Identical circuits working differently. Board design issue?

[Szewczykm]

I have a circuit that involves a signal, resistors, optocoupler, decoupling caps, and 74hc4050 hex buffer.

I have 5 input signals.  I'm using a Lite-on LTV-844s for 4 of the signals and an LTV814s for the 5th signal.

The only difference between the optocouplers is that one is a 4 channel, the other a single channel. All specs are the same otherwise.

I've attached a sketch of the schematic.

Like I said, the circuit is the same for all 5 channels.  The only difference circuit-wise is that 4 channels are on one optocoupler and 1 channel is on the other smaller optocoupler.

If I connect my signal to channels 1-4, I get a nice signal from the output of the 74HC4050.

If I connect the exact same signal to channel 5, I get no output.

I've checked all of the physical connections.  They are all good.  I get all the values I expect between points.

If I change the resistor going to the input of the optocoupler on channel 5 from 4.7k to 2.2k, I then get output on Channel 5 through the 74hc4050.

I'm perplexed as to why, when all other things are equal, I need to drop from 4.7k to 2.2k on the input to get an output.

I have a theory.  I'll link a picture of the board.

You can see, the physical layout on the board is different.  There is ground plain on top and bottom of the board.  Am I getting capacitance from how I laid the board out?  Why would the lower value resistor overcome it?

I have so many questions about this.  How should I test?  I have a Rigol scope.  Will putting a scope on there negate the built up capacitance at all if it's small?  Did I violate a basic board design rule with having a ground plane on top and on bottom?  Did I violate a rule by having that long trace running in parallel to the top ground plane?  Am I being dumb thinking that the LTV844 is the same as the LTV814 except for channel count?  Am I being dumb thinking that the capacitance would amount to anything that would have an affect on the circuit?

Thanks for any input.

Mike

[AlfBaz]

Here's my guess

According to the data sheet the current transfer ratio can be between 20 and 300%. Or for 1mA on the input you get anywhere between 0.2 and 3mA collector current in your output for a Vce of 5V.

The 4 channel opto probably all have very similar CTR's as they are all produced in the same process on the same die, where as the single one will very likely have a completely different CTR

[nessatse]

What is your input signal level.  I haven't looked at the datasheet but I expect you would need at least around 10ma to light up the led's.  With a 4k7 resistor, that would mean something above 47V.  You need to lower your input resistors to get decent current in.

[Szewczykm]

I'm reading up on that now.  It's very interesting.

What I'm using as an input signal varies.  I'm pulling signals off of a pinball machine.

Flashers = 20v DC
Insert Light Matrix = 6.3v AC
General Illumination = 6.3v AC
Switch Matrix = 5v DC

The matrix is pretty noisy.  I found that if I raised the resistance, it helped squelch out some of the noise.  The input is ending up on a microcontroller so what I think I'm going to do is bring the "squelch" down to something like a 1k resistor on the input and then use software to ignore the rest of the noise.

I thought that I'd need in excess of 10ma too but I've found that 1 or 2 ma is enough to get a signal.  But like you said, I think I need to lower them to improve the current at 5v but try to keep the resistor high enough to not get blown out at 20v

I was just leaving too small a margin of error using such a high valued resistor.  It seems to make sense.  I hope this works out in reality.

Thanks!

Mike