Purpose of electrolytic capacitor across an opto coupler?

[permal]

Hi,

Could someone please explain to me the purpose of the electrolytic capacitor C26 (lower left corner) in this schematic?

I'm guessing it is meant as some sort of protection, but aren't regular capacitors used for that? Also, when U13 closes to ground, C26 will be shorted; that can't be good?

I might be totally wrong, but as I see it all that capacitor accomplishes is to increase the rise and fall times of the input!? 

[Andy Watson]

I might be totally wrong, but as I see it all that capacitor accomplishes is to increase the rise and fall times of the input!? 

What are the optos monitoring?
I think the capacitors are there to form a low-pass filter, for example, to hold the output stable if the input to the optos is A.C. (mains ?), or perhaps to eliminate button-bounce on switch contact inputs.

[Benta]

The time constant is relatively long (~100 ms). It's for filtering 50/60 Hz AC inputs to provide a stable signal for the logic inputs.

[permal]

What are the optos monitoring?

Could be anything afaik. Is is the schematic of a general purpose I/O card.

The time constant is relatively long (~100 ms). It's for filtering 50/60 Hz AC inputs to provide a stable signal for the logic inputs.

Yes, it makes sense that if AC is applied, the optos would turn off at each zero-crossing. Having an electrolytic capacitor to keep it steady makes sense.

So what you're saying, Benta, is that due to the long time constant, the capacitor isn't discharged fast enough to damage it?

[Benta]

It's not a nice design, but probably within the peak collector current specs of the optocoupler.

[permal]

It's not a nice design, but probably within the peak collector current specs of the optocoupler.

Yeah, it feels rather brutal, even for a newbie like me. If you have a suggestion on a better solution, I'd love to hear it!

[Zero999]

You could omit the capacitors, ensure the software knows that the inputs will be 2×mains frequency waveform and accept that the LEDs will flicker very slightly.

[Benta]

Looking at Fig.6 in the LTV354 datasheet, the design is borderline and relies on "typical" V_CE vs. I_C behaviour.
A simple fix would be a 100 ohm resistor in series with each 10 uF cap. That will not interfere with operation, but limit collector current through the coupler.

[permal]

Looking at Fig.6 in the LTV354 datasheet, the design is borderline and relies on "typical" V_CE vs. I_C behaviour.
A simple fix would be a 100 ohm resistor in series with each 10 uF cap. That will not interfere with operation, but limit collector current through the coupler.

Awesome, thank you for your insights. 

[BrianHG]

You could omit the capacitors, ensure the software knows that the inputs will be 2×mains frequency waveform and accept that the LEDs will flicker very slightly.

You are assuming that the AC pulse is turning on the opto long enough to always detect that pulse in software.
Agreed that the 10uf caps are a cheap functional solution around this problem.

[f5r5e5d]

transistor output optoisolators are inherently limited in collector current by the CTR - you can't turn it on hard enough to fail the transistor Ic_max spec without blowing the LED

which is different from saying you can't exceed the Pdiss_max with high enough Vcoll

[permal]

transistor output optoisolators are inherently limited in collector current by the CTR - you can't turn it on hard enough to fail the transistor Ic_max spec without blowing the LED

which is different from saying you can't exceed the Pdiss_max with high enough Vcoll

My question is about inputs, but your statement is still of interest. Are you saying that the LED on the inside of an opto coupler will blow before the transistor on the output side is fully open? what does CTR stand for?

[Cervisia]

Are you saying that the LED on the inside of an opto coupler will blow before the transistor on the output side is fully open?

What do you mean with "fully open"? In every transistor, the output current is limited by the input current.

what does CTR stand for?

Current Transfer Ratio. See any optocoupler datasheet; it's essentially the same as a transistor's h_FE, but measured in percent because the number would be ridculously small otherwise. For many common optocouplers, the guaranteed minimum CTR is less than 100 %.

You can blow a 'normal' transistor by sending too much current either through the base, or through the collector. With an optocoupler, it's unlikely that the second case happens because you will not be able to get enough photocurrent into the base.

[permal]

What do you mean with "fully open"? In every transistor, the output current is limited by the input current.

I don't know what I was thinking

Thanks for the explanation re CTR.