Re:Audio Optocouplers
by MisterTea on Saturday September 24, 2022 @05:35PM
(#32966728)
I've used the IL300's to interface signals in a legacy machine to a PLC and found they were not very linear. Don't see how these 4N25's will be any better. Wound up using texas instruments/burr brown isolation amplifiers instead (ISO122 or something).
[ Parent
| on HN ]
Re:Audio Optocouplers
by exmadscientist on Saturday September 24, 2022 @05:51PM
(#32966850)
Those are probably great if you can afford them. I've never had the budget for them in a production design, so haven't played with them much. Standard practice these days is just to toss a full ADC on the other side of the barrier and then isolate the SPI bus (or what-have-you). A few ADCs have the isolation built in, but for others there's stuff like TI's ISOW7841, which have 4 isolated digital lines plus isolated power in one package. It's all kind of expensive when added up, but it's cheaper than the old style analog isolation amplifiers, and you'll usually get more than one ADC channel once everything's together, so you win as soon as you want two signals to cross the barrier.
This was 10+ years ago and was a smaller internal project for two older machines so we didn't care they were $20 or $30 each.
And the isolated digital trick is used in the new industrial isolators. The neat part is since there's a micro in them they add in scaling and conditioning options which is nice.
This is why I'm glad I don't do automation one-offs and remind myself to have respect for those that do. It's such a different solution space!
Did you put one of the diodes in the feedback loop to linearize them ?
Yes but there were still linearity issues. The senior engineer played with it for a bit until he said screw it, spend the money and be done. Project was small enough to justify the cost.
Do you remember numbers ? I'd guess somewhere in 0.5-1% THD range for simple single opamp servo ? I was kinda mildly interested into doing it for sensors altho just microprocessor blasting serial via optocoupler onto other side seems like easier solution...
Their application note (page 9) details circuits most audio people would just look at it and go "screw it, i will just buy audio transformer like rest of us"
https://www.vishay.com/docs/83708/appnote50.pdf
I don't doubt they achieved the claimed values but the circuit to do it had 4 precision opamps (bipolar one have 7) and a bunch of other components so I'm not surprised he didn't want to bother
Sorry, cant remember specifics. But if I were doing this again, today, I'd use an isolated serial ADC. I had a lot of fun with analog stuff but fixing signal issues in code is easier than solder.
Feedback is frowned upon in audiophile circles:
https://www.audioholics.com/audio-amplifier/amp-myths-negati...
It depends on the audiophile circle, but yes. There's a circle of folks that swear by single-ended triodes, class A operation, no feedback. I would love to hear a setup like that, but I haven't bought one (yet?), so I won't comment until I do.