Author Topic: No brand/5500W inverter disaster porn  (Read 7842 times)

0 Members and 1 Guest are viewing this topic.

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
No brand/5500W inverter disaster porn
« on: April 26, 2021, 10:32:12 pm »
A couple years ago i bought one of these EGS002/8010 based inverter boards, great deal for a 48v backup inverter. Am sure i tested it after i got it cause a few of the necessary control wires had solder on and i faintly remember making sure it worked. Recently after several years my power jack control board quit. No blown fets just no output. Can be repaired i think but needed a quick solution so got this spare off the shelf, after all thats why it there.

It took a while to fit it into place secure it and arrange necessary connections that were not now on board terminals. All connected i decided to precharge and power on before closing the lid. Precharge took little time as the capacitor bank is only 8000uF. Charge indicator came on and i closed the DC breaker. I didnt even get a chance to turn it on before an explosion ensued, immediately, im thinking fets and usually theres no point in saving anything once the noise starts but then another horrendous bang!! Then another!! Like popcorn only x1000!!! By the time i got my wits together and tripped the 100A DC breaker 6  of 8 had gone off!! Of course i somehow hooked a wire up backwards i thought....  Double and tripple checked the wiring. No problems there. Polarity was correct. DC volt meter read 52, caps stamped 63V. Ive no idea why the caps blew the way they did other than they had been sitting unused for maybe 2 years and dried out. Or the obvious...dragon brand...  Fitted some suitable replacements to the board and it powered up the toroid with zero noise or hum and 121/242VAC
no adjustment necessary. Now to fix the PJ no wait... its spring time there are other garden/patio/building repair jobs to do first.
Fun times!!
LH
Health Warning: May contain traces of nut!
LH

Offline noneyabussiness

  • Full Member
  • ***
  • Posts: 155
  • Karma: +7/-0
  • I think it works....
Re: No brand/5500W inverter disaster porn
« Reply #1 on: April 27, 2021, 03:40:55 am »
Ive had that happen to a high frequency board... " 63v " caps and same result... probably find they are 35v caps " re-labeled " work for a short time sometimes, then boom... drying out a bit probably didn't help ether, but should not have mattered, unless temp extremes... same fix worked for me too, replaced and all good...

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #2 on: April 28, 2021, 02:11:00 pm »
35v cap in place of 63, yeah that would do it. :P  Reminds me of a story years ago when a friend and i built a 4meter wind genny, first week of it being up his dog chewed the cord and unplugged it. Fortunately the caps and diodes all came apart shorting the 3/phase winding which brought it to a stop and saved the mill . Caps and diodes dont like limits disrespected Ha!

this is my first experience with the 8010/egs002 it uses less power to idle, like around 15w only which is great! Its also very unaffected by changes in load power.
The not so great is the AC output volts oscillates by a volt or so at about 2 hz and the AC output also changes with battery voltage.  At 52V the output is 121V and at 46.5V its about 115V.

Im sure those two issues can be corrected by tinkering with feedback loop a bit. Since battery volts changes very slowly and a plc already monitors battery volts I could simply drive a relay contact to adjust AC output volts based on batt level.

Its not a perfect solution but it is serving the purpose of a spare inverter which im grateful for!

LH
Health Warning: May contain traces of nut!
LH

Offline noneyabussiness

  • Full Member
  • ***
  • Posts: 155
  • Karma: +7/-0
  • I think it works....
Re: No brand/5500W inverter disaster porn
« Reply #3 on: April 28, 2021, 02:56:49 pm »
That's strange,  I've been using a bigger version of that board for years, and its rock solid, even pulling a solid 9kw for hours... i would change all the smaller feedback caps, check the solder for " dry joints " etc. , the eg8010 board is very good for feedback usually...

Specially if the bulk caps did what they did, i would suspect all of em...

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #4 on: April 28, 2021, 09:51:16 pm »
. "i would change all the smaller feedback caps, check the solder for " dry joints " etc. ,"

Very good advice!

 it also occured to me that the resonant freq. Of the toroid and choke likely is not 90hz as it needs to be. I never checked this one. Never noticed an issue with the PJ boards driving it.  Maybe its much lower than 90  and causing the voltage to fluxuate as the wave is influenced by rolling addition and subtraction of the actual resonant frequency. Kind of like the volume of old am radios used to do under the right conditions.

 Courtesy to Poida's research or i wouldnt even think about it.

The wave shows a little distortion from incorrect LC filtering.
Health Warning: May contain traces of nut!
LH

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #5 on: April 29, 2021, 02:56:04 am »
I just bought a dead 8000watt 24 volt powerjack.
I have repaired the mosfet board, new mosfets and replaced the very fiddly 47 ohm surface mount resistors. The rest looks fine.
Just wondering if you folk who play with inverters lots can tell me what the waveform from the driver board should look like.
I read somewhere here that folks test the driver board output before connecting the mosfet board just to make sure that the smoke doesn't get out.
The man I bought this inverter from said he replaced the mosfets and it blew up on him again. I am guessing that the dead 47 ohm resistors may have been a big part of the problem.
This model I have does not have the removable mosfet boards, I would have to remove the whole heatsink and main power board to fix it if the smoke got out, so I am hoping to test the  driver board output first.
Anyone know the wave shape I am looking for and if it is possible to test without the mosfet board in place.
Pete

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #6 on: April 29, 2021, 06:54:47 am »
Hi Pete, yes it is and it does save a lot of grief and ruined fets. From memory  :P the procedure i used was to unplug the ribbon cable from the fet/heatsink board and then cut wires 9&10 and 1&2.  Connect negative 12/24/48 (your flavor) to wire 9 or 10 leading to control board and then connect positive to 1 or 2.

this allows you to power up the control board without energizing or connecting to the power board. If you can do the same thing without cutting the wires, go for it. You just need to get ribbon disconnected from power board and get power and ground to control board.  Next put scope probe on each pin 3/4/7/8 (i think these are the 4 H bridge drive wires. Of course ground pin on scope goes to negative on pin 9 or 10. You will also need to plug in the long flat connector so you can switch it on/off.  Since it wont be generating a sine wave for feedback it will fault within a few seconds of each power on cycle but immediately after each power on cycle you should see a decent square wave, it wont be a continuous duty cycle so it looks strange but the basic thing to look for is that the voltage goes high enough, the transitions look fairly square and all 4 are very similar. Usually all 4 do not have same problem so just comparing the 4 will get you there. Frequency is gonna be somewhere north of 20khz range i think. Its been a few years now so i'll see if i can post a photo and better info this eve. Good luck. One other cool solution is just to buy a driver daughter board for $25 and replace it. They still have those, i think the new version still uses the same board.
have fun!

(CAUTION pin numbers were edited if you read this earlier today, power and ground pins were reversed)

i had a couple wrong. If you want to get picky, the probe neg is different for the high side drivers, each has its own negative pin 6-and4+ are a pair,  5-and7+ also a pair. 9/10 are - for pins 8 and 3 which are the two bolted to the heatsinks with the AC connections. I dont think i fussed about the correct ground though. If you just use the 9/10 ground for probing all 4, two will be inverted and look a bit wierd but there will still be two symmetrical pairs that look alike if all is well.
LH
Health Warning: May contain traces of nut!
LH

Offline noneyabussiness

  • Full Member
  • ***
  • Posts: 155
  • Karma: +7/-0
  • I think it works....
Re: No brand/5500W inverter disaster porn
« Reply #7 on: April 29, 2021, 04:50:19 pm »
Even easier , take the bulk caps off board ( very important) and power it up through a 20 ohm resistor... if any issues there won't be enough juice to blow mosfets.. you can even hook up to a transformer,  but you will need to play with resistor value , start with 20ohm, then 10, then 5 etc....

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #8 on: April 29, 2021, 05:43:18 pm »
Thanks Lighthunter, I do have a spare driver board for the running inverter I have, but unfortunately the dead one has a different driver board. The ones that powerjack sell these days have one row of pins underneath to plug in and the one on my dead 8kw inverter have two rows.
So far the transistors on it look fine, not sure about the optocouplers. Mine has a plug in cable that goes from the driver board to the heatsink/mosfet board so I was thinking of just unplugging it and testing the output of the opto's.
I will read your description through a few more times before I do that.
Thanks again
Pete

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #9 on: April 29, 2021, 05:46:07 pm »
I just read the second way you suggested. That may be worth a try too. Like you say better than cutting cables etc.
I remember a tech at a place I worked who used to power up TV sets he repaired with an incandescent globe in series with the supply to rob the set of enough power to die if there was a fault.
Thanks will let you know how it goes
Pete

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #10 on: April 29, 2021, 09:38:07 pm »
Sorry the second post was by Noneya.
I have removed the caps, will take the plunge later and see how it all goes then.
I realised that it appears the main board is fed battery power from the Fet/Heatsink board so my original plan would not have worked.
Did not mean to hijack the thread, it seemed that you folks have a lot of experience with these things.
Pete

Offline lighthunter

  • Sr. Member
  • ****
  • Posts: 323
  • Karma: +13/-0
Re: No brand/5500W inverter disaster porn
« Reply #11 on: April 29, 2021, 10:03:18 pm »
yeah, noneya's way may be easier what could be tricky is if you have one or more wounded fets, it will distort the wave shape. Thats still a win because youve found a problem but you dont know  its your driver or a fet. If you have removable fet boards that might be easy to separate though.
I dont remember if PJ is unipolar or bipolar in other words, bipolar is switching all 4 fet banks at 24ish khz, unipolar is switching  two banks at 60hz and the other two at 24khz. I thought the pj control boards ive worked on were bipolar.

So heres an example of a unipolar 8010 gate drive
First a photo of a 60hz gate drive then a photo of a 24khz drive
showing the pwm 60 hz component, then a photo of the 24 khz pulses. All you are really looking for is the full 12v drive strength and fairly straight square transitions.
Health Warning: May contain traces of nut!
LH

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #12 on: April 30, 2021, 01:55:57 am »
Thanks Lighthunter, I used Noneya's method, took out the large caps and put a resistor in series.
I  had already changed all the AC Mosfets over, i used Hy4008 mosfets. I also changed all the 47 ohm resistors on the mosfet board as well. Surface mount stuff sure is fiddly but I managed it in the end.
So firing it up all looked good until it screamed at me, No caps so no big bang. I checked the output ( pin6) of the optocouplers and it was rubbish. So delved a bit deeper and changed the 4 surface mount transistors on the driver board.
Then checked and found that there was a short still on two transistors from emitter to collector.
I pulled the TLP 350 opto couplers out of their sockets and the short went.
I have some on order so will have to wait until they arrive to go further.
I did order one of those inverter boards you talk about in the article and will play with it when it arrives. As long as we are not being punished too much for our politicians big mouths.
Thankyou both you saved me a stack of fets and lots of work
Pete

Offline noneyabussiness

  • Full Member
  • ***
  • Posts: 155
  • Karma: +7/-0
  • I think it works....
Re: No brand/5500W inverter disaster porn
« Reply #13 on: April 30, 2021, 02:09:28 am »
Good to here that helped, not my idea, I was suggested it by Oztules years ago when I was playing with em... the PJ 's are bipolar, I would check the low side driver totem pole transistors.. by memory " MY " and " NY "  surface mount ...( nevermind you already did )

The PJ's are tough, but can be fiddly... why I changed to the Eg 8010 boards... simple and VERY tough

Offline Pete

  • Sr. Member
  • ****
  • Posts: 385
  • Karma: +19/-1
  • Mount Barrow Tasmania
Re: No brand/5500W inverter disaster porn
« Reply #14 on: April 30, 2021, 03:01:41 am »
Hi Noneya, yes I have changed the NY and MY transistors. I am keen to try the inverter module when it arrives. This powerjack has two transformers in it. As their 8000 watt rating is so far over the top I am toying with pulling one of the transformers out and using it as a test bed for the 8010 inverter board.
Haven't seen any posts by Oztules for quite a while, I wondered what happened to him.
thanks again.
Pete