Opamp acting up - weak and distorted output

[Coriolis]

Hi folks

I am currently working on a box that utilizes two of these:


It's a phase reversal circuit from one of Craig Andertons books, with a few bits added (the 1k input resistor and the 47pf cap to ground, the first resistor after the input is a 1M pulldown - not currently installed). The back-to-back diodes across the inputs on the opamp are supposed to be input protection - I got that idea from R.G:
http://www.geofex.com/circuits/what_are_all_those_parts_for.htm

Thing is; it doesn't quite work.
I used an audio probe to narrow it down and:
On the output of this circuit, I get a very weak signal, which is somewhat distorted (not gated or anything, just weak and distorted)
I've checked the board for faults, solderbridges, etc several times. I've got the right parts in it.
Oh, and I'm using a MAX1044 to power the TL074 with (+/- 9V).

The two other opamps in the TL074 are being used for an input buffer/splitter and a mixer respectively, and those work just fine.

What could I be doing wrong here?

EDIT: Just checked with R.G's debugging guide, looking under "blatty or gated sound" http://www.geofex.com/fxdebug/bias_prob.htm
which states that inv input, noninv input and output must be within 10 mv of each other. Well, my noninv input is at 0.25V and my inv and outputs are around 8.40V!
That can't be good then?

Best,

C

[R.G.]

You left out a very important part of the circuit - the power supply. What power supply did you use?

If it's 9V and ground, and you don't establish a bias point in the middle, yes, you'll be biased incorrectly.

[Coriolis]

Eh..my psu is bipolar - MAX1044 putting out +/- 9V, and i've checked that the opamp has power.
No half supply "virtual ground" needed, right?

C

[R.G.]

Eh..my psu is bipolar - MAX1044 putting out +/- 9V, and i've checked that the opamp has power.
No half supply "virtual ground" needed, right?

No, but the + input must always be referenced somewhere. Solder a 1M or bigger resistor from the + input to ground and see if things don't improve.

[Coriolis]

Thanks for the tip, but ...will the circuit not lose it's ability to reverse the phase if I do this?
Can't hurt to try it though, so I'll do it and see.

C

[grapefruit]

Can you check that you actually have -9V on the VEE pin of Op amp and +9V on the VCC Pin?


Stew.

[Coriolis]

I just checked that (again) and I have -9.18 and +9.52 respectively. Oh - and on the correct pins too 
I'm just stumped about the whole thing - I had the phase circuit working on veroboard previously, but the +input wasn't referenced to ground there either...

C

[Coriolis]

Ok - I added 1M resistors from +inputs to ground like R.G. said, and that brought the sound back, nice and clear.
However, my phase reversal switches as per the schematic above now work as channel mute switches instead. Damn.
Something is strange about that Anderton circuit. Wish I could make it work though - that would let me do the entire box i'm working on with just one quad opamp.
Anybody see a solution I don't?

Now I might have to use two opamps per phase switcher (one inverted, the other not) and switch between those instead.
Maybe I'll settle for phase reversal on just one channel though - having two is luxury I guess, but could have been handy at times...

Thanks for the help though. 

C

[GibsonGM]

Why can't you just use a single, center-biased inverting opamp stage for each phase reversal? Wouldn't it be simpler? 

Try simplifying it; remove the diodes, maybe they are doing something re. rectifcying and distorting (sorry R.G., I'm sure they  have a good purpose!)  There's probably something glitchy we just can't see on 1st glance, but I'm sure it will work!
Try moving the 1M to VRef to the left side of the  10K biasing resistors maybe?

[R.G.]

Ok - I added 1M resistors from +inputs to ground like R.G. said, and that brought the sound back, nice and clear.
However, my phase reversal switches as per the schematic above now work as channel mute switches instead. Damn.
Something is strange about that Anderton circuit. Wish I could make it work though - that would let me do the entire box i'm working on with just one quad opamp.
Anybody see a solution I don't?

From Geofex, 2001: http://geofex.com/Article_Folders/polarity_reverser/polarity_reverser.htm

Put a 1M to ground across the two throws of the switch. Put a 10K in series with the switch pole to the + input of the opamp. Check your power supplies for good right at the opamp pins. Diodes should not matter as the + and - input voltages are and must be within millivolts of each other for normal operations.

[Coriolis]

Why can't you just use a single, center-biased inverting opamp stage for each phase reversal? Wouldn't it be simpler?

Try moving the 1M to VRef to the left side of the  10K biasing resistors maybe?

Mike, if I'm reading you correctly, you are suggesting that I run the circuit on single supply, or? I could do that of course, but I prefer to get the extra headroom of the dual supply, to make this circuit a little more universally useful than just something for guitar.

R.G - you mean like this?

Looking at it, I don't quite grasp what this does...

EDIT: I tried the above modification - still only gives me a channel mute instead of phase reversal.
A bit of explanation is in order:  The purpose of this box is to split the input signal in two, run them through external effects (for parallel fx) allow for phase reversal of each channel, and mix them together in the end. So, to test if the phase reversal works, I just run jumper cables from each channels send, to it's return. This should enable me to cancel out the signal entirely, with one channel in phase and the other out of phase, right? Well, right now I get a volume boost with both channel switches in one position, and no signal at all in the other position.

C

[R.G.]

R.G - you mean like this?
Looking at it, I don't quite grasp what this does...

Yeah.

Here's what it does. The gain of an opamp if you put the signal into the resistor to the inverting (-) input and ground the noninverting (+) input is Rf/Ri. The gain if you keep the same Rf and Ri, but put the signal into the + input and ground Ri is 1+Rf/Ri. If you put two signals into the opamp, one into the + side and one into the - side is just the instant by instant difference in the two signals.

If you make Rf=Ri, and put the same signal into both inputs by connecting the + input to the outside end of the Ri, you get (1+Rf/Ri) times the signal from the + side, and Rf/Ri subtracted from that from the - side, giving a total of (1+Rf/Ri-Rf/Ri) = 1, or a noninverted replica of the signal. Grounding the +input gives only the inverted -Rf/Ri version. So flipping the + input from the Ri input to ground should give gains of +1 and -1.

I've used this several times, including building a flux gate magnetometer using it. It works well. Something else about your setup is keeping it from working well, or its possible that your sensing method is keeping you from knowing it works.

EDIT: I tried the above modification - still only gives me a channel mute instead of phase reversal.
A bit of explanation is in order:  The purpose of this box is to split the input signal in two, run them through external effects (for parallel fx) allow for phase reversal of each channel, and mix them together in the end. So, to test if the phase reversal works, I just run jumper cables from each channels send, to it's return. This should enable me to cancel out the signal entirely, with one channel in phase and the other out of phase, right? Well, right now I get a volume boost with both channel switches in one position, and no signal at all in the other position.

And you're using bipolar supplies, right? Have you measured the DC on the pins in both switch positions?

Have you tried listening for signal in both switch positions for ONE channel at a time, just to be sure you're not getting some quirky cancellation?

[Coriolis]

Ok - my findings so far, after measuring voltages and audioprobing the output of both phase reversal circuits - in both switch positions:

Switches in the non-inverting position (+ input to ground)

CH1:  +input =  -0.12V (yes - negative voltage)
         -input =  -7.65V
Sound: Yes

CH2:  +input = 0.08V
         -input = 0.27V
Sound: Yes

Switches in the inverting position

CH1:  +input = 0V
         -input = 0V
Sound: No


CH2:  +input = 0V
         -input = 0V
Sound: No


Sound from one channel alone is the same amplitude as the whole box in bypass mode (as it should be).
There is a bit more clarity in the signal though, since the signal is buffered.
Both channels (when on) mixed together yields the same clear sound - only louder (as would be expected).

I'm still staring hard at my pcb layout and my wiring to find any clues, but nothing yet...

C