shanghailoz

...and a Sunsynk 8.8KW.  have emailed.

I'd be interested in 20 of the 455w.

done

done

Having run my own forums before, it is a tireless thankless task.
Get some moderators and hand it over to them also
 
Many hands make light work!
Happy holidays to you and family!

Not asking for the software code or how it works (trade secrets copyright etc) and not asking for the firmware code either, those are all propriety, copyrighted and not opensource
The comms protocol to read data and manipulate settings is most certainly not copyrighted and should not be top secret and should be available for end users to use as they please.
As stated on my post/reply many battery manufacturers and BMS manufacturers already make that available to the end users, so my choice of battery will always be the ones that I can fully own and control without being held ransom by the manufacturer or supplier.
"can break the BMS and Void the warranty" - Users do that already to their batteries and inverters.
At the end of the day the end user has bought a product that is restricted .i.e. they do not really fully own what they bought.
Similar to cellphones that get locked to networks, and servicing of certain mobile phone brands and or devices by non-approved repairers(courts have fixed or are busy fixing that).
Similar to the motoring industry warranty servicing rules (courts have fixed that already, in SA anyway)
At the end of the day it's going to be the end user who needs to affect the way these companies operate.
Support the ones that are open and allow you full ownership of your purchase

Hi all, I thought I would document my findings of the Tian Power BMS protocol used in my Narada 48NPFC100 battery whilst I was attempting to establish why it did not work with my Deye inverter. Of note is that the exact same battery model is known to use a Shinwa BMS as well, but mine uses a Tian Power one, specifically TP-ND1530.
Firstly I wanted to thank @shanghailoz for documenting the same thing here for the Revov batteries, which also use a Tian Power BMS, without which I would have struggled a lot more. Also thank you @zivva for pointing me in the right direction for the BMS manufacturer.
I also peeked at the code of the BMS software to decipher how certain fields are calculated.
The protocol follows.
 
Preamble
Firstly, all data sent appears to have a header, body and CRC check. For example:
7e 01 01 00 fe 0d The first byte is either 7c or 7e according to the software. In my case it is always 7e.
The second byte is the DIP switch setting on the battery and corresponds to the following in the software:

The third byte is the function code, which you will see several examples of below, such as "read data", "read BMS time", "read BMS version", "read serial number".
The fourth byte is the length of the payload supplied as part of the function call - most "read" type functions will not include a payload and this parameter will be 00, but the "write" functions would have a payload. If there is a payload, it would follow this byte.
The last two bytes are the CRC check.
The response is structured in a similar way:
7e 01 01 56 ... f2 0d The first three bytes are an echo of what the request was, the fourth byte indicates the response length followed by the response bytes, and finally we have the CRC check.
 
Function 1 - Read data
7e 01 01 00 fe 0d This allows you to read almost everything that is displayed on the first tab of the BMS software.
The response will look as follows:
7e 01 01 56 01 0f 0c e1 0c e2 0c e2 0c e3 0c e3 0c e5 0c e5 0c e4 0c e4 0c e4 0c e2 0c e2 0c e2 0c e3 0c e5 02 01 75 30 03 01 16 74 04 01 27 10 05 06 00 44 00 45 00 45 00 44 40 46 20 45 06 05 00 00 00 00 00 00 00 00 00 00 07 01 00 81 08 01 13 54 09 01 27 10 0a 01 00 00 f2 0d The response body for this function is a data map structured as follows: data entry index, number of high/low byte pairs, the high/low byte pairs.
Data 1 - Cell count and cell voltages
01 0f 0c e1 0c e2 0c e2 0c e3 0c e3 0c e5 0c e5 0c e4 0c e4 0c e4 0c e2 0c e2 0c e2 0c e3 0c e5 The second byte is the number of cells - this is 15 for a 15 cell battery like mine or 16 cells for Revov. The individual cell voltages follow as high/low pairs which need to be divided by 1000, for example, 0ce1 maps to 3297 which means 3.297V. Since my battery has 15 cells, there would be 15 individual cell voltages.
Data 2 - Current
02 01 75 30 This is a tricky one. 7530 hex maps to 30000 decimal. The BMS software works it out as:
current_in_amps = (30000 - value_of_item_2) / 100 This item therefore has an offset of 30000 and a scaling factor of 0.01. So when the charge current is 0, the BMS will return 30000, therefore (30000 - 30000) / 100 = 0 amps. When the charge current is 10 amps, the BMS would return 29000, therefore (30000 - 29000) / 100 = 10 amps. When the charge current is -10 amps, the BMS would return 31000, therefore (30000 - 31000) / 100 = -10 amps. Furthermore, charge/discharge time remaining is calculated based on max battery capacity in amps divided by current in amps.
Data 3 - Remaining capacity
03 01 16 74 This value needs to be divided by 100 to get Ah. In this case I have 1674 hex or 5748 decimal, so 57.48Ah left of my 100Ah battery, therefore the SOC is 57.48%.
Data 4 - Full capacity
04 01 27 10 Divide by 100 to get full battery capacity in Ah. 2710 in hex is 10000 in decimal, so I have a 100Ah battery.
Data 5 - Temperatures
05 06 00 44 00 45 00 45 00 44 40 46 20 45 In my case there are 6 temperatures corresponding to the following screenshot:

In the Revovs the number of temperatures listed are different. The last two appear to be important ones.
Each temperature value is calculated as:
temp_value = (bms_temp_value & 0xFF) - 50 Data 6 - Alarm bits
06 05 00 00 00 00 00 00 00 00 00 00 These can be mapped to (a) very specific alarm code(s). I can get these if there is interest.
Data 7 - Cycles
07 01 00 81 This represents the number of cycles of the battery, so 129 in my case.
Data 8 - Voltage
08 01 13 54 The total voltage of the battery. The number needs to be divided by 100, so 49.48V in my case.
Data 9 - State of Health
09 01 27 10 The SOH of the battery. Needs to be divided by 100, so 100% in my case.
Data 10 - ALM bytes
0a 01 00 00 This corresponds to the ALM light on the battery and what it means.
 
Function 67 - Read protection parameters
Request:
7e 01 43 00 fe 0d Response:
7e 01 43 94 00 02 0e 10 01 02 03 e8 02 02 0d ac 03 02 0a f0 04 02 03 e8 05 02 0c 1c 06 02 15 4a 07 02 03 e8 08 02 14 c8 09 02 11 94 0a 02 03 e8 0b 02 13 88 0c 02 1f 40 0d 02 03 e8 0e 02 1b 58 0f 02 1f 40 10 02 03 e8 11 02 1b 58 12 02 00 69 13 02 0f a0 14 02 00 5f 15 02 00 32 16 02 0f a0 17 02 00 3c 18 02 00 69 19 02 0f a0 1a 02 00 5f 1b 02 00 32 1c 02 0f a0 1d 02 00 3c 1e 02 00 8c 1f 02 0f a0 20 02 00 87 21 02 00 0a 22 02 00 0f 23 02 03 20 24 02 01 f4 de 0d This is another data map which ultimately gets translated into the following, with the necessary scaling factors applied:
{ "cell_ov_start": 3.6, "cell_ov_delay": 1000, "cell_ov_stop": 3.5, "cell_uv_start": 2.8, "cell_uv_delay": 1000, "cell_uv_stop": 3.1, "pack_ov_start": 54.5, "pack_ov_delay": 1000, "pack_ov_stop": 53.2, "pack_uv_start": 45.0, "pack_uv_delay": 1000, "pack_uv_stop": 50.0, "charge_oc_start": 80.0, "charge_oc_delay": 1000, "charge_oc_stop": 70.0, "discharge_oc_start": 80.0, "discharge_oc_delay": 1000, "discharge_oc_stop": 70.0, "cell_ot_start": 55, "cell_ot_delay": 4000, "cell_ot_stop": 45, "cell_ut_start": 0, "cell_ut_delay": 4000, "cell_ut_stop": 10, "env_ot_start": 55, "env_ot_delay": 4000, "env_ot_stop": 45, "env_ut_start": 0, "env_ut_delay": 4000, "env_ut_stop": 10, "mos_ot_start": 90, "mos_ot_delay": 4000, "mos_ot_stop": 85, "capacity_low_start": 10, "capacity_low_stop": 15, "volt_diff_start": 800, "volt_diff_stop": 500 }  
Function 51 - Read BMS version
The request is as follows:
7e 01 33 00 fe 0d The response body is a string, such as:
7e 01 33 18 54 50 2d 4e 44 31 35 33 30 2d 31 35 53 31 30 30 41 2d 56 31 2e 30 2e 30 2e 0d Which in my case maps to: TP-ND1530-15S100A-V1.0.0
 
Function 66 - Read PCB barcode
Request:
7e 01 42 00 fc 0d Response body is a string as above.
 
Function 220 - Read serial number
Request:
7e 01 dc 03 06 00 00 c2 0d Response body is a string as above.
 
Function 69 - Read BMS time
Request:
7e 01 45 00 fe 0d Response:
7e 01 45 06 16 07 08 14 3b 16 48 0d To convert the above to a valid date time, prepend "20" and then concatenate the rest of the bytes which represent yy, MM, dd, hh, mm, ss, so in the above example: 2022-07-08 20:59:22.
 
There are many "write"-type functions too but I would be wary of using those.
I hope the above is useful to someone. I will post more useful ones if I find any.

Thank you, @shanghailoz!

I believe that I have installed it correctly.

I followed the following steps:
1. Execute SSH installer script option via Putty (SSH) on my CCGX as per Wiki (How to Install) - installing Louis latest release of the driver
2. Download zip file of the revov branch containing the updated dbus_serialbattery folder and files
3. Overwrite files on CCGX using FileZilla (SFTP) to transfer the files from my desktop
4. Reboot CCGX and voila! SerialBattery appears on device list

Read the instructions for the repo. https://github.com/Louisvdw/dbus-serialbattery/wiki/How-to-install
Once installed, overwrite with the updated files from here from the REVOV branch - https://github.com/csloz/dbus-serialbattery
You'll need ssh access to your PI or CCGX
 
Files to update / add from his release are mostly in these 2 files
https://github.com/csloz/dbus-serialbattery/blob/revov/etc/dbus-serialbattery/revov.py (the partially completed driver)
https://github.com/csloz/dbus-serialbattery/blob/revov/etc/dbus-serialbattery/dbus-serialbattery.py (some mild updates to add a link to load the revov.py code).
 
Bit busy on work atm, if you get stuck shout.
 
 
 
 
 
 
 

Modbus Packet : [7C:01:01:76:01:10:0D:28:8D:20:0D:29:8D:37:0D:45:8D:3A:4E:0B:8D:CC:0D:2C:8D:10:0D:2B:0D:28:8D:16:0D:2A:8D:1D:0D:27:02:01:75:30:03:01:27:10:04:01:46:50:05:03:00:49:00:49:20:4B:06:05:00:00:10:10:00:00:00:00:00:01:07:01:00:85:08:01:15:34:09:01:27:10:0A:01:00:01:0B:01:00:00:83:00:0C:01:00:00:37:AE:0D:01:00:40:00:00:0E:01:00:40:00:00:0F:01:00:1A:3A:04:10:01:00:09:57:AD:2E:0D] (118 Bytes)+6 bytes head/foot =  [7c:01:01:76] + [data] + 2 bytes [2E 0D] Checksum / LF
[HEADER] 7C:01:01:76 Modbus bits  (Cut when passed)
[0..1] 01:10   BatteryModule? | Cell Count 
[2..33] [0D:28:8D:20:0D:29:8D:37:0D:45:8D:3A:4E:0B:8D:CC:0D:2C:8D:10:0D:2B:0D:28:8D:16:0D:2A:8D:1D:0D:27 [ Per Cell Data ] 
[34..55] 02:01 75:30 03:01 27:10 04:01 [46:50] 05:03 00:49 00:49 20:4B 06:05  11 doubles [18000 = Full Bat Cap?]
[56..65] [00:00:10:10:00:00:00:00:00:01] Alarm 
[66..81] 07:01 [00:85] 08:01 [15:34] 09:01 [27:10] 0A:01 00:01  [133=Cycles?]  [5428 = 54.28 V_SUM?] [10000 = 100.00]
[82..97] 0B:01 00:00 83:00 [0C:01] 00:00 37:AE [0D:01] 00:40 [3073 = 3.073] [3329= 3.329? V_AVG/MIN etc?]
[98..113] 00:00 0E:01 00:40 00:00 0F:01 00:1A 3A:04 10:01
[114..117] [00:09 57:AD]
[118..119] [2E] 0D Checksum? ,  LF
 
Some more guesswork above.  Will implement as noted, then see if the inferred values are guessed wrong or not.
Another modbus packet to compare
[7C:01:01:76:01:10:8D:20:0D:17:8D:21:0D:2F:8D:3B:0D:32:CD:E3:0D:B3:8D:23:0D:0D:8D:24:0D:24:0D:11:8D:21:0D:18:8D:21:02:01:75:30:03:01:27:10:04:01:46:50:05:03:00:49:00:48:20:4B:06:05:00:00:10:10:00:00:00:00:00:01:07:01:00:85:08:01:15:24:09:01:27:10:0A:01:00:01:0B:01:00:00:83:00:0C:01:00:00:37:AE:0D:01:00:40:00:00:0E:01:00:40:00:00:0F:01:00:1A:3A:04:10:01:00:09:57:AD:BA:0D]
02:01 75:30 03:01 27:10 04:01 46:50 05:03 00:49 00:48 20:4B 06:05
[00:00:10:10:00:00:00:00:00:01] Alarm
07:01 [00:85] 08:01 [15:24] 09:01 27:10 0A:01 00:01 [Cycles] 54.48 Bat Voltage?] [soc state?]
0B:01 00:00 83:00 0C:01 00:00 37:AE 0D:01 00:40 [these haven't changed, so probably not V_min/avg]
00:00 0E:01 00:40 00:00 0F:01 00:1A 3A:04 10:01
00:09 57:AD
 
I think cycles and Battery voltage look correct.  Will need to check in morning or later if soc state is correct as batteries should be full until after sunset.
 

New version 0.1.3
/*
Cell Voltage Implemented
Hardware Name Implemented
Hardware Revision Implemented
Battery Voltage added (but not correct!)
Added additional binary logging so I can try spot what bits are used for RED cell errors (not the Alarm error, which I'll get to later)
Added return false for the read_cell_data in case of bad packet.
To do:
SOC, Error Codes, Other variables
*/
Example Logging
 

revov.py
 
I think this is good enough as a beta to ask Louisvdw to pull to github.
I'd like testers if anyone else has a revov, so I can check whats different between units/revisions
 
You'll need an RS485 cable, and a PI (or CCGX)
 
Also available at https://github.com/csloz/dbus-serialbattery/tree/revov
I have asked for a merge to the main repo.
Further code updates will be published at the repo.
Enjoy!

Some screen shots of this in action on my Venus PI
Now with per cell monitoring (added the qml file on the current repo)

Dead right. We installed 500kWp grid tied PV at work ealrier this year and used the following to effectively get 28% back from the taxman. You would get the same through depreciation over time, but the provisions of Section 12B brings the cash benefit forward.
"Section 12B of the Income Tax Act No. 58 of 1962, as amended (the ‘Act’), provides for a capital allowance for movable assets used in the production of renewable energy. More specifically, it allows for a deduction equal to 100% basis in respect of any plant or machinery brought into use in a year of assessment for the first time and used in a process of manufacture or any other process which is of a similar nature. It is important to note that the allowance is only available if the asset is brought into use for the first time by the taxpayer. In other words, the allowance is not limited to new or unused assets. The wording merely prevents the taxpayer from claiming the section 12B allowance twice on the same asset."

Sovereign Mini Bloemfontein was kind enough to lend me the Mini SE for the weekend. 
Some pics:
Charging in my driveway. 

Pretty normal Mini Cooper interior apart from the special SE "wrap" on the dash. 

Driver cluster shows the electric usage and re-gen. 

 
a R650k electric car. Almost 200k more expensive than the Ice equivalent. Is it worth it. As so many reviews have stated... No it is not. If you charge only from solar the ROI is 42 years at R20/L. Okay so this will come down as fuel prices increase but still a very long term "investment". 
But is it worth it if you have the coin and want to use it for urban env only? Then yes, very much. I was able to drive it from the morning 7h30 till the evening 19h30 on a single charge. Managed 190km with some highway, some property viewing. Bought some plants at the nursery and they fitted in the rather small boot. In urban settings, the car excels and if money was no object I'd prefer this to the bigger EVs from BMW, Jaguar and Merc. Audi seems to have some cool fun cars upcoming in the very near future. 
The Mini is in its 3rd year but is still a conversion rather than a bespoke developed platform. This will change in late 2023/2024 I understand. 
Fun factor: The Mini SE is the purest form of fun I've driven. I lived with an 06 Megane F1 2.0T for 7 years and that was a fun car. The mini takes it to a whole new level. The 135kw instant power made Golf 7 GTi look slow from the line. 
The regenerative braking makes driving the car a breeze, at first you let go too much and get a lot of negative acceleration but as you learn you feather the throttle pedal to get smooth braking. 
187km of range sounds very low/short but I was gatvol at the end of the day trying to empty the battery. 
Will it work in current Config as a direct replacement of your Ice car? No, the range is not enough especially on open road to get to nearest charger eg. Gariep from Bloemfontein. As a urban car it's epic and could provide commuting for a week on a single charge. If you have solar it would take around 1.5days to fully charge on solar only using the 3kw charger included on a 5kw PV array. Eskom is still cheap but charging at a fast charger station can cost up to R5.50 a kW. You do the math. 
Although the mini only has a 32kwh battery of which 28kwh is usable, it results in overnight 100% charging or Saturday Sunday charging on Solar from even below 50%.
Cons:
The one I drove has no cooling seats but has the leather seats. The seats get so hot when parked in the sun that the aircon (which is amazing) cannot cool your back and you arrive at destination with a sweaty wet back. Not cool tbh. 
Being a conversion the dash and interior has not been remodeled to make use of extra space like no exhaust tunnel underneath the car so you still have that tunnel in the car. Removing this in the next model will make the mini much more spacious. I could not complain about the current space for the driver though. Felt great to me but I'm short at 174cm. Rear seats are for short people or younglings only apart from the odd "must transport friends" scenario. 
Was able to fit months groceries from Makro into the car but only after folding down the rear seats. 
If your keen on electric mobility, do yourself a favour and get a taste by test driving the Mini Cooper Se. 
 

I want to explain my findings on how my Smart load output functions.
In my opinion, there are discrepancies in Youtube videos from Sunsynk as well as their newest Installer Manual explaining this function.
I have a 150 litre geyser connected to mine and it operates as follows :
1) The solar power setting has nothing to do with the max power output - which is how they explain it. It is purely the PV power limit at which the relay will close and power the aux load Or open to disconnect, if it falls below the setting for a certain duration.  It does not limit the power output.
2) If you have grid available and the total load - normal loads + Smart load - exceeds the PV available, it draws the balance from the grid and not from the battery.
3) If grid is not available it draws the balance from the battery - Starting when battery reaches the "Aux load ON Batt" setting which is 100 % in my case.
   It switches off when the battery reaches " Aux load OFF Batt" - 95% for me.
   It will not come on again - even if PV surpases the Power setting (4000w for me)  until the battery reaches the ON Batt setting again.
4) On grid always on  will always supply power to the Smart load output, when the box is ticked and grid available.
   This can only be useful if manually enabled, after a rainy day, to "top up" your geyser.
Hope this is useful to someone.

That looks like the latch for if you want to remove the contact module from the push button module it might not be clipped in properly that’s why you don’t get continuity.

Yup, that seems to have been the issue.
Moved it right, the button comes off, checked that, worked ok on the meter.
Put it back, moved the orange thing all the way to the left, after pushing in, now the button is making contact and making switch sounds.
So much for them checking things at the factory!
 
I can now see voltage on the secondary display, so some progress.  I guess next week time to move batteries upstairs and do a spot of rewiring!

Almost there!
2 x 300mm² Busbars for battery connections behind backing board.
DC wiring done. A/C wiring next up.
I'm thinking to wire from the "load" connection in the inverter (essential power) straight into my house DB i.e. everything will be on essential load - with a changeover switch to grid supply if battery power is low. If I utilise the grid connection with CT coil on incoming live with zero eport set, the battery would charge via mains right? Not sure I want that to happen.
Any advise would be welcomed please. I'm uneasy with what to do next to be honest. Island (with changeover) / hybrid.

Shanghailoz - thank you for your advise regarding those DC breaker positions - no amount of knowledge  beats experience. I moved them. The battery cables will now enter direct into the fuse box from below and I've got plenty room on top to bend the cable into the compression gland. I think you saved me from a lot of frustration....
I've sent Freedomwon an email as you suggested. I'm not impressed with their technical backup to be honest. Took them 3 days to read the mail and still waiting for a reply....  I'll wait till tomorrow then make waves. When you spend well over a R100k buying a product from a local company you at least expect a reasonable response time for a very simple query. Not so happy.
Now - 2nd question: the battery is fitted with 2 pairs of 95mm² cables. Sunsynk suggests a minimum cable size of 35mm² for these 5kW inverters. Ohm's law for 5500W and 48V suggests that the max current will be around 115A. A 50mm² cable rated for 207A and less than 3m should surely be ok to use?
 

FYI -
For advanced support please contact [email protected] or on WhatsApp at +27 60 968 5145 during office hours (08h00 to 17h30 CAT Monday to Friday). Assistance on weekends and public holidays may be arranged for urgent circumstances.
 
Let me know how you go.
 
You may need to rethink the positioning of the DC breakers on the board, might be a bit tight as is.

Have to chip in with my 2cents here - Going off-grid with solar and batteries can be done.
Been there, done that, for a good couple of years now!
If you have a worst case scenario of a few days of rainy crap weather, there are options.  
- One is to use a generator to top up on the days where it's too low.  
- Or to just not use power (I've done that - turned off everything non-essential so that I scrimp and save enough juice to get by).
- Or, to add sufficient generation that even in crappy days, its enough to charge the batteries (thats where I'm at now).
Throwing money at batteries is also another option, but it doesn't make sense yet, especially financially, and especially when adding more generation is cheaper.
 
 
As for OP-
 
I'd probably recommend a Sunsynk 5KW for you, and more batteries than 5KW if you can afford, or more batteries and more panels.
Minimum for offgrid would be 10KW of battery imho, even with really low usage.   
 

Lots of questions there.
 
Different muni's will allow different equipment connected.  
Axpert does not have NRS97 certification so cannot be legally connected in some muni's - eg CPT, and is not on the councils approved inverter list.  More muni's will be enforcing regulations rather than less, so I'd stick to equipment that is certified, vs just cheap.  Not sure about Pretoria, but I'd stick to NRS approved stuff..
(Others - Please update me if I'm incorrect on Axpert certification, its been a while since I last checked).
In summation -  you aren't really saving money buying a cheap inverter that you'll have to replace later.
If you can afford it get something that will scale to your future needs.  The Sunsynk 5k or 8k seem to be in the sweet spot at the moment in terms of product for price, and are legal to use here.
If you can stretch to the 8KW model, I'd do so vs the 5K, as Tariq has said.
Roof temp - I wouldn't worry too much - there isn't much you can do.  Panels will derate a percentage rating for every c above their nominal temperature of 25c.  How much that is will be on their data sheet.  Temps in SA will easily hit 40c on the roof, could be higher.
As an example - if they derate 1% every 1c over 25c, then you'll lose 15% of rated capacity at 40c ( 40c - 25c = 15), so your 400w panel may only output 340W at peak temp.  It's not that bad in real life ratings examples, but do read the data sheets to make sure your panels are suitable for hot climates.
Your roof interior on the other hand will be cooler, as the panels will be taking up the heat and covering your roof, so less a/c bills in summer (assuming you have an aircon!).   You should really paint your roof with a light color to minimize heat (as well as insulate the roof space so you have less heat in summer, keep more heat in, in winter).
I wouldn't worry about panels melting, they're mostly made of glass (front glass), silicon (the panels), silver (the interconnect wires) and aluminium (frame), need far far higher temps to melt those!
If you can stretch to more panels I'd add more first.  6 x 365w is too small.   Why are you looking at those anyway?  Price/ buck sweet spot at the moment are the JA-Solar 405w Mono Perc.

Given that the mounting and labour will cost the most (even if diy) I'd buy more panels, and get that done one time.
Double that 6 panels at least.  I'd recommend 12 panels min, and suggest go for higher watt panels.
 
Double check that the inverter can handle the volt/amp load for the panel strings though.  eg 2 strings of 6 perhaps.  Read the panel specs to work out the numbers, and compare against the inverter manual for its sweet spot.  
Inverter will have an efficiency rating for a given voltage usually, so you want to stick your strings volt sizing or amp sizing in its preferred range.
 
 
 
 
 

If you want to spec a system you should have the electrical knowledge of where to put fuses and wire the batteries so that no cable in the system goes over its rating and use the proper fuses in the right place. Things like that are taught to you at university and almost any competent electrical engineer worth his salt will know what to do. In the electrical industry no knowledge is same, but a little bit is a very dangerous thing because you don't know what you don't know. Just because it works doesn't mean it is right.   

Nobody is dissing one product to make another look good. There is no comparison between Pylontechs and Hubble’s when it comes down to cycles vs C rating and that’s because a 1C Pylontech Battery doesn’t exist. If and when Pylontech produce a 1C battery, we can discuss this then. 
 
Maybe question how does Pylontech supply a 4AWG/21.1mm2 cable pack which only THEY rate at a constant draw of 100A? No objective third party testing done there. Interesting to see that the Victron wiring manual states a max current rating of 75A for a 25mm2 cable. Pylontech also state that you can use up to 4 x US3000C in a stack with a single cable pack and each battery can deliver 37A. That’s a total of 148A on a 100A cable or 75A, depends where you look. Seems like world class engineering to me.

Lots of questions there.
 
Different muni's will allow different equipment connected.  
Axpert does not have NRS97 certification so cannot be legally connected in some muni's - eg CPT, and is not on the councils approved inverter list.  More muni's will be enforcing regulations rather than less, so I'd stick to equipment that is certified, vs just cheap.  Not sure about Pretoria, but I'd stick to NRS approved stuff..
(Others - Please update me if I'm incorrect on Axpert certification, its been a while since I last checked).
In summation -  you aren't really saving money buying a cheap inverter that you'll have to replace later.
If you can afford it get something that will scale to your future needs.  The Sunsynk 5k or 8k seem to be in the sweet spot at the moment in terms of product for price, and are legal to use here.
If you can stretch to the 8KW model, I'd do so vs the 5K, as Tariq has said.
Roof temp - I wouldn't worry too much - there isn't much you can do.  Panels will derate a percentage rating for every c above their nominal temperature of 25c.  How much that is will be on their data sheet.  Temps in SA will easily hit 40c on the roof, could be higher.
As an example - if they derate 1% every 1c over 25c, then you'll lose 15% of rated capacity at 40c ( 40c - 25c = 15), so your 400w panel may only output 340W at peak temp.  It's not that bad in real life ratings examples, but do read the data sheets to make sure your panels are suitable for hot climates.
Your roof interior on the other hand will be cooler, as the panels will be taking up the heat and covering your roof, so less a/c bills in summer (assuming you have an aircon!).   You should really paint your roof with a light color to minimize heat (as well as insulate the roof space so you have less heat in summer, keep more heat in, in winter).
I wouldn't worry about panels melting, they're mostly made of glass (front glass), silicon (the panels), silver (the interconnect wires) and aluminium (frame), need far far higher temps to melt those!
If you can stretch to more panels I'd add more first.  6 x 365w is too small.   Why are you looking at those anyway?  Price/ buck sweet spot at the moment are the JA-Solar 405w Mono Perc.

Given that the mounting and labour will cost the most (even if diy) I'd buy more panels, and get that done one time.
Double that 6 panels at least.  I'd recommend 12 panels min, and suggest go for higher watt panels.
 
Double check that the inverter can handle the volt/amp load for the panel strings though.  eg 2 strings of 6 perhaps.  Read the panel specs to work out the numbers, and compare against the inverter manual for its sweet spot.  
Inverter will have an efficiency rating for a given voltage usually, so you want to stick your strings volt sizing or amp sizing in its preferred range.