How to ignore the spike from relay

I have a circuit that will check whether the current in a wire exceed a threshold. Something like an overcurrent detector.
That circuit use a simple current transformer (CT) to convert the current to AC voltage signal. The signal from the CT will then be smothed and converted to DC singal by a low pass filter (diode, resistor and capacitor). The DC signal will be feed to a comparator to check whether it reach the threshold.

This circuit work normally. However, when the wire is connected to a switch or a relay, when it is switched off, spike is found at the output of the CT. This make the signal exceeds the threshold.

How can I ignore these spikes? I have tried to use a "larger" low pass filter with bigger capacitor and resistor, but it doesn't help. Also, I do not want to use a very big low pass filter because it will decrease the response time a lot.

I found someome suggested "median filter", how can I implement it if it help?

Thank you very much.

Is diode directly across coil?

If so, the current is outside sense shunt loop so there should be no spike by conduction, rather , it could radiated noise or crosstalk.

Are coil currents shielded or twisted pair? Is current sense on long wires with same twinax or twisted pair?

The spike from the inductance of a relay coil can be thousands of volts if the load resistance is high. The high voltage can destroy switch contacts or a driver transistor.
We use a diode parallel with the coil of the relay to conduct when the spike occurs. The diode normally is reverse-biased so it does nothing. When it conducts during the spike its voltage goes only about 0.7V higher than the supply voltage. The diode slows the release time of the relay a little.

A simple question, I'm just checking:

Current transformers require a burden resistor in parallel with the secondary....do you have that resistor in place?

schmitt trigger said:

A simple question, I'm just checking:

Current transformers require a burden resistor in parallel with the secondary....do you have that resistor in place?

Click to expand...

Yes I have an resistor in parallel with it, and the value is matched with the CT so that the output voltage is in desire range.

- - - Updated - - -

SunnySkyguy said:

Is diode directly across coil?

If so, the current is outside sense shunt loop so there should be no spike by conduction, rather , it could radiated noise or crosstalk.

Are coil currents shielded or twisted pair? Is current sense on long wires with same twinax or twisted pair?

Click to expand...

The spike is generated at the relay contact side rather than on the coil side in my case. Let me explain more. The CT is checking the current on the load of my circuit. However the load is not part of my circuit, it is normally a heater with a thermostat. Therefore when the heater is full or overheat, it will switch off automatically. I found that when this happen, my CT will generate some short impulse and it exceed the threshold of my comparator.

First of all you should determine how much of this spike is due to the load of the relay itself.
Does it happens at the same rate either with inductive or resistive load ?

eepty said:

Yes I have an resistor in parallel with it, and the value is matched with the CT so that the output voltage is in desire range.

- - - Updated - - -



The spike is generated at the relay contact side rather than on the coil side in my case. Let me explain more. The CT is checking the current on the load of my circuit. However the load is not part of my circuit, it is normally a heater with a thermostat. Therefore when the heater is full or overheat, it will switch off automatically. I found that when this happen, my CT will generate some short impulse and it exceed the threshold of my comparator.

Click to expand...

Have you tried anything?
e.g. suitable RF cap across contacts? Twisted pair on contact feeds? Shielded wire on sensor cables? LPF on Comparator? CM Choke on load wires.

This thread is a little bit old but I want to share my solution in the end.

I have tried to add snubber filter across the contact or the load. However I found that the size of the component has to be very large to filter the spike. At last I added a timer circuit for the comparator so that a short spike will not trigger the output.

eepty said:

This thread is a little bit old but I want to share my solution in the end.

I have tried to add snubber filter across the contact or the load. However I found that the size of the component has to be very large to filter the spike. At last I added a timer circuit for the comparator so that a short spike will not trigger the output.

Click to expand...

Yup. Sometimes if you can't suppress a spike then just ignore it. :wink:

Many thanks for the update.

Most of the times, we provide advice to a specific question, but seldom do we know whether this advice was effective or not; or whether a combination of cures were required to solve the problem.

In my opinion, that is what a forum like this should be about. Everybody would gain a little more experience, even the veterans and gurus here.