Optocoupler - wide input voltage range

[PeLa]

Hi everyone,

I am hoping that one of you can help me with something that I initially thought would be quite simple... but then found out that it's not... for me.
I have an optocoupler that receives a signal from a sensor and forwards it to a microcontroller.
I would like to connect different sensors with a relatively wide range of signal voltages (3.3V - 24V).
Therefore, the optocoupler should turn on at sensor Voltages above 3.3V. The maximum sensor Voltage is 24V.
The LED in the optocoupler has an absolute maximum forward current rating of 60mA. 2mA is enough to reliably turn it on.

This is the optocoupler: https://www.vishay.com/docs/84181/tcmt4100.pdf

Ideally, I'd like to keep the component count to a minimum and make it work over a temperature range between 20 and 60C.

If you have any ideas, I'd be all ears!

Best regards,
Pete

[DC1MC]

Put a resistor in series that gives you the smallest current to trigger it reliably at 3V3, let's say 2mA, at 24V it should not produce more than 50mA.
Ca. 800ohms should be OK.

DC1MC

[Benta]

Use a current regulating diode:

http://www.linearsystems.com/lsdata/datasheets/J500_J501_J502_J503_J504_J505_J506_J507_J508_J509_J510_J511_Current_Regulator_Diode.pdf

You can also make one yourself using a JFET.

Doesn't get any simpler.

[DC1MC]

Use a current regulating diode:

http://www.linearsystems.com/lsdata/datasheets/J500_J501_J502_J503_J504_J505_J506_J507_J508_J509_J510_J511_Current_Regulator_Diode.pdf

You can also make one yourself using a JFET.

Doesn't get any simpler.

I would say a resistor in series is simpler 

[PeLa]

Can't argue with that! Thanks mate! 

[jmelson]

Use a current regulating diode:

http://www.linearsystems.com/lsdata/datasheets/J500_J501_J502_J503_J504_J505_J506_J507_J508_J509_J510_J511_Current_Regulator_Diode.pdf

You can also make one yourself using a JFET.

Doesn't get any simpler.

I would say a resistor in series is simpler 

But, the simple resistor will not work.  OK, let's try some values.  Most optocouplers need about 5 mA to trigger reliably.  So :

(3.3 - 1.7)/0.005 = 320 Ohms   (3.3 v minium supply minus 1.7 V drop of the LED)

Now, how does this work at 24 V?  (24 - 1.7) / 320 = 70 mA!  Most optos have a maximum LED current of about 20 mA.

Now, if you find a sensitive opto that can reliably turn on with much lower current in the LED, the resistor can be made a larger value.

Working backwards :

(24 - 1.7) / 0.020 = 1115 Ohms

Now, at 3.3 V you'd have :  (3.3 - 1.7) / 1115 = 1.4 mA  If your opto will turn on reliably at 1.4 mA, then this will work.

Jon

[DC1MC]

Well, the OP said that the reliable turn-on current is 2mA and I've based my calculations on this value. In any case, before going on further, a quick test with a resistor will actually take minutes.

[wraper]

But, the simple resistor will not work.  OK, let's try some values. Most optocouplers need about 5 mA to trigger reliably.

Simple BJT optocouplers do not trigger to begin with. You have a CTR (current transfer ratio) spec which depends on LED current and is not linear, and usually drops to a very low value at low LED current. You have an analog output in a nutshell.

[Benta]

Can't argue with that! Thanks mate! 

It seems several other posters can. Also, you'll need a 1 W resistor at least, which generates quite some heat.

Thanks for the  by the way.

[Zero999]

How well the opto-coupler triggers or not, depends on what's connected to the other side.

The minimum CTR of the TCM4100 is specified at 50% at 5mA. It will be lower at lower currents: look at the data sheet.

A resistor is definitely the simplest option, but not ideal.

[T3sl4co1l]

Easily furnished with a current source.

JFETs and CRDs aren't so popular anymore, but depletion MOSFETs are readily available.  Indeed, a DN3545 will give you a whopping 450V of input range, albeit not continuous given the power dissipation at that extreme (4mA * 450V = 1.8W).

Also, if it works reliably at 2mA, it's probably wise to double the operating point to 4mA, since CTR drops with age and temperature.  Or use a somewhat more sensitive or stable type opto, or use a lower threshold on the output side (which can also be sharpened with a current source load, or an amplifier), assuming you don't need much speed of course (optos are very slow and need every mA they can get, to go fast).

Examples:
https://electronics.stackexchange.com/questions/368365/opto-coupled-eia-422-receiver-understanding-suggested-circuits
https://electronics.stackexchange.com/questions/355159/wide-voltage-range-input-regulation-24v-3-3v
https://electronics.stackexchange.com/questions/262011/mosfet-current-source-current-limiter-request-for-review

Tim

[Gyro]

Why not spread the dissipation with an intermediate clamp.... Series resistor, Zener clamp (maybe 8 - 12V), second series resistor, Opto.

It would still be a cheap and minimal solution, which would extend the input range without punishing the Opto LED too much at the upper end. Cheaper than a FET.

[Benta]

Cheaper than a FET.

An MMBFJ112 from ON Semi costs around $0.10 and it's just one part. Compared to three parts plus mounting cost.
BTW, the DN3545 is around $0.70. Hmmm.

I dunno.

[Gyro]

If it's smd, and the board isn't crammed I think I win. When I was designing high volume consumer stuff, I'd move heaven and earth to leap on an opportunity to save an easy $0.10 on the bom. 

The resistors and additional placement are down in the noise, it's the actives cost and sourcing that matter.

[Benta]

Good. You win. 

'nuff said, the OP has seen the options.

[Gyro]

Good. You win. 

'nuff said, the OP has seen the options.

Sorry, I didn't mean it that way.

[S. Petrukhin]

US$0.1206
ON Semiconductor NSI45030AT1G

https://lcsc.com/product-detail/Constant-Current-Regulator_ON-Semiconductor-ON-NSI45030AT1G_C150066.html

[Smokey]

Bump for some interesting solutions to a common(ish) problem.