# How to know the minimum LED current of an optocoupler?

I am going to need to use an optocoupler to isolate two circuits in two situations. In both situations, a MCU GPIO will be driving the LED of the optocoupler to send a signal on the other end.

I am planning to use this optocoupler as its was the cheapest I could find that has the shorted rise/fall time. Since an MCU will be driving the LEDs, I would like very little current to be drawn on the GPIO; how do I know what is the smallest current I can drive the LED of the optocoupler?

The first circuit fastest switch frequency is 250ms the second circuit though is a bit more sensitive as it will pull low for 2ms and can have an interval as low as 0.5ms. Will this optocoupler be able to switch that fast at low LED currents?

• Is your criteria for battery power savings or heat rise and logic level switching toggle speed? THat is not clear in your question, as the pullup does not need to be 100 or 1000 ohms. Must your solution be inverting or non-inverting? Or is cost the real criteria for high volume Oct 6, 2021 at 23:00

There is a weakness in all BJT transistors when Ic is < 1 mA. The hFE starts to decline rapidly, which might see in the CTR=70% curve I added in the 1st plot.

Remember you are using the opto as a switch and not a linear amplifier,so current gain is reduced significantly.

It also declines sharply when Vcb becomes forward biased or in other words when Vce < Vbe which is the edge of saturation region. We know transistors drop to about 10% of the max. hFE when saturated near rated current. so they standardize the Vce{sat} for the nearest value of 10% linear hFE for Ic/Ib. Usually values of 10 or 20 to 50 are used.

This opto PT's is no different and indeed CTR drops off quickly at 1mA. In fact you will rarely see any transistor hFE vs Ic curve go below Ic=1mA.

This device has a maximum linear CTR of 6 and min of 3. This effectively after the PD coupling loss the effect of the variation in hFE, so hFE or CTR will drop rapidly (again) if Ic < 1mA and Vce< 0.7V.

So the answer to your question is the minimum LED current is the worst case value that will guarantee a Logic level "1" for "emitter follower" or "0" for "common emitter" using your interface logic supply of 5V.

Using a load value of 10k will give a Tr,Tf < 30 us which satisfies your speed requirement.

## Recommendation

If min = 1 mA , Re = 10k

• any questions?? Oct 7, 2021 at 2:51

Max Forward voltage of the LED is 1.4V, with 1k resistor and 3V3 GPIO that gives a minimum 1.9mA LED current.

Check "current transfer ratio" in datasheet:

Minimum is 300% (ie, 3) and this means with 1.9mA LED current, the phototransistor will pass at least 5.7mA. At this current it will not be fully saturated, so 1K value for R8 is too low.

So if you want 5V on the FET gate, to be safe it would be better to increase R8 to a higher value like 5-10k.

Datasheet mentions max rise/fall time of 18µs, so no problem for your 0.5ms requirement.

• I see so the If rating of the optocoupler is the lower LED current that i should always use? is that right? Oct 6, 2021 at 22:48
• LED current should be at least: minimum current you need through the phototransistor, divided by current transfer ratio. Also read Tony's answer. Oct 7, 2021 at 8:20

The specs say that the minimum current transfer ratio is 300 %. Your diagram shows that the output current needs to be 5 mA. Therefore, the minimum input current would be 1.6 mA. The "Relative CTR vs. Forward Current" curve on page 6 of the specs implies that the CTR is still that at 1.6 mA. So, 1.6 mA looks like a safe minimum LED current.

Will this optocoupler be able to switch that fast at low led currents?

0.5ms? Probably. The specs state switching times at 2 mA LED current, close enough to the 2 mA.

• CTR is in % and not 300. Treat this as a linear hFE of 3 and a saturated ratio much less than 3. LED current can never be < 5% of maximum unless using PWM due to long term aging effects. Your answer is off by almost 50x Oct 6, 2021 at 22:49
• Thank you for the correction. I corrected my answer accordingly. Oct 6, 2021 at 23:44