0
\$\begingroup\$

I've been given a circuit to test and pointed out that the potential reverse voltage across the led in the opto will exceed the spec. The cct is a dual input with an opto on each channel, which feed into a bus buffer device for then reading form the motherboard that this plugs into.

enter image description here So there are TVS diodes on the 2 inputs (La+ and Lb+) that protect against transients and clamp at 15V. But the problem is that at 15V we are exceeding the 6V reverse rating of the opto LED, which is fed from 3V3 through a resistor. To be honest the design is supposed to work up to 24V so it is worse. I could use a voltage shunt across the LED to clamp the reverse voltage seen by the LED, but this will mean that in the inputs inactive state (a voltage of either 5, 12 or 24V on La/b+) we will draw current form the input/consume power. As we have up to 32 of these channels in the design that's not really ideal. I also cannot affect the level of input voltage signals as this device is only supposed to 'listen' in to the signals which are part of another system. I thought about using a schottky on the input signal to block the voltage as we're only interested/active when the input is pulled to GND. But without the LED conducting there is no circuit the other side of the schottky so it doesn't do anything.
Any ideas gratefully received - thx

\$\endgroup\$
6
  • 1
    \$\begingroup\$ Could you post a higher res version of the schematic? Also, I don't think you have given a big enough overview of what you are protecting from. \$\endgroup\$ Apr 6 at 15:48
  • \$\begingroup\$ Please share a better version of the schematic \$\endgroup\$
    – kruemi
    Apr 7 at 8:42
  • \$\begingroup\$ Thanks - that image was pretty poor - sorry about that. I've changed it. So the idea is that the circuit works with both wet and dry voltages - so wet voltages of 5, 12 and 24V where the active level is 0V, and also works with a dry contact pulling to GND. The idea of using the opto was to provide the switching levels for the buffer/downstream cct but also to provide an isolation barrier to protect it. The TVS diodes were to provide some protection against transient voltage spikes. \$\endgroup\$
    – mikey_jp
    Apr 7 at 9:17
  • \$\begingroup\$ The cct works fine for the dry contact and the 5V input, but the 12V and 24V inputs will push the LED in the opto beyond it's capabilities. The idea of a voltage shunt across the opto LED works fine and fixes the problem but results in current draw from the input that I'm not sure is allowed at the moment. Other ideas I've had were to use a linear regulator on the input but this won't work with the dry contact. Other idea was to use a MOSFET to switch on the input (could replace the opto as well) but I'm not sure you can get FET's that would accept such a large gate voltage (24V) - Vgs. \$\endgroup\$
    – mikey_jp
    Apr 7 at 9:20
  • \$\begingroup\$ If +3V3D is connected to both sides of the optocoupler, then it's not isolating anything. Why are you using an optocoupler then? \$\endgroup\$
    – bobflux
    Apr 7 at 12:02

1 Answer 1

0
\$\begingroup\$

You could put in signal or schottky diodes in series with the opto LED and in the same direction. Then, put a large resistor across the opto LED. Select the resistor so the maximum reverse leakage results in less reverse voltage than the spec (6v).

A silicon diode would typically have lower reverse leakage (enabling the use of a fairly large resistor), but about 0.7V drop in operation. A Schottky would have a lower forward drop, but higher leakage requiring a larger resistor.

If performance outweighs (a little bit of) cost, a diode parallel to the opto LED but in the opposite direction could be used in place of a resistor, still using the series diode to block reverse current.

\$\endgroup\$
3
  • \$\begingroup\$ Thanks Cristobol - this would def work but I'm not sure I can draw any current from the input at the moment. Trying to establish that with the customer. This system is supposed to just listen in to another system, and not affect it in any way. \$\endgroup\$
    – mikey_jp
    Apr 7 at 9:21
  • \$\begingroup\$ @mikey_jp If you can't draw any current from the inputs, you're going to have a hard time activating these optocouplers in any case. You'll have to at least be able to sink current, and the amount you'd source would be well under 1mA with this proposed setup. \$\endgroup\$ Apr 7 at 14:00
  • \$\begingroup\$ Yes Cristobol - we are able to sink current into the input I believe. The resistor isn't sized well at the moment - a forward current of 1mA through the LED is not high enough. But the whole cct design is poor. I think I'm going to recommend starting again - understanding the exact system requirements is my first step. \$\endgroup\$
    – mikey_jp
    Apr 8 at 10:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.