ETI Audio Phaser Project 447

[RickL]

I was looking through my box of finished but not working projects this weekend (you all have a box like this don't you?) and found two(!) copies of the ETI six-stage audio phaser Project 447. I haven't a clue where I downloaded it from so I can't point you to it.

I am wondering if anyone has successfully completed this phaser. There is at least one error on the PCB layout with the bias trim pot connected to the wrong place. I've fixed this on both boards but I continue to get loud static no matter where the trimpot is adjusted.

The clock appears to be working and I get a faint (compared to the static) dry sound through it

This project uses a 4049 hex inverter as the variable resistance element. The auther claims that the inverters are complementary FETs and that the resistance can be controled by applying a voltage its gate. Can someone explain how this works? In the schematic all the inputs to the inverters are attached to the clock signal, the outputs go to each of the allpass stages and both Vcc and Vdd (and pin 16 which is marked "no connection" on the data sheet) are connected to Vref (which in this case isn't exactly half of 9 volts, it uses a 10k/15k reference). Does this make sense?

[Rob Strand]

I haven't a clue where I downloaded it from so I can't point you to it.

I think it was from the CAG (Cloned Analogue Gear) site.

I am wondering if anyone has successfully completed this phaser. There is at least one error on the PCB layout with the bias trim pot connected to the wrong place.

I've never build the whole thing on the PCB but I've build my own verions with similar blocks - the idea does work.  I seem to remember the article, or a errata note from the magazine, mentioning the pot isn't as shown on the schematic but it should still work (I think the pot on the PCB went between +V and gnd, instead of Vref and gnd).

I've fixed this on both boards but I continue to get loud static no matter where the trimpot is adjusted.

So you aren't even getting clean signal through?  It sounds like the whole show if off the air.  You should at least be getting clean signal.  Try temprarily lifting the resistor between the all-pass stages and the transistor stage - I think it's 5.6k.  If don't get clean signal now it's time to debug the clean path.  Measure the voltages on the all the opamp outputs and the transistor pins.  From that you should be a able to see where things are pooping out.

Can someone explain how this works?

It's clearer to say the inverters are being used as individual N-channel MOSFETS - the weird connections are all about disabling the P-channel devices and pilfering the N-channels out fo the inverter.  Once you see that yu will find it isn't much different to a JFET design.  JFETs and MOSFETs can both operate as variable resistances - basically the gate voltage controls the resistance of the channel.

Vref (which in this case isn't exactly half of 9 volts, it uses a 10k/15k reference)

Yes that's OK, it should still work.

[Mike I.]

Hi Rick,
Take a look at:

http://www5b.biglobe.ne.jp/~houshu/synth/mike_irwin

There is some explanatory info there plus a highpass filter design that uses the 4049 as a bank of identical voltage-controlled resistors. The same  applies to the ETI phaser design (but with a much simpler control circuit).
Regards, Mike

[RickL]

I've explained myself badly. I am getting clean sound out of it but it's overlayed by either steady static or pulsed static depending on the adjustment of the trim pot. This is at a *way* louder volume than the clean signal. The clean sound is about as loud as the bypassed sound. The speed pot appears to be working because the  pulses or static bursts change speed with the position of the pot. I think if I can get the inverters to actually work as variable resisters it will work.

I'll try removing the 4049 to see if that removes the noise. If so, I guess that narrows it down to the 4049 circuit as causing the problem.

Thanks for the replies.[/i]

[RickL]

It is indeed at CAG. Here:

http://omega.tellus.vallentuna.se/anders/pdf/eti447.pdf

[Mike I.]

Rick,
It is possible that at certain settings of the trimpot/LFO the op amps in the audio path go into high frequency (ultrasonic) oscillation. This might be what is causing the noise. Best way to check is to look at the op amp outputs with an oscilloscope. TL071 are much more prone to oscillate than 741 in these circuits (in case you are not using 741). A remedy is to put a small capacitor - say 30 pF to 100 pF - across each of the op amp feedback resistors in the allpass stages to reduce the high frequency gain.
Regards, Mike

[Rob Strand]

(I'm assuming you are using the original layout, so there aren't any wiring problems etc)

The reason you get a lot more noise on the we signal path is because that circuit amplifier the wet signal by a factor of 10 (5k6 vs 56k resistor). So any junk in the wet signal is made worse.

OK, with the 4049 out you should measure the voltage on the output of each opamps in the wet signal, pin 6 should be at about 5.4V or so.  If you aren't getting this there's a problem somewhere with the opamps.  Start at the input side and work through the opamp chain. A bad voltage on the output is usually a sign that the opamp has a problem (soldering or dead).

By the way, you have to use an unbuffered 4049UB device for this circuit.

Put the 4049 back in a re-check the voltages, where the voltage poops out check your soldering and the part values.  I'm not sure what else could be wrong.  When you do this you could try pulling R22 to remove any confounding information caused by the oscillator.  With R22 out see if you can find a point on the trimpot where things are good or bad and measure the voltage on the wiper of the trimpot and on the other side of the 330k resistor.  While these tests aren't solving the problem they are trying to look where the problem is so you can take it from there.