Schematic?
What kind of "faults?"

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! https://www.youtube.com/watch?v=BKorP55Aqvg
FAQ731-376: Eng-Tips.com Forum Policies forum1529: Translation Assistance for Engineers Entire Forum list http://www.eng-tips.com/forumlist.cfm

Vexing.
As I understand your post, The contactor is held closed all the time and the current is switched on and off by another device.
A couple of points strike me:
The first is the inrush. You are using a factor of 2.5 for the inrush. An old rule of thumb for incandescent inrush was a factor of 10.
Contactor rating. The contactor should have an incandescent rating.
This contactor appears to be rated for incandescent loads up to 18 Amps so you should be good with an 8.3 Amp lamp.
This contactor is approved/rated for voltages from 200 Volts to 690 Volts AC. I don't see any ratings for lower voltages DC.
But why is the contactor failing open?
Back in the day when we first started working with solid state logic devices, a rule of thumb, borne out by practical experience was that contact film rendered pilot switches unreliable when used at voltages below 100 Volts.
This may be mitigated to some extent by contact wipe on better quality devices.
Note that it was voltage and not current that we considered when faced with contact film issues, albeit with much lower current levels.
Contact wipe:
A device with contact wipe is designed so that the contacts close with a slight sliding action. The contacts are designed so that there is a slight movement of wiping action between the position of first contact and the final seated position.
In a contactor this may be accomplished by arranging the contact carrier so that in the open position the contacts are slightly off parallel. As the contactor closes the contacts first make slightly off parallel. As the contactor moves to the seated position, the contacts move to a parallel position with a slight wiping action.
There are other means of providing contact wipe.
By keeping the contactor closed, you are not getting the benefit of either contact wipe nor the burning action of the inrush that may be of the order of 80 Amps.
Your observations seem to support this theory.

Bill
--------------------
"Why not the best?"
Jimmy Carter

Is the coil getting the full rated voltage? It may be that the contacts aren't closed with sufficient force.

Thank you for the responses so far. Point by point:

I do not have a schematic, I'll see if I can mock up something quick.

The "faults" come from a current detection circuit looking at lamp current. If current is less than ~1A for more than 0.5s while the lamp is on a fault is generated.

The inrush is due to the DC power supply which has active current-limiting. The PSU runs a total of 3 lamps. The current is limited to ~50A, 50/3=17A.

I didn't look at DC voltage ratings for this contactor, but it should work. The main issue with DC current and relays/contactor is the pull-out, since any arcing will not necessarily self-extinguish, unlike AC power which does so every half-cycle.

I cannot speak to the internal construction of the contactor; presumably all of this has been taken into account.

Before the program change the contactor did still switch every cycle, it was just switching off back on while the lamp remained off.

The coil is an electronic coil that draws much less current than a standard coil. This allows direct wiring from a standard PLC discrete output rather than through an interposing relay. I have just verified that it is seeing 24V. Also it continues to make the typical loud "clunk" every time it actuates.

Quote (OP)

The main issue with DC current and relays/contactor is the pull-out, since any arcing will not necessarily self-extinguish

I have seen that. I was called to a small electric crane with the complaint that the hoist brake would not apply.
The brake coil was energized by a small relay.
Someone replaced the relay with an AC rated relay. This was a good quality double pole, double brake relay, but not DC rated.
Both poles were in series.
What happened is that the relay struck four arcs across the four circuit break points. The current was low but the persistent arc eventually melted the silver alloy contact material. When the contact surfaces melted the metal flowed together and completed the circuit.
At that time the arc was shunted by the molten metal bridges which then cooled and solidified.
I no longer have the photograph of the four small silver alloy hour glass shapes.

Bill
--------------------
"Why not the best?"
Jimmy Carter

However, this is supposedly happening without the contactor actually opening at all.

Nevertheless, there does seem to be something about DC currents that do bad things with contacts; I've had similar problems with a torchiere floor lamp that had a DC driven bulb, and after a certain amount of operating time, the bulb connection to its socket would degrade. I had thought that one might be able to switch the DC every so often to mitigate that, but never got around to it.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! https://www.youtube.com/watch?v=BKorP55Aqvg
FAQ731-376: Eng-Tips.com Forum Policies forum1529: Translation Assistance for Engineers Entire Forum list http://www.eng-tips.com/forumlist.cfm

It's oil or contamination on the contacts. Possibly, even from their own manufacture or internal factory lube. Running them as dry contacts (original scheme) never allowed the oil to burn-off so eventually they became 'insulated'. Now that you're running them 'not dry' that can not happen.

Keith Cress
kcress - http://www.flaminsystems.com

Rough simple schematic. Some of the symbols may be a bit off, the MOSFET may be backwards, but you get the idea. https://res.cloudinary.com/engineering-com/image/upload/v1502896990/tips/IS32_res_measure_dojhvq.bmp

My EYES!!!!!




Otherwise thanks for the schematic. Glad using the contactor is working for you. Happy I don't have to listen to it. I think I'd have used a light sensor to tell if the filaments had fried. LOL

Keith Cress
kcress - http://www.flaminsystems.com

I'm wondering why you are doing this. If you are testing the life span of the lamp this may not be real world conditions. The lifetime of the lamp when subject to the normal x10 inrush may not be as great as when the inrush is limited to x2.5.

Bill
--------------------
"Why not the best?"
Jimmy Carter

Hey I said it was rough!

We are actually interested in the contacts going to/from the lamp, not the lamp itself. We have done separate tests in the past to measure lamp degradation. Yes the lamp is part of the measuring circuit but only out of necessity.

I think you have your answer already. If your contacts are not switching they will eventually fail open when used for low voltage DC.
If you want to measure the contacts under real world conditions it may be better to actually switch the full inrush current.

Bill
--------------------
"Why not the best?"
Jimmy Carter

Waross, This is the full inrush current in this application. We use that same DC power supply in production, so the inrush is always limited.

I agree, I think I have my answer, although I'll have to ponder this a little longer to fully digest it.

Thanks everyone!

One option might be to try and reverse the current polarity through the contactor every n cycles.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! https://www.youtube.com/watch?v=BKorP55Aqvg
FAQ731-376: Eng-Tips.com Forum Policies forum1529: Translation Assistance for Engineers Entire Forum list http://www.eng-tips.com/forumlist.cfm