Growatt SPF 12000T DVM-MPV MPPT Failure

[Gannam]

I have confirmation from Sim at Growatt that the replacement MPPT control board that I ordered is incorrect. (Amos assigned the case to Simphiwe, who goes by Sim.) I will be shipping it back to Ian and am looking to get a replacement from Growatt directly.

The connection that essentially blew off the board was the positive battery connection. There is a 2-wire lead from the primary battery connection to the MPPT power board. (One pair for the positive and one pair for the negative.) Closer examination of SolarAssistant's logs show that the power flow went from ~755W into the batteries to an ouput surge of ~3.1kW and then a spike to ~8.5kW before power was cut. (8.5kW is ~60A from each of the 3 batteries. This is well within the discharge spec for the cells so the BMS's weren't tripped and each battery has a 250A fuse, which of course didn't trip either.) The discharge occurred for a ~1minute period. The replacement MPPT power board, and the remaining good boards in the inverter, all have a conformal coating, so I'm leaning towards an internal board fault being the cause of the failure at this point. The failed board having the problem section blown to bits, and not having a schematic, makes a diagnosis difficult.

I will update as I have more information.

 

[Gannam]

I've run in to a dilemma. I went to return the incorrect MPPT control board to watts247 and FedEx wants $102 to send the board out...the board cost $105. I'm looking into other couriers on Monday, but I suspect that the cost will be similar. My brother flies out to the US next month, but that's past the 30 day return policy. (I'll email Ian asking if there can be an extension due to the circumstances.)

I did try reflashing the firmware to see if it would detect the board correctly, but no such luck. The inverter shows the firmware version as '511 220' but the MPPT firmware is reporting as '000 000.'

I am looking into alternate options as I wait for Growatt to assist with ordering the correct MPPT control board. (I expect that this could take months with the way that things have been going.) I've essentially narrowed it down to two options that can be put into place without having to pull new wiring from the roof-mount panels to the MPPT charge controller(s). They are 1x Growatt or 2x Victron MPPT charge controllers:

• Growatt
  • SC48120-MPV
    • There is a possibility that I might be able to have this recognized in SolarAssistant, but I'm not counting on it.
    • I don't have to alter my strings at all as this is essentially the same as what's in the SPF 12000T DVM-MPV.
    • Lowest cost option at ~$400 before shipping.

• Victron (For all Victron calculations I used my panel's specs and went below the minimum record temperature here as a buffer.)
  • SmartSolar MPPT 250/70 Tr
    • The 5s2p string can connect to this with no modifications.
    • The PV power ratio is reported as 80% (undersized) using Victron's MPPT calculator.
    • Cost is ~$480 before shipping.
  • SmartSolar MPPT 150/35
    • The 4s1p string will have to be converted to 2s2p. This can be done on the roof with two branch connectors. (I have an additional fuse that I will install to the additional string.)
    • The PV power ratio is reported as 64% (undersized) using Victron's MPPT calculator.
    • Cost is ~$183 before shipping.

• SmartSolar MPPT 100/20 (alternate option for the small smaller string)
  • The 4s1p string will have to be converted to 2s2p. This can be done on the roof with two branch connectors. (I have an additional fuse that I will install to the additional string.)
  • The PV power ratio is reported as 110% (oversized) using Victron's MPPT calculator.
    • This will clip generation on both low and high temperatures and is not ideal. Technically okay according to Victron, but I'd rather pull all available power.
  • Cost is ~$90 before shipping. (5% coupon available)

Points I've been using to weigh options:

• I have a once-bitten-twice-shy with Growatt's MPPT boards. (I'm VERY happy that I have everything installed in the detached garage. I had this as a requirement when designing the setup and would recommend others install their equipment somewhere outside of their houses as well if at all possible.)
• The Victron equipment is slightly more efficient, but there's no way that I'd be able to interface it with SolarAssistant at this time. I would have to use Venus on a Rasberry Pi and have it and SolarAssistant report to HomeAssistant for a unified interface.
• When I eventually get the part(s) needed to repair the 12kW inverter, I can either put the arrays back on it or continue to use the stand-alone MPPT controllers. The unused controllers can either be used when I expand my array, or saved as backup units.
• I can get two of the SmartSolar 250/70 Tr units and not have to alter the smaller string at all either. This is of course a much more expensive option, but may potentially be more flexible in the future if used in the larger planned array. With that being said, I would likely keep them in place once setup.
• I will have to pull the funds from boat upgrades, which really bites. (I'll have to apologize to my wife and kids for not replacing the marine head this summer.)

Any input on the above or options that I've not thought of would be appreciated. It's always nice to have another set of eyes on a problem/solution.

 

[Kornbread]

I have an older sc48120, 150v pv max. It doesn't have the extra port so it does not talk with the growatt spfxxxk, but Solar Assistant sees the JK bms and reports the correct battery power input, but the other charts are not correct. The newer sc48120 with the extra port is 'supposed' to communicate with the spfxxxt. Communica with Solar Assistant says if the spfxxxt and sc48120 are communicationg, SA will show the correct values.

 

[shopman]

You can set up the Growatt charge controller in Solar Assistant as an SPF inverter. Solar Assistant will correctly report everything that both MPPT's are generating but that's where it ends. Trying to adjust the controllers settings via Solar Assistant will not work since it's not an inverter. It's strictly a read-only device in this setup. You will need a DB9 serial to USB converter(preferably with the FTDI chip) to make the connection with the supplied Growatt cable. Don't bother with setting up communications between the controller and the AIO. It does work but offers no advantage as you can not make adjustments to the SC48120 through the inverter. I have the SPF8000T DVM MPV AIO and the SC48120 working happily this way under Solar Assistant in open loop. Haven't had any problems with the inverter in two years but I do open it up periodically and blow the dust out of it since it has no filters. Too bad we're on opposite ends of the continent as I would love the chance to examine the failed board in person to see what went wrong. If the inverter portion runs without the MPPT/controller installed I would probably just pick up the external charge controller and run with it until the inverter dies, but I'm not in quite the same situation as you are.

 

[Gannam]

You can set up the Growatt charge controller in Solar Assistant as an SPF inverter.

Does it connect alongside your inverter on the same instance of SolarAssistant, or did you need to purchase an additional licence and Pi? (I thought you could only have multiple inverters listed if they were the same and configured in parallel, but it's been a bit since I've actually looked over the documentation.)

While trying to make a decision, I've dusted off a no-longer-used Pi2 and have installed VenusOS on it. I should be able to get it to interface with HomeAssistant for a pooling of resources.

 

[shopman]

Does it connect alongside your inverter on the same instance of SolarAssistant, or did you need to purchase an additional licence and Pi? (I thought you could only have multiple inverters listed if they were the same and configured in parallel, but it's been a bit since I've actually looked over the documentation.)

While trying to make a decision, I've dusted off a no-longer-used Pi2 and have installed VenusOS on it. I should be able to get it to interface with HomeAssistant for a pooling of resources.

All on the same instance. Since it's connecting as an SPF device, Solar Assistant just sees it as another SPF series inverter.

Home Assistant is awesome and interfaces nicely with Solar Assistant via MQTT. I recently put Wombatt on a Pi Zero 2W and interfaced it with my 6 EG4 batteries to send that data to Home Assistant via MQTT as well. Now I can see the entire system all from one app.

 

[Gannam]

I know someone going to the US next month who's agreed to take the incorrect MPPT Control Board with them and ship it from the US. This will surely be FAR cheaper than trying to send it from here. (The expense of shipping it from here makes the return refund a wash.) Ian is thankfully acommodating an extended return window due to the circumstances. It looks like Growatt has sent him an incorrect shipment, as all of the MPPT Control Boards for the SPF 12000T DVM-MPV that he has in stock are the wrong boards. As Chineese New Year was on the 10th, neither of us will likely hear from anyone until the 19th at the earliest. (It's a public holiday for 7 days, but is often celebrated for just over 2 weeks.)

Curiosity got the better of me, and as the MPPT control board is dead, I snipped all the pins to reveal the board's silkscreened printout. I haven't the foggiest why they chose that location for the silkscreening instead of the back of the board that has lots of available space.

It looks like I'll have to get the board from China directly. I expect shipping to take ~50 days unless I pay out the nose for expedited shipping. I think I'm going to have to pull the trigger on a dedicated MPPT solution while I wait. Once I have the replacement board and have confirmed that the inverter's MPPT circuits are functioning again, I'll make a decision as to which MPPT I'll use as a primary and which will be backup.

Before shipping and duty, the Victron MPPT solution will cost ~$830 while the Growatt MPPT solution will cost ~$525. (Note that this is before shipping and duty costs are factored in. The Victron option will likely cost more to ship as it'll require the purchase of more hardware to implement than what the Growatt option will require.) The Growatt option will be a faster setup, but I'm trying to predict the future and determine if the $300+ is worth it for the Victron gear in the long run.




@Kornbread and @shopman

What did you use for your PV connection(s) and battery connections to the SC48120? Do you have any pictures?

It looks like 10awg UL certified ring terminals should be able to handle the PV input lines. I'm looking at some that are gold-plated-copper rated up to 48A. My theoretical open circuit cold weather peak is ~205v @ 20A. I've realistically only seen up to ~183v @ 18A at the MPPT's. I'll go with proper lugs if needed, but I'd have to see if I can use someone's crimper as mine bottoms out at 8awg.

For the battery cables I'll go with 2awg welding cables with 2awg tinned copper lugs. I'll likely put a fuse between the battery bus bar and the 2awg wire. There is no fuse between the battery connection and the MPPT in the SPF 12000T DVM-MPV, but I suspect that it is probably a good idea to have some protection for the SC48120-MPV's 2awg line.

 

[Quattrohead]

Before shipping and duty, the Victron MPPT solution will cost ~$830 while the Growatt MPPT solution will cost ~$525. (Note that this is before shipping and duty costs are factored in.

EG4 have a good unit for less than these prices https://signaturesolar.com/eg4-solar-charge-controller-mppt-500voc-100a-mppt100-48hv/

 

[Gannam]

EG4 have a good unit for less than these prices https://signaturesolar.com/eg4-solar-charge-controller-mppt-500voc-100a-mppt100-48hv/

Thank you for the suggestion. The SC48120-MPV is comparable price-wise. I've added in the additional hardware that I'll need to properly implement the addition of one or two MPPT's in my pre-shipping estimates.

Additional hardware needed differs slightly between the two options that are being considered:

• 2AWG welding cable to connect the MPPT(s) to the battery's bus bars
• Tinned 2AWG lugs
• MC4 branch connectors and 10AWG PV wire if altering the 4s array to 2s2p for the SmartSolar 150/30
• 1050 strand tinned 10AWG wire to connect to the Victron MPPT's (I'll use a short run, so that it'll be the 19-strand PV -> MC4 -> 1050 strand.)
• Ring terminals for the PV inputs if using the SC48120-MPV
• 150A fuse(s) with holder(s) to protect the 2AWG line(s) from MPPT(s) to battery bus.

I looked briefly at the EG4 MPPT100-48HV and another high voltage MPPT unit, but unfortunately my smaller 4s string won't get the voltage high enough until a short mid-day window, and not at all on cloudy days. (On clear days it generally hovers around 120v.) The 4s string puts out a Voc of 163v (137v Vmp). The 5s2p string could be put into a 10s string and work well with the MPPT, having a Voc of 408v (342v Vmp). The EG4 MPPT100-48HV has a PV Array MPPT Operating Voltage Range 120-450Vdc.

I wish I could put more panels on the roof, but the only space that I have is North-facing. My 5s2p array is South-facing at a ~30° angle, while my 4s array is sitting at a ~5° angle. The lower PV operating voltage ranges of the SC48120-MPV, or SmartSolar MPPT 150/35 and SmartSolar MPPT 250/70 will allow me to squeak out more generation from the day's sun with my setup. When I eventually add more panels, I'll have to have them either on a pergola or carport. Panel installation here should be rated as close to 140mph winds as possible to account for hurricanes. (My roof mounted setup is rated to 175mph winds according to the local company I bought the panels and racking system from.) Most of the carport framings that I've found peak at ~134mph wind ratings. That should be okay as it'll be installed in a more sheltered area, but I'll need to have it approved by the planning department before proceeding.

My record generation for a day so far with my 4.48kW array was on May 7th, 2023 where I managed to pull in 28.5kW. I've attached a shot of the beautiful solar generation curve.

 

[Daddy Tanuki]

Yes it does cost to go with quality... what are your thoughts now a month or two later with a bricked inverter that still does not work? Not quite ford focus to rolls royce... But infinitely better than the options you chose.

for how many years? i can still get replacement boards for an inverter that was first sold in 2004... can you do that with any of the Chinese products?

For your 4x Tri-Star MPPT60 charge controllers at almost $700 each, and your Magnum 4448 at ~$2000 each, I'd hope they would give you a foot massage while replacing the boards themselves on-site. I would need 3 of the Magnum 4448 inverters in parallel (~13.2kW constant), and 3 Tri-Star MPPT60 charge controllers after re-configuring my array, needing more wiring as well as channelling out concrete and stone walls to install larger conduit. (I would need to split my 5s2p array into 2 separate arrays. My 5s2p array is angled at about 30°, while my 4s array is angled at about 5°. This cannot be changed as it is roof-mounted as per government regulations.)

I would need to spend $8,100 for the inverter and MPPT controllers, before shipping + wharfarage fees on everything with additional ~$1,500 duty on the inverters. Ignoring the shipping costs, I would be spending an additional ~$5,300 on the inverters and MPPT charge controllers alone. I find the argument akin to comparing a Ford Focus to a Rolls-Royce and a moot point concerning my issue, but thank you for your input.

 

[Gannam]

Yes it does cost to go with quality... what are your thoughts now a month or two later with a bricked inverter that still does not work? Not quite ford focus to rolls royce... But infinitely better than the options you chose.

With the options and funds available at the time it was one of my best options. The fault that I've had appears to be an isolated incident. (If you can find more reports of a failure like mine, please let me know. I'm genuinely curious.) Most failures that I've seen reported for the MPPT power boards on the SPF 12000T DVM-MPV look to be related to over-paneling, with most killed by PV voltages at or over the unit's spec'd limits. My PV feed is under voltage and amperage limits for the unit, it just looks like bad luck more than anything.

Yes, being down for this long sucks, but I expect there to be delays when components need to be sourced from China. Power usage at the house now is ~28 - ~35kwh/day. It's still cool, so the AC's haven't been running but the dehumidifiers have been. One of the main pushes to get it up is the arrival of my daughter next month. I won't have time to mess around with the solar while looking after 2 kids and helping with the baby. The Growatt has gotten me through several hurricanes, 2 with major with long widespread outages, as well as several random outages during non-weather events. After one hurricane our parish was without power for just over 2 days and I was able to run everything as normal. The AC's are a god-send after a hurricane passes. The humidity's high and the air is often still. I had neighbors coming over to charge their devices and a few were able to get a shower. (We run either well pumps or on-demand pumps as all fresh water is caught on our roofs and stored in large tanks usually below the house. There is a company that provides water, but it's expensive and not available everywhere and often isn't functional after a major storm.) I was hoping to have everyting repaired before the baby arrives, but that's the way the cookie crumbles and the reason for adding alternate MPPT controller(s) now. I will need to expand in the future to allow for more panels, but didn't want to start doing that so soon/before permits are granted.

If I had cash to burn during the original install, I would have likely gone with Victron SmartSolar MPPT's, 4x Quattro 5k units configured in parallel split-phase, and the battery setup that I built. As this would have been almost 3x my allocated budget, I went with the SPF 12000T DVM-MPV. I was hoping to get 5-8 years out of it before having any trouble, but c'est la vie.

I still find the SolArk 15k and EG4 18kPV enticing for the eventual expansion. Their ability to back-feed is desireable, but I'm not sure they'd be able to handle the high surge currents of the well pumps (one for fresh water and one for the brackish well water used for flushing the toilets), the AC's, etc. I suppose I could always eventually put the 4s array on a grid-tied inverter for the $35/month discount for back-feeding plus the feed-in-tarrif. If I can remain off-grid, the 4s array going to grid-tied might negate the utility power bill entirely. (The last bill was just under $700 for the month. I'll be happy to get rid of that. I average between $500 to $650 each month for power. If I can go fully off-grid, the completed install with the expansion will be paid off in under 5 years. My main limitation now is time and the allocation of funds.

 

[Kornbread]

@Kornbread and @shopman

What did you use for your PV connection(s) and battery connections to the SC48120? Do you have any pictures?

xxx

 

[Gannam]

xxx
View attachment 195729

Did you go from PV to welding cable (high strand count) for the SC48120 MPPT connections? Did you use standard lugs or ring terminals in the SC48120 for the PV input? The manual doesn't indicate any specifics about the supported thickness of the lugs or ring terminals that should be used.

 

[42OhmsPA]

I know someone going to the US next month who's agreed to take the incorrect MPPT Control Board with them and ship it from the US. This will surely be FAR cheaper than trying to send it from here. (The expense of shipping it from here makes the return refund a wash.) Ian is thankfully acommodating an extended return window due to the circumstances. It looks like Growatt has sent him an incorrect shipment, as all of the MPPT Control Boards for the SPF 12000T DVM-MPV that he has in stock are the wrong boards. As Chineese New Year was on the 10th, neither of us will likely hear from anyone until the 19th at the earliest. (It's a public holiday for 7 days, but is often celebrated for just over 2 weeks.)

Curiosity got the better of me, and as the MPPT control board is dead, I snipped all the pins to reveal the board's silkscreened printout. I haven't the foggiest why they chose that location for the silkscreening instead of the back of the board that has lots of available space.

It looks like I'll have to get the board from China directly. I expect shipping to take ~50 days unless I pay out the nose for expedited shipping. I think I'm going to have to pull the trigger on a dedicated MPPT solution while I wait. Once I have the replacement board and have confirmed that the inverter's MPPT circuits are functioning again, I'll make a decision as to which MPPT I'll use as a primary and which will be backup.

Before shipping and duty, the Victron MPPT solution will cost ~$830 while the Growatt MPPT solution will cost ~$525. (Note that this is before shipping and duty costs are factored in. The Victron option will likely cost more to ship as it'll require the purchase of more hardware to implement than what the Growatt option will require.) The Growatt option will be a faster setup, but I'm trying to predict the future and determine if the $300+ is worth it for the Victron gear in the long run.




@Kornbread and @shopman

What did you use for your PV connection(s) and battery connections to the SC48120? Do you have any pictures?

It looks like 10awg UL certified ring terminals should be able to handle the PV input lines. I'm looking at some that are gold-plated-copper rated up to 48A. My theoretical open circuit cold weather peak is ~205v @ 20A. I've realistically only seen up to ~183v @ 18A at the MPPT's. I'll go with proper lugs if needed, but I'd have to see if I can use someone's crimper as mine bottoms out at 8awg.

For the battery cables I'll go with 2awg welding cables with 2awg tinned copper lugs. I'll likely put a fuse between the battery bus bar and the 2awg wire. There is no fuse between the battery connection and the MPPT in the SPF 12000T DVM-MPV, but I suspect that it is probably a good idea to have some protection for the SC48120-MPV's 2awg line.

If you're considering spending that much for the mppt, why not buy an entire inverter and gain the additional AC output?
You could probably find 230V even cheaper.

$502.70 49%OFF | SRNE 5000W 48V Hybrid Inversor Built-in 80A MPPT Solar Charger 110-120Vac PV 500VDC 50Hz/ 60Hz 40A Battey Charger Support WIFI

https://a.aliexpress.com/_msm8wfk

 

[Gannam]

If you're considering spending that much for the mppt, why not buy an entire inverter and gain the additional AC output?
You could probably find 230V even cheaper.

$502.70 49%OFF | SRNE 5000W 48V Hybrid Inversor Built-in 80A MPPT Solar Charger 110-120Vac PV 500VDC 50Hz/ 60Hz 40A Battey Charger Support WIFI

https://a.aliexpress.com/_msm8wfk

Unfortunately this unit has a PV input voltage range that is too high to take advantage of my smaller array. (Similarly to the EG4 MPPT100-48HV mentioned above.) I'm not sure how I would use the power it would ouput unless I paralleled two units for split-phase and ran the office off of it utilizing an interlock kit on the subpanel.

I'm not familiar with the SRNE brand and with a quick look it's hard to find reviews. I did go through your post showing your setup with the SRNE ASF48100U200-H 10kw. I'm impressed how well the unit handled your test loads, but I feel that I would run it a bit too hard too often. There are a few occasions where I've run just over the 11kW mark for 30-45 minutes, but I generally try to keep below 9kW especially when it's hot out. I definitely lean more toward low frequency inverters due to the loads that I often run. One of the limits of the 12kW Growatt is that it cannot be run in parallel. There have been times where I could have used more power, but managed to get by easily enough with a little bit of planning around running large loads.

 

[Kornbread]

Did you go from PV to welding cable (high strand count) for the SC48120 MPPT connections? Did you use standard lugs or ring terminals in the SC48120 for the PV input? The manual doesn't indicate any specifics about the supported thickness of the lugs or ring terminals that should be used.

Used 4/4/2 triplex aluminum urd service entrance cable for solar arrays to first din rail. The reason for such large gauge wire is the ~200' home run and a limit of 150v max mppt. Smaller 4gauge copper thhn from first row of din terminals to middle din rail breakers. Third din rail terminals are for directing/combining solar inputs to their appropriate mppt. From there, 4gauge thhn to growatt spf6000t (barely fits) and 6gauge fine strand weld cable from two pv arrays to both inputs of growatt sc48120 (barely fits). sc48120 to battery buss bar is 1gauge weld cable w/150amp t fuse. 2/0 weld cable from spf6000t to battery buss bar w/200amp t fuse. Wire into the units and din rail items were bare strand. Wire to everything else was crimped lugs.

 

[Gannam]

@ Daddy Tanuki - I think you'll find this entertaining. I'm going to upgrade the system, but will do so in more bite-size chunks if at all possible.

I've decided to repair the Growatt's MPPT to be used as a spare, if needed, and proceed with the Victron MPPT's. (I don't have to do anything with conduit modifications to implement them for my current arrays.) Down the road I'll upgrade the 12kW Growatt inverter, either when it fails or I expand my array, to either a Victron or Schneider inverter/charger. (I'm hoping to get a few more years out of the Growatt inverter.)

I like the ability that the Schneider has to push power back to the grid. I've done the configuration for a 2x split phase Schneider system with AC coupled panels and a stand-alone battery array for a client. (A local solar company called me in to assist.) Schneider's support can be tough to get ahold of on the phone, but it's do-able...eventually...depending on the day. I'm not a fan of their warranty proceedure though. One of the paired units sounded like something was touching a fan blade when inverting. It had to be shipped out and a replacement shipped in. They would not allow the unit to be opened to check it without voiding the warranty. (The shipping was NOT cheap. I'm glad it wasn't my expense.) The transformer coil humm is VERY noticeable with the XW Pro.

As for the Victron inverter's side, I have no hands-on experience. I'm hoping that the Multiplus-II 5kVA 120v unit and/or Quattro 10kVA gets UL certification before I need to/intend to potentially place an order, but I'm not holding my breath with how long it took for the Multiplus-II 3kVA unit to get UL1741 certification. I'm still waiting for the utility company to confirm requirements for net-metering to see if UL1741 certification is needed.

I like the option to expand that both have, and know that 2x Multiplus-II 5kVA units will only give me ~8000W of usable power. The Schneider units look to be better for surges, but I'm not a fan of their software. I would likely expand to 4-6 Multiplus units or 3 Schneider units eventually as power needs change, but I would start off with 2 of either option initially.

My current dillema is in how to protect the Victron MPPT's battery connection. I am having trouble finding t-class fuses for the units. The 150/35 needs betwen 40A-45A and the 250/70 needs between 80A-100A. The busbar that the inverter is connected to is protected by a 400A t-class fuse and each battery has a 250A t-class fuse. I'm looking at something like MidNite Solar's MNEDC series breakers and wanted to know if there are potentially better options that I haven't considered.

• SmartSolar 150/35
  • MidNite Solar MNEDC40
• SmartSolar 250/70
  • Midnite Solar-MNEDC100
• Both breakers housed in the Midnite Solar Breaker Box MNEDC Quad

Edit: Would ANL fuses be adequate between the MPPT's and the bus bar? (There's a 400A t-class fuse between the batteries and the bus bar.) It would be far cheaper, but I don't want to skimp out. The fuse values for the MPPT's are as recommended by Victron.

 

[CharlesWhitlatch]

this is why a AIO is not the way to go. multiple components make for easier maintenance.

I have 4 SCC. they are all running at about 35%. one burns up I put the extra panels on one of the other units and continue on until replacement parts arrive.

Two inverters, either one can run the entire house. one goes south, put one on full duty until parts/replacement arrives. (I now keep spare boards, so that would be like 5 minutes to swap out).

BMS? 4 of them only two running, guess what, swap out in five minutes. each of the individual items are less that one AIO unit. my system overall is more expensive not counting spares, but even counting spares its better than an AIO unit.

You have good points but realize Chinese AOI units are cheap although it appears spare parts are absolutely not available (for EG4 Luxpower and Voltronic). I am talking to SS and EG4 about this.

In my prior life, we did both. We had redundancy and spare parts. We continuously inspected and maintained our electronic equipment. We had duplicate systems. A long time ago, even the manufacturer (U.S. based) could, and did, ship the largest of items next-day, by air.

Electricity generation is similar. The local utility almost never fails. You are seeking this reliability with your system. Alternatively, redundant AIO units could work if a failure is not too severe, affecting surrounding equipment.

I am looking into failure modes of HF AIO units. Depending on how the inverters are designed, single or multiple points of component or board failures can destroy AC equipment with DC voltage from the battery. A large AC transformer or motor on the AC side would likely mitigate this by tripping a circuit breaker.