Author Topic: power jack lf8000/48v blew up and smoked  (Read 19678 times)

0 Members and 1 Guest are viewing this topic.

Offline techitout

  • Newbie
  • *
  • Posts: 15
  • Karma: +2/-0
power jack lf8000/48v blew up and smoked
« on: November 23, 2014, 09:32:42 pm »
4 months light, continuous duty from my powerjack, then, today, I noticed the hum sounded slightly louder, I tapped the case and it went quieter, I then gave it a second tap, it got louder and smoke came out (emergency shutdown carried out).

Mods I carried out were extra inductor for low idle consumption, and bypass voltage sense divider to stop over voltage shut down ( caused by charging voltage spikes? ). I had not yet fitted the temperature gauges to the heat sink, and transformer, and suspect the main fan was not coming on ( the middle one was going on and off as it should ).

I am located in central queensland and it has been 35+ degrees ambient, so maybe heat stress has taken its toll ( I hope it's not the control card causing shoot through, because I don't want to fry another lot of transistors ).

I am currently investigating potential loose connections / shorts and replacing the 20 x  IRFB4310 with 24 x IRFB4110

I just thought I should share my experience, and will follow up with results of repair and hopefully a positive solution.

Offline techitout

  • Newbie
  • *
  • Posts: 15
  • Karma: +2/-0
Re: power jack lf8000/48v blew up and smoked
« Reply #1 on: November 24, 2014, 02:03:42 am »
now for an update ...... the a3120 (TLP250) opto isolators on the control board are U/S, so I can't get the inverter going until I get a couple more (good that they're in sockets). I am now convinced that the front fan has not been coming on and the middle fan just isn't enough to keep the heatsink cool. Also, I am suspecting the assumed over voltage divider is somehow linked to the over temperature alarm and fan fail alarm somewhere in the hardware and / or firmware. I've measured the fet drive signals from the control board on the cro, and bottom bridge signal is getting through to the fet gates, but the top bridge drive signal isn't getting past the opto isolator (bootstrap voltage ~ 15v). but at least it seems to be trying to fire up ... after 15 sec, I get an audible alarm and then shut down, so I'm not too concerned about further damage to the control board......fingers crossed.

Hope this is of help / info to you all with  lf power jacks ... cheers

Offline MadScientist267

  • Impossible Condition Curator
  • Hero Member
  • *****
  • Posts: 1514
  • Karma: +44/-4
  • Rules? What rules?
Re: power jack lf8000/48v blew up and smoked
« Reply #2 on: November 24, 2014, 11:18:34 pm »
Just my 2¢ worth... The combination of "tapping on the case"  with it "tapping back", and lack of drive on one half... doesn't sit well with me. Fair possibility I'm jumping to conclusions but may be something to it. I'm sure oztules will have a more informed position on it, just thought it worth mentioning that there's seemingly less than coincidence there in my mind.

Regardless of the cause, looks like it was quite a show!  :o

Steve
Wanted: Schrödinger's cat, dead and alive.

Offline techitout

  • Newbie
  • *
  • Posts: 15
  • Karma: +2/-0
Re: power jack lf8000/48v blew up and smoked
« Reply #3 on: November 25, 2014, 02:01:04 am »
Thanks for your input Steve, certainly, I welcome anyones input on this. I have gone through the entire inverter with a fine tooth comb looking for potential bad connections or potential shorts to no avail, why should it blow when I tapped it ... beats me, I think the outputs of the opto isolators were taken out when the fets blew, hence now no drive to the top end. New opto-isolators are on the way, so I will find out soon, I am testing on a current limited supply to be safe. Welcome anyone elses thoughts ... I'm still thinking heat, but cannot be sure as I'd been slack and not put the temperature gauges on, but open to others thoughts ..... oztules? .... cheers

Offline oztules

  • Forum Advisors
  • Hero Member
  • ****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
Re: power jack lf8000/48v blew up and smoked
« Reply #4 on: November 25, 2014, 04:24:40 am »
Oh goodness.. I've seen that scene before... usually when i try the battery charging without turning it off before disconnecting... but not this time from your explanation.

Heat ... possible if no fan, but light duty... gee it should have stayed hot but happy.

Over voltage if your charging from solar... the unit can't shut down, and if you go over 65v it will perish.

Most likely, the ribbon connector going to the main board from the control board had a poor connection, and the heat started help the  poor connection, and tapping took it over the edge.

They are very tough if the signal does not get messed up..... very very unlikely to over current the thing... but mistiming or losing signal for a fraction of a second is terminal.

the current limited supply is a good idea.... except you have a massive number of UF in those big caps, with a low impedance path to the fets..... that may blow them to hell if there is a cross over of signal... without any supply. They hold a few joules indeed.

Interesting that the totem pole transistors didn't fail.... very odd, I have never been that lucky..... be sure they are working properly before applying power to the main boards fets... you may be seeing driver voltage from the lm339 chip, and not the totem pole drivers.... usually they blow.. unless yours is a 24v system, then you may be ok.

The pwm power supply gives you the high side driver voltage for the high fets, it will be fine if the control card is waking up ( drives the housekeeping too), and the opto gives us good isolation.... and the sockets are a good thing for repair..... except if you have heat problems ... not enough to kill the fets ( >130c for light duty I expect), but enough to cause the plastic to micro- deform in the socket legs... and possibly give us poor connections.... so it is in the connector/s side of things  I would be looking at very very carefully.

I have not suffered this kind of unknown failure ( usually doing something stupid)  ie  from loose wires/connectors, but Graham on another forum was bedeviled by it on a W7 inverter ( same boards essentially).. particularly when he was plasma cutting and welding at the same time. Once he solved that, all was good regardless of load.

.................oztules
Flinders Island...... Australia

Offline techitout

  • Newbie
  • *
  • Posts: 15
  • Karma: +2/-0
Re: power jack lf8000/48v blew up and smoked
« Reply #5 on: November 25, 2014, 02:28:56 pm »
Many thanks for the input oz,

Yep, I was pretty surprised when it blew, I was hoping I wouldn't need to do any repairs like this for a few years yet, but, I guess I'm learning as I go (the Power Star circuits were a godsend).

I've ruled out over-voltage as it was morning when it happened, so voltage around 51/52, besides, my solar reg is pretty tight on tolerance and well smoothed.

I'll be super checking for bad connections on the control to power ribbon.

I'm just hoping that the current limited supply may be the difference between losing 1 leg, and 3 legs of the Bridge, at least until things are all good again. (I only have so many spare transistors)

On the totem pole transistors, I only checked for signal at the gates, so it certainly could be the lm339 o/p getting through a blown transistor, more checks, although, both high legs and only one low side leg was blown, I'll test out those transistors anyway.

I think I'll be building an external temperature proportional speed fan control for this one, I was paranoid about heat, at first, then got complacent, now I'm super paranoid.

cheers ... Dave




Offline oztules

  • Forum Advisors
  • Hero Member
  • ****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
Re: power jack lf8000/48v blew up and smoked
« Reply #6 on: November 25, 2014, 02:47:17 pm »
Yes, their algorithm for the fan is confusing and ... well I don't think much of it.

I built a small comparator and driver for the fan/s, and it holds it steady at 38c even at 4kw and more. I have now wired the hot water into it permanently, so at least 2 hrs a day it runs >3kw. Before it would run up to 60c before the fan decided to run.... then drop like a rock to 35c

These things are not perfect, but gee they are  a good start.

I would put my inverter up against anything else now for set and forget for running a house... it is just like the grid, turn on anything you want whenever you want with whatever you choose..... but it is big...... real big....,and runs the 15kw main board which has double the heat sink of the normal 8000w... but electrically identical ( no surface mount resistors, but same value through hole ones for the 20k gate resistors.).

Make sure your inductor has no loose connections, and the connections to the heat sinks are tight. The high side fets switch much more slowly than the low side, so should run much cooler. They only switch on and off with each cycle, the low side is pwm... so heat should have affected the low side first due to switching losses not rds on loss...

I'm a sook..... I like obvious faults..... such as why did I drop that screw driver in there?????


................oztules
Flinders Island...... Australia

Offline techitout

  • Newbie
  • *
  • Posts: 15
  • Karma: +2/-0
Re: power jack lf8000/48v blew up and smoked
« Reply #7 on: November 25, 2014, 11:20:18 pm »
I think they forgot to put the algorithm in my one (a friday afternoon inverter in my opinion .. LOL)

Anyhow, the plot thickens (maybe) .......

I have found that the power supplies to the isolated side of the opto isolators have blown the zener diode (shunt regulator)

The question is ... did this cause a shoot through by destroying the opto isolator output / over volting the fet gates, or is it a result of it ????????

The ribbon cable to the power board tests ok and doesn't seem to have intermittent fault (that I can find).

One of the opto isolators didn't seem all the way in the socket, maybe that was the problem, but no heat damage obvious

I hate sneaky faults.

As far as it goes, I'm with you oz, these are good value for us diehards, a good starting point for the ultimate flexibility.

Might be time for me to order another board set.

Offline techitout

  • Newbie
  • *
  • Posts: 15
  • Karma: +2/-0
Re: power jack lf8000/48v blew up and smoked
« Reply #8 on: November 25, 2014, 11:35:24 pm »
This might coincide with the adverse reaction to the technical tap????

Offline oztules

  • Forum Advisors
  • Hero Member
  • ****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
Re: power jack lf8000/48v blew up and smoked
« Reply #9 on: November 27, 2014, 03:23:13 am »
the tech tap requires a movement of something... a loose chip would qualify for that .

The zener would require that the  gate fused with the drain. This would give you high potential for the zener to swallow...
It should die short, and the fet would likely be destroyed completely.

The fact the zener dies short would suggest the fets blew before the zener, as a shot zener would shut down the fet ( short gate to source)... and that has no consequences that result in death to the fet.

Shame they don't sell just the control card like they did months ago on ebay.

If you contact the seller, they will likely sell you the control card only for about $45 plus $25 post.... they did for me before they went ebay.... better than $179 for the set. 4110 are less than a dollar each. The power card is basically indestructible, just fets and resistors... in fact if you got 2 control boards and postage, your way in front of the dual boards for $179.



.................oztules
Flinders Island...... Australia

Offline techitout

  • Newbie
  • *
  • Posts: 15
  • Karma: +2/-0
Re: power jack lf8000/48v blew up and smoked
« Reply #10 on: November 28, 2014, 01:18:41 am »
So the PJ lives again (hoorah), though I was a pessimist and have so far only put in 3 fets on each leg to test it (at low load ~ 60w), and can now see a reasonable sine wave coming out the plug hole.

I have replaced the original 8 pin opto-isolator sockets with some better quality items, replaced the zeners on the opto power supply.

I went 18V, just to protect the gates of the fets, and not lower as I first was going to as 12v zeners would have sat there generating heat.

The ribbon cable has been replaced with a 10 pin idc socket (ok, more connectors = more chance of faults, but I had to wreck the ribbon connector to remove and test it anyway, (I hate desoldering double sided boards ......).

The saga of the fan continues, and looks like beneath the hot glue goop, the assembler had put a temperature switch across the gate and source of the fan power switching fet, ensuring that the fan never turned on when it got hot. (source and drain may have worked better).

I am now trying to work out a good strategy for the fan control before putting the PJ back into service.

Oz, your comparator system sounds easiest, did you leave the fan tacho connected to the control board?. ( I think my over voltage problem may have been a fan fail alarm due to no tacho signal from the fan)

I am still concerned about the lower totem pole drivers, and wish I'd ordered them with the new opto isolators, although they did test ok with the multimeter, but they may be degraded and fail at a later time. Maybe I'll just swap 'em anyway. thoughts?

Thanks for your support and knowledge on these PJs, oz, I am learning a bit more about what makes them tick.

I've just bought a 600 x 800 electrical cabinet, so I can rebox the pj and put a few more whistles and bells on it.

cheers


Offline oztules

  • Forum Advisors
  • Hero Member
  • ****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
Re: power jack lf8000/48v blew up and smoked
« Reply #11 on: November 28, 2014, 03:37:36 pm »
Nice work there.... the machined sockets are an improvement for sure.
The 10 pin  idc is the same as the power star inverters.... should be good, it is in most pc's.

The 18v setting is a good idea, no point in driving the transformer more than necessary.

Yes replace the totems poles if you can. They are almost free those surface mount transistors, and will set your mind at  ease.

As you have no charring yet from those, it can be useful to use to92 versions, to stay away from the board surface.
The biggest problem with blowing these up while experimenting, is the burning under the totems.

If your not expecting to experiment, or push them beyond their boundaries, then this is moot. They should not fail under normal operating conditions... but your did...
somehow.

The fan and it's settings are weird... and I don't like them... at all.

I used a small fan from ebay with  3 pin  plug to satisfy the tacho needs of the control card.. put i where you wish.....you can see it hanging near the transformer at the top of the pic.
The chip needs to see it or it will shut down the unit.

Then used the 12v output that normally drives the main screen as a power supply for a small comparator and fet to drive the two main 5" fans independent of the board.

The assembler was in error... must have been a Friday thing. On the ones I have, they got it right.... same with pressing in the chip.
Quality control is not their strong point as yet from the looks of it, but it is truly a useful little control card, and you can make a very very very robust unit out of it.... and mostly easy to fix if you kill them from abuse.

Here is a shot I should not share... it was meant to be tidy, but I keep adding bits to it, and it became a scene of total chaos..... I will build a second one with all the additions planned for.. rather than tossed in
3553-0

It is only 8.30 in the am at the moment, here is the front panel shot... ( really 7.30am without daylight savings)

3555-1

The advantage of tons of solar panel.... cheap, and even before the sun actually gets to hit them at a decent angle, the batteries are getting a fair shake and are 94% charged... the rest of the day is wasted yes, but even on 8/8 cloud cover days, it still fully recovers by 4pm... including 4kwh of hot water each day ( or thereabouts)

Life is  just too easy with big solar and big inverter.... just like on the grid, set and forget..... water the batteries once each second month ( not much at all)

We can see near 1kw of grid tie coming in, so it is running the house, and providing another 500 watts of power into the battery ( apprrox10 amps ) the rest of the 49amps is coming from the48v arrays.... some of which have no real sun as yet... sun is south of east west line, it won't be until 10am or there abouts before it comes north of an  E-W line and hits the panels at a decent angle.

A shot of the pv control cabinet made from a aerosharp grid tie box. It controls up to 100 amps or more @ 60v. Simple knob at the bottom left for voltage control.
It is supplying the remaining 40 amps.... yes a new meter and shunt are on the way since the upgrade.

3557-2

The top temp is the transformer, the lower one is the heat sinks on the low side fets.

The only problem I see with this current system, is the batteries are too big now. Ideally I could get rid of 16 of the t105's, and just run 8 of them, and put the 16 into dry charged storage without electrolyte.... but because I can and do use very high loads, I don't want to sag the voltage, or distort the plates from excessive current.

I believe that with lead acid, it is a case of use it or lose it, so you are best to drive them hard, and fully charge each day, rather than lightly discharge them and recharge fully each day. I don't seem to need three days storage, as with this much solar, every day is a good day... maybe one or two days per year are dark enough to cause concern for a smaller battery system... but the rest of the time, 250ah would be enough to do it.

The super cheap solar changes the paradigm that solar engineers use... but they have not woken up to it yet. Now days 10kw of solar is $5000, and  $5000 for batteries and modified power jack, and you have the best system I have seen. Not tracking array, no solar hot water gas fridge or any other RE stuff need be tolerated.

In fact a 15kw inverter built from PJ boards from ebay etc and grid tie transformers is less than $500 all up..... leaving $4500 for trojan t105re's

It is a different world now for sure.... on a sunny day, I probably shed more than 75kwh... total waste... but it costs nothing to do. I design for worst days... not best.
Or I can run a few kw of lawn sprinklers  all day....if I had that much water available... power to burn.

...............oztules
Flinders Island...... Australia

Offline sean

  • Newbie
  • *
  • Posts: 13
  • Karma: +0/-0
Re: power jack lf8000/48v blew up and smoked
« Reply #12 on: December 06, 2014, 02:40:30 am »
hi guys new to this site i see its not just me that has a blown 8000w LF RIP PJ 29/10/2014 to 4/12/2014..... Is there a teamspeak or a skype where people can talk about there blown up inverters as im bad with typing and most of the time i get it wrong trying to type things that i dont know how to fix sorry to be a pain any help be good as im not get help from the seller


the black smoke smells so bad it going to be there for long time

Offline oztules

  • Forum Advisors
  • Hero Member
  • ****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
Re: power jack lf8000/48v blew up and smoked
« Reply #13 on: December 06, 2014, 03:27:40 pm »
If you can do a bit of elecrtickery, then you could replace the 4 driver transistors, and the 2 opto's and the 24 fets and the 24 gate resistors, and it will PROBABLY be fine.

If the driver transistor have burnt the board under themselves, then it is more difficult to achieve this, but doable with some imagination..


If you don't trust yourself to do this, then
http://www.ebay.com.au/itm/repair-parts-main-board-control-board-of-LF-Pure-Sine-Wave-Power-Inverter-/221351354856?pt=AU_Tool_Parts_Accessories&var=&hash=item338991a1e8
Replace wire for wire, and it will be a new machine... perhaps if you tidy up your new wiring, and tighten all bolts to the heat sinks well, be probably better than new.


Quality control is not their strong suit. Not uncommon to have spare nuts and bolts floating around, and the wiring all over the place.

They do wax in the connectors which is probably a good thing.... but they could spend a bit of time doing a lot of things better.... but they are cheap and very reliable when you fix up their loose bits and pieces.... not all are poor, but there is a sprinkling of real doozies that turn up.

The actual electronic design is very good and stable as an inverter only.

.......oztules

Flinders Island...... Australia

Offline Norm

  • Sr. Member
  • ****
  • Posts: 475
  • Karma: +26/-0
  • Today is the day you worried about yesterday.
Re: power jack lf8000/48v blew up and smoked
« Reply #14 on: December 07, 2014, 07:40:39 am »
Tapping ......   
Gee I believe a Novel could be written on the subject .
I Knew a foreman once taking readings on about a 100 gauges
was never satisfied with each gauge until he had tapped it at least 5 times
Not that I'm against tapping....just saying like in this instance it can be carried to excess .....like Steve said .....If it "taps back"  Whoa !
Hope you can appreciate my humor here : )
Norm.