TTL vs. CMOS inputs?

Professor Chaos

Hi all -

I am using a BS2 to automate some functions of an AC-powered model railroad. I am using a circuit (see attachment) with an AC optocoupler to detect the presence of the train; the train wheels complete an AC circuit which turns on the optocoupler. I connect the transistor end of the optocoupler to a BS2 pin. Because the AC signal is very noisy (intermittent contact between track and wheels), I have a 47uf capacitor (in series with a 2k resistor) across the optocoupler output to smooth/debounce the signal.

This circuit works very well when the outputs are connected directly to the Stamp. I tried connecting the outputs to a SN74HC165 shift register instead, to see if I could conserve input pins. Operation is a little erratic with the shift register - the signal goes low from time to time when it would stay high if directly connected to the Stamp pin.

I am wondering if this is a CMOS v. TTL voltage issue? The optocoupler is described as "transistor output" which I am assuming equates to TTL, and from what I gather the BS2 is a TTL input. The SN74HC165 is a CMOS device though. Is it possible that the problem is hooking a TTL output to a CMOS input? Could I solve it by using a TTL version of the shift register?

Thanks for your thoughts.

Mike Green

I think part of the problem is the circuit. I would reverse the whole output circuit with the 10K resistor in the collector lead and the Stamp pin connected to the collector. The emitter should be grounded. I would put the filtering on the LED side. I'd get a simple bridge rectifier (like from Radio Shack) with the capacitor at the output, a current limiting resistor between that and the optocoupler. If you need a current limiting resistor from the track to the bridge rectifier, use a much smaller one than you have. You want the capacitor to charge very quickly from the track (without excessive arcing) and discharge slower through the optocoupler (maybe at 10ma).

Professor Chaos

Mike, can you explain why those changes are desirable? I should note that the circuit seems to work fine when connected directly to an input pin; it just is a little flaky when connected to a shift register.

One issue with filtering the LED side is that I have several of these circuits together, using 4-channel optocouplers. It would be kind of bulky to have 8 or 16 bridge rectifiers stuck in.