How to sense a switch with both opto-isolator and logic?

[arvidb]

I have a problem. In the circuit below I want to read the state of the switch with a logic IC - from the GND and SIGNAL lines.

Now there are a few problems. I cannot really control the things on the left of the dotted line. V1 is not sure to be 24V (but 24V is max). I can't even be sure the opto coupler to the left will be in the circuit. The things on the right of the dotted line will sit on a PCB I'm cadding.


1) I could use a diode followed by a pull-up resistor (as drawn). The diode is needed so as not to activate the opto coupler to the far left (it would see 3.3V as ground if the 3.3V were allowed to sink current). But then the low level for the SIGNAL will be 0.7V, only 0.1V below the max voltage for a low logic signal. And the cables to the switch and back are several meters... not very much headroom for noise and component tolerances.

I'm not sure if the schmitt trigger would solve this; after reading some schmitt trigger data sheets it looks like they also guarantee low level up to 0.8V only.

2) I could use an opto coupler on my board. But I have 12 inputs, and this would get prohibitively expensive, not to mention clumsy (most opto-couplers seem to be hole mounted and single per package). Also since I cannot be sure that V1 is 24V this might not work very well.

Any ideas? This is driving me nuts...

Arvid

[arvidb]

Hmm, now I got an idea. If I moved the diode to the Vcc "leg", it still cannot sink current, and the high level would be 3.3-0.7 = 2.6V. That is 0.6V above the lowest permissible high voltage for LVTTL logic. And there's no long cables since the pull-up is right next to the device. Should work?

Arvid

[ESjaavik]

Exchange D1 with a resistor and put a couple of diodes in series from behind it to GND to limit the voltage. The R should have high enough restistance to make sure it is not able to draw enough current through the diode to trigger it, but low enough to make sure it triggers the schmitt even if volts are a bit low.

BTW, you can get surfmounted optocouplers. One example: Elfa 75-300-74

[Al_The_Man]

Could you use the old favorite of mine, the 2n7000 it has very high impedance and can tolerate a wide range of gate to source and gate to drain voltages, and it is only 10c per.
Although the gate will switch at 3.3v it will allow a maximum of +- 40 volts gate-source.
It is really intended for logic to high voltage output, but I see no reason why it would not work in this case??.
Al

[pminmo]

arvid,
Putting the diode in the vcc line will possibly damage the gate. If it were me I'd simply use a pnp small signal transistor. What is you logic voltage, 3.3V?

Phil

[Al_The_Man]

Originally Posted by pminmo arvid, . If it were me I'd simply use a pnp small signal transistor. Phil

I think the base current would introduce too much leakage through the opto, I was looking through the app notes on the 2n700 and it shows a very similar senario as to what you want to achieve. It also shows connecting the drain to 5v and a 1k to 10k resistor in the source to gnd and feeding the gate with high voltage 12~24 in, the top end of the resistor would then connect to your logic and voila, no load imposed.
I could run a simulated test tommorrow if I get the chance.
Al

[arvidb]

Einar: I would still need a pull-up in case the opto coupler was not included in the circuit. I'm not sure how/if that fits into your suggestion?

Phil: yes, my logic voltage is 3.3V. The IC is 5V compatible though, so it can take up to 5V on its logic inputs.

Al: I'm reading up on the 2n7000. So I would use the gate as an input (from the external circuit)? What happens when the gate floats (when no opto-coupler is connected in the circuit)?

Another solution would be to use a schottky diode in my original schematic (only 0.4 V drop).

Thanks for all your help so far!

Arvid

[ESjaavik]

Arvid: Ohh. I didn't know the opto was not there all the time. Guess I didn't watch the schematic long enough to notice.
A pullup directly on the gate would work. Just keep it high resistance.

[arvidb]

*LOL*

To be really sure the opto coupler doesn't activate, what would be a reasonable max current? 0.1 mA? In that case I'd need about 200kOhm resistance on the input. I guess that would put my pull-up resistor in the MOhm range. Would this work?

edit: forget that, the total input resistance to Vcc should be 200k. So perhaps 20k input resistor and about 200k pullup would work?

edit2: but that would give me ~0.3V low level voltage. About the same as the schottky diode, without the added benefit of overvoltage protection the schottky/pullup combo gives. Hmm...

Arvid