Linear optocouplers?

[phasetrans]

Has anyone ever played with linear optocouplers, like the Vishay IL300, as e.g. element for gain control?

[PRR]

https://www.vishay.com/docs/83708/appnote50.pdf
https://www.vishay.com/docs/80130/spiceil300.zip
https://www.vishay.com/docs/83710/appn54.pdf

[Rob Strand]

I worked on a non-audio project where they were used as intended for analog isolation.   They did work.  I remember some issues with low-level tracking/offsets.   For the early stages of the project they worked well enough that we didn't need to address the offsets.   I can't remember exactly what bandwidths the devices were passing.

[phasetrans]

I worked on a non-audio project where they were used as intended for analog isolation.   They did work.  I remember some issues with low-level tracking/offsets.   For the early stages of the project they worked well enough that we didn't need to address the offsets.   I can't remember exactly what bandwidths the devices were passing.

I presume they're handy for current sensing and the like...

For here, I'm thinking about abusing the servo loop to make distortion, or a compressor, or similar. Maybe the CTR nonlinearity is interesting as an effect?

[Rob Strand]

I presume they're handy for current sensing and the like...

For here, I'm thinking about abusing the servo loop to make distortion, or a compressor, or similar. Maybe the CTR nonlinearity is interesting as an effect?

The details of the project aren't so clear anymore because it was a bit of a mess and quite some time ago.  A colleague and I were put on the case to try to recover it as best as we could.  We did have things like current limits - multiple implementations  .

I have pondered using those devices for a compressor more than once since they are a bit more precise than your average opto.  The low level tracking thing always stuck in my mind though.  I can't remember the linearity of the opto part itself.

[phasetrans]

I presume they're handy for current sensing and the like...

For here, I'm thinking about abusing the servo loop to make distortion, or a compressor, or similar. Maybe the CTR nonlinearity is interesting as an effect?

I have pondered using those devices for a compressor more than once since they are a bit more precise than your average opto.  The low level tracking thing always stuck in my mind though.  I can't remember the linearity of the opto part itself.

The IXYS app note I have for linear optos gives "8bit" resolution for photoconductive mode of operation, and "13bit" for photovoltaic mode.

[PRR]

.....I can't remember the linearity of the opto part itself.

Besides which, the IL300 part is *intended* to be wrapped in an opamp to reduce errors.

Not that a LED/optoDiode has such a lot of nonlinearity. Like a BJT except 10X the price. And the dark current (hiss).

If you need to break ground it can be a valid approach, especially at hundreds of volts ground difference.

It is not at all like a PbS photoresistor.

[Rob Strand]

Besides which, the IL300 part is *intended* to be wrapped in an opamp to reduce errors.

Not that a LED/optoDiode has such a lot of nonlinearity. Like a BJT except 10X the price. And the dark current (hiss).

If you need to break ground it can be a valid approach, especially at hundreds of volts ground difference.

It is not at all like a PbS photoresistor

That's the basic idea of the device which lends itself to some form analog signal isolation.  No, it's not a photo resistor - there's been a few threads on the forum about the differences between solid-state photodevices and photoresistors already.  The idea behind say a compressor is to modify the transfer characteristic so the analog gain is can be made variable while keeping the feedback loop.   The simple photoresistors are inherently isolating but there's no real need to have isolation.   When we throw out the requirement for isolation and go for solid-state we end-up with an OTA solution.  The question is can the opto idea bring any improvements to that.   Maybe not.

Some interesting points in this article.  Way back in 1976 and basic technologies.  Stumbled across it in my stash of stuff. Not a widely quoted article.  (John Vanderkooy is a highly respected guy in audio and acoustics.  He doesn't often contribute to magazines.)
https://keith-snook.info/wireless-world-articles/Wireless-World-1976/Wideband%20compander%20design.pdf

[amptramp]

You can use two normal optoisolators to create a single linear circuit:



Some time ago, Unitrode came out with their UC1901 series that allowed a feedback signal to be amplitude modulated on an RF carrier so that feedback for a power inverter could be isolated.  This is an application note:

https://bigdiode.co.uk/pdf/UCx901__unitrode.pdf

[phasetrans]

Some time ago, Unitrode came out with their UC1901 series that allowed a feedback signal to be amplitude modulated on an RF carrier so that feedback for a power inverter could be isolated.  This is an application note:

https://bigdiode.co.uk/pdf/UCx901__unitrode.pdf

Thanks for the fascinating read.

[phasetrans]

...Not a widely quoted article.  (John Vanderkooy is a highly respected guy in audio and acoustics.  He doesn't often contribute to magazines.)
https://keith-snook.info/wireless-world-articles/Wireless-World-1976/Wideband%20compander%20design.pdf

Rob,

I enjoyed this article quite a bit, especially as someone who has spent almost no time thinking about compressor design. I can see how the IL300 would trigger the memory.

I have a different question about Vanderkooy article, namely in the control voltage circuit, not the gain circuit.

How common is his configuration of A3 and R1 for draining the C1 capacitor in other designs? Similarly, how common is the bridging D5 diode for faster response to dramatic level changes?

[Rob Strand]

How common is his configuration of A3 and R1 for draining the C1 capacitor in other designs? Similarly, how common is the bridging D5 diode for faster response to dramatic level changes?

I'd call it a known but not that common configuration.

The discharge actually depends on the input level.  For the same average discharge behaviour the resistor values needs to be a smaller value than a simple resistor to ground connection.  The question is, once you do that, does it sound better or worse?   There's so many minor variations on these things.

A lot of compressors deliberately have a non-precision diode drop in the rectifier to get a softer compression knee.

[phasetrans]

How common is his configuration of A3 and R1 for draining the C1 capacitor in other designs? Similarly, how common is the bridging D5 diode for faster response to dramatic level changes?

I'd call it a known but not that common configuration.

...

A lot of compressors deliberately have a non-precision diode drop in the rectifier to get a softer compression knee.

Rob, is there a canonical discussion of control voltage rectifiers and attack release approaches?

2 decades behind mixing desks, and most of the things that matter in a compression are in the profile of the control voltage attack and release.

[Rob Strand]

Rob, is there a canonical discussion of control voltage rectifiers and attack release approaches?

2 decades behind mixing desks, and most of the things that matter in a compression are in the profile of the control voltage attack and release.

Not really.

For rectifiers it's all over the place.   The basic forms are some electronics texts, Linear Tech and Analog devices applications.  There were quite a few examples in magazine articles and books by Jerald Graeme from the 70's.   These are just rectifiers in isolation.   To keep things simple just pick a form and run with it.

A detailed analysis of compressors is in this article.   However he only goes through a few scenarios.  This is a "pure" design without adding diode drops to mess with the control profile.
https://web.archive.org/web/20160607164247/http://ajoliveira.com/ajoliveira/gen/pdf/preprints/paris88.pdf

The NE571, NE570 datasheets and appnotes are always good value.  So too are a lot of stuff from That corporation.  It can take some time understanding the rectifiers in those because they are tricky ways of getting rms converters.

If you look at all the variations it's actually an enormous amount of effort to evaluate.   I did a lot of stuff in the 90's.   The control profile is very different for feedforward and feedback.   A simple linear rectifier can produce infinite compression on feedforward but you need a very sharp characteristic to get that with feedback - for example the Dyna Comp rectifier uses the VBE characteristic,  very sharp on/off, and it produces a limiting character (with practical signal levels it reaches something in the order of 10:1 as the exponential characteristic never gets to be vertical.)