draw, regen. From Q5 drain I sent two signals via a 223 cap + 150K each to the input and to Q3 base. A lower resistor and it begins howling. The feedback to the input sends the sweep more to the treble side, and to Q3 base to the bass side. A couple of independent pots, say 1M-A may add controllable regen.

mac

[mac]

Thanks for the replies. I'll have to check all those suggestions but I beg the help of someone with 2 hands and free time
My fingers still look like a pack of sausages

1.The idea of 1 or 2 LDRs worth the try.
2. Current draw is about 12ma, IC: RC4558 does not get hot.
3. Fets with Vgs > 4.5v could be a problem. MPF102 can be above that limit according to datasheet. Maybe a BF245B or 2N5952 to be safe, although I have different brand MPF102 and they are all below that limit. What about 18V?
4. A darlington using 10Ks at C and E may have enough input impedance to be used instead of fets. Something like this, 10ks at C and E, 4M7 from B to Gnd, 2M2 from C to B. This would set C at near 6V and E at 3V for a wide range of darlingtons despite their hfe.
I tried a 2N3904 at Q5, using a 2M2 from B to C, 10k at C and 1K at E, and it worked fine. Maybe even 2N5088 or BC549C can be used in all stages, 10k at C and E, 2M2 from B to C, 4M7 from B to Gnd, Vc=6.2V and Ve=2.8V.
In other words, maybe stages Z-in need to be a few hundreds Ks.
5. I tried 472, 103, 223 and 473 caps, kind of univibe, nice.

IMO LEDS are kind of hard to dim in a controlled way.  If I was to build this I think I would look into making a voltage controlled PWM drive led drive circuit to mimic the univibe lamp time constant and maybe other leading and trailing edge ramping up and/or down time adjusts contol

I had to put R28 and C8 to tame the LEDs, like the 47nf in the P90 from the 3M9 to Vref. R24 also control the brightness.

What about using a secuencer with those LEDs? If it is mixed with the LFO, it probably could be done! Just some ideas...

         

It is good to see something not a ff or ts.

Rickl

I tried connecting Q1 drain, source or base to the volume pot wiper via R21 and a 223 cap. Or to Q5 base. Also from the junction of C1 and C9 to Q5 base. All work, but sound different. I liked this way. It also gets rid of 223 caps.
I guess that the input to Q5 base is more P90-ish.

Since there is some loss in every stage, the input signal, no matter if taken from the input or Q1 drain, is higher than what reaches Q5 base, so R21 takes car of this situation. At min there is more dry signal than shifted, at max more shifted than dry.
Q5 boosts the weakened signal. At max volume it is above unity gain.

mac

[mac]

A new version with multiple regeneration and a deeper effect.



New sample soon.

****

I replaced all fets for BC549C since BJTs are more predictable, and it is working fine.

mac

[mike_a]

Does all of the LED's flash at the same intensity?

It seems strange to me that they would, without a resistor in series with each of them...
the series resistor would also let each LED reach it's OWN peak intensity, rather than the intensity that corresponds with the LED which has the lowest forward voltage.

[slideman82]

I don't think the LEDs intensity will affect the sound or function very much... do all the LDRs measure the same?

[mike_a]

I think that it will.

LEDs doesn't have the EXACT same V_on (forward voltage), just like transistor's hfe varies within the same models.
When you connect LEDs in parallel the way it's done here, you only let the voltage get to that of the LED with the lowest V_on.
That means that the other LEDs doesn't reach their OWN V_on.
That means that you don't get the maximum value-swing from those LEDs, which in turn means that you won't get the maximum resistance-swing from the LDRs.

One might say that a couple of millivolts wouldn't make such a difference, but one must remember that the LED (and all diodes for that matter) have an exponential current-voltage curve, and therefore every millivolt does count!

Mike.

[mac]

I thought to put  a 100R in series with each LED but I considered variations in LDRs, the fact that if one LED does not reach its peak intensity the LDR may be at its min resistance, the amplitude of the swing from 100k to 470R or so, etc.
I also remember posts were it was discussed that LDRs and LEDs are not tricky as control resistor as fets.
Just connected as is an hoped for the best, since all diodes and LDRs are form the same batch.
But I can add the resistor.

mac

[bioroids]

What if you use one of those big round leds that look like bugs, and put the ldrs around it, univibe style... Then you can forget about the led variations.

And if someone opens the pedal, it will look like you scraped the circuit from a crashed alien ship!

Regards

Miguel

[slideman82]

Yeah, you're right again, Miguel! Maybe a 4-ldr-in-a-box like the 'Vibe will be better, less current draw.

[mac]

That's a good idea. I'm going to try it with the BJT version which BTW it's working as good as the FET version.

mac

[GonzoFonts]

That sounds REALLY good.

To me, at times it sounds more like a Univibe instead of a phaser (especially the first 20 seconds or so).

Have you tried to play around with the caps to get it to sound more like a Univibe?

GF

[mac]

yes: 472, 103, 223 and 473 for C1 ... C4   C L!
maybe 222, 472, 103 and 223 for less bass sweep.

mac