How reliable are safety relays?

[Neale]

Strange question, maybe, but...

I have a Pilz PNOZ safety relay in my control box, bought "new but old stock" from eBay. Unused, in rather tatty box. Been working fine for some time, but recently something has been a bit flaky about the machine. I can turn it on, but the "reset" button that should energise the safety relay has no effect. I spend some time checking connections, power to the relay, switches, etc, but nothing is wrong. The pilot LEDs on the relay do not light and there is no "clicking" from inside it. It just doesn't work! I have even had the safety relay out and tried to dismantle it, but it's pretty difficult to get into it far enough for real fault-finding. I wander away and look at its data sheet, which has a diagram of its internal connections, to try to get more ideas about testing. I go back to the machine and it works perfectly, even though it's been switched off during that time and I have not changed anything. I can't absolutely tie the problem to the safety relay but it looks that way. At one time I thought that the problem might be associated with the CSMIO controller but now that's been replaced with a PV Automation ET6 and the problem still occurs. Very unpredictable about when, and maybe once or twice every couple of months.

This afternoon's fault was slightly different. I turned the box on, hit the "reset" button, and the safety relay energised but the relay switching mains to the driver PSU did not come on. I could trip the mains relay and the machine worked, indicating that there was a problem with the safety relay contacts feeding the mains relay, confirmed by voltage measurements. In fact, I swapped the mains relay feed to an unused set of contacts on the safety relay and all worked as it should. Looks as if one of the safety relay internal contacts had failed (although I'm only ever switching 24V to another relay plus front panel LED so hardly stressing it).

The internal diagram of these safety relays shows a bunch of interconnected relays with all sorts of latching/set/reset contacts and external connections. However, a simple interpretation of its function is that it consists of two relays wired in parallel, with suitable latching contacts, but with the relays' N/O contacts wired in series. So both relays have to latch on, and the latching circuit is arranged so that if one relay fails to operate, the other will also be disabled. In principle, either a relay coil failure or the failure of any relay's contacts will prevent system operation. Even if one relay's contacts weld shut, the other relay will cause the external circuit to disconnect if the estop button is hit. The safety relay actually has 5 or 6 relays internally to allow for more sophisticated operation.

I think that the effects I'm seeing are consistent with some kind of internal safety relay fault, albeit intermittent, and while it fails safe, it's a pain in the b****y neck! Has anyone else seen anything similar? I would expect these relays to be built from fairly reliable and conservatively rated components and contacts.

Although it's working at the moment, having swapped relay connections, I'm now scouting eBay for a replacement. Any sensibly-priced more modern alternatives that anyone can suggest as what's available all seems to be ex-equipment? As one possible solution, I'm thinking that I could pick up a 4pole c/o relay plus base from somewhere like Rapid for a tenner or so, use one contact set for latching in series with estop buttons, and still have the equivalent of 2N/O plus 1N/C that I need for my machine. Sound like a good way to go?

[cropwell]

Like this one ?2920029201

[Neale]

Similar but different! No, mine is the older version, big and clunky in appearance.

I was thinking about it overnight and realise that the only reason I used it was because I needed a relay of some kind - NVR-type function, multiple contacts to switch driver PSU, signal motion controller, feed enable inputs on drivers - and I stumbled across this one cheap. In practice, I could get the same functionality minus the "fail safe" aspect for a lot lower price and had I not found the Pilz, that's what I would have done. So I might as well do that now! There are plenty of plug-in relays available new with DIN rail fitting bases that would only need trivial wiring changes.

In fact, two such relays with latching contacts cross-connected and contacts wired in series or parallel would provide similar capability to the Pilz in this situation but I don't feel I need to go that far...

[JAZZCNC]

I've had 2 Pilz relays do similar things, both bought used, so can't know how used or worn. But now I don't bother anymore and just build my own latching safety circuit using vanilla ice cube relays, they are cheap and easily replaced if they go faulty. Which to be honest I rarely have problems with, even with the cheap Chinese relays.

[Neale]

Think that answers my question! Thanks for that - the fact that sometimes I could go to the machine and find that it just didn't start up and for no reason that I could find was frustrating, to put it mildly - and especially when you could walk away without doing anything and come back to find it all working.

Cheap relay and base to be ordered shortly, then.

:beer:

[JAZZCNC]

Think that answers my question! Thanks for that - the fact that sometimes I could go to the machine and find that it just didn't start up and for no reason that I could find was frustrating, to put it mildly - and especially when you could walk away without doing anything and come back to find it all working.

Cheap relay and base to be ordered shortly, then.

:beer:

Yep, I was getting exactly the same things happening. It would work fine I'd cut a job or four, E-stop the machine for changing tools, etc or dinner, and then it would be dead, NaDa nothing.!! Power down and start back up and would work fine for days or weeks then start doing it randomly for no apparent reason. Took me bloody weeks to track it down to the relay because I thought it wouldn't be the Pilz as they are quality units, or supposedly.

The last one I fitted was second-hand but came from a friend who fitted it new to the machine(not CNC related) which did only 3 months of work and then was stripped out. I had it another 6 months on my machine.! . . . .That was the last straw and I've never fit another since.

[Neale]

I can understand that - I've spent hours trying to track down what is an intermittent problem. And when everything pointed to what I assumed was the most reliable component in the machine, I just didn't believe it. Until I had definitive proof when one of the contacts failed!

Your moral support much appreciated.

[Mud]

If they're used to pass a decent amount of power, then later changed to signals then they can be unreliable - the contacts get dirty at the higher power level and somewhat self-clean, but signal levels won't clean the contacts.

[m_c]

Safety relays are generally pretty reliable.

Failing to switch on, is usually down to a problem with the enable/e-stop circuit. First thing I'd do, if they appear dead, is to bridge the relevant terminals, and see if they kick into life. If they still don't, then it's dead. Although voltages may measure OK on the offending circuit, some relays increase the current when switching on, so they may still be the fault, which is why bridging terminals directly is a more reliable diagnostic.

As Mud mentions, relay contacts have a minimum current rating, as they rely on arcing to clean the contacts (sometimes referred to as contact wetting). If you regularly switch below the minimum current, the contacts eventually oxidise with their resistance gradually increases until they won't switch the load.

[Neale]

I have a background in electronics - I'm not really up with the latest semiconductors, but I'm OK with good old-fashioned relays! I checked both the loop resistance through the estop switches (measured as expected) and also tried the "short the estop terminals" trick. No go. Also checked that the momentary reset switch was supplying 24V measuring across the relay terminals, and that looked OK. No unexplained voltage drop indicating high resistance on the supply side. Came back half-hour later and all was working.

Then the actual contacts on the "output" side failed, definitively proved by swapping to the unused but identical contact set. That was switching a 24V relay coil used to switch power to the driver PSU, so should be enough contact to keep them clean. I had heard about the need to switch more than trivial currents before, one reason why 20mA current loop was popular, i was told, The internal interconnect contacts all switch internal relay coils, so I would expect them to carry enough contact, aided by a bit of back-EMF from the relay they are switching. Anyway, new relay from Rapid should be here tomorrow, so all my problems will be over...

Except that this afternoon, the machine refused to go all the way through its homing routine. Stopped either while homing Z, or when homing X+A. No idea what that is and had to leave it there, but I'm about to go out to the workshop and either fix it, or bang my head on the wall. One or the other...

[m_c]

but I'm about to go out to the workshop and either fix it, or bang my head on the wall. One or the other...

Could always bash your head on the machine. One stone, two birds... :joker:

[Neale]

Could always bash your head on the machine. One stone, two birds... :joker:

Unfortunately, I've done that one already! I bent down to pick up something and brought my head up under one of the brackets that hold the steppers. Ended up dripping blood in A&E from a scalp wound...

But I appreciate the labour-saving thought! And my new relay and base arrived earlier so I'm off to fix that later.

[routercnc]

I went through a few relay options on my early control panel including automotive style relays. But I didn’t like the look of any of them (I didn’t finish building the complete controller just didn’t like the look of them). Then Dean suggested those cheap plug in coil types with DIN rail mounting and I’ve used them ever since without issue. They would not pass industrial use but at home where the designer is also the user then worth trying.