Classic running warm, now dead

[offgridQLD]

One thing I was thinking was that if something like this were to
happen again, a noise maker/buzzer or additionally even a relay/contactor of some sort could be connected
to the to one of the Aux outputs in a mode that would turn on when the battery voltage got above some preset
value for at least a second or two or three.  That would keep the plumping from happening

The issue I see with that is wouldn't you still be relying on the classic  to activate the Aux output. If a rare total meltdown system failure has the PV connected unregulated to the battery and the brains of the unit all upset. Then I'm not sure you could trust the Aux output would be functioning correctly.

I think double redundancy like I have controlled by a external device is the only way to safe guard 100% (well not 100% as that could fail to but you have to draw the line somewhere).  Rare events like this just show how important it is . Lifepo4 is reasonably forgiving and a safe chemistry. You need something dramatic abuse to upset it and some extra hard abuse to make it vent or go pop. Though they are are not as forgiving as flooded lead acid when it comes to chronic overcharge as they wont just boil away and makes bubbles like the FLA's. That said even 48v FLA being fed unregulated full output pv at 76v in 40C ambient weather could end in flames.

Kurt

[wrt]

Hi Bob,
I have got the photo I promised of the FETs. The unit was purchased late 2012. The battery temp at shutdown was 48.7c. Whle this does not appear warm it needs to be noted that these cells rarely approach 2c over ambient.

A few of the posters have expressed concern about how dangerous it may have been that the battery reached 76v. I think the scary thing is that they reached this voltage in the time it took me to head out to the shed and pull the fuses on the array after the inverter shut down.

It needs to be considered that my array is around 120voc and the controller would have kept driving the cells to this had I not been home.

The other thing worth noting is that the unit had already shut all its fans OFF by the time the FETs reached 67c temp, therefore removing any heat protection it had. They only came back to life after the temp had reduced into the high 50s and the LEDs stopped flashing.

The place I am driving to tomorrow is called DC solutions in Melbourne. They are the Victorian distributor for Midnite products, and the repair agent. It's a five hour trip but at least I know it won't be lost in the post 

[boB]

WRT, I don't see anything weird in the FET picture.  At least I don't see any scorching.

The Classic now has battery over-temperature shut down but is normally disabled.  Your version
probably didn't have that but even with the latest, it is hidden in the modbus until the new firmware
coming up in a few days.

I think you said that the MNGP display showed the high voltage on its display ?  If so, the Aux output would
have worked to show over-voltage before it crashed or whatever happened.

Are you using your own BMS ?  I think I saw something about that ?  There is a Classic Aux 2 input mode
that is made for a certain LiFePo BMS where  it tells the Classic to shut off when going high or low as well.
Can't remember the particular BMS though.  Let us know how the visit with DC Solar goes so we can
fix you up if that doesn't pan out.

boB

[wrt]

The marks that have only recently appeared are the dark stains between the metal tabs that overlay the white background on the edge facing the camera. These were not there last time I checked inside and looked to my untrained eye to be scorch marks, as I have seen these before on other electrical items.

Hopefully the trip went well. After a fruitless discussion about warranty periods and my retailer stepping in for me they are looking to repair it under warranty unless they find something that voids it.

They have no idea of a time period yet as they have no way of knowing what parts are needed or how long it will take to get them, I am happy happy with that.

I have requested that they note what went wrong with the unit and what was required to repair it so I can share with you.

As for the BMS, as offgrid has stated, I had yet to fit a hi voltage cutoff to the system and was relying on the Midnites reputation for reliability until I got it done. It was just bad luck I got caught out by an unknown fault. I am happy to share what I can though so we all can learn 

[boB]

Ahhh yes !!!   I DO see that mark in the picture.  that is definitely where the smoke came out of the FET.  That FET is probably
split in two.

So, what will happen, I am pretty sure, is that your Classic (in the picture) will come back to us from DC Solar and they
will do something to satisfy you.  I am not sure exactly how that works ?  Or, at least I can't remember, but I have
a feeling we will see your unit.  I would like to know what came first...  A blown FET or bad processor or both or just
one of them ?

Thanks for the information !  The more info is always better.

boB

[Westbranch]

Am I seeing it right, that there may be as many as 4 FETs involved, going by the dark marks?

[boB]

If the input is shorted, this usually means at least 2  dead.  Takes 2  to short.  Pretty much the same for most mppt CCs.

[kitestrings]

First off, truly sorry to hear of your experience.  I know how much work goes into planning and setting things up...

The issue I see with that is wouldn't you still be relying on the classic  to activate the Aux output. If a rare total meltdown system failure has the PV connected unregulated to the battery and the brains of the unit all upset. Then I'm not sure you could trust the Aux output would be functioning correctly.

I think double redundancy like I have controlled by a external device is the only way to safe guard 100% (well not 100% as that could fail to but you have to draw the line somewhere).  Rare events like this just show how important it is.

Kurt, I know with wind systems it is pretty common for folks to employ an alternate means of protection for overvoltage (from the turbine); as the risk is high.  We're using an inexpensive voltage sensing relay - still unproven in our case, but it's based on a design used effectively by Rob Beckers on grid-tied systems with his Aurora inverters.  You may find it of interest.  If you search the OP board you'll find more.

It was just bad luck I got caught out by an unknown fault. I am happy to share what I can though so we all can learn 

This is a rare, and remarkably positive attitude given your circumstances.  From my opinion these guys have some of the best customer support out there.  There will be something to have learned here that, may benefit many others.  All too often there's finger pointing and nothing learned from the fact that sometimes, despite our best human efforts, things go south.

Good luck with it.

~ks

[offgridQLD]

+ 1 about wrt's calm attitude . If my new lithiums had been fried. I wouldn't have needed to get on the forum to share it. As my screaming fit of rage would most likely make international TV news   ...headlines (man destroys city in a godzilla like fit of anger over swollen batterys)

Yes I have always admired the open and honest attitude that the Midnite crew have maintained even as they grow. Try and find another company there size that  involves there customers so much in there products on a personal level. Only good things and a ever improving  product can come from that.  I sure am keen to find out how the rare even was triggered even if it just turns out to be bad luck and a 1 in a million component falure....might help midnite employ was to protect against it happening again

Ok now the brand loyalty speech is out of the way

Kurt, I know with wind systems it is pretty common for folks to employ an alternate means of protection for overvoltage (from the turbine); as the risk is high.  We're using an inexpensive voltage sensing relay - still unproven in our case, but it's based on a design used effectively by Rob Beckers on grid-tied systems with his Aurora inverters.  You may find it of interest.  If you search the OP board you'll find more.

Yes when I purchased my 16x400ah  lithium cells they came with 16 individual cell monitoring cell top boards and a BCU box that has two large 200A contactors inside it (one for charge power in and the other for loads out) the electronics in the BCU box pull the appropriate contactor based on if its a HIGH voltage event or LOW voltage event signal sent from the cell top boards. The box also has a small  lcd display with capacity meter, SOC, AH IN , OUT and load - charge values (not really needed now I have the (WBJR). Not a bad little unit covers all bases. (I did a small over view of the BLUE EVpower brand  BCU in my youtube channel)...link below in my signature.

Kurt

[wrt]

Hi guys,
Just a quick update on what is happening with my classic.
I have phoned the AU distributor to find out what is happening with the CC.

They state that the problem was caused by one FET overheating, failing, then causing a cascade down the other FETs as each took on an ever increasing load that they could not handle given their already stressed state. All but two failed.

The only problem I have with this diagnosis was the system had been on float for an hour and was running at ~700w I would think the remaining FETs should have no problem accommodating 700w without failure. Although this does link with the fact that the higher the voltage difference between panels and battery, the higher the temp developed within the unit (120v on float compared to 90v under heavy charge output)

As for warranty, the tech stated they are considering sending out a new replacement, however the boss is away until Monday so this cannot be confirmed until then.

As for those discussing my attitude toward the situation...My opinion is that more people need to use logic and common sense when attributing blame to product failure. I believe that any company can have something unexpected happen. They should not be judged by this because it is usually not predictable, they should be judged by how they handle it.

Although the failure of the midnite product did result in the damage to my battery, I had not taken the precautions that should reasonably be expected of someone running LiFePo4 cells against this occurrence, therefore the damage to my battery is in no way Midnite's responsibility or problem.

The failure of the midnite itself, installed to spec, is something they are responsible for and I am very happy with the way it is being handled and with the effort they are making to both rectify my situation and continue improvement to make their product the best available.

To Bob, if this unit does end up coming back to you for a look, I can PM you my details so you know who it belongs to and you can feel free to ask anything you think will help in identifying the cause of failure.

[Halfcrazy]

WRT
Please ask your contact at DC Solar to email me ryan@midnitesolar.com I will give him the OK to send out a new one asap

Ryan

[boB]

Thanks WRT.   Yes, if you don't get results with DC Solar (which I'm sure you will though), we will back you up as well of course.

What happens in 99.9%  of cases when one FET shorts is this...    That FET, that WOULD normally connect the PV input to
the battery terminals through the RELAY contact,  would then short the BOTTOM FET and from the almost un-limited
battery current going back into the bottom FET that turns on, that bottom FET will then short, turning OFF your
battery circuit breaker and stopping any more bad things from happening.

In addition, (99.9% of the time), when both top and bottom FETs have shorted out the PV input, the PV will
be at 0.0 volts because of that short circuit of PV + to PV -   So if this did not happen, this is extremely
rare.

Normal operation is (at 25 kHz) the TOP FET turns ON and then the TOP fet goes OFF and the bottom FET turns ON.
then the bottom FET turns OFF and the TOP FET turns ON....  Repeat 25 thousand times per second, two cylinders.

So, both of the converters are actually in parallel.  If one "cylinder" shorts, the other cylinder should operate
but if one cylinder is shorting the PV plus and minus, the remaining cylinder cannot run.  Even if only two
FETs in one cylinder are shorted, the other two cannot operate with their co-horts shorted... UNLESS the
blown FETs actually OPEN up.  This would be that 0.1% of blow-ups but we rarely, if ever see that
happen.

Then, in addition to all this, even if a FET goes OPEN, the software should see that the battery voltage
is way over voltage and shut things of.   In addition, there is a RELAY...  If that relay happens to
not be able to open up, or the contacts weld together (also very rare), then what happened could
happen.  There is quite a bit of redundancy and along with over-current protection by way of
the battery circuit breaker, you would think that would be enough.

So much for the department of redundancy department.

Waiting to see what exactly happened, circuitry wise on this one.

boB

[wrt]

Thank you Ryan, I have phoned and passed on the info to the tech, though from his non committal response I don't think it will make too much difference to his decision to wait fir his boss's go ahead 

On the subject of the cause of the problem, I am interested in the temp limits of the classic, what temp does it shut down at? What temp would you normally see on the FETs on a35c day, or even on the odd 40c day (we get up to 7 days a year at this temp)

I guess what I am asking is, under normal circumstances should the classic have no problems under these conditions with panels putting out up to 120v or will I need to have my array rewired for 2s and maybe the classic string tool be altered to reflect this for future customers. Something along the line of 3s only for climates where max ambient is 30c. Above this, limit to 2s.

It should be noted I have never had temp problems under big charge loads where input volts are lower, only float/low charge rates where they rise.

Or perhaps non of this makes a difference and I was just unlucky to get a crook FET that was waiting for the right combination of high temp and volts to fail.

Obviously I don't want to have to reconfigure if it will no difference to the classics lifespan 

[Westbranch]

WRT, with PV output normally rated at 25* C, as the temps rise what amount of de-rating (volts) are/were  you observing?

My personal preference is to keep below that 80% of Vmax for the Cl150, and I see  you are at 120V/80%...

Mine will be stepped down from nom. 48V to 24V, I see you re going from 120V to 48...  maybe too big a stepdown...??

add: for clarification:  nom.

[wrt]

The volts on mine vary with load as well as ambient temp. They range from around 120v on low load and a cool day to around 90v on a hot day with a big load.

FET temps vary under the same conditions. Cool day and big load they hover around the 48c. Cool day and light load they will rise to around the 52c mark. On a hot day and big load they are around 52c with this rising to around the 56c mark as the load drops off.

I note that your panel volts are at 48. Keep in mind my volts are at actual out put, not nominal, if you are using 2s x 24v nominal rated panels you may find that they actually run around 70v-80v in practice, and your step down to 24v is greater than mine.

I am running 3s x 24v nominal panels down to 48v