Safety clarification request: AEK-POW-BMS63EN + AEK-POW-BMSHOLD with 7-cell configuration

[kaleed]

Dear ST AutoDevKit Support Team,

I am working with the ST AutoDevKit BMS setup and would like to request your technical guidance before powering the system again, because I want to avoid damaging the AEK-POW-BMS63EN board, the L9963E IC, the AEK-POW-BMSHOLD holder, or the cell sense wiring.

My hardware setup is:

• AEK-POW-BMS63EN BMS node board

• AEK-POW-BMSHOLD 14-cell holder

• SPC58EC / AutoDevKit environment

• LG INR18650 MJ1 cells

• Rigol DP711 bench power supply used in CC/CV mode for charging, supports up to 30 volts (7 cells constrains)

• Rigol DL3021 electronic load for discharge tests

I would like to operate the system with 7 cells only, using the following physical configuration in the 14-cell holder:

• First 5 cell positions populated

• Middle positions unused

• Last 2 cell positions populated

The intended logical L9963E cell mapping would be:

• Active cells:

  • CELL0–CELL1

  • CELL1–CELL2

  • CELL2–CELL3

  • CELL3–CELL4

  • CELL4–CELL5

  • CELL12–CELL13

  • CELL13–CELL14

• Unused/dummy channels:

  • CELL5–CELL6

  • CELL6–CELL7

  • CELL7–CELL8

  • CELL8–CELL9

  • CELL9–CELL10

  • CELL10–CELL11

  • CELL11–CELL12

I understand from the L9963E documentation that the device can manage fewer than 14 cells, but I also found information suggesting that the AEK-POW-BMSHOLD may be wired specifically for exactly 14 cells in series. I would like to confirm the correct and safe implementation for this exact case.

During previous testing, I shorted some unused sense nodes with thin wires in the BMS board pins. The measured DC voltage difference between the unused sense nodes was around 0.02 V, but when connecting or switching the DP711 supply, a sense wire burned and the L9963E IC became very hot. This suggests that the issue may involve the physical holder current path, transient current, incorrect sense wiring sequence, or sense wires accidentally carrying charge current.

Could you please clarify the following points?

1. Can the AEK-POW-BMSHOLD safely be used with only 7 physical cells installed, specifically using the first 5 positions and last 2 positions? (or is there anyway to make it work??)

2. If middle cell positions are empty, does the holder provide any high-current continuity across those empty positions, or must those empty cell positions be bridged with current-rated conductors?

3. Should the unused L9963E cell sense inputs be directly shorted together, connected through resistors, or connected according to another recommended topology?

4. For the above 7-cell mapping, exactly which CELLx pins should be shorted or connected, and should the unused group CELL6–CELL12 be tied to CELL5, CELL12, or both?

5. In AutoDevKit Studio / BMS GUI configuration, which cells should be enabled or disabled for voltage measurement, diagnostics, undervoltage/overvoltage checks, open-wire diagnostics, and passive balancing?

6. Should passive balancing be completely disabled on unused/dummy channels? Is it safe to balance active cells adjacent to the dummy region, especially CELL4–CELL5 and CELL12–CELL13?

7. What is the recommended power-up and power-down sequence when using the AEK-POW-BMS63EN, AEK-POW-BMSHOLD, a bench power supply charger, and an electronic load?

8. If the L9963E became hot during a previous miswiring event, is there a recommended diagnostic procedure to check whether the IC or input network is damaged before reconnecting cells or/and the BMSholder?

My main concern is that I do not want the sense wires or CELLx inputs to become part of the main charge/discharge current path. I would appreciate a recommended wiring diagram or exact connection table for a safe 7-cell setup if this configuration is supported.

If the first-5-plus-last-2 physical configuration is not supported with the AEK-POW-BMSHOLD, please confirm the safest alternative configuration, using only contiguous 7 cells, or using a custom 7S holder/harness connected to the AEK-POW-BMS63EN.

Thank you very much for your support.

PS: here some images of the sense part got connected and the other one showing the exposed part that got burn

[SoTenhoM]

I would like to find out how to use the kit with less than 14 cells too.

I know you can make changes to the configure.xlm file to disable cells and adjust the pack over-voltage and pack under-voltage to the number of cells. It compiles and debugs with no problems.

I'll wait for answers to this post!

[kaleed]

Just to follow up, I am using 6 Cells now as attached below, connecting cell 1-4 and 13-14 

 

• Soldering Cell 4 to the unused one (5, 6, 7, 8, 9, 10, 11, 12) also shorted S5, B6_5, S6, S7, S8, B8_7, S9, B10_9, S10, S11, B12_11, S12 with Cell 4 (checked using DMM)

• The unused cells show approximately:

  Cell5–Cell12 ≈ 0.01–0.02 V

  Software configuration

  In the AutoDevKit configuration, I enabled only the active cells:

  Cell1, Cell2, Cell3, Cell4, Cell13, Cell14

  I disabled:

  Cell5–Cell12

  I also modified the application code to prevent balancing on the unused cells. In the GUI, balancing appears active only for the enabled cells and OFF for the unused cells.
  
  

  Problem

  When the board is connected but the L9963E is not active, there is no obvious heating.

  However, when the MCU activates the BMS / L9963E starts operating, the board begins to heat. The heating appears to start around the JP1 area, especially around the RG / CGS / RPD jumper region. and also around P3 in the other edge and around senses of Cell 13 and 14

  I tested different JP1 positions and even removed the JP1 plug, but the pins/area still become hot when the L9963E is active.

  Measurements observed

  When the L9963E becomes active:

  JP1 to PACK− ≈ 0 Ω

  Also, measuring across used cell sense paths gives high resistance, around:

  39–43 MΩ

  But across the unused/bypassed region, I measure approximately:

  0–2 Ω

  In particular:

  Cell4+ to Cell13− ≈ 0.5 Ω

  Pack+ to Pack− is not readable in resistance mode.

  Question

  Could you please for the AEK-POW-BMS63EN / L9963E clarify the correct unused-cell connection for this exact configuration?

  The configuration is:

  6 physical cells:
  Cell1, Cell2, Cell3, Cell4, Cell13, Cell14
  
  Unused:
  Cell5, Cell6, Cell7, Cell8, Cell9, Cell10, Cell11, Cell12

  Specifically, I would like to confirm:

  1. For unused Cell5–Cell12, which exact CELLx / Sx / Bx_x / Cx pins must be shorted together?
  2. Should the balancing pins of unused cells also be shorted, or only the cell sense pins?
  3. Are some balancing pins intentionally left unconnected in this configuration?
  4. Is a near-0 Ω path between Cell4+ and Cell13− expected for this bypassed unused-cell region, or does that indicate an incorrect bypass?

  The board communicates correctly, the GUI connects correctly, and the measured active-cell voltages look reasonable. The issue is specifically the heating near JP1 / RG-CGS-RPD when the L9963E becomes active with the unused-cell bypass installed.

  Any official clarification or recommended wiring diagram for a non-contiguous 6-cell setup using cells 1–4 and 13–14 would be very helpful.

  Thank you.
  
  PS: attached the GUI which shows the FAULT even though I disactivate the unused Cells! Any help kindly appreciated!

[kaleed]

Update / resolution:

I wanted to share the result in case it helps others using the AEK-POW-BMS63EN + AEK-POW-BMSHOLD setup with fewer than 14 cells.

The issue was solved without changing the physical wiring from what explained above! The hardware connection stayed the same. The required fix was on the firmware/application side.

Final working pack configuration:

• 6 cells in series
• Active cell positions:
  • CELL1, CELL2, CELL3, CELL4, CELL13, CELL14
• Unused positions:
  • CELL5-CELL12

Because of the holder/L9963E topology, this is a non-contiguous active-cell configuration. The important firmware point was to make sure that all measurement, protection, and especially balancing logic respected only the real active cells.

The active-cell mask used in the application is:

#define AEK_POW_BMS63CHAIN_ACTIVE_CELL_MASK 0x300F

This corresponds to:

CELL1  -> 0x0001
CELL2  -> 0x0002
CELL3  -> 0x0004
CELL4  -> 0x0008
CELL13 -> 0x1000
CELL14 -> 0x2000

The unused cells CELL5-CELL12 must stay disabled and must never be selected for balancing.

The main firmware changes were:

1. Configure the project for the actual 6-cell pack topology.
2. Disable unused cells CELL5-CELL12.
3. Add an active-cell mask so every balancing request is filtered through the valid cells only.
4. Force balancing OFF by default during startup/testing.
5. Prevent manual balancing commands from enabling unused cells.
6. Prevent automatic balancing from selecting unused cells.
7. Add safety checks so balancing is blocked if BMS communication, VREF, or measured cell voltages are invalid.
8. Add checks so automatic balancing does not run at low cell voltage or during invalid/stale measurement conditions.
9. Keep the UV/OV thresholds consistent with the real 6S pack behavior.
10. Add diagnostics to confirm the actual selected balance mask, SPI/BMS status, and measured cell values.

After these firmware changes:

- The same hardware wiring worked.
- The heating problem disappeared during normal operation.
- Cell voltage readings became realistic and stable.
- VREF was around 4.97 V.
- SPI/BMS communication status was healthy.
- Manual and automatic balancing worked, but only on the valid active cells.
- Unused cells CELL5-CELL12 were no longer selected for balancing.

So the root cause in my case was not an incorrect physical connection, but that the demo/application firmware needed to be adapted for the non-contiguous 6-cell configuration.

For safety, I would still appreciate ST confirmation that this is the recommended way to configure a 6-cell pack on the 14-cell holder when using CELL1-CELL4 and CELL13-CELL14, especially regarding unused-cell handling and balancing protection.

I also documented the firmware changes and experimental setup here for anyone interested:

https://github.com/khaledalamin/aek-pow-bms63en-6cell-balancing

This repository is not official ST material and not a certified BMS design. It is an experimental/educational reference showing the firmware changes I used for the 6-cell configuration, including the active-cell mask, unused-cell balancing protection, serial balancing control, safety checks, and diagnostics.

I hope this helps others who are trying to adapt the AEK-POW-BMS63EN / L9963E example to a non-contiguous 6-cell setup. I would still appreciate ST confirmation regarding the recommended handling of unused channels and balancing protection for this kind of configuration.

[db16122]

Thanks for your valuable test data and sharing. I also have one AUTODEVKIT at my pocket but it seems not easy to get the board running smoothly with limited resource on the websit as well...