Relay termination

[warpie]

Just wondering what is a good practise for the unused pins on a relay.

Say for example that you only use one side/ switch (SPDT) in a DPDT relay, do you terminate the second unconnected SPDT somehow?

 

[Bo Deadly]

Par

Just wondering what is a good practise for the unused pins on a relay.

Say for example that you only use one side/ switch (SPDT) in a DPDT relay, do you terminate the second unconnected SPDT somehow?

Parallel them. Relay contact resistance will eventually increase over time more or less depending on various conditions. So if one half would otherwise just go unused, parallel the contacts for better reliability.

 

[warpie]

Par

Parallel them. Relay contact resistance will eventually increase over time more or less depending on various conditions. So if one half would otherwise just go unused, parallel the contacts for better reliability.

That sounds like a good idea. TBH, I've never thought of it but after reading online a bit it seems to be not very uncommon. Are there any disadvantages by doing this?

Many thanks for the suggestion!

 

[warpie]

After a bit of testing I noticed the following...

When I switch the relay (only one set of contacts), I get a bit of an audible "click" even when there's no audio signal present. I believe what I hear is the leak from the coil to the contacts when it's energised. It's pretty quiet but it's there.

However, when I parallel the contacts, the "click" noise increases by approx. 5dB, which I think it makes sense since now I'm getting the coil leakage to both COM contacts.

Hm, so it's a some kind of a compromise like always I guess...

 

[ruffrecords]

Which relay are you using?

Cheers

Ian

 

[Bo Deadly]

After a bit of testing I noticed the following...

When I switch the relay (only one set of contacts), I get a bit of an audible "click" even when there's no audio signal present. I believe what I hear is the leak from the coil to the contacts when it's energised. It's pretty quiet but it's there.

However, when I parallel the contacts, the "click" noise increases by approx. 5dB, which I think it makes sense since now I'm getting the coil leakage to both COM contacts.

Hm, so it's a some kind of a compromise like always I guess...

It's almost certainly not "leak from the coil". If you're talking about a physical "click" from the relay itself and not in the signal path, that is normal. If you're talking about a popping noise in the signal path, that is from the circuit itself. If there is any DC buildup across the contacts that discharges when they close, you will get a pop. This can happen if you do not have drain resistors on coupling caps for example. If you post a schematic, we can probably see where the problem might be. In some scenarios it might not be completely avoidable. If it's a pad on a mic input for example, even a tiny disturbance in the signal will be amplified by the mic pre and probably audible depending on how you're listening. In practice, at normal gain settings, this particular example scenario should not be a problem. Again, post your schem and explain the circumstances ...

 

[warpie]

It's almost certainly not "leak from the coil". If you're talking about a physical "click" from the relay itself and not in the signal path, that is normal. If you're talking about a popping noise in the signal path, that is from the circuit itself. If there is any DC buildup across the contacts that discharges when they close, you will get a pop. This can happen if you do not have drain resistors on coupling caps for example. If you post a schematic, we can probably see where the problem might be. In some scenarios it might not be completely avoidable. If it's a pad on a mic input for example, even a tiny disturbance in the signal will be amplified by the mic pre and probably audible depending on how you're listening. In practice, at normal gain settings, this particular example scenario should not be a problem. Again, post your schem and explain the circumstances ...

I'm not talking about the physical "click", that's fine. It's not DC offset either. Even without any signal connected to the pins (only the probe of the oscilloscope on the COM pin) I can see/ measure the spike. The schematic is rather simple and we had discussed it in the past. The only addition to the schematic is a 1n4148->33kOhm with 100nF to ground after the switch and before R3.

Switch---->1n4148---->33kOhm----|----> to R3
_________________________________|
_______________________________100nF
_______________________________ |
_______________________________ Gnd



It's currently built on a breadboard. I'm pretty sure it's leakage from the coil. In fact, I'm tempted to try some other relays but I have a bunch of these.

 

[warpie]

Following my previous post here are some measurements.

As I said, no audio signal is connected to the relay. Just the probe at the COM.

Single connection (On and Off)




Parallel connection (On and Off)

 

[ccaudle]

I'm pretty sure it's leakage from the coil.

If that is the case is should be proportional to risetime. Slow down the signal driving the coil a little bit and it should reduce any coupling.

 

[Newmarket]

I'm not talking about the physical "click", that's fine. It's not DC offset either. Even without any signal connected to the pins (only the probe of the oscilloscope on the COM pin) I can see/ measure the spike. The schematic is rather simple and we had discussed it in the past. The only addition to the schematic is a 1n4148->33kOhm with 100nF to ground after the switch and before R3.

Switch---->1n4148---->33kOhm----|----> to R3
_________________________________|
_______________________________100nF
_______________________________ |
_______________________________ Gnd



It's currently built on a breadboard. I'm pretty sure it's leakage from the coil. In fact, I'm tempted to try some other relays but I have a bunch of these.

You should run that test with a defined voltage on the input pins. Otherwise thosr pins are paying in the electronic wind and you don't want that for repeatable results. So tie directly to OV or have a pull down resistor to 0V or bias them to some fixed voltage using a voltage divider.
wrt "cool leakage" - are you thinking about a leakage current there ? Or capacitive coupling or some other mechanism ?

 

[ccaudle]

Or capacitive coupling or some other mechanism ?

From the scope pictures it is either capacitive or inductive coupling. Although that will be sensitive to physical layout, it could be coupling from the supply wires to the signal wires. What is the physical arrangement of relay power and signal wiring? The current from the relay coil is getting injected into the node shown as gnd in your schematic. What other currents share that path?

 

[Matt Syson]

Depending on the construction of the relay it is possible that the magnetising current can induce s 'spike' into the contacts. This would be rare but not impossible and would heavily depend on the impedances of the circuits cfonnecting to the various contacts. If it were switching a mic signnal perhaps, with plenty of gain following it, you need to be careful of the physical type of relay used. Direct capacitive coupling has already been mentioned. A scope probe when unteerminated will pick up stray impulses so you have to test it as a 'completed' circuit, not just in isolation.

 

[Audio1Man]

I would disconnect the scope lead and connect it to the PS GROUND. This will remove an inductive pulse from the scope ground.

 

[Newmarket]

I would disconnect the scope lead and connect it to the PS GROUND. This will remove an inductive pulse from the scope ground.

But the scope probe was used to show us waveforms. The probe defines an impedance to 0V and so is likely to give a realistic result. Waveforms at x1 and x10 settings might be informative if available.

 

[Bo Deadly]

I'm not talking about the physical "click", that's fine. It's not DC offset either. Even without any signal connected to the pins (only the probe of the oscilloscope on the COM pin) I can see/ measure the spike. The schematic is rather simple and we had discussed it in the past. The only addition to the schematic is a 1n4148->33kOhm with 100nF to ground after the switch and before R3.

It's currently built on a breadboard. I'm pretty sure it's leakage from the coil. In fact, I'm tempted to try some other relays but I have a bunch of these.

It's definitely not leagage from the coil. Not with that circuit. That circuit is going to limit the current pulse through the relay (which is completely unnecessary BTW).

Note that if you post plots, it is important to tell us what the grid lines correspond to. It's not at all obvious what the size of those peaks are.

Also, when I referred to the schematic, I meant the signal part of the circuit and not the relay.

However it doesn't matter because I know what the problem is ... it's the breadboard ground. When you breadboard something you have to be VERY careful about noise (like high current pulses) through signal 0V. Adjust your breadboard grounds so that the relay 0V runs back to the 0V of the first large filter cap out of the power supply over a separate conductor. Meaning you want two 0V, one for signal and one for relays. They should converge on the 0V of the first filter cap at the supply.

And no loops. The should be completely separate 0V and only be connected together at the 0V of the filter cap.

 

[warpie]

Thank you all for the suggestions.

I'll try to play around a bit more during the week or next weekend.

Bo Deadly, the vertical scale is 200mv/div and it shows on the right hand side. Is this what you mean?


Also, you lost me here.

That circuit is going to limit the current pulse through the relay (which is completely unnecessary BTW).

What circuit is completely unnecessary?

 

[Bo Deadly]

What circuit is completely unnecessary?

The RC transistor driver isn't really necessary. You can just dump 24V across the relay coil. If you want to drive remotely over a wire then I would use a transistor driver but I would put the series resistance BEFORE the switch at the power source to minimize the chance of something getting energized with 24V. And I would use supply-follows-return rules. Although personally I would almost certainly not use wires at all. Certainly not for a new design. The circuit should be integrated into the panel. Or better still use a micro and separate your peas and carrots entirely.

 

[Mr. Moscode]

Any common grounds between the coil and the signal path? Use a different relay and see what happens.

 

[Mr. Moscode]

Can you take a pix of your layout and post it, please?

 

[warpie]

So, I did a quick test last evening...

I disconnected everything (transistor, voltage dividers, etc...) and I just energised the coil straight from the 24V supply. I probed the COM with the two input pins floating (A) and with input pins connected to ground (B).



With the pins connected to ground there are no spikes so no point in uploading the graph.

With the input pins floating (A) the results are not very charming as you can see below.

When the relay is "ON":





Does this mean that my measuring method was wrong since the contacts will be connected to some audio signals anyway?


FWIW, if probe is set x10, the results are pretty similar.

"ON"


"OFF"