maxmaia

Hi all,

I wanted to share our recent experience with a Deye firmware update that went wrong and how Deye support resolved it. Hopefully this helps someone in a similar situation.

---

Date: 2026/04/28
Model: Deye SUN-8K-SG01LP1-EU

---

🔵 Original Firmware (Before Update)
Protocol Version：V0.2.0.1
Lithium Battery Version Number：V0.0.0.0
HMI：VC.3.4.D
MAIN_1：V3.8.8.1
MAIN_2：V0.7.1.7

---

🔴 Failed Firmware Update - Resulted in COMM Error
Protocol Version：V0.2.0.1
Lithium Battery Version Number：V0.0.0.0
MAIN：1400-0017
HMI：0000-C384
LCD Type：0000

What happened: Deye remotely pushed a firmware update from their end. After the update, the inverter displayed a persistent COMM error and would not recover. We followed the standard troubleshooting steps of switching off PV, AC and battery power, waiting 5-10 minutes then restarting but the error would not clear.

We contacted Deye Support at [email protected] with the serial number, firmware details and photos of the LCD screen. They responded promptly and pushed a corrected firmware remotely.

---

✅ Latest Firmware - After COMM Error Resolution by Deye Support
Protocol Version：V0.2.0.1
Lithium Battery Version Number：V0.0.0.0
MAIN：6027-1724
HMI：0000-C384
LCD Type：0000

---

The inverter is now back to normal operation. Full credit to the Deye support team for resolving this quickly.

If you encounter a similar COMM error after a remote firmware update, my advice is:
1. Don't panic and don't replace the control board just yet. Check the forums first.
2. Contact Deye Support directly at [email protected] with your SN and firmware screenshots.
3. They can push a corrected firmware remotely to resolve the issue.

Hope this helps someone!

Date: 2026/04/28
Model: Deye SUN-8K-SG01LP1-EU
---
Original Firmware (Before Update)
Protocol Version：V0.2.0.1
Lithium Battery Version Number：V0.0.0.0
HMI：VC.3.4.D
MAIN_1：V3.8.8.1
MAIN_2：V0.7.1.7
---
Failed Firmware Update — Resulted in COMM Error
Protocol Version：V0.2.0.1
Lithium Battery Version Number：V0.0.0.0
MAIN：1400-0017
HMI：0000-C384
LCD Type：0000
---
Latest Firmware — After COMM Error Resolution by Deye Support
Protocol Version：V0.2.0.1
Lithium Battery Version Number：V0.0.0.0
MAIN：6027-1724
HMI：0000-C384
LCD Type：0000

Just a small update on this:
the setup is in place since 12days and it works very well, cell are pretty ballanced, stays under 0.004V normally and when at 100% charge it raise to around 0.02V. The total voltage between batteries is a bit different by 0.1V. Probably due to wiring/internal battery resistance/bms measuring accuracy.
Set the inverter to voltage control, using following settings (winter settings):
Battery float charge voltage 51 V
Battery absorption charge voltage 52.5 V
Battery equalization charge voltage 52.5

@fhocorp I've run with it for a while now just to see how it works as one of the Pylontech cells is fcked ... but I realized the passive BMS not good enough for balancing so I started going the other way around. Installed JK BMS on the Pylontech pouches ... that goes a bit better :)
Although JK is not the best when measuring SOC it still better to balance.
Now I am in the process of replacing ALL the Pylontech BMS boards with JK.
In order to be able to control both JK and Pylontech at the same time , I have installed one iBMS for JK batteries and one iBMS for the Pylontech, this way I sum up the data and send one stream to the inverter so everything works together.
Link to iBMS : https://shop.jamestronics.com/

Will make a review in a short time :) and explain the functions and how all works together.

Thanks, will keep an eye on cell differences. Currently when at 100% charge, Pylontechs stays below 40mV difference.

The goal is to get the cell delta between the pack cells as close as possible <50mv.
Below delta between the 2 packs 3mv

Just check the Pylon voltage at 100% SOC and then charge DYI to the 100% SOC and that same voltage before connecting these two set of batteries together.

Just for information purposes not Pylontech with other 15s but rather FW e- tower paied with Svolt rack mount at the identical capacity and same bms Pace. As a pilot project i pushed the anvelope to see if i could successfully pair the FW etower with Svolt. To my amazement it actually turned out quite successful.
The setup: Victron Multiplus-II Gx paired with FW 51.2v 16cell( Master) and Svolt 51.2v 16, cell( slave) both incorporating Pace bms. Capacity combined 200ah.
I updated the Svolt pace bms with the same firmware version using Pbms tools via Rs 232 serial communication.
The outcome a perfect hand shake between the bms's via can communication and daisy chained between FW and SVolt.
It goes to show you would not know the outcome until you experiment.

Just use any 15S LiFePO4 cells, then connect the new battery in parellel with Pylontech and it will work okay.
Before connecting in parallel for the first time it's better to get all the batteries to the same voltage in order to avoid sparks and disbalance. For example:
Charge Pylontech to 100% SOC, then disconnect.
Connect DYI battery, charge to 100% SOC.
Then connect both batteries in parallel.
Do not forget to set the same CC and CV target voltage for charging DYI like you have set for Pylontech.
The only issue is, that the BMS of DIY battery will not be able to daisy-chain with Pylontech BMS. If you want to daisy-chain comms, then you will need to use salvaged Pylontech BMS board for the DYI battery, plus use the cells with the capacity equal or slightly larger than what that BMS board is set to (50Ah, 74Ah, 100Ah).

Yes no problem. Should JK detect some error, like overvoltage/undervoltage etc. it will disconnect DYI battery. Very same like when you have JK with an inverter that does not support BMS comm.
If possible, I would leave Pylon CAN connected to the Deye and connect JK to the Home Assistant in order to be able to monitor it comfortably.

Yes

Is it true that Us5000 version can support up to 16 batteries without the need of a hub? Not sure if I get i tright from their manual, here attachef.
Pylontech_us5000_series.pdf

Hi guys,
I've realized that my first Pylontech battery will be 6 years old this March, so I've pulled some stats on the SOH and Cycles to share together with my thoughts.
CC = 52.5V
CV = 52.5V
8x US3000
System SOH : 96 % Power Volt Curr Tempr Tlow Thigh Vlow Vhigh Base.St Volt.St Curr.St Temp.St Coulomb Time B.V.St B.T.St 1 50955 -260 23000 20000 21000 3390 3399 Dischg Normal Normal Normal 100% 2025-01-30 22:16:53 Normal Normal 2 50965 -256 23000 20000 21000 3390 3400 Dischg Normal Normal Normal 100% 2025-01-30 22:16:52 Normal Normal 3 50970 -260 23000 20000 20000 3392 3402 Dischg Normal Normal Normal 100% 2025-01-30 22:16:52 Normal Normal 4 50948 -252 23000 20000 20000 3387 3399 Dischg Normal Normal Normal 100% 2025-01-30 22:16:52 Normal Normal 5 50969 -256 23000 20000 20000 3389 3399 Dischg Normal Normal Normal 100% 2025-01-30 22:16:52 Normal Normal 6 50975 -262 22000 20000 20000 3392 3400 Dischg Normal Normal Normal 100% 2025-01-30 22:16:52 Normal Normal 7 50959 -268 22000 19000 19000 3391 3399 Dischg Normal Normal Normal 100% 2025-01-30 22:16:52 Normal Normal 8 50945 -260 21000 18000 18000 3388 3398 Dischg Normal Normal Normal 100% 2025-01-30 22:16:52 Normal Normal CYCLE Times : 267 CYCLE Times : 518 CYCLE Times : 521 CYCLE Times : 523 CYCLE Times : 529 CYCLE Times : 520 CYCLE Times : 511 CYCLE Times : 501  
8x US3000C
System SOH : 98 % Power Volt Curr Tempr Tlow Thigh Vlow Vhigh Base.St Volt.St Curr.St Temp.St Coulomb Time B.V.St B.T.St MosTempr M.T.St 1 50890 0 22100 19900 19900 3387 3396 Idle Normal Normal Normal 100% 2025-01-30 23:52:12 Normal Normal 20600 Normal 2 50903 -203 22300 19900 20200 3378 3397 Dischg Normal Normal Normal 100% 2025-01-30 23:52:10 Normal Normal 21000 Normal 3 50897 -256 23300 19700 20100 3386 3396 Dischg Normal Normal Normal 100% 2025-01-30 23:52:08 Normal Normal 21300 Normal 4 50906 -256 22300 18900 19300 3386 3396 Dischg Normal Normal Normal 100% 2025-01-30 23:52:08 Normal Normal 20600 Normal 5 50895 -258 22000 19000 19400 3385 3395 Dischg Normal Normal Normal 100% 2025-01-30 23:52:08 Normal Normal 20200 Normal 6 50896 -266 21900 18300 18600 3385 3397 Dischg Normal Normal Normal 100% 2025-01-30 23:52:08 Normal Normal 19900 Normal 7 50896 -260 20900 17900 18200 3386 3395 Dischg Normal Normal Normal 100% 2025-01-30 23:52:08 Normal Normal 19200 Normal 8 50901 -264 20600 17100 17500 3387 3397 Dischg Normal Normal Normal 100% 2025-01-30 23:52:08 Normal Normal 18900 Normal 9 - - - - - - - Absent - - - - - - - 10 - - - - - - - Absent - - - - - - - 11 - - - - - - - Absent - - - - - - - 12 - - - - - - - Absent - - - - - - - 13 - - - - - - - Absent - - - - - - - 14 - - - - - - - Absent - - - - - - - 15 - - - - - - - Absent - - - - - - - 16 - - - - - - - Absent - - - - - - - CYCLE Times : 140 CYCLE Times : 141 CYCLE Times : 246 CYCLE Times : 233 CYCLE Times : 240 CYCLE Times : 249 CYCLE Times : 249 CYCLE Times : 247  
At a first look, it's not that bad with 96% SOH on a 6yrs old battery. But on the other hand, it's been 4% down for just roughly 500 cycles it made in that time. If linear, then after around 2500 cycles made in total, it will be at 80% SOH. If 500 cycles took me 6yrs then after 2500 cycles done in total the battery will be 30 years old. No way the battery will last for that long. Based on this, I would say that the battery will die of age well before it dies on cycles.
What do you think?
Youda

Technically the inverter is not sending voltage, but it's sending current. It's so called "CC mode". The current is being split between all the batteries based on their internal resistance, interconnection resistance and willingness to accept charge that changes with SOC of the individual cells. For the same voltage, different batteries will have different charging current. The 53,5V is the voltage value that the inverter is looking for to stop sending current (stop charging).
Some of the events that signalize end of charging (CC mode) and transfer to the CV mode:
If the battery reports 53,5V the inverter stops charging, even if the cells would still accept current. If the cells stops accepting current, the charging is finished. This might occur even before the cells reach 53,5V. If the BMS reports 100% SOC, inverter will stop charging. If the BMS reports "Advise Charge = 0A", inverter will stop charging. If the BMS reports one of the many possible Errors and Alarms, inverter will stop charging. Most of the time CMOS will be activated by the BMS, so even if the inverter would still continue with the charging, the actual current will be zero amps. Pylontech BMS is really doing nothing to the current that is flowing from the inverter. It has two protective MOSFETs (DMOS and CMOS), that are being operated in the ON/OFF state based on alarms. No analog current modulation feature is in the BMS.

Also, it is important to understand that physically, there is nothing like 100% SOC for the cell.
A LFP cell is just like an air-filled birthday party balloon:
When is the balloon 100% full? Is it the state when you blow another air molecule into it and it pops? If so, how does it depend on the balloon's temperature? If that 100% full mark is somewhere lower, where more additional molecules would safely fit inside, can I blow some more air into the balloon? How much? The "curse" of LFP is that they are so energy efficient that you can damage them even with just 1mA of charging current, if applied for a long time. With Lead-Acid batteries that would be absolutely no problem, but LFP stores lithium ions in the hard 3D lattice. If there's no more space for incoming ions, the lattice will crush and gas-mix of O2+CO2+phosphate will be produced.
And that is the reason, why some cells are swollen. Either they had manufacturing defect (most of the time it's the the separator failure), or something was hitting them with just a few miliamps of current for a long time, so the lattice started to break.
 
Yes, it would be great to have a BMS that can perform sophisticated monitoring of the cells, but in reality all the BMS-es are just measuring temperature, voltage, current and doing time-based integration of Coulombs in order to "guess" SOC. But keep in mind that more complicated electronic device is, more likely will it fail. Not to mention that it will cost more, consume more energy and produce more heat.
While the Pylontech BMS is far from being perfect, the solution is not to improve and over-engineer the BMS itself, but to configure whole solar system in a way that the margin between "balloon full" and "balloon popped" states will be reasonably wide and safe.
 

The 92 % might be way too hi,. Its good to note you have implemented this control system and look forward to what you  will discover and test. 
 

On @BritishRacingGreen suggestion, I setup now a automation in HA to reduce charging speed to 10A when one of batteries reaches 92%. It works fine, still noticed a 3.56V for one cell, but again, very brief, under a second. Now lowered the charging threshold to 5A and will monitor how it is, it should be good, we'll see...

Yes, if you have access to SOC metric in HA, you can leverage this well to affect even better control, along with the pack voltage. 
Care needs be exercise though that one must limit HA to change settings of the inverter  to  not exceed  30 changes per day. You might exceed the wear limit of the EEPROM nonvolatile storage  in say 10 years time given a 100000 write cycles of EEPROM. @Coulomb might be able shed some light here on wether modern EEPROM / FLASH  still has tight constraints. 

From the PIP-4048MS etc index:
[url=https://forums.aeva.asn.au/viewtopic.php?p=63091#p63091]Eeprom wear and saving parameters[/url] 
Basically, the Eeproms are rated for a million cycles, so that should not be a problem. Whether that rating is realistic or not, I have no idea. There are a number of glitches that can be attributed to Eeproms, but I suspect that they are firmware race condition issues. I doubt that frequent writing to eeprom would noticeably increase the chances of such a glitch. 

Thanks for sharing this. Well, I was asking as noticed that using BMS link, from time to time, I can see cell voltages as 3.57V for very short period of time. And overall battery sometimes reaches 53.4V. These are just very short peaks, these voltages never stay longer than a second or two. I'm just recording such maximum values with Home Assistant connected to Solar Assistant via mqtt. Thus wondering if it make any sense to use some predefined voltages in Deye configuration instead BMS...