Causality 6 phaser Vref problem <SOLVED!>

[ElectricDruid]

Hi All,

I've got a Causality 6 phaser built up on stripboard for some experiments, but I'm having problems and I don't think it's my fault!

<EDIT>SOLVED! It was my fault, of course...I should have known The solution is a few posts down...</EDIT>

Here's the schematic:


Now, the thing is, each 13700 phase stage has a little chain of resistors between Vref and ground: 1K8, 27K, 10K. That's a total of 38.8K. If you parallel up four stages like that, that resistance is effectively halved and then halved again - 9.7K. That 9.7K is in parallel with the bottom leg of the voltage divider that's supposed to give us a 4.5V Vref. Since that divider uses 220K resistors, the resistance of the lower part is dominated by the 9.7K resistor, and Vref should be way out of whack.

That's what I see on my stripboard prototype. Vref is far too low. What I don't understand is how it could ever have worked properly, or perhaps it works even with the very low Vref?

I notice that the EH Small Stone phaser (where this OTA-based phase stage was borrowed from) uses a op-amp to buffer the Vref. That would fix it, but I'd rather not add another op-amp if I can avoid it.

Has anyone who's built the Causality 6 seen this problem? What's the Vref reading in a working pedal?

Thanks,
Tom

[antonis]

Μake 220k/220k resistors 10k/10k to see what happens then..

[ElectricDruid]

Μake 220k/220k resistors 10k/10k to see what happens then..

My thoughts exactly! Well, ok, not *exactly*...I actually swapped them for 47K/47K, and got 1V for Vref (as expected). 10K should give me roughly one third of the 9V, since the value of the bottom resistor would be halved.

[antonis]

 

OK.. Now make upper 10k resistor 4k7 for Vref slightly higher than 4.5V or 5k6 for Vref slightly lower than 4.5V..

[anotherjim]

I'm not sure actual Vref is so important.
Look at it another way...
27k into 1k8 is input attenuation.
10k is emitter follower load termination. Low source impedance at that point.
27k from the output is negative feedback.

[ElectricDruid]

I'm not sure actual Vref is so important.

I agree the exact value isn't crucial, but it needs to be reasonably in the middle of the 0-9V range. With 220K/9.7K, you get Vref = 0.36V, which is so close to ground the input op-amp and output mixer will be clipping.

Which is what I don't understand - it seems to me there's an error here, but no-one else ever seems to have noticed, and I'm not sure how you could *not* notice. So what's going on?!?

Look at it another way...
27k into 1k8 is input attenuation.
10k is emitter follower load termination. Low source impedance at that point.
27k from the output is negative feedback.

All true, but all true when Vref is nice and central too!

[PRR]

....each 13700 phase stage has a little chain of resistors between Vref and ground: 1K8, 27K, 10K.

No.

Look inside the box.

It has 1k8, two 27k, going to two emitter followers biased (nominally) at Vref. And to keep the emitter follower alive, there is 10k to ground.

The 1K8 & 27K conduct nominally zero current. (Few nA to feed Bases.)

Cas6-guts.gif

[anotherjim]

Given this ^, try to find whatever else could be pulling Vref down so much.  Although it does look like a 220k divider was meant to bias an opamp for active Vref and the opamp was left out. I always view high divider resistor values with suspicion anyway. If I built one I'd put 10k's in.

I'm sure Pauls explanation is on the money but... could there be a power-up condition going on? that is, Vref starts low anyway behind the +supply (another cap has to charge) and the circuit acts to keep it low? I mean the same kind of thing that some opamps like the LM833 do with a weak Vref supply and then another cap (same as to 0v) on Vref to the +supply can cure it as Vref rises at +V/2 in step with the rise in +V.

[Rob Strand]

Given this ^, try to find whatever else could be pulling Vref down so much.  Although it does look like a 220k divider was meant to bias an opamp for active Vref and the opamp was left out. I always view high divider resistor values with suspicion anyway. If I built one I'd put 10k's in.

I'm sure Pauls explanation is on the money but... could there be a power-up condition going on? that is, Vref starts low anyway behind the +supply (another cap has to charge) and the circuit acts to keep it low? I mean the same kind of thing that some opamps like the LM833 do with a weak Vref supply and then another cap (same as to 0v) on Vref to the +supply can cure it as Vref rises at +V/2 in step with the rise in +V.

When everything is working there's no problem with the OTA stages loading down the weak (2x220k) Vref but when something goes bung the assumption is no longer true.

Measure the DC voltage on the 10ks to ground.   One of those will be off for sure (unless there's a bigger problem and there's a massive offset at the inputs).   However since it's DC coupled you will have to measure all stages and deduce where the fault is.  You might even have to disconnect the feedback.

(Something to think about 0.1V drop across the 10k translates to a much larger drop across Vref.)

[ElectricDruid]

Thanks Paul, Jim, Rob.

Initially I was checking voltages before I put the chips in (always a good idea to save magic smoke!) but in this case, it throws things out completely, for the reasons given. Having convinced myself there was something funny going on, a low voltage with the chips in matched my expectations. So I fooled myself, not for the first time.
It looks like I'm probably just looking for a basic short or wrong value on the Vref somewhere (how ordinary!). I'll let you know what I find.

[bean]

I notice that the EH Small Stone phaser (where this OTA-based phase stage was borrowed from) uses a op-amp to buffer the Vref. That would fix it, but I'd rather not add another op-amp if I can avoid it.

On a few occasions with OTA circuits I've run into a similar problem and buffering Vref solved it, FWIW.

[ElectricDruid]

Ok, found it! There were two problems which made it a bit tricky. Problem one was one of the 13700's was fried. It has probably had a tough life on my breadboards, poor thing. The Iabc pins were reading 0.4V and 0.6V on that one (1.35V is good) so I likely gave it too much Iabc current at some point. It's easily done. Problem two was I'd forgotten the connection between the OTA output and the darlington input (pin5 to pin 7, and 12 to 10). Doh!

It seems to be happy now. I'll get some sound through it tomorrow, but for tonight all the voltages check out finally.

Thanks everyone for all the help. Sometimes I can't see past the end of my own nose, but I knew talking to you all about it would shake my ideas up, and so it proved.

[Rob Strand]

I'm pretty sure you would have worked it out.

The other day I was doing some fine tweaking by adding resistors in a parallel with a 3k3 1% resistor.  I finally hit the 2k7 point and since I want to tweak the value very finely I put in a 2k7 1%.   Then it seemed like the value was far too small.  After wasting half a day doing measurements, I thought stuff this I'm going back to the 3k3.  So I measure the 2k7 and it was 2k2.    Turns out a 2k2 1% got into my 2k7 1% stash.  It also happened the second band was a little thinner on the 2k7 and hard to read.  If I look at the 2k2 now I still see 2k7 50% of the time.  So the next morning I put in a real and measured 2k7, go through the test again and now everything make sense.  Haven't been screwed over like that for over 10 years - I hate those darker blue 1% resistors 

[bluebunny]

I was going to chip in with a "I built a C6 and it works fine" comment, but by the time I got all the way down here, it looks like you have it sorted!      (I can take some measurements if you need.)

[anotherjim]

Problem two was I'd forgotten the connection between the OTA output and the darlington input (pin5 to pin 7, and 12 to 10). Doh!

Don't feel too bad over that one. I think the brain assumes it's an internal connection already from the way it's usually drawn.