eevBLAB #64 - Tesla Solar Panels Are CATCHING ON FIRE!

[bdunham7]

But I also see carbon tracking on a panel's back sheet. The cells must get overloaded during a mid-array ground fault and melt the back sheet?

I'm not sure what happened there, but it is the middle 1/3 of the cells or interconnects that have burned through, or one section out of three that is wired in a unit and bypassed with one diode.  A shorted diode would cause the cells to be shorted, but they should withstand that.  An open would subject the cells to high voltage if they were shaded.  I'd really like to know what happened here as well. 

[bdunham7]

I meant to say there are pictures of connectors that look like MC4. I can't (yet) tell the difference between the real MC4 and something similar. Before this thread started I thought any connector of that style was a MC4 and could just plug in to any mating connector. This thread has been a real eye opener about solar specific connectors, proper installation, and fires.

I've got MC4 panels and MC4 inverters that were produced before H4 and alternatives became popular.  However, if I had gotten one or the other in H4 and the spec sheet said "MC4 compatible" I would have plugged them in without a second thought--but I'd be thinking about it now!  Everyone with a DC system and mixed or unknown connectors ought to be taking a look at this point.

[SparkyFX]

Looks impressive, but lets not forget the arc is just the effect of something breaking that was not supposed to break. If these installations worked flawlessly several years before failure they most likely had been under peak load several times, but something changed/wore out and made it fail catastrophically.

Based on some pictures at https://safersolar.com.au/gallery/ it looks like reduced contact force lead to arcing and fire there. That could have been caused by loss of spring resistance in switches (either in the spring or the plastic  frame it is mounted in), work hardening (does thermal expansion and contraction cause work hardening?) or an oxide layer formed (e.g. through water ingress) and pushed contact surfaces away from each other. In case of DC isolator switches and connectors it might have been caused by enclosures that were not as weatherproof as they should have been, some pictures show how far the green copper oxide creeped.

These might be fixed installations, but vibration from wind and thermal expansion almost call for requirements of mobile applications.

[floobydust]

... lets not forget the arc is just the effect of something breaking that was not supposed to break.

It's not always something breaking, the insulation can degrade as well. I see the connector plastics cooked to carbon and then they conduct to the aluminium bar it is tie-wrapped to, and an arc starts to finish the party.

Amphenol is very heavily into their RADSOK (RADial SOcKet) spring-contact technology, to get 52A ratings on H4.
I wonder if it actually ages well or if manufacturing QC is good enough for the "hyperbolic mesh"? It does not look easy to make/assemble in a pin. No mention of the metals used and I would suspect the connection between the two sleeves does not last.

Another overlooked aspect for DC connections is ion migration at the connection, the connection's metallurgy changes with age, it does oxidize/corrode more than AC. I would guess tin migrates away. That would be seen as a polarity issue for the failures, they might favour a cathode end.

[SparkyFX]

Amphenol is very heavily into their RADSOK (RADial SOcKet) spring-contact technology, to get 52A ratings on H4.
I wonder if it actually ages well or if manufacturing QC is good enough for the "hyperbolic mesh"? It does not look easy to make/assemble in a pin. No mention of the metals used and I would suspect the connection between the two sleeves does not last.

Thanks for pointing their technology out, other material so far did not show these details.

Another overlooked aspect for DC connections is ion migration at the connection, the connection's metallurgy changes with age, it does oxidize/corrode more than AC. I would guess tin migrates away. That would be seen as a polarity issue for the failures, they might favour a cathode end.

I wonder if those are coated, and if so before or after stamping the shape. The specs say it is tinned copper.

I've got MC4 panels and MC4 inverters that were produced before H4 and alternatives became popular.  However, if I had gotten one or the other in H4 and the spec sheet said "MC4 compatible" I would have plugged them in without a second thought--but I'd be thinking about it now!  Everyone with a DC system and mixed or unknown connectors ought to be taking a look at this point.

English Wikipedia states Amphenol H4 as compatible to MC4 (name derived from manufacturer Multi-Contact, now Stäubli, Barrel diameter 4mm; although different diameters available), as well as some list their system as MC4 compatible.

But, there was also a statement of our TÜV Rheinland in 2015 (sorry, german) that strongly advises to only use connectors from the same manufacturer throughout, by lack of standardization and plenty of field defects, found through inspection, reports and subsequent internal long time testing. No certification that allows cross compatibility was actually issued, although stated by companies that had their products tested by TÜV Rheinland. Tests only covered partial aspects, but no conformity was issued.

They state onesided change in dimensions, contact material, outgassing and chemical stability would otherwise not be given or break compatibility overtime.

In other words it is an old hat and yes, i come to the conclusion that government issued inspections are a good thing. 

[floobydust]

Wow, these MC4 connectors have a serious thermal de-rating! Thanks to Stäubli for giving us some data, and saying "don't mix and match" with other connectors.
I'm not sure what typical rooftop temperatures are, but the rated amperage is to 30C and I would cut it in half. Bet people aren't doing that and then add a Y to double input current.
I could find no data from most PV connector manufacturers on deratings.

Amphenol H4 TÜV certificate R 72123201 is no longer valid, there is only one for Amphenol Technology (Shenzhen) TÜV certificate R 50388083. edit: added H4 datasheet thermal derating curves.

Weidmuller has PV-Stick push in (no crimp) the bare wire type of connector, rated 30A. I wonder what magic is inside.

Why can't they just make these connectors a bit bigger and give them true current ratings?

[orion242]

If these installations worked flawlessly several years before failure they most likely had been under peak load several times, but something changed/wore out and made it fail catastrophically.

Water intrusion in the connectors due to improper tightening would explain that quite nicely.  Not a problem at first, yet given time nearly 100% fail rate is almost certian.

Wonder if the teams all had these plastic tools on installs and equipment with mis-matched connectors.  They break the wrenches shortly after they start and walk in wally for channel-locks or resort to hand tightening there on.

[SparkyFX]

I could find no data from other PV connector manufacturers on deratings.

Amphenol has them in their specs too, looks the same. It is not said that other types of connectors (e.g. CEE) do not derate similar, as those temperatures can exceed the maximum 70°C for PVC type insulation or connectors and tests in that direction might therefore be disregarded as improperly rated.

The MC4/H4/.. are however tested according to norms including solar use cases and so on. Solar does have its own set of problems.

Why can't they just make these connectors a bit bigger and give them true current ratings?

Thats exactly how the problem started. The market called for connectors that are safe to be put together by workers not being certified as electricians, and in open environment, whilst there are plenty of other options to connect equipment with similar power (CEE connectors, or simple cable lugs), but those need to be made properly, have a protection against disconnection under power or require permanent installation, which code would forbid to be installed by a layman. Some might work, but need an additional shroud or cover.

Don´t get me wrong, they will do the job and many ideas - that the other options were not built for - went into them, but there seem to be many caveats.

Weidmuller has PV-Stick push in (no crimp) the bare wire type of connector, rated 30A. I wonder what magic is inside.

Maybe a pre-tensioned clamp, similar to WAGO clamps.

[orion242]

PV-Stick push in (no crimp) the bare wire type of connector, rated 30A. I wonder what magic is inside.

They offer free samples so easy enough.

30A 1500v water proof stick on connector has my interest as well.

[thm_w]

Wow, these MC4 connectors have a serious thermal de-rating! Thanks to Stäubli for giving us some data, and saying "don't mix and match" with other connectors.
I'm not sure what typical rooftop temperatures are, but the rated amperage is to 30C and I would cut it in half. Bet people aren't doing that and then add a Y to double input current.
I could find no data from other PV connector manufacturers on deratings.
...
Why can't they just make these connectors a bit bigger and give them true current ratings?

Its not really that crazy de-rating since its based on ambient temperature. But I think you have a good point here, if the connector is exposed directly to sunlight or is in direct contact with the back of the solar panel, it is feasible for it to reach 60C+ which is a ~75% de-rating. So maybe ambient is a bit deceiving to use as a gauge.

[bdunham7]

I think this is all a bit crazy since the implication seems to be that the connector is OK until the contact pieces reach 105C and at lower ambient temps it is just fine to run enough current through them until they heat to 105C.  A little extra resistance and fwoosh! 

I wasn't able to quickly find info on the UL listings for these, but I believe the UL ratings are all much lower.  They recently listed some of the newer connectors for 1kV, but they are 1.5kV elsewhere.  AFAIK, the UL rating for the basic MC4 is 20 amps, period.  And that seems OK.  Running 104 amps through a 4mm connector (if I'm reading this right), no matter how "robust", with an 8ga cable, is just way, way beyond what you get away with in any other area.  There's no way that should be rated for more than 40 amps and I still wouldn't be happy with the connector. 

I didn't know that non-electrician installers were stringing together parallel systems at high current this way.  I wonder if these ratings or listings have been misrepresented to local building code enforcers.  The last I knew--which has been a while--is that the temp help can bolt the panels on and clip the string wiring together, but each string is connected to a combiner box by a real electrician using real wiring and then that electrician does the balance of the wiring, again with real, enclosed-in-conduit wiring.  I guess I'm hopelessly out of touch.  The IBEW should be all over this since you now have seven (eight plus with Amazon) fires possibly caused by crap work.

[sibeen]

I suspect the sheen is wearing off Elon.

https://www.vanityfair.com/news/2019/08/how-elon-musk-gambled-tesla-to-save-solarcity

[floobydust]

From early Amphenol marketing:
"The Amphenol Industrial Helios H4 Solar Connectors are fully intermateable with industry standard (MC4) connectors and are ready for field assembly with no small parts to lose. The RoHS compliant Helios H4 connectors are engineered to meet UL, TUV, IEC/CEI, NEC, and DIN V."

Misleading because they would not be UL approved when cross-mated. Who did the IP68 test and current ratings/derate with H4/MC4 mated? OOPS

Sunday Amphenol gave a statement:
“We have no reason to believe that Amphenol’s products are the cause of any issues related to the claims filed by Walmart against Tesla,” Wallingford, Conn.-based Amphenol said in a statement Sunday. “We stand behind the quality of our products, including the H4 solar connector which was manufactured to meet established industry specifications, and certified to meet those specifications by UL, an independent third-party testing service.”

[orion242]

each string is connected to a combiner box by a real electrician using real wiring and then that electrician does the balance of the wiring, again with real, enclosed-in-conduit wiring.  I guess I'm hopelessly out of touch.

There might be some areas that this isn't the case, but the majority here will require licensed electricians.  Homeowners can get permits to do work on their own homes in some places, but not commercial installs.  Going to need permits, which almost always start with a master electrician.  Last state I was licensed in, >$600 requires permit and anything over class 2 or running conduit requires licensed electricians.  SC certainly could have had a few master electricians that where nothing more than permit handlers and monkeys doing the actual installs.  Wouldn't be much AHJ oversight adding a solar install to an existing building like there would be with new construction.  Pull permit, half arsed install, call for final inspection.  Many times you don't even need a representative present when they inspect.

[orion242]

Would be another interesting look into this.  Permits are likely public or FOIA.  Would be a crap load of work sorting that out over 240+ sites.  Subcontracted out or did SC have a few puppet master electricians that pulled all these?

[floobydust]

each string is connected to a combiner box by a real electrician using real wiring and then that electrician does the balance of the wiring, again with real, enclosed-in-conduit wiring.  I guess I'm hopelessly out of touch.

It's a good point, commercial buildings require all (mains) wiring must be enclosed in conduit. Rooftop air conditioners are wired this way.
I wonder why the NFPA allows loose wires and zip ties for solar systems.

[SparkyFX]

The layout (how many panels are in series or parallel) and therefore required rating per connector is not in the hand of the fitters, there must be plans by someone being responsible -- except they really f´d it by connecting panels in the wrong way and every control failed to notice. But that would be kind of obvious for fire investigators to check how the panels were connected.

60A alone sounds like there might be a lot of safety, a single, state-of-the-art panel generates around 15A@20V max?
Therefore how many panels in parallel to exceed the rating? More than three? The thing with the derating is, the hotter the sun shines, the more you need to derate as the panels generate more current and what works in the northern part of the country might not work in the south.

[orion242]

I wonder what magic is inside.

Heavy spring that appears to snap down on the conductor once inserted and less fiddly tightening procedure from the looks of it.  Your supposed to hear a click when you push it in enough and the bottom edge of that spring is not in the initial conductor path.  Something might break off the "wire guide", that most inner part of the stack up, which looks to be holding the spring up initially.  The spring has a fair bit of force, but would like to see how much contact area there really is with the conductor and the inner pin/spring.  Doesn't look like much on the surface.

8-3-19.jpg is the spring contact inside with the plastic "wire guide" removed.

Looks like the spring is held back by the wire guide and snaps free when the conductor is jammed in.  I pulled apart one of them to get the other picture and don't have suitable cable here to try the unmolested one.  Once the conductor is in, its just hand tighten to seal.  There are knurled bits where the nut meets the main body to try and retain things from slipping loose.  Tightening one down all the way and looking in at the cable seal, its maybe an 1/8" opening so I would think that's a tight seal against a proper conductor.

Have to dig around for a suitable chunk of cable.

These don't have appear to have UL listing, so not something we could use here.

[tautech]

each string is connected to a combiner box by a real electrician using real wiring and then that electrician does the balance of the wiring, again with real, enclosed-in-conduit wiring.  I guess I'm hopelessly out of touch.

It's a good point, commercial buildings require all (mains) wiring must be enclosed in conduit. Rooftop air conditioners are wired this way.
I wonder why the NFPA allows loose wires and zip ties for solar systems.

Here, spec demands solar wiring is to same standard as mains and in conduit.

[orion242]

Been a while since I was up on NEC code here.  When I was...

Class 2 is 50v & 100va and under, no conduit required in most cases.  Above that is class 1 and requires "conduit" of some fashion generally.

That said, each cable type / use / location has numerous exceptions.  For example, romex in residential.  No conduit required where its "protected" in residential.  That's typically in walls, attics that are used not for storage, or protected such it cannot be readily damaged by normal use.  Protected in the open, typically means the cable is in the middle of whatever wood members your passing through.  So a garage without any drywall on the interior, you can use romex as long as its not on the surface of the studs.  Still exposed IMO, but meets code.  You can run 240v 100A romex without any conduit all day long in this case.

SO cord can be used for 480v 3 phase and many times 100va, in certain use cases and locations.  Elevators don't use conduit to make the connections to the passenger car.

Never dealt with solar and far removed from current NEC now, but I would expect there are certain exceptions for solar panels.  I know solar has its own sections in the NEC, never paid any attention to them or was required to.  For starters, have you seen a panel with an enclosed box for the connection and conduit knockouts?  Bit of an issue if NEC requires 100% conduit then.  Can't put a splice in conduit under any condition, so what exactly is the market for these connectors if 100% conduit is required then?

It may be that most inspectors are clueless that some of these are being installed in series/parallel strings at fair voltages and currents.  They are just generally not aware of whats really going on, make assumptions and pass things especially when its a well known name doing the install.  I see a good number of these solar installs and never really paid alot of attention to them.  Until now, I would have never expected to see >300v volts @ ~30A on a innocent little connector like this with free air cable.  Opening that live is going to be exciting I expect.  That stuff just gained a lot more respect/interest next time I'm walking by.

With all the BS OSHA safety crap is around arc flash hazards today, I'm bewildered how a connector like this can be rated for 1KV @ 30A without requiring permanently embossed warnings, high viz colors and LOTO provisions.  1kv AC has an absolute $hitload of safety requirements at any moderate power level.  Internally in a building, your going to have to accommodate that with special construction in most cases.

[David Hess]

With all the BS OSHA safety crap is around arc flash hazards today, I'm bewildered how a connector like this can be rated for 1KV @ 30A without requiring permanently embossed warnings, high viz colors and LOTO provisions.  1kv AC has an absolute $hitload of safety requirements at any moderate power level.  Internally in a building, your going to have to accommodate that with special construction in most cases.

Unlike the AC power line, solar panel outputs are effectively impedance protected by the voltage to current curve of the solar array which limits voltage and current to values not much higher than the typical operating conditions.  So you can get hazardous voltages and currents but power and energy are limited preventing anything like arc flash.

A short just reduces the voltage to zero and raises the current 10s of percent.  The power ends up dissipated in the panels which have plenty of surface area.  If that increase in current causes a connector to fail, then it was defective already.

[SparkyFX]

Opening that live is going to be exciting I expect.

They are built to snap in and "require a tool" to be opened again. No idea if that actually makes them safer, but at least someone thought of that.

I'm bewildered how a connector like this can be rated for 1KV @ 30A without requiring permanently embossed warnings, high viz colors and LOTO provisions

They probably leave that to the person that specs the particular install to place warnings wherever necessary. The connector rating is just it´s maximum,
LOTO probably starts on the DC isolator switch. The thing to remember is... danger is highest when the sun shines, it´s a generator after all.

[tautech]

Been a while since I was up on NEC code here.  When I was...

Class 2 is 50v & 100va and under, no conduit required in most cases.  Above that is class 1 and requires "conduit" of some fashion generally.

Two 300W panels and you're over......actually one, one 300W 9A panel in a 12V installation is 108 VA !

[orion242]

They are built to snap in and "require a tool" to be opened again. No idea if that actually makes them safer, but at least someone thought of that.

Its trivial by hand push the two locking tabs in and separate it.  Tool required on these?  Fail...

LOTO probably starts on the DC isolator switch. The thing to remember is... danger is highest when the sun shines, it´s a generator after all.

I don't typically see isolators per panel.  If they are only installing them at the end of a series string....that's a fair number of connectors with little way to kill the power.

[floobydust]

6" 72-cell commercial panels seem to be around 330-390W, Isc almost 10A or 9.88A full sun, IMPP 9.37A, Voc 48V and VMPP 40V typ.

An arc occuring mid-string would short out the lower (grounded) panels and cause the top panels get +ve overvoltage injected from other parallel strings.
A combiner box has a fuse+disconnect for each string, so there is protection.
Junction box, couplers, splitter has no fuse, there is no fault protection. So a panel can see whatever parallel array current is available which seems to be very high and explains the back sheets burning up, a panel's conductors are sized only for the panel's output current.

I don't think the NFPA/NEC etc has considered this difference between AC mains systems and solar.
A ground fault on a branch conductor can result in higher current than the branch can supply or was designed to.


I realized in a big way, Tesla's solar-shingle electrical connectors are going to have great difficulty not burning up.
It's high heat, water, cold on 100's of connections. Forget about the laminations and the sandwich, it's the connectors that will be his undoing.
Announced in 2016, in 2019 the shingles are supposedly in production but I think vaporware? He has a waiting list, pre-orders etc. and those poor engineers better pull off the miracle soon.