info on optocouplers needed

Correct. Your LED drive level has to allow for aging. And if you want the LED to last longer, don't drive it as hard.

The worst current transfer ratio I saw in that series was 20%.

Datasheet here:

With your load, for 20% CTR, you do need about 3mA drive minimum, plus aging margin.

10 years = 90.000 hours. Aging depends on LED duty cycle. Choose accordingly. 10 years continuous duty? Cut the current! Use the 50% min. CTR version, and increase the load resistor.

A back-to-back LED isolator needs no reverse protection diode, if that's what you were asking.

It's not clear what you want. What does the signal driving the opto look like electrically, and how do you want the opto to respond?

(Is the signal source a voltage or current source? Or sink? Of fixed, or variable, or unknown polarity?, etc.)

HTH, James Arthur

Thanks James

I need a lowest cost solution so i'll use a quad optocouler . The type is LTV-846S .It has a 50% CTR @ 5mA. The load i'll fix to 0.8 mA . This means i'll need 1.6 mA .I'll keep it safe at 3 mA led current @ 12V .So finally this is 6 mA @ 24 Volts.With aging it should be Ok since until the light intensity divides by two should last continously for 10 years . Am i correct ?. Now regarding polarity i meant what happens if accidentally somebody connects the led input in reverse ?.In that case is it absolutely necessary i use a diode in shunt with the led Or the reverse breakdown voltage is related to current also so with 6mA we are ok? Now regarding configuration i meant how can somebody detect a contact closing with common to ground or +V easily. Imagine an array of n optocouplers with n inputs detecting n number of switches and one common for all switches. How can we detect without extra connections GND or +V common connection?

James Arthur wrote:

(un top-posted. James)

michael Nikolaou wrote:

years. > Am i correct ?.

That's the right method, but we have to check on the LED aging rate. Also, you've not given the duty cycle. The problem is trivial if the LEDs are only on intermittently, possibly serious for 90.000 hours of continuous 12mA illumination, probably okay at 6mA--not sure.

For comparison, last time I faced this requirement I used 100% CTR optos, 22k pullups to +5v (220uA load), and 2mA drive.

Either in shunt or series, but yes, the protection diode is needed. 12v reverse will kill the LEDs. Unacceptable.

This is what I did, with jumpers to reconfigure each input:

FIGURE 1 Contact closure to ground. (view schematic in Courier font) ======== LTV-846S ___ . . . . . . . . . . V+ >------>>---|___|---+-------. .----- R1 | . | | . --- . --- |/ . D1 / \\ . \\ / ~~> | . --- . --- |>. . | . | | . | . | '----- O-----------------+-------' . . . . . . . . . . . O---- GND

FIGURE 2 Contact closure to V+. ========

O----V+ LTV-846S ___ . . . . . . . . . . O---------|___|---+-------. .----- R1 | . | | . --- . --- |/ . D1 / \\ . \\ / ~~> | . --- . --- |>. . | . | | . | . | '----- GND--------------+-------' . . . . . . . . . . .

Cheers, James Arthur

A resistor-zener-resistor TEE would both reverse protect and remove the sensitivity to drive voltage. Many very inexpensive transistors have base-emitter breakdown that can be useful here (because zeners are more costly).

James The problem is if you want to use an array of inputs , with one common reference either V+ or Ground .So for n inputs you require n+1 cable connections .You have to fill the board with jumpers.Is there another alternative ? Anyway James your help was great . I appreciate that.

Michael

That's a nice idea in this way you can work with higher voltages and constant led current. The thing is is there a sot-23 device with two zener's inside ? . Another factor is the cost of this device .Anyway the idea is fine

A resistor-zener-resistor TEE would both reverse protect and remove the sensitivity to drive voltage. Many very inexpensive transistors have base-emitter breakdown that can be useful here (because zeners are more costly).

[concerning two resistors-and-zener input limiter]

DigiKey lists MAZC062D0LCT for $112/thousand, a dual

6.2V zener. MMBT3904 variants are $30/thousand, and the base/emitter breakdown is about 7V, should do fine.

The optoisolator is considerably more expensive, and a service repair is much more than that.

Nice idea.

Cheers, James Arthur

No alternative that I see with your isolators and signal sources. In my case, I designed in jumpers to configure my inputs as needed.

If you wanted to spend the money you could do something like this (below) and sense contact closures to either rail:

FIGURE 3 ======== LTV-846S ___ . . . . . . . . . . V+ >------|___|--------+-------. .----- R1 | . | | . --- . --- |/ . D1 / \\ . \\ / ~~> | . --- . --- |>. . | . | | . CONTACT | . | U1a '----- INPUT O----------------+-------' . | . . . . . . . . . . | | | . . . . . . . . . . +-------. .----- | . | | . --- . --- |/ . D2 / \\ . \\ / ~~> | . --- . --- |>. . | . | | . ___ | . | U1b '----- GND-----|___|-------+-------' . R2 . . . . . . . . . .

Decode the outputs of U1a-b to detect contact input status.

Regards, James Arthur

Googling "optocoupler LED degradation" turned up several sources, but, basically, optocoupler IRLED degradation at 10mA drive appears to be minimal. Fig. 6 of this source:

shows 2% LED degradation with 100.000 hours' operation at 10mA.

Fairchild's H11L1 datasheet only suggests a 10% guardband (overdrive) to compensate for LED degradation over time.

By comparison, just 1.000 hours' operation at 30mA and +55°C degrades an emerald green LED by 28%:

So, all of this reminds me that my earlier referenced application had visible LED indicators in series with the optocoupler LEDs, giving me different aging considerations (visible LEDs age faster). I was also trying to save power, much of which went to the many channels of isolated inputs.

You haven't mentioned those requirements, so even 6-10mA wouldn't seem to be a problem for you.

Best, James Arthur