Determining forward voltage/current of several in series LEDs ?

[LinuxHata]

Hello.
I don't know whenever this is the proper place to ask, so forgive me, if not.

The single led has characteristic current/voltage dependence curve, based on which, it is quite easy to determine the working voltage, even without knowing the working current - just begin slowly increasing the voltage, while monitoring the current, and when there's a sharp increase in current, here we are.  LCR-T4 and similar testers can do this for a single led quite reliably.  But I very often encounter strings of unknown number of leds, mostly 3-30 in series, which might have any (also unknown) working current, starting from 5mA to 1A most commonly.
I do have some high voltage lab power supplies, which I'm using for that measurement purposes, but it requires considerable amount of time, so I'd like to have some automated tool for that.

Are there any readily available or simple to make solutions for that?

Thanks in advance!

[DavidAlfa]

Same as a single one, just do VF * Number of leds.

If a single led does 1.7Vf@5mA, 1.9Vf@100mA, 5 leds in series will be 5x that.

Of course assuming they're all the same.

Also  that's why power leds are usually driven with constant current, not constant voltage, VF gets lower with temperature, risking thermal avalanche.

[Echo88]

To automatically derive the forward voltage of an unknown number of leds in a string you´d have to know the color or more accurately the semiconductor-mix that is used in the string (green leds can be based nowadays on other semiconductors than 20 years ago afair).
If you have a power supply that is remotely controllable then you could write a small Python script where you tell it the number of leds in the string or the color and it should be able to determine the forward voltage/current.
Or just get a 100V 10mA current source (safety precautions observed...), observe forward voltage of complete string, then divide by visible number of illuminated leds.

[LinuxHata]

The idea is to completely have this process automated. Leds might be epoxy potted and not visible independently, so can't count them to determine the voltage.

[mariush]

Get a step up led driver that can boost to a high voltage - higher than the sum of all forward voltages - and set it to limit the current to some sane value like 5-10mA. 
Start the test process, the led driver will keep boosting the voltage up until the leds consume that 5-10mA of current ... now you can measure the voltage across the leds with a multimeter or some microcontroller (after bringing it down with a voltage divider or something to sane levels)

For example TPS61196 can boost up to 120v and have a bunch of protections built in that can be useful : https://www.digikey.com/en/products/detail/texas-instruments/TPS61196PWPR/3880034
It's 6 channel driver, but you can use only one channel to test a led strip.

[LinuxHata]

Well, as I said above, I know how to technically do it by "hand". I want automation - I connect the LEDs, hit button and device display shows Forward voltage and current.

[Psi]

If you know the led type and its typical forward voltage then you could have a MCU apply a 20mA current source and then read the voltage and divide that by the known led forward voltage to obtain an estimation for the number of leds in the string.

If you have many led types all in series then it gets tricky.
If they are different colors/types you can't tell if it's many reds leds or fewer blue leds for example.
If they are all the same color but that color could be different then you could have a switch to tell the MCU what color leds you are testing. Then it can apply different logic to the calculation.  But its only an estimate since the same color led can exist with different led chemistry and forward voltages

[shapirus]

The idea is to completely have this process automated. Leds might be epoxy potted and not visible independently, so can't count them to determine the voltage.

Not knowing the number of LEDs makes it one equation with two unknowns, which means an infinite number of solutions. And that's if we assume that all the LEDs are identical, otherwise the number of unknowns will increase.

[ivo]

Get a programmable power supply up to the highest rated voltage you need, you might also need something that is able to programmable-y log voltage / current (current is easily logged using voltage over a resistor), and you can hack this capability up in a python script.

But always the proper way to drive any LEDs, is a constant current source that has the required compliance voltage. That only requires figuring out a nice current for the LEDs to run at.

If someone is yelling at you that they have all these crazy strings of LEDs that they have no idea about, I would just be charging commensurately for use of my time to figure it out for them.

[LinuxHata]

It does not generate "infinite" number of solutions, because LED has very specific voltage/current related curve, and if you "catch" the moment when say, increase of voltage by 1% causes current increase by, say 10%, then - here you are.

[shapirus]

It does not generate "infinite" number of solutions, because LED has very specific voltage/current related curve, and if you "catch" the moment when say, increase of voltage by 1% causes current increase by, say 10%, then - here you are.

It does.

For n LEDs connected in series, assuming each of them has an equal drop of V_f (which is the parameter we want to determine) at a certain current which is a fixed condition of the test, we have the following equation:

V_f * n = V_tot

Where V_tot is voltage drop across all the LEDs combined.

you can measure V_tot, and if n is known, then V_f can be easily calculated. But if n isn't known, then there is an infinite number of combinations of V_f and n that satisfy the equation.

[dietert1]

I think a lab power supply that supports constant current mode will do the job. It needs an automation interface like GPIB, RS232 or USB. For example the OWON P4603 will do current and voltage measurements. So you wire it to the load  and a PC and write a little application that steps constant current like 1, 2... mA and records voltage. It can stop when the "10% more current at 1 % more voltage" condition is fulfilled, if that is your criterion.

Regards, Dieter