Optocoupled PNP Trannies-Rather than optocoupled FETs

[ExpAnonColin]

I recently bought a fairly large network router box, for the enclosure, at Quinn's.  In order to use it for my purposes, I had to take out the PCB, which wasn't anything really special...  Lots of resistors, a few trannies, a few LEDs, and some network jacks, but one thing caught my eye:  There are about 8 6-pin devices.  I recognized them as possible optocouplers and did a search for the part number, CNY17F, and it turns out they're optocoupled PNP's.

I've had smashing success using H11F3's as voltage to resistance converters and other resistance-based applications, but I'm not sure what I would want to do with the CNY17F's.  On the datasheet it appears as though basically you could use it as a unity follower, using the LED as a very simple VU meter and the output of it as the voltage followed signal.  Do you guys think that's a decent thing to try, an envelope follower?  It would be sort of pointless because whatever you used to drive the LED would probably have to be another follower with little ripple...

Does anyone have any ideas?  I understand FETs better than PNPs, and without the base connection I am lost as to how to make amplifiers or anything out of them.  In an ideal world I would like to be able to create a well biased VCA with them, with the LED + being the CV input, but that could be a stretch.  Anyone have any designs or ideas?

Thanks.

-Colin

[Peter Snowberg]

Just look at any circuit that controls a CLM-6000 or a Vactrol with an LFO. That will show you how to bias the LED to almost-on and vary it as needed.

*** NOTE ***: Don't drive the opto too hard, the LEDs typically found in them are very low power.

In terms of the PNP for an output, just remember that PNP transistors are current based devices so to get a voltage change across them you need to have current flowing through them. The easiest way to do that is with a simple current source (10K resistor) hooked up as a follower and then processed by an opamp to get your result. You could also hook the opto up like a typical class A gain stage with two resistors, one being bypassed by a cap to set the gain and the bass response.

Some of those optos actually have a base connection going to a pin and you need to add bias resistors to bias the output off with no LED drive.

Keep in mind that some optos are made to be digital control devices with darlington outputs and the like. Some won't be easy to control.

The thing that strikes me offly is you say this is a network board? Could it be a phone system board?, because you won't find optocouplers in most network devices. There are lots of little pulse transformers in some network gear that look similar to optos so I just wanted to mention that.

[ExpAnonColin]

I've actually already designed quite a few things with the H11F3's as well as vactrols, so I'm good in terms of biasing the LED... the problem was really the tranny.

This particular one doesn't have a ground connection...  and I'm positive it has to be some sort of network router, all it was was a long box with 8 network jacks in the back.  And a google search for the part number yields that it really is an optocoupled NPN of some sort.  Not a darlington... here's the datasheet:
http://www.fairchildsemi.com/ds/CN/CNY17F-1.pdf

The class A amp idea sounds cool.  I'll look into that.

-Colin