kuba.cz

This is RS485, not RS232, also the pinout and the protocol is different. It won't work.

I would check for earth leaks on the PV panels. Or generally any leaks between the panels if they are not earthed. It's only one of many possibilities, but easy to verify.

This looks similar to Anern. Both Anerns that I have are not user-updateable. They don't have a bootloader so they can't be updated through the serial port. It could be possible to replace the DSP on the control board with a new (unlocked) one and flash the firmware using JTAG, but unfortunately I don't have any firmware for this particular model.

They are on the main board, on pin 15 of the 50-pin IDC connector. But I don't think that any Axpert firmware can work with this board. The ADC and GPIO connections are completely different. And most likely the chip does not have a bootloader anyways, so it can't be updated through the serial port. The only reasonable option in this case is to get the firmware somehow (from the manufacturer?) and flash it into a new unlocked chip using JTAG.

It's a link. Just click it.

Error 99 means no bootloader. The firmware checks for a magic number at a specific address in the bootloader. The address may differ among various models. Some models do have a flag for grid-feed in the EEPROM settings. Use Communication Tool to verify/change it: QFLAG command to check whether the flag is there. Search for a lowercase "d" letter in its output. If it appears after "E", the flag is enabled. If it appears after "D", it's disabled. PEd to enable solar feed to grid. PDd to disable solar feed to grid.

Some more schematics here. These are from Easun SMW 8kW which is, I believe, identical to Axpert Max 8kW. In this case the control board did have part references printed so most of those in the schematic should be correct. It's a multilayer board but only power planes are on the inside layers so it's not that hard to trace. The main difference, compared to Anern, is that this board uses multiplexers for 8 of the ADC's inputs, increasing total number of available analog inputs to 24. I also made a schematic of the SCR board and some fragments of the main board. I'm not going to make schematics of any power supplies as these are already pretty obvious. Just note that the topology in this case is different from Anern's. There's no intermediate SPS_IN (150V+) branch. The main power supply and bus softstart are located on the main board and powered directly by the battery voltage through a diode. The grid/AC power supply and solar power supply are on a separate board, both of these output around 63-66V and also provide power to the main power supply through diodes. Unlike Anern's, this topology does not need a relay in the softstart circuit. I made a note of some critical parts (MOSFET type, shunt resistor value and how they deal with voltage spikes): main SPS: JCS86N25WT, 0.15R, RC snubber 2x47R (parallel), 2.2nF softstart: SIF13N50C, 0.62R, RC snubber 2x22R (parallel), 220pF grid SPS: SVF9N90F, 0.47R, RCD clamp 15k/5W, 10nF solar SPS: SVF9N90F, 0.47R, RCD clamp 15k/5W, 10nF Original KiCad files and PDF schematics are available on my GitHub: https://github.com/solar-pwr/axpert/ There will be errors as usual, so every feedback is appreciated.

Hi, you can use ST-Link (or any other SWD probe) to reflash the chip with the right firmware. Check this thread for SWD pinout: https://powerforum.co.za/topic/14829-easun-axpert-max-8kw-firmware/page/5/#findComment-140679 No external power is needed, the SWD probe will provide +3.3V for the device. First of all, make a full backup of the flash memory. You'll need it to extract the bootloader which is not included in the upgrade file. I like to use OpenOCD for this but you can choose any other tool except STM32CubeProgrammer, which won't probably work with a GigaDevice chip. After successfully connecting, run the backup using telnet: The file will be stored in the directory where OpenOCD is running, /OpenOCD-20251211-0.12.0/share/openocd/scripts/ in my case. Size of the flash memory (0x40000) may differ in your case, check the datasheet of your chip. 0x8000000 is where the flash starts. It's the same for all STM32's and clones. Once you have the backup, post it here and I'll merge the bootloader with the proper firmware you need so you can flash it into the chip.

Most part references around SG3525 seem to be the same as in this schematic: https://forums.aeva.asn.au/viewtopic.php?p=88434#p88434 Got one on my desk too 😀

@Coulomb All schematics are here https://powerforum.co.za/topic/32491-anern-102kw-schematics/ in my first post. Part 2 was added later. I've just uploaded original KiCad files to my GitHub as well: https://github.com/solar-pwr/anern . BTW, yesterday, while working on an Easun SMH, I realized that this Anern is a rare case which can only measure charging current on the DC/DC converter's Hall sensor. The output of the op-amp is not shifted to +1.65V so discharging current will always be read as zero on the ADC. Easun SMH does have a voltage divider between the output of the op-amp and +3.3V (taken from a voltage follower, not directly from the power rail). I think it will be similar in all MAX models.

I think the only explanation is that battery 2 disconnects because of cell overvoltage. Otherwise it's not possible to see different voltage on batteries connected in parallel. Differences in internal resistance of cells might be the reason, causing balancer to operate the wrong way during charging.

I'm not sure what was the reason to remove the snubber, but the power supply seems to run fine without it as long as the output current stays within normal range. But when there's a short circuit (and that can be on the COM port also), the voltage spikes can be high enough to cause arcing on the transistor drain or transformer pins. If UC3845 failed, don't forget to check R253. The power supply may run even if this resistor is open-circuit, but UC3845 will stay at 50% duty cycle, just switching on and off according to the voltage feedback. In this mode, the spikes on the primary will be extremely high. Currently I'm running two inverters with RC snubber added between source and drain of the MOSFET without any issues (that's not an RCD clamp as show in the schematic). I used 100pF/3kV MLCC capacitor and 33R/5W resistor as already mentioned.

I only have firmware for two different 10.2kW models (single MPPT and dual MPPT).

It's in the schematic, 0R5, like 0.5 Ohms. 5W rating is just a guess, it's probably less, maybe 3W. You can also use two 1 Ohm resistors in parallel. If it's burnt, you need to check other components as well. Especially Q71, U10, R253, R250, R249, D49. You may also want to add an RC snubber or RCD clamp to prevent the same failure happening again (see my earlier post - the one with board photos).

These are two unrelated problems. The update via COM port fails probably because of a weak/slow optocoupler or some other hardware problem which corrupts the data. When uploading the firmware update via JTAG you get F99 which means "missing bootloader". That's expected. The firmware checks for a signature at the end of sector A where the bootloader resides. And it's not there because the firmware update does not contain the bootloader. When flashing via COM port, sector A is never erased or overwriten but the JTAG utility by default erases all sectors. I've merged the patched firmware with part of your original file. This one should work if uploaded by JTAG. Do not use this with the original update utility. dsp_374.04_patched_05_wboot.hex

Only low voltage (145V) models convert PV input to battery levels. On high voltage (500V) models, PV input goes to the high voltage side. Check the schematic here: https://powerforum.co.za/topic/14666-repair-of-axpert-inverters-a-journey-started/#findComment-138439 edit: Or better this one: The first one doesn't have a PV input, actually.

I've added schematic part 2/2 of the DSP board to the first post in this topic, hopefully it's complete now. Also I made some minor changes in part 1/2. Most importantly, R85 and R86 values in the inverter DC offset measurement circuit were wrong (10k), these are 1k resistors.

Here are some photos. Not great but I tried my best. To avoid confusion, I have to note that I've made some modifications on the main board. I have three of these boards (got them as not-working) and all of them had signs of arcing around the drain of Q71 or TX9's primary winding. On this particular one, U10 and R255 was faulty. There were some unsuccessul repair attempts from the previous owner. After replacing the parts (including Q71 which still measured OK), I decided to also remove R249/D64 and install the unpopulated RCD clamp to reduce the voltage spikes. Used 3.3nF capacitor for C143, 150k/3W resistor for R245 and some diode that I don't remember. That didn't work out, the same parts blew again after a few power ons (exactly when the soft start turned off and SPS_IN raised to the bus voltage). BUT most likely due to my mistake that I didn't check R253 which was already open-circuit. Wrongly suspecting the RCD clamp's diode to be too slow, I deciced to try an RC snubber instead, consisting of a 100pF/3kV MLCC capacitor and 33R/5W resistor, connected between source and drain of Q71. Only then, watching the waveforms on the oscilloscope, I realized that the duty cycle is constant 50% and the power supply keeps switching off and on by the optocoupler feedback. That lead me to the R253 1k resistor. After replacing it, everything works fine and the waveform on the drain of Q71 looks nice so I left the RC snubber there, not trying to revert to the RCD clamp. So that's the current state of the main board on the photo. Also I swapped the DSP, the orignal one was TMS320F28062PZT, this one is TMS320F28066PZT (differs in memory size), but this is only related to my experiments, not a malfunction of the inverter.

Not exactly 175V but I can see two limits related to grid voltage in 64.xx firmwares. The first one depends on setting 03. For APL, grid is considered "lost" at 90V and recovers at 100V. For UPS it's 180V/170V. Then there are the following limits for exporting to grid which are set depending on the country: lower limit upper limit India 195.5V/200.5V 253V/248V Germany/South America 184V/198V 264.5V/250V These are hardcoded in the firmware and cannot be changed by any serial command. The only way to change them is to modify the firmware, which would only be possible if we had the original one for this inverter.

Not yet. I'm still at the very beginning of drawing part 2/2, hopefully I'll be able to finish it later this year. But I do have the board on hand so I can provide some part values/pinouts if needed.

I checked 78.07 and 73.63 and both seem to use 1:7.3 transformer ratio (surprisingly for me - I thought they always used integer ratios). This means there should be around 473V on the bus while charging which is still ok and corresponds to what you are seeing. The overvoltage threshold which temporarily disables the boost circuit is set to 490V and responds within one PWM cycle (like 50 microseconds), which is good. Some other versions keep waiting for multiple cycles before turning the boost circuit off. Then there's another protection which throws the "Bus overvoltage" error if 500V is exceeded for more than 1 seconds. That should never happen as long as everything is working correctly. So it seems quite possible ("safe" is probably not the right word 😀) to try to increase the charging current gradually, even with 18S batteries.

If the transformer ratio of this inverter is 1:8, then charging the battery at 62V would ask for 62*8+20=516V on the bus, which is way too high all by itself. It will probably trigger the overvoltage protection, which is usually at 490V, turning off the boost circuit temporarily. I'm not sure about the exact thresholds in this particular firmware, hopefully I'll find time to look at that later. However, it seems that your inverter is already working on the edge of its capabilities even without the derating disabled.

I don't think these are interchangable. However I happen to have 373.63 modified already. Lada.cb asked me for this one about 10 days ago and reported it working correctly in this topic. Firmware file in the attachment. Thanks in advance, I'll have a look at that. dsp_373.63_patched_05_200V_derate_460V.zip

Okay, so due to high demand 😀 I'm posting it here. Hopefully it doesn't cause more harm than good. As always, make sure your battery (BMS) doesn't disconnect while charging - that seems to be the main risk here. Also I'd recommend to use at least two or three battery packs in parallel. Charging current over 100A would be a lot for a single 280Ah battery. dsp_278.07_patched_200V_derate_460V.zip

Not this one, but I've modifed 278.07 for another member too. I'll send it to you in a message. So far, I have no reports of anyone testing it. Consider it to be highly experimental.