1 hour ago, wael_fathe said:

can you guess their  function

No, sorry. It's between the two sets of 8 MOSFETs, with many parts not populated. It doesn't seem to correspond to anything I know; my guess is that it's something that is specific to VM models.

Connecting to a 100 μF capacitor in the main power supply seems weird. Usually those would have DC only.

Edit: Oh, wait, it's cut off, looks like that's a 100 Vdc capacitor. Even weirder.

Edited July 11, 20233 yr by Coulomb

4 hours ago, Coulomb said:

No, sorry. It's between the two sets of 8 MOSFETs, with many parts not populated. It doesn't seem to correspond to anything I know; my guess is that it's something that is specific to VM models.

Connecting to a 100 μF capacitor in the main power supply seems weird. Usually those would have DC only.

Edit: Oh, wait, it's cut off, looks like that's a 100 Vdc capacitor. Even weirder.

the  problem is  that many  people  in this and other  forum report it to be open  circuit when  there is  primary fet short

it seems to  do n active or  partly active  role ....but it looks   so passive in  the edge of the board 

hello  colomb ..it is  getting interesting  here

qmax5000 watt

i knew the  secondery side  igbt circuit assumed to be good    

all fets  in primary  are out of circuit

 

i  decided to short the   sg3525   optical copler

 

i did it and i get  a  nice square wave   so sg3525  is good

 

but  why  the cpu not  send on command   ,,,,,i   tested  across the diode  side of the copler

 

one  pin  was  tied to  12vdc  the bottom one  should  be  dragged  down to ground    sg35xx to work 

 

but    i was  shocked  with 14vdc  at the bottom,  leg of the diode  side of the copler

 

in short.  the copler is   reverse  biased   by the cpu board ,,  the cpu  wait for 30 second then issue   09  error  

and in this 30 second it never  turn  on the  sg3525  ...i changed  the   uln2003  driver ic  with used  one  but 

 

it is  the same  no improvment   still no  square signal

 

 

can you guide me  a bit  ,,,,,what else  can prevent the cpu .....i  even  thought that uln2003  have no ground  connection 

 

reaching but find it  grounded   

 

 

should i do factory reset?  should i replace un2003 again?..i have  1.5k with same board should i test it  with it?

should i clean the cpu  with  solution may be  some  dirt stuck  on legs?

 

i tested the 3.3 5vdc 12+  -12  all good   i totally lost   

15 hours ago, wael_fathe said:

but    i was  shocked  with 14vdc  at the bottom,  leg of the diode  side of the copler

14 VDC with respect to what? Perhaps you are measuring with respect to -12V?

The opto coupler circuit is very simple. There is no way you should be seeing 14 VDC with respect to earth at CN11-6. Maybe with respect to -12 V, but then the opto or R65 would have to be faulty.

Perhaps you are talking about the transistor side of U19? If so, I assume that Vref is about 5 V positive with respect to -12 V, so around -7V with respect to earth. I still can't see how you get 14 V.

That 14 V has to be your problem. Have you checked R65 and the diode drop across the opto coupler? I'm guessing about 0.9 to 1.2 V would be normal (with a multimeter, not when conducting hard).

3 hours ago, Coulomb said:

14 VDC with respect to what? Perhaps you are measuring with respect to -12V?

The opto coupler circuit is very simple. There is no way you should be seeing 14 VDC with respect to earth at CN11-6. Maybe with respect to -12 V, but then the opto or R65 would have to be faulty.

Perhaps you are talking about the transistor side of U19? If so, I assume that Vref is about 5 V positive with respect to -12 V, so around -7V with respect to earth. I still can't see how you get 14 V.

That 14 V has to be your problem. Have you checked R65 and the diode drop across the opto coupler? I'm guessing about 0.9 to 1.2 V would be normal (with a multimeter, not when conducting hard).

He mentions an error 09  occuring, so If the DSP performs the bus soft start and if that test fails, then the DSP does not enable the battery mosfets . So I dont think there is a fault around the mosfet drives. 

Hi  @Coulomb can you comment on this capacitor bank module of the Infini 5kW inverter ? , link below 

Its been a number of weeks , had a bit of a winter's break , and ready to take the Infini inverter on again. 🙂

 

As far as my partial schematics of the Infi is concerned , I have very recently discovered that the Infini 10kW service manual   has two interesting  partial schematics.  They are shown below :

here is the  battery module :

and here is the main board : Note that the 10kW is a 3 phase machine hence the 3 half bridges as shown.

When I traced my own partial schematics for the 5kW (there exist no service manual on the public internet) I had not been aware of these schematics. They would have certainly helped me to trace easier, but fortunately they are an indication that my trace work did not turn out to be inaccurate at all.

Below is my own :

there are a number of components not shown in my schematic , eg. I did not trace the mppt , as my only concern is to repair the faulty infini battery module , and not to trace the entire machine.   The 10kW machine schematic of the MPPT confirms my original finding that the MPPT is not split supply , as can be seen on the schematic. So my theory remains that the large split capacitor bank module provides us with a split supply for MPPT around the Neutral midpoint.

 

 

Edited July 15, 20233 yr by BritishRacingGreen

15 hours ago, BritishRacingGreen said:

Hi  @Coulomb can you comment on this capacitor bank module of the Infini 5kW inverter ?

Yes, I think you have it correct; they go to a lot of trouble to provide a mid-point for the DC bus, which they can then connect to neutral in a 3-phase inverter, and the neutral can source and sink current, just like the centre point of a star connected transformer.

It has the same use with American style split phase systems, where the neutral could also be called upon to source or sink significant power to or from an unbalanced load.

There is also the advantage that PV- (which has to connect to BUS-) doesn't have lethal potential with respect to neutral, and hence also earth and puny human bodies. Now something you said back then makes sense now: for the 230 V single phase model, the two outputs are in synch, and are literally paralleled. This is because with a double bus voltage arrangement, each output can go positive or negative with respect to neutral, so there is no need to connect neutral alternately to bus positive and bus negative. They could use a single half-bridge to generate the active output, but they seem to have chosen to use transistors with half the current rating in parallel instead. They get to use the same printed circuit boards for both single phase and split phase models.

Another advantage of this arrangement is that there is far less likelihood of weird problems when panels get wet and start leaking to earth. For example, the problem some Axpert users have reported with bus overvoltage faults and residual current breakers tripping.

Against these modest advantages a safety advantage that they don't seem to be talking about, there is the expense of double voltage silicon, and a large number of physically large and moderately expensive electrolytic capacitors. This is possibly a major reason that the Infini models are generally considerably more expensive than similarly rated Axperts; the circuits are far more complex, and the components more expensive. I think it makes the Infinis more bulky than the Axperts, too.

Edited July 16, 20233 yr by Coulomb

33 minutes ago, Coulomb said:

Yes, I think you have it correct; they go to a lot of trouble to provide a mid-point for the DC bus, which they can then connect to neutral in a 3-phase inverter, and the neutral can source and sink current, just like the centre point of a star connected transformer.

It has the same use with American style split phase systems, where the neutral could also be called upon to source or sink significant power to or from an unbalanced load.

There is also the advantage that PV- (which has to connect to BUS-) doesn't have lethal potential with respect to neutral, and hence also earth and puny human bodies. Now something you said back then makes sense now: for the 230 V single phase model, the two outputs are in synch, and are literally paralleled. This is because with a double bus voltage arrangement, each output can go positive or negative with respect to neutral, so there is no need to connect neutral alternately to bus positive and bus negative. They could use a single half-bridge to generate the active output, but they seem to have chosen to use transistors with half the current rating in parallel instead. They get to use the same printed circuit boards for both single phase and split phase models.

Another advantage of this arrangement is that there is far less likelihood of weird problems when panels get wet and start leaking to earth. For example, the problem some Axpert users have reported with bus overvoltage faults and residual current breakers tripping.

Against these modest advantages a safety advantage that they don't seem to be talking about, there is the expense of double voltage silicon, and a large number of physically large and moderately expensive electrolytic capacitors. This is possibly a major reason that the Infini models are generally considerably more expensive than similarly rated Axperts; the circuits are far more complex, and the components more expensive. I think it makes the Infinis more bulky than the Axperts, too.

Thank you for a great answer and explanation. 

On 2023/02/02 at 9:40 AM, Coulomb said:

I just got the repaired inverter back into position today; there is so much gear packed into the metal case that it's a pig of a job. There is still an hour's work to wire it in. It was late, so I decided I'd just turn on the already-wired inverter that has been running the house on its own for the past week, and blow me down it did the ker-clunk routine. Maybe something got dislodged, but it sure looks like the symptoms that BritishRacingGreen has been running into with the 1000 μF capacitors drying up.

I'm wondering if we'll see a even more of a spate of these as these cheap capacitors reach their end of life. Though I have to wonder about the other dozens of electrolytics. Maybe they don't dry out as fast because they don't suffer the same repetitive surge currents that the high frequency power supply capacitors do.

At least it had the decency to hold on until the other inverter was repaired. I'll just have one night of utility power, and it was a poor enough solar day that the battery would not have lasted the whole night anyway. I wonder if powering it up causes a bigger surge than running continuously.

They dry out simply since it is very hot (to hot) in this area..I've suggested to add a FAN connected to CN12 (?) on top of the mainboard for efficient cooling of the upper left part of the machine.

 

Kind regards,

holm

thanks  colomb   i solved the problem

 

i took   the  wrong  repair  path

 

all the time   asking why the   cpu not turn  the dc to dc  

the  entire inverter  seems  ok  except the primary side  fets

the   cpu refused to   turn on  because there  was   fault  in  secondery caps

 

when i removed   all  fets   and turned  on     error 9  appear 

when i removed all   igbts and  fets   error 52  appear  and  another good thing    ,,,cpu tutn the dc to dc   for one second

 

this   tells  me that   sg   25xx working  cpu boaard ok  and some thing else is wrong

i  test all  things nothing bad

 

as  last   dispaired  attempt  i   changed the caps   and  fitted new  mosfets   when  i turn the device  i  was shocked  

that the   device  not  make  arror sound  

 

the  screen was turned all  the  way down  when i looked  at it i saw

a  230vac    

so lets  anylise  why that happend

 

the  device come in  for me to repair  with   good igbts  in secondery  and tottally  destroied   mosfets

i  failed to   thing  properly  it was   quite  logical to suspect the caps   right a way

what could blow  mosfets    if no  bad  commponent in secondery side  shorted 

 

since  all   parts  in secondery ok   the caps  are prime  suspect   

 

they pulled   large current   and  failed the soft start   to work  

may be  they  eventually   go   almost  shorted  and   stressed the dc to dc   to blow the mosfets

 

 

i  failed  to   analyse the failure  and lost   5  days  looking for  trouble  that  does not exist

 

thanks

On 2023/07/15 at 8:24 AM, BritishRacingGreen said:

He mentions an error 09  occuring, so If the DSP performs the bus soft start and if that test fails, then the DSP does not enable the battery mosfets . So I dont think there is a fault around the mosfet drives. 

true  that  what   happen   i  followed the  unlogical path .............could colomb  give  us   the  checking process order

 

 

i know   the   inverer checked  for fan first   what else it does   if we know  the order we know  how to trouble shoot

 

On 2023/07/15 at 4:41 AM, Coulomb said:

14 VDC with respect to what? Perhaps you are measuring with respect to -12V?

The opto coupler circuit is very simple. There is no way you should be seeing 14 VDC with respect to earth at CN11-6. Maybe with respect to -12 V, but then the opto or R65 would have to be faulty.

Perhaps you are talking about the transistor side of U19? If so, I assume that Vref is about 5 V positive with respect to -12 V, so around -7V with respect to earth. I still can't see how you get 14 V.

That 14 V has to be your problem. Have you checked R65 and the diode drop across the opto coupler? I'm guessing about 0.9 to 1.2 V would be normal (with a multimeter, not when conducting hard).

thanks  for taking the time     i actuallly  followed the wrong repair  path  the problem  was in soft start  and   a  case of  almost   shorted  cap that  alert the cpu   to  send 09  error  with the caps  repaired  all things work ok 

So it was the two large bus capacitors? They had gone very high leakage, nearly short circuit?

13 hours ago, Coulomb said:

So it was the two large bus capacitors? They had gone very high leakage, nearly short circuit?

they appear ok  ,,,as  i checked  all things    and as  i  removed  all  fets  and igbts  still  cpu only for  a  fraction of second  does  give the fet and  active rectifiers  signal  drive  i suspscted the caps  fail the cpu  test

 

when i   replace them  the unit work  ok   but   i have small question 

 

 

the   newly installed  caps  are  400  volt ....will they endure  the orginal one were rated 500

another  question  mppt  board   does charge those caps  the man install 6 seiies solar panels  around 250 vdc  voc

what voltage you would  expect across the caps  once the mppt   works  and try to charge them i imagine more than 400?

as mppt will try  and boost the 250 to 400vdc  so that h bridge can turn it to 220vac

-

last queston 

 

during the repair  i sus[ected one 500  cap is bad  insteaad of  replacing them i decided to  remove  the b+  bus  wire  that connect   the mppt  board to the   inverter board  and when i turn on the inverter  fried 2 power igbt and  boost diode

 

those parts   are some what connected to the   bus  but what they would short  like that  ??  they are not in switcing state  ?

they are  for mppt  and  i not connect the inverter to sun or any thing its  just 50vdc  power supply?

 

 

 

Edited July 18, 20233 yr by wael_fathe

I think it isn't a very bright idea to change the 500V caps against 400V.

The command QPIGS from my PIP4048 displays a BUS Voltage of 422Volts currently...

 

Besides of that, those HV caps are almost never..ever bad.

 

Regards,

Holm

Edited July 19, 20233 yr by holmoe
addition

8 hours ago, holmoe said:

I think it isn't a very bright idea to change the 500V caps against 400V.

The command QPIGS from my PIP4048 displays a BUS Voltage of 422Volts currently...

 

Besides of that, those HV caps are almost never..ever bad.

 

Regards,

Holm

there is no  slight  doubt  thhat they  are bad   as the moment  i changed them  the unit works ok   the 400  units  were  just  temporary  fix   thanks  but  the  fact  that they not  go bad is just  according to you own  experience  

On 2023/07/19 at 9:05 AM, wael_fathe said:

the   newly installed  caps  are  400  volt ....will they endure  the original ones were rated 500

I would definitely replace them with at least 450 V rated parts, but 500 V if you can possibly find them. Even though you might only need 425 V now, one day the customer will add extra panels forgetting this repair altogether.

On 2023/07/19 at 9:05 AM, wael_fathe said:

what voltage you would  expect across the caps  once the mppt   works  and try to charge them i imagine more than 400?

Yes, the MPPT will try to boost the panel voltage to at least about 380 V, and if the panels are at say 410 V, you'll get very close to 410 V on the bus. A boost converter can't buck even a little bit.

On 2023/07/19 at 9:05 AM, wael_fathe said:

during the repair  i sus[ected one 500  cap is bad  insteaad of  replacing them i decided to  remove  the b+  bus  wire  that connect   the mppt  board to the   inverter board  and when i turn on the inverter  fried 2 power igbt and  boost diode

those parts   are some what connected to the   bus  but what they would short  like that  ??  they are not in switching state  ?

That's pretty weird. All I can think of is that the MPPT input capacitors retained some charge, and there was enough energy stored in them to destroy the boost transistor and diode when the MPPT connected and there was no load on the MPPT, so nowhere for the inductor energy to go. Also, the DSP would be using its own bus measurement to decide how much boost the MPPT needs, and since it saw no bus voltage rise, it commanded more and more boost.

That's pretty weird. All I can think of is that the MPPT input capacitors retained some charge, and there was enough energy stored in them to destroy the boost transistor and diode when the MPPT connected and there was no load on the MPPT, so nowhere for the inductor energy to go. Also, the DSP would be using its own bus measurement to decide how much boost the MPPT needs, and since it saw no bus voltage rise, it commanded more and more boost.

 

the soft start circuit does  exist on the mppt  board  conifirmed  ...maybe  when i seperate  the   cables  there  was not enough capacitance to    mooth  the resulting bus  voltage a nd   thus that damaged  the booost   diode and booost  transistor igbt>?  could be  a possibility?

13 hours ago, wael_fathe said:

there is no  slight  doubt  thhat they  are bad   as the moment  i changed them  the unit works ok   the 400  units  were  just  temporary  fix   thanks  but  the  fact  that they not  go bad is just  according to you own  experience  

Yes, that's my own experience..but not only from repairing some inverters, I'm repairing industry-related electronics like VFD for motor control  to make a living, that since 2007.

Before that I was Unix system programmer and administrator. I only had exactly one (1) bad HV Cap from Siemens-Matsuhita in a large spindle drive of a CNC milling System.

So you can have your own opinion, but I have mine.

 

Regards,

Holm

 

 

Edited July 20, 20233 yr by holmoe

@wael_fate:

Check the Gate drivers and their power supplies from the bad Mosfets.Test the drivers and look at their output waveforms before you put any load on them. How this can be done is described in detail in the large Thread at https://forums.aeva.asn.au/viewtopic.php?p=97123

 

Regards,

Holm

On 2023/07/20 at 5:06 PM, wael_fathe said:

the soft start circuit does  exist on the mppt  board  conifirmed

Oh. Well that could explain it. If the bus soft start power supply is located on the solar charger, then it will generate high voltage from the battery, and the DSP will stop it when the voltage it measures on the bus is high enough. But if you cut that connection, then it would not know when to stop, and the bus soft start power supply could apply excessive voltage to the solar charger circuit. The voltage would also rise very quickly without the bus capacitors to slow it down.

On all models that I'm familiar with, the bus soft start power supply is located on the main board, not the solar charger board. I think that MAX models have two power supplies on a smallish printed circuit that is attached to the solar charger board; one of these may be the bus soft start circuit. Though I can't think why they'd do that.

On 2023/06/21 at 2:38 AM, Coulomb said:

That might be 90VA and almost no real power. I don't know if the Infinis calculate power properly, I assume that they would, since Axperts do. If you're basing the 90 "watts" on utility current measured by a clamp meter, that's almost certainly the case. 

It might also be powering losses, which get powered by solar during the day. Some oddity of the circuit might cause that to come from utility at night. 

Nope, no clamp meter of any sorts.  I base my statement only on the EmonCMS monitoring graphs that show numerous little ridges at night that disappear during the day.  They look a lot like fans running for a short period of time, which corroborates your theory of internal powering losses.

 

 

A number of members have already mentioned that their inverters display the value 120V for the battery , some permanently , others at times. Please if you experience this , switch that inverter off and get it fixed.  I  was on the receiving end today of a catastrophic failure .

I have been requested by a member to check out  a faulty 5kW Axpert clone inverter  that displays the value of 120V for the battery voltage. The inverter raises the error 03 (battery voltage too high)  and of course switches off after a number of seconds. I looked forward to this repair as I am a bit tired of igbt/mosfet failures and driver repair  , I mean how difficult can this one be ? 

@Coulomb has made on number of occasion mention of the measurement resistor strings that can and will go open circuit, So that was the first check . But all resistors is fine . I shorted the battery bus and measured 8 Mohm ( 4+4) on the 2 pins going into the DSP board.  Fortunately I obtained a second DSP board from someone I know ,  but the same 120V and error 03.

So logic instructed me to verify the 5V , 12V and -12V that feeds into the DSP. And guess what , the -12V read +1.23V ! Oh dear , I check on the -12V linear regulator (7912 type) and measured the same . Removed the 3 pin device , check for load resistance not shorted,  and replaced with a new one. Switched on , and the inverter was fine , producing 220VAC from my current limited PSU , no grid , no PV. Voltage reading on display matches the actual 'battery' voltage.

This was way too easy , and while I was satisfied with the quick fix , I was worried why a supply line could go faulty like that ? , without a short circuit on its load output . I am so use by now that when I bring up a 5kW Axpert via PSU , the device draw just short of 1A at 52V VDC when it start inverting 220VAC output , no load. And same  here , no excess power requirement.

So I decided to let the machine burn-in  and I kept it on . The 7912 gets warm on its little heat sink , but I recall this as typical. After 20 minutes I was so confident that I posted a nice whatsapp to member with display image , showing his machine in operation.

After about an hour I decided to switch off , and back on , just for the good order. Perfect , but after a couple of minutes the machine suddenly died silently  , fortunately without an explosion , but the PSU went into hard current limiting mode. Short on the battery bus.

3 x  MOSFET short circuit , 5 x IGBT short circuit .  some drivers gate resistors blown open circuit  Wow !!! 

Turns out that this -12V supply of ours might not be as mission critical as the 5V and 12V , but it feeds the  3525 PWM controller ic  , which in turn drives the mosfets and the battery side igbts.  The problem is it does not feed the circuit on its own , but is combined with the 12V in order to affectively provide a 24V supply for the 3525 . This is somewhat nasty in that should the -12V fail  with a reasonably low impedance wrt to GND , the 12V rail can still power the 3525 , but at out of spec conditions. 

I am of the opinion that either the 3525 or its load circuits was failing , causing the -12V to sag . Meanwhile this same -12V is used on the DSP to power one side of the measurement op-amp , this causes a so called high battery voltage , and the DSP fortunately switches off, albeit all for the wrong reason.

So  the second time this rail failed while the inverter was inverting , it was less forgiving , it destroyed the power mosfets and igbts.  Here I am of the opinion that the failure mode of the -12V  caused the 3525 to limp with the +12V , but the out of spec pulses to the mosfets was non-deterministic . It turned the mosfets on at the wrong times , or unable to turn them off in good time.

Note that if the -12V regulator is already damaged , there will not be catastrophic failure on startup. This is because error 03 will be generated before the DSP enables the 3525 PWM ic.  So on face value this looks like a simple fault on the measurement side , but the catastrophic failure is hidden and abstracted from us , until you 'rectify' the 'fault' . Very unforgiving. Repairing of high power  Inverters is not for everyone. It can break your spirit at times. 

Will investigate whether this failure is viable to repair  , but that is for another day.

 

 

 

 

 

 

 

Edited July 29, 20232 yr by BritishRacingGreen

& so the journey continues...Live to fight another day & not all was lost. A failure of a component by another component can lead to something. But yes I agree it must be hard to swallow. 

Very intriguing read. 

Thanks @BritishRacingGreen, I had exactly the same 120V battery voltage fault on my old 5kVA clone. I think it was caused by running my panels in 3S with 136.6Voc. It would always trigger the error in the evening or early morning when the PV was switching on and off in low light and cold temps, so i figured it was the panels.

At the time I was told it was probably my gel batteries because "the plates can bend and short circuit causing high voltages" which turned out to be a load of hogwash causing me to waste another 22k on a new lithium battery.

Could high pv voltages cause this failure at all or would that be more likely to fry the mppt board?