EHX Octave Mutiplexer signal detection question

[Fancy Lime]

Hi there,

I have been thinking about and looking into suboctavers again and found that the EHX Octave Multiplexer seems to track particularly well. So I tried to find a correct schematic (there are some with fatal errors out there) and happened upon this article:

http://blog.randagodron.eu/index.php?post/2017/12/30/EHX-Octave-Multiplexer-Analyse

My French is not what it used to be but do I see this right that the thing they call "zero crossing detector" is actually not a zero crossing detector but a positive and negative peak detector? And one that uses less parts while tracking better than the one in the OC-2? I must say, this is one of the most ingeniously elegant pieces of circuit design I have seen in a while. My utmost veneration to whoever came up with and designed this and fine-tuned it decades before Spice.

Took me a while to wrap my head around and the confusing way in which the factory schematic is drawn does not help. Nor does the fact that there is an important looking but completely functionless 100k resistor that looks like it biases the comparators but doesn't need to because they are DC coupled to the previous stage. Also, I'm pretty sure there is a cap missing somewhere in series with the Range switch. The way it is drawn, closing the switch would bias the entire peak detector close to ground and seriously mess up the peak detection. Or maybe that is the point? If so, the surplus 100k resistor makes sense again. Not sure *why* one would criple the detector like that but hey, maybe it sounds good? Can't really tell from the videos I found.

Cheers,
Andy

[Rob Strand]

My French is not what it used to be but do I see this right that the thing they call "zero crossing detector" is actually not a zero crossing detector but a positive and negative peak detector? And one that uses less parts while tracking better than the one in the OC-2? I must say, this is one of the most ingeniously elegant pieces of circuit design I have seen in a while. My utmost veneration to whoever came up with and designed this and fine-tuned it decades before Spice.

I'm seeing the same as you.   Looks quite neat.    There's a few EHX pedals with novel detector circuits.

completely functionless 100k resistor

The 100k discharges the peak detector otherwise it will get stuck on the largest peak.   You want to get a pulse on *each* peak of the fundamental eg. for a decaying note you want to track down the decaying peaks.   The 18ms time constant is in the right ball-park for that.

The 560R is a little quirky.   At first sight I'd be more inclined to use two such resistors around the opamp outputs but maybe in the end they could skimp with a single resistor in the "ground" leg.

[StephenGiles]

The marked "zero crossing detector" is of course an Adaptive Filter, in fact almost all if not all of the front end up as far as the LM311 was used in the EH Rack Mount Guitar Synth. So what you should have at the input of the 311, acting as a squarer, is a very compressed pure sine wave which as the designer told me back in 1980, will convert any input into a sine wave!

[Fancy Lime]

...

completely functionless 100k resistor

The 100k discharges the peak detector otherwise it will get stuck on the largest peak.   You want to get a pulse on *each* peak of the fundamental eg. for a decaying note you want to track down the decaying peaks.   The 18ms time constant is in the right ball-park for that.
...

I'm not sure we are talking about the same thing. Unfortunately, the component numbers in the factory schematic are largely illegible, so I'll be refering to numbers in this one:
http://blog.randagodron.eu/public/stomp/octave_multiplexer/octave_mult_simu_sch_01.PNG
The resistor I meant is R11 (100k). The discharge resistor for both halves of the peak detector is, if I understand the hole thing correctly, R10 (180k), no?

The marked "zero crossing detector" is of course an Adaptive Filter, in fact almost all if not all of the front end up as far as the 3011 was used in the EH Rack Mount Guitar Synth. So what you should have at the input of the 3011, acting as a squarer, is a very compressed pure sine wave which as the designer told me back in 1980, will convert any input into a sine wave!

I have questions:

How is it an adaptive filter? I am not too sure of the terminology here but I don't see how to interpret this as anything other than a positive and negative peak detektor acting as a frequency detector.

What 3011?

Which EH Rack Guitar synth are you referring to? IIRC there were several versions but all schematics I could find have pretty much nothing in common with the Octave Multiplexer, especially not the frequency detector. I remember dimly however having read an interview with someone from EHX, who said that there was something very special and awsome about the way the EHX synths detect frequency. One schematic of the Rack Synth (http://schems.com/Schematics/Guitar%20Synth%20and%20Misc%20Signal%20Shapers/Electro%20Harmonix%20Guitar%20Synth.gif) has a frequency detector, which I do not fully understand but that bears more resemblance with the one in the Boss OC-2. The microsynths have almost exactly the same detector as the OC-2. Either way, these things all produce a square wave, not a sine wave, don't they?


BTW, some intresting info here:
http://schems.com/Schematics/Guitar%20Synth%20and%20Misc%20Signal%20Shapers/


Cheers,
Andy

[StephenGiles]

3011 of course should be LM311 - ever tried putting text into an iPhone holding a lead with 34 kilos of greyhound on the end in the other??

Now I promise you that the original circuit does label this circuitry as an Adaptive Filter which was described to me by the designer at EH UK in 1980. This presents a sinewave version of any input to the 311 acting as a squarer then frequency doubled by a cmos 4030 and presented to one of the 4047s in the frequency detector loop - my name, which does need to be calibrated in a very particular way to work properly, and by Clapton it did!!

Help, I can't find my paper copy of the circuits just now to prove my point. Oh well back to photographing the planet Venus!!

[Rob Strand]

I'm not sure we are talking about the same thing. Unfortunately, the component numbers in the factory schematic are largely illegible, so I'll be refering to numbers in this one:
http://blog.randagodron.eu/public/stomp/octave_multiplexer/octave_mult_simu_sch_01.PNG
The resistor I meant is R11 (100k). The discharge resistor for both halves of the peak detector is, if I understand the hole thing correctly, R10 (180k), no?

Yes, some part of those EHX schematics are hard to read.   

FWIW, the time constant is 9ms not the 18ms like I mentioned previously.

I see what you are referring to now.   I suspect that 100k forms a deliberate x0.9 divider with the 10k (R7 on your link) between the filter and the detector.   The idea is the peak detector opamps need to swing one diode higher than the filter opamp.  The divider drops the level so if the filter opamp clips (which is likely) the peak detector is still operating as a peak detector.   [Another theory is the schematic has an error and the schematic should look more like the OC-2 which has a tapping along the filter before the detector.    Not a lot of evidence pointing this way though.]

[Fancy Lime]

3011 of course should be LM311 - ever tried putting text into an iPhone holding 34 kilos of greyhound in the other??

Now I promise you that the original circuit does label this circuitry as an Adaptive Filter which was described to me by the designer at EH UK in 1980. This presents a sinewave version of any input to the 311 acting as a squarer then frequency doubled by a cmos 4030 and presented to one of the 4047s in the frequency detector loop - my name, which does need to be calibrated in a very particular way to work properly, and by Clapton it did!!

I wouldn't know. I don't have an iPhone, my dog is not a Greyhound and only weighs 24 kg. But a piece of work all the same 

I see now where the misunderstanding lies:
The Guitar Synth (http://schems.com/Schematics/Guitar%20Synth%20and%20Misc%20Signal%20Shapers/Electro%20Harmonix%20Guitar%20Synth.gif) has a thing called Adaptive Filter in the schematic, which feeds into an LM311 and probably does what you say. I can't tell because I don't understand it. Am I right in assuming that "1048" stands for the EH1048 OTA and that that has the same pinout as the CA3094? In that case I will try to wrap my head around this thing. By the looks of it, it seems to me that it is a very clever fundamental extractor that works by shifting the cutoff of a low pass filter (implemented with OTAs), using a frequency-to-voltage converter (741), driven by a frequency detector (2x1458). But I really have to look at this some more to fully appreciate it.

The same Adaptive Filter arrangement appears in the DELUXE Octave Multiplexer (http://www.freeinfosociety.com/media/images/4113.jpg). But the DELUXE Octave Multiplexer shares almost nothing except the name with the (non-deluxe) Octave Multiplexer. I was talking about the latter, which has neither the adaptive filter, nor any LM311. For simple frequency division, the OCatve Multiplexer scheme seems sufficient to me and I really like it's elegant simplicity.

I'm not sure we are talking about the same thing. Unfortunately, the component numbers in the factory schematic are largely illegible, so I'll be refering to numbers in this one:
http://blog.randagodron.eu/public/stomp/octave_multiplexer/octave_mult_simu_sch_01.PNG
The resistor I meant is R11 (100k). The discharge resistor for both halves of the peak detector is, if I understand the hole thing correctly, R10 (180k), no?

Yes, some part of those EHX schematics are hard to read.   

FWIW, the time constant is 9ms not the like I mentioned previously.

I see what you are referring to now.   I suspect that 100k forms a deliberate x0.9 divider with the 10k (R7 on your link) between the filter and the detector.   The idea is the peak detector opamps need to swing one diode higher than the filter opamp.  The divider drops the level so if the filter opamp clips (which is likely) the peak detector is still operating as a peak detector.   [Another theory is the schematic has an error and the schematic should look more like the OC-2 which has a tapping along the filter before the detector.    Not a lot of evidence pointing this way though.]

Hmm, that could be. I had not thought about the opamps operating close to the rails. Thanks!

Cheers,
Andy

[StephenGiles]

That awful version of the EH Guitar Synth circuits was a scan of a tracing I did from my paper copy of the factory circuits - yes on tracing paper!!!!!! back in the 1980s, the scan was most probably done in the 1990s, thus it's appalling quality!!

I have a fairly recent scan of the relevant part of the factory circuits where you can just about make out adaptive filter under the 2 x 1048 (CA3094) section. It is not surprisingly quite faded now after all these years.

Circuits taken off, you've had time to download!!

I will put in dropbox if requested.

I found my paper copy of the Guitar Synth Circuits - here is the Adaptive Filter label !!

[sergiomr706]

https://lectric-fx.com/ made a version of this octaver, the Downtown and another sub octaver, the dod meatbox, soy sauce. Maybe schematics are cleaner there.

[StephenGiles]

"my dog is not a Greyhound and only weighs 24 kg. But a piece of work all the same"

Our greyhound just got into a temper and threw her bed and toys across the room because my wife would not get off the sofa!!
 

[Scruffie]

That awful version of the EH Guitar Synth circuits was a scan of a tracing I did from my paper copy of the factory circuits - yes on tracing paper!!!!!! back in the 1980s, the scan was most probably done in the 1990s, thus it's appalling quality!!

I have a fairly recent scan of the relevant part of the factory circuits where you can just about make out adaptive filter under the 2 x 1048 (CA3094) section. It is not surprisingly quite faded now after all these years.

Circuits taken off, you've had time to download!!

I will put in dropbox if requested.

I found my paper copy of the Guitar Synth Circuits - here is the Adaptive Filter label !!

Per chance do you have the other halve of the full sized version you originally posted up, Stephen? Once this quarantine business is over I'm getting a broken unit in for repair and I suspect those margin notes are going to come in handy... if I can decipher them.

[StephenGiles]

Here you go

https://www.dropbox.com/transfer/AAAAAL8Bi3vGZHAblo1gSQwcb93BJAo2NmYdRNuICDA5R28zjLX4QLI

The notes are very faded now unfortunately.

[Scruffie]

Thank you kindly 

Oh, you weren't kidding, well, plenty of free time to play about with photo software... perhaps a redraw is in order.

[Fancy Lime]

Here you go

https://www.dropbox.com/transfer/AAAAAL8Bi3vGZHAblo1gSQwcb93BJAo2NmYdRNuICDA5R28zjLX4QLI

The notes are very faded now unfortunately.

Thanks, Stephen! This really is a fantastic beast.


May I propose a quarantine time passing project? How about a thorough functional analysis of the Guitar Synth? It seems pretty modular so it should lend itself to module-wise collaborative exploration. I was going to try to redraw, Spice, and analyze the Adaptive Filter anyway but there are plenty more parts, which I don't even begin to understand at the moment. Is anyone else up to this?

Cheers,
Andy

[StephenGiles]

Good idea Andy, I look forward to group words of wisdom on this!

[Scruffie]

Here you go

https://www.dropbox.com/transfer/AAAAAL8Bi3vGZHAblo1gSQwcb93BJAo2NmYdRNuICDA5R28zjLX4QLI

The notes are very faded now unfortunately.

Thanks, Stephen! This really is a fantastic beast.


May I propose a quarantine time passing project? How about a thorough functional analysis of the Guitar Synth? It seems pretty modular so it should lend itself to module-wise collaborative exploration. I was going to try to redraw, Spice, and analyze the Adaptive Filter anyway but there are plenty more parts, which I don't even begin to understand at the moment. Is anyone else up to this?

Cheers,
Andy

The adaptive filter's not too complicated, guitar input > compressor > then it's just a state variable filter made with those two 1048 (CA3094). The actual tracking bit is done with one rectifier coming from the high pass, another from the low pass which both feed an integrator that compares the two signals and decides if the filter is 'unbalanced' and needs to move up or down to get back to its centre point.