Practical differences between octocuplers

[waspinator]

I'm interested in using optocouplers to isolate and shift 24v signals to 3.3v, and I've picked out a couple to try, but I'm not sure what the differences are. Besides one being 4 channel and the other being 1. Are there two LEDs in the 4 channel, and would that mean that I could sense AC signals? Why would I want to connect to the base of the receiver on the 1 channel, but not on the 4? Could you use optocouplers for output (3.3v to 24v) or would I want a relay driver (ULN2803) instead?

Thanks

[dmills]

The base connection allows you to trade CTR for speed by adding a resistor between it and the emitter, seldom done in practise as there are much quicker parts readily available these days.

The LTV part will indeed work with AC, but be a little careful as the isolation between the individual LEDs is often not up to mixing mains and low voltage on different inputs to the same opto package, but for 24V in an industrial control situation, these have much to recommend them.

You can use optos for outputs, but they are not particularly robust, so you would need a protection network, and they are polarity sensitive, photomos relays are better if you need the isolation and reasonable speed in a solid state device, or go for the real thing and use a miniature relay.

Do remember to design for worst case CTR, it is not what you would call a well controlled parameter.

Regards, Dan.

[MrAl]

Hi,

One big difference is the switching speed.  This can vary by a lot.  They are generally slow anyway, but some are much slower than others.  Check the data sheet carefully if you need fast speed.

[Cervisia]

The base is usually not used. (Once upon a time, nobody made DIP-4 packages, so it did not cost anything to connect the base to some otherwise-unused pin.)

Many optocouplers cannot switch much current. Depending on how much current you need, you might be required to use the optocoupler to switch a transistor or other driver.

Also see What are the differences between optoisolator output styles?

[Ian.M]

Also, to allow for LED aging, take the worst case CTR and halve it!
If you are right on the bleeding edge when freshly built, it *will* fail as it ages . . . .

[Kjelt]

For CTR and explanation:
http://electronicsbeliever.com/opto-coupler-ctr-explanation/

[MrAl]

Hi,

I almost forgot some of this stuff myself which i read here now and i will outline below.

Yes, the CTR which is the current transfer ratio is an important spec because it tells you how much current you need on the input to get the output current up to par.

The aging is another good point.  When the LED ages it can no longer produce the same CTR so after some time the application will fail.  Since LED's age MUCH less when run at one half the rated current, it makes sense to run the LED only as high as half the rated current for long life.  If i remember right the LED lasts ten times as long (half life) when run at one half the current.

The other small point is the transistor base connection.  To get maximum CTR you cant use the base, but that also leads to the slowest turn off speed.  If you want faster speed you have to use a base resistor and then try to figure in how much CTR you lost because of that.  For example, with a 100k base resistor you might lose one half the CTR which means you dont get as much output current.

[Cervisia]

[…] use a base resistor and then try to figure in how much CTR you lost because of that.

This is hard to do in practice because the phototransistor's h_FE is not specified. You'd have to use an optocoupler with a much higher CTR, just to be sure.

In any case, a base resistor does not really give you much of an improvement. In most cases, you would be better of using a logic-output optocoupler with a guaranteed speed, such as the H11L1 (1 Mbps) or the 6N137 (10 Mbps).

[Zero999]

[…] use a base resistor and then try to figure in how much CTR you lost because of that.

This is hard to do in practice because the phototransistor's h_FE is not specified. You'd have to use an optocoupler with a much higher CTR, just to be sure.

In any case, a base resistor does not really give you much of an improvement. In most cases, you would be better of using a logic-output optocoupler with a guaranteed speed, such as the H11L1 (1 Mbps) or the 6N137 (10 Mbps).

I wonder if a Baker clamp as well as a pull-down would help to increase the speed?

If speed is required I wouldn't suggest an opto-isolator. I'd recommend a digital isolator, such as the Si8620, which can do 150Mbps.
https://www.silabs.com/documents/public/data-sheets/si861x-2x-datasheet.pdf

If there's no DC in the signal, then it could be coupled with a transformer or Y capacitors, if a little capacitance bridging the gap is not a problem.

I'm interested in using optocouplers to isolate and shift 24v signals to 3.3v, and I've picked out a couple to try, but I'm not sure what the differences are. Besides one being 4 channel and the other being 1. Are there two LEDs in the 4 channel, and would that mean that I could sense AC signals? Why would I want to connect to the base of the receiver on the 1 channel, but not on the 4? Could you use optocouplers for output (3.3v to 24v) or would I want a relay driver (ULN2803) instead?

Thanks

There are other types of opto-couplers such as: photo-diode, opto-TRIAC, FET, photovoltaic, MOSFET driver, etc. It depends on what you want to do.

If no isolation is required then you might be better off with the ULN2803. If you don't need to drive the full 24V, say 18V is acceptable, then go for the CD4504 and a suitable voltage regulator.

[Cervisia]

I wonder if a Baker clamp as well as a pull-down would help to increase the speed?

The 6N137 already clamps its output transistor.

For analog optocouplers, the results are mixed (below: both with I_F = 5 mA); it depends on the characteristics of the phototransistor, which are not specified.