Plaasjapie

I found this thread because I have exactly the same model of inverter (Voltronic Axpert MKS 5kVA) which exhibited the same fault condition. I want to say thank you for all the posts here because this helped me to fix my inverter quickly and it didn't cost me anything.
I am electronic engineer in South Africa with 30+ years experience in hardware and software design. I have also designed a 1kW switch mode battery charger using active power factor correction. I performed my own solar inverter installation around 5 years ago with the above mentioned Axpert 5kVA. It also has the same printing on it as per @Plaasjapie 's post (TheSunpays.co.za). It suddenly exhibited the same fault. I disassembled and found the two guilty capacitors C78 and C79. It would seem that this is a common failure point on this design.
From my analysis of the reason for failure of the components is either because they have insufficient voltage rating or they have insufficient ripple current rating, or both. The capacitors are used for smoothing on the output of what appears to be an auxillary power supply supplying +5V and -12V to the circuit. This can be deduced from that switch mode transformer nearby, the output diodes feeding the capacitors and the 7912 and 78M05 regulators. The 7912 regulator is a standard -12V linear regulator and given the dropout voltage required for it to operate the input voltage would need to be around -14V or lower. The capacitors on my board were 16V rated. This is way too close to the maximum rated working voltage of the capacitors. They should be at least 25V rated or higher. Just to respond to the comment by @Coulomb about operating electrolytics at 80% of rated voltage. That is not quite correct:
To begin with the way the maximum rated working voltage of an electrolytic capacitor is determined during testing is they crank up the voltage and measure the rate of spark through events on the dielectric. When the rate of spark through events reaches a certain value that is how they determine the maximum safe working voltage. So in essence at the maximum rated voltage the capacitor is already failing, just very slowly. So it's definitely a good idea to have quite a margin when it comes to working voltage. Also the value tolerance of most electrolytic capacitors is +/-20%. No circuit should be designed using an electrolytic capacitor where an accurate value is required. In this case the electrolytics are used for smoothing so the value chosen should be "enough" to perform adequate smoothing but doesn't need to be an exact value, more is better. The choice of value might also affect the corner frequency of the control loop for the switch mode, but again the design should allow for a wide tolerance of capacitance value. But what is more important is the capacitors should be of the low ESR type (low equivalent series resistance). All capacitors used in switch mode circuits where high frequency ripple currents are involved must be chosen for low loss and have sufficient ripple current ratings for the design. In addition the ripple current rating is reduced at high temperature. Another indicator that the capacitors are handling high frequency ripple currents are the small ceramic chip capacitors solder directly between the pins of the capacitors on the underside of the board. They are there to handle the very high frequencies where the electrolytics impedance starts to climb because of parasitic inductance. But this doesn't mean that they handle the bulk of the high frequency ripple currents.
The capacitors I found on my board were 16V rated and appeared to be made by a company called yst. There are no other markings to indicate which series they are. I googled yst and the information is sketchy on the manufacturer. I replaced the capacitors with some Rubycon YXF series 1000uF 35V capacitors I had in my stash and the inverter is working fine now. I just didn't have 25V rated capacitors but it didn't matter. Rubycon is a reputable manufacturer and the YXF series are low ESR 105C rated capacitors with long life. I've used Rubycon YXF for years and they've never failed me. I do not want to have to replace the caps again in 5 years time. You can use any capacitors from a reputable manufacturer that are specifically Low ESR and high temperature rated, such names as Rubycon, Nichicon, Panasonic, TDK Epcos, BHC Aerovox, Wurth etc come to mind. They only cost a little more than standard grade but its worth it. My guess is the yst capacitors used on the board are just standard grade capacitors and with insufficient voltage and ripple current rating, thus the failure and it's clearly happening on a lot of these inverters so it's a design flaw, not a random failure.
I hope this explanation helps.

Also just want to say thanks to all the contributors of this thread. 
My E&E engineering son returned back to his flat in Stellenbosch this week only to find the 4 year old Axpert King 1 inverter won't start up after the holiday.  As this specific inverter requires a connected battery before it will startup, we originally thought it was just the battery which ran flat (he left it on ...).    He first borrowed a 48V charger and charged the battery to 51V which then started the inverter , but with the same flashing LCD and LED lights problem as shown in this thread.  
Showed him this thread with the possible solution and he then proceeded to take apart the whole inverter, bought 2 new CAPs, soldered them in as replacements  and put the whole inverter back together the next day.  No screws left behind, he was hopeful ..., and VOILÀ (!) ,  inverter started back up functioning perfectly as before.  Hopefully fit for a good few more years of service.
R10 for the CAPs vs R12 000 for a new inverter, now that's a good deal.  Reminded me of my days when I was his age when I fixed my PC's by just replacing the busted caps on the motherboard .  It's a good feeling !
 
 

Also just want to say thanks to all the contributors of this thread. 
My E&E engineering son returned back to his flat in Stellenbosch this week only to find the 4 year old Axpert King 1 inverter won't start up after the holiday.  As this specific inverter requires a connected battery before it will startup, we originally thought it was just the battery which ran flat (he left it on ...).    He first borrowed a 48V charger and charged the battery to 51V which then started the inverter , but with the same flashing LCD and LED lights problem as shown in this thread.  
Showed him this thread with the possible solution and he then proceeded to take apart the whole inverter, bought 2 new CAPs, soldered them in as replacements  and put the whole inverter back together the next day.  No screws left behind, he was hopeful ..., and VOILÀ (!) ,  inverter started back up functioning perfectly as before.  Hopefully fit for a good few more years of service.
R10 for the CAPs vs R12 000 for a new inverter, now that's a good deal.  Reminded me of my days when I was his age when I fixed my PC's by just replacing the busted caps on the motherboard .  It's a good feeling !
 
 

Also just want to say thanks to all the contributors of this thread. 
My E&E engineering son returned back to his flat in Stellenbosch this week only to find the 4 year old Axpert King 1 inverter won't start up after the holiday.  As this specific inverter requires a connected battery before it will startup, we originally thought it was just the battery which ran flat (he left it on ...).    He first borrowed a 48V charger and charged the battery to 51V which then started the inverter , but with the same flashing LCD and LED lights problem as shown in this thread.  
Showed him this thread with the possible solution and he then proceeded to take apart the whole inverter, bought 2 new CAPs, soldered them in as replacements  and put the whole inverter back together the next day.  No screws left behind, he was hopeful ..., and VOILÀ (!) ,  inverter started back up functioning perfectly as before.  Hopefully fit for a good few more years of service.
R10 for the CAPs vs R12 000 for a new inverter, now that's a good deal.  Reminded me of my days when I was his age when I fixed my PC's by just replacing the busted caps on the motherboard .  It's a good feeling !
 
 

I'd say 12-13k is a fair price given the lack of warranty

Lithium Batteries SA has always been very open which cells and BMS they use in their batteries. 
Also, some overseas battery manufacturers is open with this as well.
I don't know why most of our SA resellers are so shady ...
Why are our suppliers not like this one ? 👇
 

The ESP32 is powered by 3.3V . As a result its GPIO inputs are also on 3.3V logic specs . Feeding a ESP32 input with 5V logic is non-deterministic and may eventually lead to damage.  However there are 3.3V microcontrollers that are 3.3v but are 5V tolerant , meaning there inputs has been designed to entertain the higher levels of 5V logic. Also there outputs is guaranteed to meet the minimum 5V high  logic level spec.  Otherwise its just trouble ,  But the ESP32 is not 5V tolerant.

Lithium Batteries SA has always been very open which cells and BMS they use in their batteries. 
Also, some overseas battery manufacturers is open with this as well.
I don't know why most of our SA resellers are so shady ...
Why are our suppliers not like this one ? 👇
 

This post is a good example of why I rage against these unscrupulous sellers of batteries in SA.  If they won't tell you what BMS and Cells are inside, run away as fast as you can !   

This post is a good example of why I rage against these unscrupulous sellers of batteries in SA.  If they won't tell you what BMS and Cells are inside, run away as fast as you can !   

Esener, Taico, Volta, Junlee, GenixGreen, The BMS comes from www.vkingner.com, or www.cleverbms.com 
the software to connect via RS232 - TOPBMS 6.1
 

Volta is just a SA spec'd rebrand of this Chinese system builder ... 
https://www.genixenergy.com/power-storage-wall/5KWH-ES-BOX12-51.2V-Wall-Mounted-LiFePO4-Battery.html
 
Volta refuses to disclose the BMS they use inside their batteries
Volta refuses to disclose the cells they use ... It could be cheap 2nd life cells inside, sold as new.   Who would know ?
Volta does not want to allow me to monitor the individual cell voltages or temps inside the battery on a dashboard like Home Automation ...
Volta does not provide a LCD screen on their batteries to show SOC, individual cell voltages, temperature,  etc ...
Volta does not provide a DC cut-off switch on their batteries
Volta include cheap tin-coated (not 100% copper) DC battery cables with their batteries
 

Volta is just a SA spec'd rebrand of this Chinese system builder ... 
https://www.genixenergy.com/power-storage-wall/5KWH-ES-BOX12-51.2V-Wall-Mounted-LiFePO4-Battery.html
 
Volta refuses to disclose the BMS they use inside their batteries
Volta refuses to disclose the cells they use ... It could be cheap 2nd life cells inside, sold as new.   Who would know ?
Volta does not want to allow me to monitor the individual cell voltages or temps inside the battery on a dashboard like Home Automation ...
Volta does not provide a LCD screen on their batteries to show SOC, individual cell voltages, temperature,  etc ...
Volta does not provide a DC cut-off switch on their batteries
Volta include cheap tin-coated (not 100% copper) DC battery cables with their batteries
 

So to be clear, the OP wants a UPS, not a solar setup.  Axpert works perfectly fine for that.  I've been using an Axpert in that configuration since 2015. I've also installed Axpert in a couple of family/friend situations for that.  And they are sold as power trolleys for a reason.  DO NOT use their solar, it sucks, but as a UPS they are really solid
https://www.fullcirclesolar.co.za/product/5kva-vp-5000w-unit-vat-incl-copy/
Then battery options:
https://www.fullcirclesolar.co.za/product/pylon-tech-li-ion-up5000-battery-pack-vat-incl/ https://www.fullcirclesolar.co.za/product/pylontech-li-ion-battery-pack-3-5kwh-vat-incl/ Goes without saying I trust that fullcirclesolar.  I've bought from them a number of times.

You should ask Hubble a few questions in return :
Who stands behind their warranty on their batteries ?  Can you have a copy of the guarantor's financial statements to see if they have a solid balance sheet to honor warranties ?  A guarantee/warranty is only as good as the person/company who stands behind it ...
Who manufacture the cells used inside the battery ? Is it new Grade A cells,  and how many cells are there inside ?
Who manufacture the BMS used to protect the cells inside the battery ?
( Long term testing done in Australia has shown that most battery system failures are due to BMS faults or failures ...) 

Success !

The radio interference is gone. I improved the inductance and added some capacitors.
In order to run several turns trough the ferrite core I went to the motor and transformer winding shop to get some 6mm² winding wire. But they don’t have such. We finally agreed on 2mm² wire that I would use in 3 wire bundles. I bought 10m of it (R40). I also went to the electronic shop to buy 4 0.33µF 310V capacitors (R40).
I first thought to have one core for each polarity. But then I realized that the 30A PV current could saturate the core and reduce the inductance. So I decided to run both polarities trough the same core. The two magnetic fields would then cancel each other and prevent saturation. I managed to run each polarity 3 times trough the core. Since the core can be opened it is easy to wind each half separately. I separated the polarities with a piece of card board. One challenge was to strip the 12 wire ends of the tough isolation varnish. The outgoing bundles are hold together with shrinking tubes. I tightened the two half cores together with a strong cable tie to prevent a gap between the core halves. The finished winding is shown in picture below.

I had many thoughts how to connect capacitors to the 48V M6 terminals. I finally found the solution shown below. I soldered the capacitor leads to 6X22 washers with a 60W soldering iron. The washers are then installed under the M6 nuts of the terminals.

The final installation of inductance and capacitors is shown in picture below. The outgoing leads of the inductance are directly wired to the PV input terminals together with the leads of the capacitors. I had to solder little pieces of installation wire to the capacitors leads because the terminals would not clamp the tiny wire of the capacitors. The other ends are crimped to the earth wire.

IMO, don't fall for the carrot.
Bought 2 recent LFP batteries, one from LBSA and the other from Volta.  Both 1C charge and discharge (100A) batteries.
Volta looks good, come with DC cables, but don't have DC cut-off switch and don't come with a LCD screen and menu buttons.  Bad customer and back-up support.
LBSA, local manufacturers of system, very good local support and they call YOU back !   Battery come with DC cut-off switch and LCD screen en menu buttons.

It seems to me that it's been mentioned a bunch of times, but only by you. The charge curve does not really provide much insight into why you would choose a lower voltage. It all depends on what the chemistry does when you do go over this voltage, and in the case of LiFePO, the answer is not much (up to around 3.65V).