Shadders

@BritishRacingGreen yes I do repair units but methinks Harare, Zimbabwe might be a bridge too far for you 😁

I have found that water ingress isn't too bad unless it has burnt tracks. I find the biggest issue for me is the DC caps, they are not easy to come by and my parts supplier can't source them. I've had to pull from several old machines to get good enough ones.

If the machines are less than 8 months old and weren't running at high power conditions you will most likely get away with the repair not having to replace the caps. Message me if you need any help and I'll try my best.

You don't want to try only replace the faulty ones, it's asking for trouble. I unscrew the IGBTs from the heat sink using a Z screwdriver and then desolder them. Removing the heatsink is a lot of effort. You might have to remove a capacitor or two to get good access but I normally manage without having to do so. Once the faulty componenst are off check the DC mosfets and all driver circuits again. To be honest if the inverter is more than a year old I don't recommend repair unless you replace the DC capacitors as well because I've had several repaired inverters fail again soon after repair. They passed full load tests so I can only think that capacitor aging is not buffering the surges well enough.
 

@Shadders this is not the soft start circuit. Values are correct. Here in attachements the soud of TX9 after powering on the circuit via the power button. Also D54 get very hot just in few seconds after powering on via the power button.
31 mars à 00.18​.aac

Thank you @Coulomb and @Shadders.
Can confirm that this works on my Kodak 7.2 kW by following Shadders PDF and using the communication tool link in the same doc. 

I shared the SOP I wrote up using the information linked above
 

Hey, neat! I have linked your file at the top of the calibration post. Thanks for doing that; it was sorely in need of a clean up.

Check the resistors on the gate of those IGBTs. Do you have an oscilliscope to check the driver circuits?

The bus softstart circuit might also be damaged, a quick check would be if the MOSFET driving the primary of TX2 is short circuit.

I use a high voltage IGBT/transistor/LED tester on the IGBTs from the DC-DC circuit to the H-bridge. I have had some pass the basic multimeter test but their break down voltage is out of specification when tested. It's saved expensive failures of replaced parts. See below of failure vs pass. (it's also very handy for checking zener diodes). It was USD20 odd from Aliexpress. EEVblog forum has some technical discussion on it here https://www.eevblog.com/forum/reviews/completely-non-sketchy-mini-high-voltage-transistor-tester-(emeco)/

Check the resistors on the gate of those IGBTs. Do you have an oscilliscope to check the driver circuits?

The bus softstart circuit might also be damaged, a quick check would be if the MOSFET driving the primary of TX2 is short circuit.

I use a high voltage IGBT/transistor/LED tester on the IGBTs from the DC-DC circuit to the H-bridge. I have had some pass the basic multimeter test but their break down voltage is out of specification when tested. It's saved expensive failures of replaced parts. See below of failure vs pass. (it's also very handy for checking zener diodes). It was USD20 odd from Aliexpress. EEVblog forum has some technical discussion on it here https://www.eevblog.com/forum/reviews/completely-non-sketchy-mini-high-voltage-transistor-tester-(emeco)/

Check the resistors on the gate of those IGBTs. Do you have an oscilliscope to check the driver circuits?

The bus softstart circuit might also be damaged, a quick check would be if the MOSFET driving the primary of TX2 is short circuit.

I use a high voltage IGBT/transistor/LED tester on the IGBTs from the DC-DC circuit to the H-bridge. I have had some pass the basic multimeter test but their break down voltage is out of specification when tested. It's saved expensive failures of replaced parts. See below of failure vs pass. (it's also very handy for checking zener diodes). It was USD20 odd from Aliexpress. EEVblog forum has some technical discussion on it here https://www.eevblog.com/forum/reviews/completely-non-sketchy-mini-high-voltage-transistor-tester-(emeco)/

A common low tech method is to tie a PET bottle (with some sand or weight so it just floats) on a string inside the tank with the string running through a hole to the outside. Length of the string = height of water tank. Tie a small weight (plus maybe visual aide) at the end string. When the tank is full the weight/aide is on at the bottom, when it is 50% full the weight is halfway and when the tank is empty it is at the top.

***DISCLAIMER: Use at you own risk. You may brick and/or damage your batteries
How to upgrade firmware of the Pylontech batteries
If it works okay, do not touch it! If it does not work okay, contact your dealer. If your dealer is not helpful, contact Pylontech support. If Pylontech support is not answering, then you can try to upgrade the firmware. Connect laptop PC to the Pylontech battery:
1) First, you'll need to make (or purchase) a serial cable in order to connect a laptop to CONSOLE port of the battery.
Older models of Pylontech batteries are using RJ-11, while newer models are equipped with RJ-45.

Wiring on the right is suitable for all the new models, including US3000C, US5000C, Force H1 and Force H2:

2) A lot of people are struggling with making a working cable, because in some versions of Pylontech user-manual there's a missing information on the GND pin for the RJ-45 console port. Other people are unable to connect since they swapped TX and RX. So, here's the actual pinout of console port for RJ-11 and RJ-45:

3) Grab a Windows laptop PC equipped with a physical DB9 serial port and connect it to the battery stack via the cable above. As an alternative, you can use cheap USB-to-SERIAL converter, for example FTDI-based.
WARNING: Console port is RS232, with positive and negative voltage levels. Therefore, you have to use true RS232 serial-port interface, NOT UART 3.3 or 5V!
4) Download and unpack Pylontech_Tools.zip from the link bellow.
The password for the ZIP file is: Youda
5) Start the BatteryView software:
For batteries with a very old firmware, BatteryView 2 works the best. For new batteries, use BatteryView 3.0.28 or newer. Select the respective COM port and use 115200 baud-rate 6) Now you can perform diagnostic tasks, or update battery's firmware.

Updating Pylontech firmware:
1) When updating firmware, the best is to power-down whole stack, remove all the LINK cables between the batteries and then turning-on just one battery at a time and perform the FW upgrade on it. Then repeat the process for the next battery. Updating batteries while online in a stack works too, but you will get alarms and red lights.
2) There are several models of Pylontech batteries and the firmware is INCOMPATIBLE between most of them:
If you flash a wrong FW in the battery you will brick it. FW numbering is INCONSISTENT between the models. Fox example: For an old battery with certain PCB the FW2.4 might be the most-recent, while for a new battery with a different PCB and chipset the most-recent version would be FW1.9. In other words - higher number does not automatically mean that the firmware is newer, nor better! 3) If possible, it's preferred to update FW via BatteryView 3.0.28 while using following rules:
For updating US2000C, US3000C and US5000 select the whole ZIP file that includes two BIN packages inside and perform update. The BW3.0.28 will be able to pick right BIN file inside the ZIP package automatically. DO NOT select BIN package manually. When updating US2000 and US3000 then you must select the correct BIN file manually, for a shame. 4) Due to the silicon chip shortage Pylontech changed the BMS chip for some of the produced batteries. Therefore, for some models there are two different firmware branches. One for the original chip and the other for the new chip. Luckily, when upgrading FW via the ZIP method desribed above, the BW3.0.28 will choose the correct branch (BIN file) automatically.
 

5) This list indicates firmware version suitable for the each model and what file to flash:

Model: US2000plus
FW: V2.9
FLASH: us2000b_v2.9_Crc.bin
Model: US2000plus95
FW: V3.4
FLASH: us2000B_Plus_V3.4_Crc.bin
Model: US3000
FW: V3.4
FLASH: us3000a_V3.4_Crc.bin
Model: US2000C (original chip)
FW: V2.8
FLASH: NT1.7+2.8.zip
Model: US3000C (original chip)
FW: V2.8
FLASH: NT1.7+2.8.zip
Model: US2000C (new chip)
FW: V1.7
FLASH: NT1.7+2.8.zip
Model: US3000C (new chip)
FW: V1.7
FLASH: NT1.7+2.8.zip
Model: US5000 (original chip)
FW: V1.3
FLASH: US5000 ST+NT 1.3.zip
Model: US5000 (new chip)
FW: V1.3
FLASH: US5000 ST+NT 1.3.zip
 
6) If you have a bricked battery, you can use Pylontech Upgrade Tool V1.0.9 from the Pylontech_Tools.zip to recover it via flashing a correct firmware. The process is as follows:
Connect the debug cable to this software and the battery (attention: at this time please do not switch the battery on), then 1. open the software, click Immediate Update. 2. Click Connect. 3. Click Browse to select the correct firmware. 4. Then click Program and switch on the battery by hard switch and the red soft start button immediately. This will bring the battery back to normal. 7) When updating firmware, it's the best to turn-off all the batteries in the stack and remove all the LINK cables.
Then power-on a single battery and perform FW update on it.
Repeat for the remaining batteries in the stack.
Reconnect all the LINK cables and start the stack as normal.
Although it is possible to perform FW update while the battery is running in the stack, you will get alarms and red lights when you'll do it that way.

8 ) Firmware packages mentioned above are packed in this archive:
Download and unpack the ZIP budle from the link bellow.
The password for the ZIP file is: Youda
 
Youda

Pre-paid meters (at least in Zimbabwe) come in two parts; the main part that is connected to the utility directly and power runs through to your house/loads and a second human machine interface (HMI) unit that allows you to enter recharge codes, check balances etc. The HMI is typically plugged into a free socket and communicates via the socket wiring back to the main unit. You do not need the HMI connected at all times but to check balances, enter voushers etc it needs to be connected to the main unit. When an inverter is inverting and supplying power, it may or may not break the connection back to the utility depending on inverter design and/or anti-islanding requirements. I suspect this is why the contractor said you need an Eskom only socket.
Hopefully your installer included a bypass switch for your inverter so that you can remove the inverter from circuit and run solely on Eskom? This would accomplish the requirement of having a Eskom only socket and allow communication between HMI and main unit when it's needed. It is a good back up to allow your system to run while the inverter is being maintained or if there is a failure. You might want to have an Eskom only socket fitted near your utility connection to leave the HMI plugged in for ease of use?
Regarding your new battery:
Double check warranty and wiring regulations regarding fusing and battery connection. Some manufacturers require independent fuses and connections to a bus bar per battery I would typically recommend that the batteries are turned on before the inverter Other steps all seem fine. You should also check the maximum allowable charge current to the battery. This may need to be adjusted.

As basic as it sounds, I've had a lot of random errors caused by dry/aged capacitors on the battery SPS circuits. If the inverter in question is more than 3 years old the capacitors will likely need checking and replacing.
 
The manual looks like the unit is a VMII design. For the VM models the bus soft start circuits are on the solar controller board so when testing startup/inverting this has to be plugged in...when I first worked on one I spent 2 days looking for non-existent faults because I didn't have the board connected.

Thanks for the quick reply @Shadders,
the inverter is 1 year old, it has about 500 hours of Inverter work so far, mostly the complete system is switched OFF, the batteries are charged via PV as needed when/and only when arriving at the location. Only then do switch the system to All-ON. The cottage is located at the edge of the forest, a rather remote location, the cottage is used approximately once a week.
So I doubt that the electrolytes are already dried out, but I will still visually inspect/Checking/measure them on the PCB.
A friend (the owner of the cottage) checked the inverter's error history yesterday, the Error code 05 ([05 ]Output short circuited or over temperature is detected by internal converter components) also appeared (only once), and that was when the system was connected to the generator set, there was not much PV energy due to bad weather, but used a generator to supplement the batteries and enable the operation of the refrigerator and lights (tot 500Wmax, so it was not Output short circuited , so more likely an Overtemperature fault. As Input Utility, the generator enabled Inverter Bypass mode and additionally charged batteries, solar energy was not available => Program 01 SOL!
So here I see a similar problem around DC<->AC , Utility provides Bypass to OUT 230VAC, and we have energy transfer AC<->DC=>DC<->DC=>Charge batteries, so again I suspect AC<->DC IGBT Gate drive problem like @BritishRacingGreen mentioned, not correct/insufficient Gate drive/Gate ON/OFF times, Overlapping ON/OFF Gate times/too short DeadTime due to bad Gate drive/IGBT Id current in FullBridge on one or both half bridges partial current pass through , so I suspect a problem here as well around this Error code 05 like IGBT Overtemperature.
Utility charging was set in Program 11 to 15A (insted 25A default), and in Program 02 Maximum charging current: To configure the total charging current for solar and utility chargers, was set to 50A(default for PWM)
LP
Dragan

Dry Joints and Axpert Error 06
Good day, hardware hackers. It’s only human not to share all of our failures on a public forum, but this one I need to share with you. I recently repaired a 5kW Axpert that had the dreaded Error 09. While there are worse traumas in life than Error 09, it is indeed problematic. Invariably, you need to replace 40-60% of the power silicon (MOSFETs/IGBTs), and it’s almost guaranteed that some gate drivers are faulty as a result of the big bang.
I repaired the machine following my methods to bring it up in a deterministic 'soft' manner, and the initial test/verification went very well.
However, a few days after the repair, I used the same machine to test a Pylontech US3000, which had a fried BMS power chain. About half an hour later, the Axpert displayed Error 06. I reset the machine, and either Error 06 appeared immediately again, or it would fail after a random period of time, which could be as long as one hour. Error 06 indicates that the AC output voltage is too high. Since there was no Bus Voltage Too High error, I suspected that the DC-AC full bridge might be at fault. This wasn’t due to clever deduction; rather, I knew that this bridge had been severely zapped during Error 09, resulting in the replacement of some gate drivers and gate resistors. Even though the full bridge was not faulty most of the time, I suspected the quality of the gate drive performance.
I decommissioned the machine and disassembled it. The first thing I typically do is test the 47Ω gate drive resistor path from driver output to the IGBT gate pin. Guess what? One of those four paths had high resistance. It was a dry joint or a faulty resistor. The annoying thing was that it wasn’t open circuit but had an unstable value between 300Ω and 150kΩ (!!!!). I tested the resistor, which was fine at 47.2Ω, but I noticed a terrible solder joint on one pad. I tell you, I tested this circuit for continuity during the Error 09 repair. Nonetheless, I repaired the joint and checked all four circuits again. I reassembled, tested, burned in for 24 hours, and the Error 06 was gone.
So, why the Error 06? Here’s my theory: We know that the IGBT gate has parasitic capacitance, which influences the switching times due to the RC time constant, where R is the gate drive resistance. The smaller we make the gate resistance, the shorter the gate switch-on or switch-off times will be. This explains the typical gate resistor values of 22Ω - 47Ω. We also notice in typical gate driver schematics that the driver DC supplies, referenced to the IGBT emitter, are asymmetrical (+15V and -5V). This means that the turn-off delay is somewhat longer than the turn-on delay in practice, because the gate voltage must discharge from +15V down to about the gate threshold voltage of 2-3V. This takes longer than switching the gate on, which requires raising the gate voltage from -5V to about 2-3V.
This difference becomes significant when the RC time constant is quite large. The net effect is that the intended PWM duty cycle, controlled by the DSP controller, becomes distorted at the gate of the IGBT. As a result, your duty cycle ON periods will be longer than intended. This, in turn, causes the integrated voltage on the collector of the IGBT to become increasingly higher, leading to the “output voltage too high” error.
I also believe that this is one reason why some IGBTs with large gate capacitances require a resistor anti-parallel arrangement. An additional resistor is connected in parallel with the gate resistor, but via a steering diode. The polarity of the steering diode is chosen to lower the net gate resistance when the IGBT is switched off (gate discharged).
So, how do we test all this in practice? The first option is to have a dual-channel oscilloscope, with one channel connected to the DSP control signal and the other connected to the IGBT gate. However, these two signals are not in the same power domain, so I don’t typically go this route. One could connect the grounds together via suitable high resistors, probably in the order of 500kΩ - 1MΩ, but I haven’t tried that yet. The second option is to use only one channel and monitor the voltage across the gate resistor. This waveform represents the current through the gate and provides a lot of information.
An example waveform is shown in the oscillogram below:
 

 
The actual gate turn-on occurs during the positive-going pulse. This current pulse results from charging the gate capacitor to the 15V voltage level. The gate turn-off occurs during the negative-going pulse, which relates to discharging the gate capacitor. In this waveform, the period between on and off is about 6µs, so in practice, you need a 40kHz square wave to produce this as a test. Neither the gate charge pulse nor the gate discharge pulse plays a significant role due to their short durations.
What I do is remove the DSP controller and inject a square wave on the relevant control pin of about 40kHz. Then I check the current profile as shown above. Under circumstances where the gate resistance is out of spec, you will see quite wide charge and discharge pulses.
Cheers, and happy hacking!

A lot of errors can often be caused by dry and aged capacitors on the internal power supply circuits. Replace them and the unit can be returned to service. If the unit is more than 18 months old this is often the case especially if the error doesn't make sense in the installation. A quick check is to remove the cover and have a look at the top righthand side and see if any of the capacitors are swollen. Not a 100% guarantee that this isn't the problem if they are fine but if you see a swollen one you know its a problem.
As BRG stated it's mostly enthusiasts doing board level repairs. Errors that typically require major repair works and IMO probably don't justify the repair on an economic basis are:
09- bus soft start failed (usually battery mosfets, output IGBTs)
08 - bus voltage too high
32 - no coms between screen and main CPU (often damage to bus softstart circuit)
 

A lot of errors can often be caused by dry and aged capacitors on the internal power supply circuits. Replace them and the unit can be returned to service. If the unit is more than 18 months old this is often the case especially if the error doesn't make sense in the installation. A quick check is to remove the cover and have a look at the top righthand side and see if any of the capacitors are swollen. Not a 100% guarantee that this isn't the problem if they are fine but if you see a swollen one you know its a problem.
As BRG stated it's mostly enthusiasts doing board level repairs. Errors that typically require major repair works and IMO probably don't justify the repair on an economic basis are:
09- bus soft start failed (usually battery mosfets, output IGBTs)
08 - bus voltage too high
32 - no coms between screen and main CPU (often damage to bus softstart circuit)
 

Updates
 
First the sg get fairly hot and that is normal ...at working fans. Are directly facing sg and the temp never get sensed well using ur finger
 
When we do columb trick of shorting copluer u19 no fan will spin and that is where ur finger sense the temp
 
 
Another update ...use columb trick when ever possible...even if all component test good u will burn things ..here is. A case ..
Where one leaky diode test good in forward...and taste some leakage of 2.2in diode mode...i thought it be some partallel component
 
But once diode out of board i taste in both direction one gave 0.6 ok in other it gave 2.2. Bad ...good diode never test in both direction
 
My testing skill failed...once i apply trick i notic on set of mosfets give this cut waveform...which led me to test all components related to thses fet extensively and precisly and out of board
Columb trick is Day saver 

Try looking at these threads:
or
 

I've seen -100° appear on the battery page when the BMS coms to the inverter are not present, in fact it's the way I use to check that the coms is working correctly. I suspect that for some reason you have maybe momentarily lost had an interruption or corruption of the data between BMS and inverter. I have no clue why your system would be fine for two years and run into this problem though. Maybe new equipment or old equipment failing nearby that produces interference? Or possibly an aged filter capacitor in the BMS or inverter coms circuits?

This is not just an YATV (Yet Another Teardown Video).  I for one has found it not only interesting , but valuable .People like myself and other members on this forum are predominantly focusing on the lifecycle of Voltronic products from a support point of view . But it is very important to know a bit more of the other brans eg. SunSynk , Solis etc. It provides a gauge and insight to whether one product is superior to another , and if so, what can we qualify as being the reasons for that . So I am going to comment on this video in posts to follow how I see SunSynk vs Axpert as an example , and in the process trying to be as objective as we should be.
@Steve87 has put me onto this link of the video . It is courtesy of the folks from evblog.com . I have actually watch numerous videos of their 'video-man', and finds the guy very interesting .  He always grabs my attention and I tend to watch all his video from start to finish as a result. His voice is high pitched at times , I think he uses a mixture of helium and oxygen to breathe , else I think his underpants may well be a size or two too tight!
One of his members has gone so far as to trace out a basic schematic of the powerchain which I find very interesting . Firstly because it is exactly the same topology as per Axpert , except for two small differences which I will cover in posts later on.
Also interesting  is that he finds that the machine under teardown employs generic Chinese silicon , electrolytics and relays , as opposed to Deyes specific claim that they use Western specified components. In the video comments you will notice Deye themselves has commented that they were force to use whatever  was available post-covid period. 
 
 
Additional resources from evblog here: 
https://www.eevblog.com/forum/blog/eevblog-1620-deye-solar-hybrid-inverter-extreme-teardown/
 
 
 
 

Good news today..
The company gave me full schematic for both battery and mains sps  to help fix dead mains sps
I will ofcourse put them for benfit of good people ... battery sps  was really normal output of 150vdc...no doubt now ..
By the way this is the most clone design i have met..if well understood will open the gate for understanding alot of other makes and models
@Coulomb
@BritishRacingGreen
 

Just a bit of feedback that the voltronics VM (II, III and IV) range has this same design.
With the control board removed I usually check for shorts on the power rails, then check the rails and the -5.4V on the output IGBTs gates by flicking the power on and off after injecting 30V at D74. I then test mosfets and DC-DC by powering +/-12V rails through the CN10 pins (not the same as Maxo's schematic), triggering the SG3524 by shorting U19 output and then checking all the gate signals. This can be safely done even with the mosfets and IGBTs installed as the battery and DC buses are unpowered.

It's been a while since I checked up on the forums so if there's been any testing improvements that I'm missing don't hesitate to point them out. I've recently had a bunch of machines with SPS failures so thanks @BritishRacingGreen for your recent test procedure it will be very handy!
*caveat: I use the component designators from Maxo's schematic, they are not all the same on the VM models so check the circuit before using any information posted above (one of these days I hope to start work on a schematic)*

You can get a lot of general information at OpenInverter.org:
https://openinverter.org/wiki/Main_Page
Battery voltage depends largely on power.  Small scooters/golf carts will usually use 48V. Low power city cars 96V.  Most full size EVs 400V. Hypercars up to 1000V.
(Some 400V systems are 2 bank series/parallel arrangements - discharge in parallel, charge in series to reduce charging currents.)
Small inverters are free air cooled. Intermediate ones fan cooled. Large ones water cooled.
Output voltage is effectively around Vdc/sqrt(2) RMS.
Yep - no filtering - only snubbers to kill the worst of the HF components.
Most controllers are torque demand, so would be duty cycle controlled with frequency limits.
Most have peak 98% efficiency, down to 96% or so under worst case conditions.
No physical standards on the control ports (yet), although there is some standardisation happening at the logical (CAN/LIN) level - but not wide spread yet.