DIY 16 kWh home battery Q&A: cell compression, orientation, and plug-in myths

Why this video exists

After Robbert published the build video for a cheap 48V home battery, the comments filled up with technical follow-ups. Some of them were genuinely good corrections. Others were misconceptions worth unpacking on camera. This Q&A goes through them one by one: cell compression, foam sheets, orientation, stud variants, torque, system compatibility, mixing packs, and a hard "no" on plug-in batteries.

Cell compression: skip it

LiFePO4 cells swell during charging. The lithium ions migrate into the graphite anode, the graphite expands at the atomic level, and the cell physically grows a bit. The force behind that expansion is enormous. A thin sheet of foam does not stop it. A heavy compression frame does not stop it either.

The cell datasheet shows around 4000 cycles without compression, roughly 2500 with. After 2500 cycles the cell still holds about 80% of its capacity. For a household running one full cycle per day, that translates to 10 to 20 years of useful life. The 4000 to 8000 cycle figures only appear in lab tests with multiple cycles per day.

Robbert's take: the small extra lifespan gained from compression is not worth the build effort, especially since BMS hardware and battery tech will move on within those 20 years anyway. He compares it to advertised fuel economy on cars: lab numbers under ideal conditions, real-world results 20 to 40% worse.

What does matter is keeping cells in place so they cannot slide. Movement stresses the busbars, which is exactly why most LiFePO4 packs use flexible busbars that absorb the expansion and contraction.

Foam sheets vs. epoxy plates

A common comment was about foam patches between cells. Foam will not absorb cell expansion forces. The yellow epoxy plates Robbert uses serve a different purpose:

• Electrical isolation so adjacent cells cannot short
• Thermal isolation
• Prevent the cells from grinding against each other during expansion and contraction

Can cells lie on their side?

Yes and no, and there is real debate about it. In the ideal case, cells stand upright with the vent facing up. They can function in any orientation. The electrolyte argument (it will not reach all the internals) does not apply to LiFePO4: these are not flooded lead-acid batteries, the electrolyte is not pourable.

The real reason upright is safer:

• During a thermal runaway, the cell needs to release pressure through its vent. A cell lying flat can have the vent blocked by electrolyte or debris, which prevents safe outgassing.
• Stacking cells on their narrow flat side puts mechanical pressure on the wide face, which can deform the cell.

Short answer: cells work on their side, but upright is better for both lifespan and safety.

Single stud vs. double stud (EVE MB31)

The cells underneath the terminal are identical. The difference is the top:

Both work fine. Robbert used the double stud version for the build video only because the single stud cells were on long backorder when he wanted to film. He did not want to make viewers wait.

Torque the terminal screws

A loose terminal screw makes poor contact, raises resistance, gets hot, and can start a fire. Over-tighten and the terminal snaps off. Use a torque wrench. The EVE MB31 single stud spec is 6 Nm. Robbert recommends going slightly below the max, around 5.5 Nm, so the joint is tight but the terminal is not stressed at the limit.

Is a 51.2V pack compatible with a 48V inverter?

Yes. The "48V" label is a nominal value carried over from lead-acid conventions. Real inverters work across a voltage range because battery voltage varies between empty and full.

A Victron Multiplus accepts 38V to 66V on its DC input. The 16S LiFePO4 pack stays inside that range from 0 to 100% state of charge. The same logic applies to any reputable 48V inverter.

Will the BMS talk to your system?

This is the harder question. Every brand uses its own communication protocol. The JK BMS is unusually flexible here: it can speak many protocols, and the target protocol is selectable in the BMS settings. The JK pack also has a standard communication board on the bottom for a data cable. Configure it for Victron, plug it in, and the pack reports state of charge and other data to the Victron set as expected.

No balance PCB on top: why?

Higher-end cases ship with a balance PCB across the cell tops, with clip-on leads to each terminal. Cheaper cases (like the one in the build) require running the red BMS leads directly to terminal lugs. Robbert's view: the PCB version costs roughly 100 euros more. Ten minutes of extra wiring is a fair trade.

Shortening the BMS leads is fine. The resistance difference is negligible. Coiling and tie-wrapping the excess is the cleaner option.

Can you add this pack to an existing battery?

Only if the existing pack is also LiFePO4, also 16S (48V class). Capacity differences are allowed.

The critical part is matching state of charge before paralleling. A 100% pack connected straight to a 50% pack will dump current at unsafe levels: cables can melt, BMS units trip, fuses blow. Process:

1. Charge the existing pack to 100% and disconnect it.
2. Connect the new pack and charge it to 100%.
3. With a multimeter, verify both packs read nearly identical voltages.
4. Once voltages match, parallel them onto the same busbar.

After paralleling, update the inverter software with the new total capacity (16 kWh + 16 kWh = 32 kWh).

Why plug-in home batteries are a bad idea

This is the most direct part of the Q&A. The hard no is for plug-and-play units like the HomeWizard battery. The reasoning is electrical, not aesthetic.

Household wiring is sized for the breaker on each circuit, usually 16A. Plug an inverter or battery into a wall outlet and you can defeat that protection.

Example: a 2400W (10A) inverter is plugged into a circuit also feeding a dishwasher, oven, air fryer, induction hob, and a coffee machine drawing a combined 24A. The 10A from the inverter does not pass through the breaker; only 14A does. The breaker stays closed at 14A. But the actual wires in the wall carry the full 24A.

The cables are overloaded by 50% without any breaker tripping. This is also exactly why solar inverters are required to sit on a dedicated breaker.

The HomeWizard product itself adds another problem: 2.5 kWh of storage, 800W charge and discharge. On a sunny day with 2500W of solar input, the battery absorbs 800W and the remaining 1700W still feeds back to the grid. On a 2000W dishwasher load, the battery contributes 800W and the rest comes from the grid. Stacking multiple HomeWizard units multiplies the capacity but not the throughput. Each unit is still 800W.

A cleaner alternative: dedicate a breaker in the meter box, run two wires and a ground to a Victron Multiplus GX (which has the control already built in), and connect any battery pack to it.

The Victron path costs less, stores more than twice as much, charges and discharges three times faster, captures full solar production, and is modular for later expansion.

Other viewer questions

Weight: about 110 kg for one 16 kWh pack. 16 cells at 5.6 kg each is 90 kg, plus 22 kg for the steel enclosure. The table holding it during the build was not having a good day.

Dimensions assembled: 72 cm long, 42 cm wide, 27 cm high. Shipping carton is flatter: 93 cm by 52 cm by 13 cm.

Takeaways

• Compression is not worth the effort. Hold cells in place, do not try to crush them.
• Yellow epoxy plates between cells. No foam.
• Stand cells upright when you can. They work flat, they are safer upright.
• Single stud is fine. Double stud is a small upgrade. Torque either to spec.
• 51.2V LiFePO4 packs are compatible with any reputable "48V" inverter.
• Never parallel packs with different states of charge.
• Plug-in batteries trade safety and headroom for convenience. Run a real breaker and a real inverter instead.