The attached file shows 7805 and 7812 regulators being powered by a single Radio Shack AC-DC wall wart. The second attachment shows the application circuit being powered. The caps for the 7805 and 7812 were originally the 0.33uF and 0.1uF values normally recommended, the values shown in the attachment are what is currently in place; they appear to be no better or worse than the initial values used.

The +5VDC is powering a Sentron CSV-1A magnetic sensor, the +12VDC is powering a 741 Op Amp, an LM339 comparator, an LED, a 4N25 opto-isolator, and a voltage divider which provides the reference voltage Vref for the comparator. Both the Op Amp and Comparator have 0.1uF ceramic caps between the power and ground connections. As per Sentron, the magnetic detector also has a 0.1uF cap between power and ground. All unused connection on the LM339 are tied to ground.

The application attachment shows a simulation of the Sentron output via a several sine wave generators, the detector side of the 4N25 is actually powered externally, the show components are there for simulation purposes.

The application objective is to detect the comparator output changes when Vin > Vref + Vhysteresis. The comparator is connected to a 2N3906 switch which controls +12V for an LED and a 4N25 opto-isolator. The peak to peak voltage change into the comparator is only ~400mV so the hysteresis is about 260mV peak to peak.

The application circuit works 99.95% of the time, but if a noisy load in the house starts, the application circuit will react. The only connection to the 'real world' is the power (the CSA-1V only detects magnetic field changes, and the external counter card is connected via the 4N25 opto-isolator).

I can recreate the problem at will using a pencil sharpener plugged into the same power strip as the wall wart, but the usual suspects are the furnace oil burner motor and the circulating pump motor which are on a different circuit (but on the same phase).

So, are there any improvements I can make to the power supply (or application) to make it less vulnerable to external noisy loads? TIA.

Mark

 

Power it with batteries and put it all in a Faraday cage.

 

Try increasing the value of C1 significantly.

 

What was stated above.

C1 a 470uF Aluminum electrolytic in parallel with a film 1uF on the input of the 7805

A 47uF Aluminum electrolytic in parallel with a 0.1uF film cap on the output.

Add 0.1uF film caps across the supply pins to all ICs as well.

Keep leads/wires as short as possible.

Make sure the capacitance between the in and out of the 7805 is at least a factor of 10. The values can vary, but do keep the high ESR electroytics in parallel with low ESR film caps on both sides. The capacitance difference is needed to prevent the regulator from oscillations.

 

Thanks for your replies.

I implemented the changes this morning, unfortunately the symptoms continue to persist.

This led to a long day of incremental debugging and the probable root cause is that the external counter (Hobby Boards Dual Counter 1Wire DS2423) is definitely susceptible to external noise. Connecting the Counter Card +5V to the counter input and then creating a noise event is sufficient to generate false counts.

I suspect the issue might be that the +5V on the counter is so called 1Wire 'parasitic power'. Perhaps providing external power (which is allowed) will resolve the problem.

Until I can resolve the problem at the counter level, I will leave the power and application circuits alone.

Thanks again for your time.

Mark