DC isolator arcing when turned off

[swaRRR]

At work we've got an 8 panel single string set up, going into a 3000w inverter via a double pole DC isolator.

The AC goes supply > isolator > meter > inverter.

Obviously the DC usually comes from the roof, into the DC isolator, into the inverter where it's converted to AC which then goes back to the consumer unit.

Never had a problem with this and we've done thousands of them.

Today however, i came to commission the set up and the inverter wasn't recognising the modules. So i did some testing; DC correct polarity coming into the isolator, and when the isolator is turned on the load side is showing voltage, as are the MC4's from the load side of the iso to the inverter.

Tried re-making off the MC4's and their cable ends, tried changing the isolator, tried using different terminals on the iso, but to no avail. Then i realised when i was turning it off it was starting to smell like burning plastic and then i noticed the inside of the DC isolator was arc flashing every time it was turned off under load (so if plugged into the inverter or if the pos-neg were plugged together).

I chucked a fluke continuity tester on the isolator and when the MC4's are plugged into the inverter, the continuity tester is detecting AC power at the DC isolators connections.

On the display, the inverter was showing a two way feed and supply between the inverter and the grid.

Any ideas? Seems like the inverter is backfeeding AC into the DC supply which leads me to believe the inverter is playing up.

 

[fourtytwo]

The DC isolator should never arc or burn under any circumstances so it is most likely not rated for the use you are putting it too, what is the make and model number ?

 

[James]

The DC isolator should never arc or burn under any circumstances so it is most likely not rated for the use you are putting it too, what is the make and model number ?

any isolator under load will arc as contact is broken,
there is also often a smaller arc generated on closing for the brief moment as the contacts are made but not under full compression.

 

[fourtytwo]

any isolator under load will arc as contact is broken,
there is also often a smaller arc generated on closing for the brief moment as the contacts are made but not under full compression.

Yes but not a sustained arc as indicated by the OP together with "smell like burning plastic", that implies the arc is continuing after the contacts have fully opened that should not be the case if the contacts are rated for the DC voltage & current involved. It sounds more like a slow opening AC breaker to me rather than a toggle spring action DC breaker.

 

[brianmoooore]

Has a DC rated isolator been used?

 

[swaRRR]

The isolators are definitely DC and definitely rated for the load.

I changed the inverter and that solved the problem - it was backfeeding AC into the DC isolator.

 

[fourtytwo]

Very interesting, presumably a transformer less inverter and that probably means isolators should be rated for at least VPVOC (open circuit string voltage) + VGRID PK (nominal 350V), out of interest if you would not mind sharing what is your VPVOC & isolator rating ?

A transformer less inverter can rectify the grid backfeeding this onto its DClink capacitors and if the input stage is faulty apply this voltage to the PV input terminals

 

[marineboytim]

At work we've got an 8 panel single string set up, going into a 3000w inverter via a double pole DC isolator.

The AC goes supply > isolator > meter > inverter.

Obviously the DC usually comes from the roof, into the DC isolator, into the inverter where it's converted to AC which then goes back to the consumer unit.

Never had a problem with this and we've done thousands of them.

Today however, i came to commission the set up and the inverter wasn't recognising the modules. So i did some testing; DC correct polarity coming into the isolator, and when the isolator is turned on the load side is showing voltage, as are the MC4's from the load side of the iso to the inverter.

Tried re-making off the MC4's and their cable ends, tried changing the isolator, tried using different terminals on the iso, but to no avail. Then i realised when i was turning it off it was starting to smell like burning plastic and then i noticed the inside of the DC isolator was arc flashing every time it was turned off under load (so if plugged into the inverter or if the pos-neg were plugged together).

I chucked a fluke continuity tester on the isolator and when the MC4's are plugged into the inverter, the continuity tester is detecting AC power at the DC isolators connections.

On the display, the inverter was showing a two way feed and supply between the inverter and the grid.

Any ideas? Seems like the inverter is backfeeding AC into the DC supply which leads me to believe the inverter is playing up.

Errrr not a good idea using an isolator under load conditions. Switches are designed to open and close contact under load. You should be isolate ac first then open the DC isolator contacts. That's with a system that's working correctly. Sounds like you have other issues in combination with incorrect shutdown preceedure

 

[swaRRR]

Very interesting, presumably a transformer less inverter and that probably means isolators should be rated for at least VPVOC (open circuit string voltage) + VGRID PK (nominal 350V), out of interest if you would not mind sharing what is your VPVOC & isolator rating ?

A transformer less inverter can rectify the grid backfeeding this onto its DClink capacitors and if the input stage is faulty apply this voltage to the PV input terminals

The isolators are the

K&N Dual String DC Switch Disconnector 25A 220V - 11A 460V (per string)​

Votage from the string was 395v and showing 5.7a.

 

[swaRRR]

Errrr not a good idea using an isolator under load conditions. Switches are designed to open and close contact under load. You should be isolate ac first then open the DC isolator contacts. That's with a system that's working correctly. Sounds like you have other issues in combination with incorrect shutdown preceedure

I was measuring the AC on the DC isolator with both isolators turned on.

 

[fourtytwo]

Not wishing to be controversial BUT I had never heard of the DC last shutdown sequence and I am not sure the average person attending a fire would either, certainly there is no prescribed site labelling indicating this.

Going back to the OP's now resolved problem his breaker looks in the manufacturers catalogue strongly as a re-rated 3-phase AC design and as such possibly has slow contacts ? In any case it's voltage rating of 450V is inadequate to break a 400V string and faulty transformer less GTI backfeeding 350V so that would explain the arcing/burning.

As my experience has/will only be with transformer based GTI's I had not thought of this problem before seeing the OP's post and I think maybe something should be adjusted in the safety requirements regarding transformer less inverters that I must admit personally I have always viewed as unnecessarily unsafe just to gain a few percent in efficiency and save a few pounds in cost.

Rant over

 

[fourtytwo]

It is unfortunately impossible for someone such as myself to access the actual standards documents BUT I did find this reference about Australian standards doubling the voltage requirement of the DC breaker when a transformer less inverter is used (vs transformer type) so this generally supports my feelings.