Backfeeding Solar through inverter

I will presume that the batteries were full, and the loads on the inverter were far less than the 5kw provided by the grid-ties?

Ideally, yes, with the PJ being an LF inverter, power would backfeed through the transformer and FETs into the batteries.  Should theoretically be able to handle at least several kilowatts of power in reverse.

Remember that grid-tie inverters will push as hard as possible to dump as much power as possible into the "grid."  If there was nowhere for the power to go (loads or batteries), the inverter output voltage regulator will just push the throttle downwards to try to maintain the desired voltage.  Zero throttle is basically a dead short across the transformer through the FETs--which especially if they're being poorly driven, will quickly cause them to blow out.

At least that's my guess.

Let me see if I understand this correctly.

So, once the FETs are being held at zero percent (in anticipation of the voltage drop), the power coming from the solar array simply has no place to go.

Now, what I don't understand is how this would cause an explosion.  The transformer voltage would continue to rise up to the microinverter's max output (around 268 volts), but that should only equal around 15 volts on the other side....being stopped by the FETs.

I'm not sure how this would cause a failure... you would think the transformer would just sit there idling, with the FETs off.

....unless the PJ software decided to randomly start firing off FETs, causing the micro's to lose sync.  I am wondering if the loss of sync due to 0% throttle had something to do with it... especially when the FETs powered back on.  If they even did.

This inverter does not have a built-in charger, and the lead-acid batteries were at 50%.   

The plot thickens.

This inverter does not have a built-in charger, and the lead-acid batteries were at 50%.

My  8kw PJ  inverter has a charger board  but the ebay  seller put in a large note  say do not use charger ac  input  for unknown reason .  I  know on youtube people  had their PJ  inverter blow up  by using  ac charger input .  Your 8kw PJ  is easy to repair with  new mosboards  and a  rev11.1  control board  .   I  been using  my 8kw PJ  for  the hot water heater  for 2 years  no problem  and even  start a 5hp  shop air compressor .    This  ASL 4  transformer is unbelievable .  

OK, so "Throttle" is the sine table multiplier for the SPWM output that directly drives the FETs. 

If the FETs are held at 0% throttle (due to the regulator dropping the throttle to zero as a result of the grid-ties forcing power into the inverter), the FETs are literally holding a dead short across the transformer secondary...holding both leads to battery positive.  If they aren't fully turned on, they will end up dissipating pretty much all of the grid-tie's power. 

24v inverter?  (guessing from your comment about 15v)

If the microinverters reach 268vAC out, considering that PJ likes to use 230->18v spec these days, that's (268 / [230 / 18] = 20.97vAC * sqrt[2] = 29.65vDC max out.  And if the microinverters reached a full 268vAC out, the PJ definitely reached zero throttle due to the AC overvoltage condition.

Lead acid batteries are pretty "movable", and as such don't constitute a solid dump load.  See, if at normal throttle (which for a PJ is closer to 100%), the batteries aren't absorbing enough energy to keep the AC voltage in the desired space, the inverter is going to spiral the SPWM throttle down to try to regulate the desired AC voltage.  Initially this causes the FET's to work as an inadvertent boost charger, pushing the battery voltage higher and higher--but if the battery doesn't hold it's ground, there's still no place for the 5kw of grid-tie power to go.  Down the throttle spirals towards zero--and as the FETs are poorly driven, the likelihood of them dissipating a significant percentage of this 5kw is pretty high.  And that's not something they can survive for very long.

If the micros lost sync, they had better shut down instantly...

A couple of years ago I ran a 2000w xantrex grid tie with my 15kw powerjack as "grid" running the house.  Everything worked well.  I watched it very closely, as there was nothing to prevent smokey loud fet failure if  grid power exceeded batteries ability to absorb.  As Sid said, It has to go somewhere, and the fets are the weakest link in that situation.  I only ran mine for a couple hours at a time  two days running, mainly to see if I could make it work.  I was nervous as a cat the whole time.

It worked fine.  I never tried it again.  There is a definite advantage in power production, as opposed to just solar charging your batteries and letting powerjack run the loads.  But wasn't worth my worry level.

A friend of mine does this all the time.  Much smarter than I am with electronics.  He built a control circuit that measured ac "grid" voltage out of the powerjack and had it progressively short out solar panels to the gridtie inverter in sets of (I think) 4 panels at a time as voltage started to climb above 245v.  This reigned in the grid tie output when necessary to prevent powerjack (and household loads) from damage due to high voltage.  Also, this method let the grid tie inverter stay online (with reduced output) so you didn't have to wait the 5 minutes restart time each time it cut back.  It kinda simulated clouds passing over your panels. 

In Paul's situation, I wonder if part of the problem is the micro inverters.  Shouldn't make any difference, one grid tie inverter or several, but I wonder....  Maybe a group of micro inverters is too much for powerjack's sensitivities

Most grid tie inverters are either 240v only or 120v only.  Shouldn't matter.  In this situation they backfeed through the powerjack and help power loads or charge the batteries.  Hopefully you didn't have anything connected to the powerjack's ac input connections?  That might throw a wrench in the works....

Dochubert's First Rule of Powerjack;  Never connect anything to the AC Input connections.

I'd be quite curious how long the PJ inverter lasted before the magic smoke came out.  GS inverters have native GTM frequency-shift throttling (i.e. if there's nowhere to put the power, they will shift the frequency up to a max of 2.5Hz to shut the GT inverters off).  But that currently takes a second or 2--and of course will restart the grid-tie inverter's timeout for restart.  Could result in the inverter getting slammed with full grid-tie output every 5 minutes, so...

It was only about 4 or 5 seconds between the solar coming online, and the FETs blowing out.  Poor thing didn't have a chance.

Is it possible to get the GS inverters to not go under 2% throttle or so?  Obviously, they can still frequency shift if the line voltage gets way too high.... but it may eliminate that big voltage spike when the solar system gets into MPPT mode.

4-5 seconds...that should be plenty of time for a GS inverter to frequency-shift the grid-ties into off state.  Not to mention driving the FETs cleanly...they shouldn't blow out.

There will always be a voltage spike when the grid-ties suddenly turn on, and dump 5kw of power into the system.  Would be very helpful if the grid-ties supported frequency-shift throttling...that'd make it super easy to control the array with a GS setup.  Most of the older units cheaply available don't support linear frequency-shift throttling.

Of course, the GS throttle could easily be limited to a specific minimum if that is deemed necessary for surviving a high-powered assault by grid-tie inverters 🤪.