OTGW switching from Monitor to Gateway mode "by itself"

[emmeesse68]

Checking out timings. Diagnostic #4 says:

OK1A high-to-low: 10us
OK1A low-to-high: 28us
OK1B high-to-low: 5us
OK1B low-to-high: 6us

Regarding R8, the schematics on otgw.tlcode.com report a valie of 1k5, while the component list and the PCB layout on the same page report 1k2. I checked and I have actually soldered an 1k2 resistor. Should I try lowering it to... how much would it be safe, not to break the LED inside the OK1?

[hvxl]

To avoid unsoldering and resoldering resistors on the PCB multiple times, which may result in damaging the PCB traces, I would start with clamping another resistor (e.g. 5k6) in parallel with R8. That will lower the resistance and you get an indication if that is moving the figures in the right direction. Once you've determined which parallel resistor gives the best results, calculate the resulting resistance and replace R8 with a value close to that.

The LED can easily handle the maximum output current the I/O pin of the PIC can deliver, so even if you short out the resistor, that shouldn't damage the opto-coupler.

[emmeesse68]

clamping a resistor in parallel to R8 (by chance I picked a 5k6...) moved figures the opposite I was expecting. In fact I got something like this:

OK1A high-to-low: 7us was 10us
OK1A low-to-high: 32us was 28us
OK1B high-to-low: 5us
OK1B low-to-high: 6us

I was thinking a smaller resistor meant a greater curent to excite the LED, thus a shorter rise time, but I guess I overlooked something (my electronics knowledge lies beneath a 30-years layer of dust and rust ).
Anyway, I went through timing specs in OT protocol documentations and they set for hte rise/fall time some 50 us max in both direction, with a 20us as a typical value. Thus my 28us rise time lays within the protocol boundaries, even if it's a bit more than typical.

If, in your experience, this timing (or the weak symmetry between fall and rise time) could lead to poor transmission, I may try unsoldering R8 to try with something slightly higher (or I could see if I've got some spare resistive trimmer), otherwise I might just leave R8 as it is.

Meanwhile, there's something statistically interesting in the errror count, but I'll look closer to check if it's a coincidence or something to note. I'll get back in a few days with results.

cheers

M

[hvxl]

It's all just a matter of perspective. A smaller resistor makes the LED light faster, which results in faster activation of the output transistor of the opto-coupler. The output transistor pulls the OT line low. The high-to-low and low-to-high descriptions are referring to the change on the OT line.

Note that the time measured is the delay introduced by the opto-coupler, that's something different than the rise/fall time! Under normal circumstances I would not consider 10us vs 28us worth worrying about. But when testing other equipment, it's good to realise that the gateway may introduce a slightly asymmetrical delay.