Funny Noise Problem

[keebler303]

Background:

I am helping some older gentlemen in the neighborhood on their model train setup. They want to develop a control circuit so the train can travel the whole track without them having to do anything. This involves automatically switching the track in several places to keep the train going in the right direction.

I have a circuit on the breadboard which works for the most part. I am using QRD1114 optosensors to sense the train. This signal is fed through a schmitt trigger, then onto a 74ls221 one-shot. The signal then goes to a ULN2003 which is driving relays to switch the ~15 VAC to switch the track turnout back and forth. The sensor sees the train, the schmitt trigger cleans up the signal. The one-shot terminates the pulse to switch the track so the coils in the turnout don't burn up if a train stops on the sensor.

Problem:

The circuit is working fine with the track power turned off. When the track and turnouts are turned on, I get noise in my circuit. If I use the manual button to trigger the track turnout, when I release the button (turn off the coil) my one-shot outputs a pulse. The signal is absent from the input to the one-shot but it is there in the output.

I thought it might be due to noise in the power so I added a 47uF and .1uF cap to the power near the one-shot, which did not solve my problem. I am not sure how the noise is getting into the circuit as the 5V and 12V are isolated from the AC powering the turnout. The only way I see is if the leads to power the optosensor are acting as an antenna and picking up EMI from the spike in the coil voltage when its turned off. I have poked around with a scope but have not seen any significant noise. It is either of a small magnitude or such a quick impulse that I am not seeing it.

Should I add snubbers to the coils on the turnouts? Any way to increase the noise filtering on the circuit?

Sorry for the long post, needs a bit of explanation.

Thanks for any advice
Matt

[bharbour]

What are you powering the one-shot and relay from? Is it isolated from the track and turnout power source?

Is the sensor wiring physically isolated (not bundled with) the wiring to the turnout coils?

There are a couple of places that you could be picking up noise from:
1 ) power supply, like a common ground or a shared supply.
2 ) sense input. logic inputs are notoriously sensative when connected to long wires.

To troubleshoot the problem, I would disconnect the sense line from the one-shot and ground or tie the one-shot trigger input to +V so it is not triggered. If the one-shot still fires when you release the manual button, it is probably a power coupling. If it does not, you might think about putting a small cap from the trigger input to ground and maybe a small series resistor as well.

sense line --------/\/\/\------------| trigger input of one-shot
50 |
== .1uF
|
----
gnd of
logic
ckt.

An optoisolator would be another idea. Not so much because of the isolation which is good, but it makes the signal require several mA of current to transition it.

I foget if the ULN2003 has catch diodes to protect the transistor from the inductive kickback of the relay coil, but if it does not, adding one per relay would be a good thing.

Good Luck,
BobH

[Al_The_Man]

I wonder why AC is used? DC is more efficient and does not burn out if there is no pick up.
If AC devices, I would use RC snubbers across each.
You may be better off to take each supply common to a common ground from each supply, this would also be connected to the incoming service ground also.
This is a method I have always subscribed to and have never had any noise induced problems.
I have always felt there is too much stock placed in trying to isolate everything, particularly where AC is involved.
Al.

[keebler303]

Well I worked on the circuit some more tonight. I only go to work on it once a week so it is hard to test things out quickly. I think I have it working now.

The logic and relay are powered from an adjustable voltage bench supply. The AC is coming from a separate isolated transformer. The sensor wires are not bundled with the coil wires but within a foot or so.

I first tried installing some RC snubbers to the coils. I used 100 ohm and .1uF as it seemed a good baseline from what google had to say about it. Problem still existed. I started poking around with the scope some more and found the problem. I did not see it earlier as it is a super short duration spike. The spike is on the emitter of the optosensor. It is about 4V in magnitude and only several 10's of ns in duration. I am still not sure where it is coming from. I think the 6ft long unshielded wires to the sensor may be picking up emi and causing the impulse on the line. The schmitt trigger sees the spike and then all bets are off.

I solved the issue for now by installing a .1uf cap between the emitter of the optosensor and ground, that is the input to the schmitt trigger. This pretty well damps out that spike and it functions correctly now. I think when I move to a real circuit board I will put in a proper low pass filter to make sure I don't have any more problems with it.

ULN2003 already has flyback diodes inside.

Not sure why they use AC. Maybe to avoid wiring mistakes? Cheaper? Tradition?

Thanks for the help, I think I've got enough to keep it working now.

Matt