Dogy 6N137 Optocouplers?

[Zer00]

Hello everyone,

I'm working on a project where I need to address several high power LEDs. I decided to use the venerable WS2811 for this. But in order to protect the logic side in the event of some installation error or power surge, I've decided to put a 6N137 optocoupler on every input, the maximum data rate is 800Khz, the optocoupler is rated as a high-speed logic device (1Mbps).

Datasheet: https://www.tme.eu/Document/23c5dd4e02a7db41ce8ad9637e6eba2e/6N137S-L.PDF

Standard stuff, but I've run into a bit of a pickle. First of, here's my schematic for how I'm using the optocoupler:



Time to build my PCB and test it out!, Oh hey, it's not working? What's going on, let's use the scope (CH1: Input | CH2: Output):



That's... unusual, I've used this circuit before, it's not reacting as expected. Let's rule out Murphy. I went to my old parts drawer and found a couple older 6N137M, I have no idea where I got them, but they have a fancy F on them so i'm almost sure they're from ONSEMI (Fairchild): Datasheet: https://www.tme.eu/Document/724e01c6de3535b2bc18f091ff16d8f9/6N137.pdf

Let's take a look at the waveform with this optocoupler: (CH1: Input | CH2: Output):



It works just fine! What is going on?

Looking at the spec sheets, they're identical components! (Or maybe I missed something). Are the optocouplers made out of low grade chinezium? Or is there something fundamentally wrong with my schematic?

Any help would be appreciated, thanks!

[John B]

Try pulling the enable pin 7 high. Also try lowering the value of R19 to 470R

[jmelson]

Ugh!  These are open drain outputs, you may need a much stiffer load resistor to pull up the outputs faster.  Your traces show that they are really not capable of 10 mbit rates.
Jon

[thm_w]

Yeah 10M not 1M as stated in the OP.
After you try the suggestions above (enable pin and pullup resistor), let us know where you purchased the newer parts from.

[John B]

Actually something doesn't look right. Here's a 1MHz square wave, with roughly 6.7mA input current on the LED, and a 910R load resistor on the output:

This is using an Everlight 6N137. I've found them to be fine even running 2-3mA on the LED. The propagation time increases to around ~100ns vs ~50ns but if that's not a concern then the waveform will remain essentially unchanged on the output.

Dropping the load resistor to 560R slightly sharpens the transition, but no-where near your waveform. In fact I have to run the opto at 10MHz to resemble your trace.

[Zer00]

Thank you for your answers John B.

I have tried both approaches, pulling pin 7 high didn't make much of a difference, but there's a slight improvement in the waveform when switching up R19 (I didn't have a 470Ohm resistor so I soldered another 1K on top of the 1K resistor already on the PCB to give me 500Ohm).

The optocoupler was bought from here: https://www.tme.eu/en/details/6n137-l/optocouplers-digital-output/liteon/

I think the old optocouler is this one: https://www.tme.eu/en/details/6n137m/optocouplers-digital-output/onsemi/

Here's the waveform on the new optocouplers:



And here's for the old optocoupler:



Thanks again.

[John B]

Where are you probing exactly? For starters your output signal should be inverted from your input signal, if high-side switching the anode of the LED. It's difficult to make sense of how either opto signal is operating.

[Zer00]

I'm probing at the TP3 point in my schematic, and the D_O output. The optocoupler is in a non-inverting configuration.

[Zer00]

I think I made a mistake on the first schematic, sorry. This looks like what I have on my PCB, hence why it's non-inverting.

[John B]

All I can suggest is probing the pins right at the IC to ensure there's no connection issues on the PCB. It does seem like a faulty chip. On some cycles it looks like the open collector has varying impedance. What does the output settle to if you just feed in a DC voltage on the input?

[Zer00]

Admittedly, my probing setup isn't the best. I'm on an older TDS524A with the wrong probes. I grabbed some 10X probes and re-did the measurements. (Ignore the vertical scaling, the Oscilloscope doesn't recognize the probes as 10X).

I tried running a DC voltage through the IC and it works as expected, I do get a steady DC output from it. At this point my only idea is to buy lots of 6N137s from other manufacturers to test them out.

(TEK00003.png is the new IC. TEK00002.png is the old IC)

[moffy]

So the dodgy ICs are ON all the time except for a brief OFF period during a low to high transition. That looks like capacitive coupling turning it off but maybe the OFF voltage is not enough to turn it really OFF? Voltage across the driving resistor(current through LED) is a much better indicator as to the state of the driving LED.