diy solar

diy solar

Feedback on a battery pack approach

brum

Solar Enthusiast
Joined
Feb 8, 2022
Messages
495
I'm looking for feedback on the approach I'm taking with my first LiFePo4 battery. A small one, but the idea is that this will prepare me for the next one (~8-10kWh) that will provide electricity for the household).

The goal for the first battery is to replace a bulky 80Ah Exide Equipment Gel leisure battery used for camping purposes. Average daily consumption is ~30-40Ah. Most of this is used while the sun is shining, so I suppose that even a 30Ah battery will do the job, but I've decided to go with 100Ah just in case we get a cloudy day. For charging, I use an Epever Tracer 1206AN and 160W panels with 36V MPPV (higher voltage, less current, cheaper and lightweight cables).

I started with just a bit of knowledge - 4 LiFePo4 cells are supposed to form a battery that will suit my needs. Based on the positive feedback for the Docan store in Alibaba I ordered 4 cells. I ordered 100Ah plastic CALB cells as I was aware that they are reliable. Unfortunately, these were out of stock and I ended up with 105Ah EVE cells. While waiting for them to arrive I started picking up the other components.

The BMS was ordered from Aliexpress - Hankzor products look promising and the one I picked has the following parameters:
* Charging protection voltage: 3.65v
* Discharge protection voltage: 2.35v
* Charge release voltage: 3.55v
* Discharge release voltage: 2.55V
* Equalizing voltage 3.405v
* Equalizing current: 194ma
* The maximum continuous over-current value is 120A
* Maximum charging current: 60A

The equalization currently is likely going to be reduced after a few tests. The equalization voltage of 3.4V is what caught my eye. I will use a lower charging voltage (14V probably) to be a bit more on the safe side and to try to keep the cells below 90% SOC.

Next comes the fuse. EVE cells are 0.3mOhm and the maximum current is 12kA. 20kA to be a bit more on the safe side seems enough. Luckily my load is low and a cheap 32A 10x38mm fuse that has a breaking current capacity of 50kA will do the job. While looking for it I saw its bigger brother - 22x58mm fuses with a breaking capacity of 120kA. These will be used for the big battery. A lot cheaper compared to the T class fuses. I'm in Europe and I suppose this makes a difference to what I can easily find in the store. T class fuses are hardly available around, but the listed fuses are pretty easy to source.

Next comes the clamping force. The specs say 300kg. But the same load applies to the big EVE 280Ah cells that have a larger surface area. At 12psi the load should be 200kg and I plan to stick to that number. I'll use aluminum plates on both sides, 8 x m4 studs, and 8 springs. Between the cells and the aluminum plates I'll put an insulation layer. Most likely the same material that is used for making PCBs (without the copper part).

Then come the bus bars. The stock ones are straight and quite long. The width of the 105Ah cells is around 37mm. The current option that I'm considering is to laser cut thin aluminum bus bars (0.6-0.8mm) and stack 2 or 3 of them. These will be bent in the middle to remove the stress from the terminals.

Then comes the pack design. The picture shows the current idea. The BMS (green) and the fuse (blue) are on the same side as the springs. On the top will be placed an additional isolation block to protect the cell terminals and to provide a base for fixing the wires.

The battery connecting terminal will be an XT90 fixed to the outer enclosure. The current flowing from and to the battery will be no more than 10A and the XT90 is more than sufficient.

The outer enclosure will probably be made of plywood. Still checking the other options.

There is one concern I have. I'm reading that the charging of LiFePo4 cells should be terminated at 0.05C. I'm planning to charge the batteries with 0.1C to 0.05C. This opens a question - is limiting the charging voltage to a lower value going to limit the SOC? And if yes what would be the recommended value for the bulk charging and the float charging voltages?
 

Attachments

  • 1.jpg
    1.jpg
    33.8 KB · Views: 12
I found the sticky post named "Recommended Charge Profile for DIY LiFePO4 Batteries" and sorted the concerns with the charging and the floating voltages.
 
Charging voltage should be at least 50 mV above the start of the balance voltage. I believe this was listed above as "equalizing voltage 3.405"
I set my absorption time fairly short at about 15 minutes. This time may need to be extended at lower charging voltages.
 
Thanks for the advice! I do have a coulomb counter and I will try to identify what is the proper absorption time to keep the SOC ~90%.

Some updated on the design - finding suitable springs happened to be a challenging process. Rated springs do cost a lot (they don't sell just 10 pieces, there is a minimum number of springs to be ordered as they are being manufactured on demand). So I got what I found in the shop and tested a few different options. Testing was needed as they sell springs by dimensions, not specs. Finally, one that has 2.5mm wire diameter and 15mm outside diameter showed to be a good candidate. It takes ~50kg @ ~ 36mm compression and this is almost half of the total possible travel. So I'm moving to 4 rods and 4 springs.

Still wondering should I go with 200 or 300 kgF. Specs say 300kgF and maybe I'll just check email EVE for clarification. We'll see if they will pay attention to a small fish like me :) .

The outer enclosure will be 3D printed. The last drafts of the design show that the pack will be ~155mm in width and ~215mm in height. And my printer can handle up to 230/230. The cell separators will also be printed. They will incorporate small venting channels to improve heat dissipation. This is "a bit" of overengineering, but the pleasure of doing something perfect will justify it. I don't expect these cells to heat at all, as I'll be charging them with at most 10A.
 
Tested a bit the BMS and it seems OK. High and low cell voltage cutoff works as expected.

Cell balancing is a bit strange, but I suppose it will do the job. The balancing process gets activated only when there are cells above and below the threshold voltage of 3.405V. Until that condition is true the cells that are higher than 4.05V are getting discharged. The closer the cell gets to 4.05V, the lower the discharge current gets. But otherwise, the specs are correct - almost 200mA balancing current.
 
I'm used to knowing what stuff that I'm using does, so I removed the BMS cover and identified the IC inside - BM3451-BHDC-T28A. The specs of BM3451 confirmed my findings. This is how the BMS is expected to work.

I'm unable to comprehend the "during charging" detail. So I'll have to test it to confirm it. But I suspect that the first sentence statement stands correct and if all cells are above the Vbal voltage no action would be expected from the balancing circuits.

6. Balance Function
Cells’ balance function is used to balance the cells’ capacity in a pack. When all voltages of VC1, (VC2-VC1), (VC3-VC2), (VC4-VC3) and (VC5-VC4) are lower or higher than VBAL , all the external balance discharge circuits will not work. Otherwise the cell, whose voltage is higher than VBAL, will turn on the external discharge circuit and make its voltage lower than VBAL. During charging, If the highest voltage of five cells enters overcharge state and its cell balance circuit turns on, the charge control MOSFET turns off and the external discharge circuit works and makes the battery voltage fall down to VREL1 which is the overcharge release threshold, then turn on the charge control MOSFET for continuing charge .For a long enough time of charge and discharge cycles, the voltages of all cells will reach to more than VBAL, and avoid the capacity differences between batteries.
 
So far so good. The cells arrived a week ago and today I managed to assemble the compression part.

The plates are 5mm thick aluminum (some hard version, not sure about the exact type). Springs are putting a bit over 200kgf and have around 5mm more travel. The studs are 6mm and are covered with a soft PVC tube around the cells. The spacers are 3D printed and they do a pretty good job of aligning the batteries during the assembly process. Spacers have 2mm thick vertical bars and 0.5mm thick horizontal bars. The idea behind this is to allow some form of airflow (not that such is needed with the expected current of ~10A flowing through them).

The idea has evolved a bit. It's no longer going to be just a battery, but a device that includes both the battery and an SCC in the same enclosure.

On the side of the studs I'll have two more layers - the first will facilitate the BMS and the fuse, and the second will facilitate the SCC (Epever Tracer 1206AN). This will be enclosed in a plywood or MDF box with a few ports - PV input, output, RJ45 for the MP50 and a disconnect switch. I'll carve handles at both sides and there will be some mechanism for fixing this under the driver seat of a VW T5. Something like portable power stations, but without an inverter.

I'm wondering what expansion can I expect from compressed cells and if close to none - can I go with the solid busbars that came with the cells or should I be looking for a flexible option...
 

Attachments

  • cells.jpg
    cells.jpg
    22.1 KB · Views: 17
  • spacer.jpg
    spacer.jpg
    45.9 KB · Views: 17
Its great that you are reading and experimenting with a small system first before going big. Thumbs up!

My advice is that the "equalizing voltage" is far too low. You will *always* be equalizing, and when equalizing, this should be done when a cell is well up into the knee, not at the start like at 3.4.

Raise that equalizing voltage to 3.5v and try again and monitor. The whole point is that if well balanced, there is NO equalization at all, other than *minor* touchups. If things are really going well, try 3.55v.

In the past, many bms manufacturers would either sell or let you specify which eq voltage to set depending on model. (HousePower bms for you old-timers) Some started their businesses too conservatively at 3.4v, while others blindly went to the other extreme of 3.6! In the years following, most settled on 3.5 to 3.55.

Don't get too hung up on voltage determining your SOC. What determines that is capacity - 100% full.

You can achieve a 100% full SOC, by using any voltage between 3.42 to 3.6v / cell - the only difference is a matter of *time*. 3.42v / cell will *eventually* reach a full capacity. 3.6v / cell will simply do it quicker. Much quicker.

Those who argue this never actually take the time to test it and do a capacity test since charging at 3.42/cell takes a long time to reach 100% soc. But eventually it will and has presented degradation to those using LFP for ups duty sitting at full charge, even though their charge voltage was only 13.8v. Over time they got to 100% SOC and plated the lithium. Another story, so nuff' of that.

But again, thumbs up by building and testing your small setup first.
 
Agree with the fact, that voltage is not a guarantee for keeping the cells below 100% SOC. But it is at least a step in that direction. I suppose I'll still be getting to ~100% SOC due to the low charging current. The SCC and the panels produce at most 8A from what I've seen with the lead-acid battery. But even with that, I'll be getting more than a few thousand charging cycles and for camping purposes, this is more than sufficient.

Regarding the balancing voltage threshold of 3.4V - the only option to change it is by getting a new BMS. The 3.4V comes from the IC and is kind of hard-coded. I neither have the time to wait nor do I wish to throw away the available and tested BMS. So I'm staying with the 3.4 volts.

And yep, lessons learned. That was the goal for doing it the "right" way from the first time. The knowledge that I've managed to build will definitely be useful with the real thing coming up next.
 
Last edited:
Finally, got the battery running. The internals of the device is ready. The BMS and the fuse were mounted on the first layer, the SCC was mounted on the second layer. Next is the external enclosure. Likely 8mm or 12mm MDF, as this is what I have.

The first start of the battery was a bit tricky. If there is some load when the balance leads are being connected the BMS doesn't start. My only load was the SCC and to get the BMS running I had to disconnect it before plugging the balance leads. Took me half an hour to find it out.

These days I plan to balance the batteries and do a full charge/discharge cycle. I'll check the dimensions of the batteries in both cases. If they don't change I'll keep the solid busbars. The other option is 6mm wire and heavy-duty lugs.

And this is not the final wiring. I'll cut a bit the balance leads and I'll add a battery disconnect switch between the fuse and the SCC. The stainless steel M4 studs will replace the stock terminal bolts.
 

Attachments

  • 2.jpg
    2.jpg
    121.4 KB · Views: 16
  • 1.jpg
    1.jpg
    172.7 KB · Views: 16
  • 3.jpg
    3.jpg
    200.5 KB · Views: 15
Here are the 16mm M4 stainless steel studs with a red thread lock. I suppose this can be considered a permanent mount now.

The first cell took a bit over 50Ah. At around the 45th amp hour, the voltage was still 3.35V. The last few Ah took it up to 3.65V. Seems like with a low charging current 3.4V is more than enough to go over 90% SOC.

The distance between one row of studs was 115.98mm before I started charging them. After the first cell was fully charged the number stayed the same. I'll check it after each cell to see if there will be some cell expansion, but I do have some doubts that there will be such. And if so - the solid busbars should be good to go.
 

Attachments

  • 1.jpg
    1.jpg
    176.1 KB · Views: 8
And the distance is still 115.98mm with all 4 cells charged. I guess that this is either the compression fully preventing the cell expansion or charging at a low rate resulting in no expansion.

I'll use the solid busbars.
 
Busbars and wires are in the final arrangement. The items on the stud (stainless steel) are as follows:
1) silicone grease
2) tin-plated copper busbar (or lug)
3) stainless steel washer (to spread the pressure on the whole busbar area)
4) spring washer (zinc plated, without a cut, to guarantee some pressure)
5) balancing lead
6) oxidized stainless steel nut

Next is the enclosure. The ports I'm planning to have are:
1) XT60 direct to the battery (charging purposes, critical loads that bypass the SCC and its load disconnect functionality)
2) 3 * XT60 through the SCC and its load disconnect functionality
3) XT60 for PV input
4) Access to the SCC port for connecting an external display (Epever MT50)

There will be a protective 3D printed cap above the battery. All but the front side would be from 8mm HDF. The front side with the ports will be 3D printed and will be mounted with nuts on the remaining of the studs used for the compression enclosure. There will be vents on the sides. The handles will also be on the side and they will act both as handles and as vents.

I may be repeating myself, but the whole thing will be a 12V portable power station with integrated SCC. The goal as said somewhere in the beginning is to provide power when we are camping in the wild. The main load comes from a compressor fridge (~20-30Ah per day). The PV panels are 2x80W producing ~36V at MPP.
 

Attachments

  • 1.jpg
    1.jpg
    238.4 KB · Views: 14
Those springs don’t look like they have any springy springs left to go.
 
Some more progress. The enclosure is almost ready and the front panel is in progress.

The grey "caps" around the terminals are used to secure the cells from vertical movement in the enclosure. The terminals are used only to center them. They are putting pressure on the cells around the terminals. The high of these is such, that there is ~ 0.05 to 0.01mm gap between the enclosure top side and them.

The front panel has 3D printed. The switch may not be the most appropriate one, but this is what I had.

From left to right there are:
1) Direct battery output (not going through the SCC)
2) SCC disconnect switch
3) SCC communication port (for EPEVER MT50 external display)
4) Two outputs that the SCC can cut out if the battery gets low
5) PV input

This is pretty much all I need. The holes are positioned in the right place to provide ventilation for the SCC heat sink. The only possible places where somebody can interact with the cells are the handles and this will require some (minimal) care to avoid incidents.

The whole thing is disassemblable and if maintenance is needed it will be performed with ease.

Next - paint job (likely) and soldering behind the pannel to get everything connected.
 

Attachments

  • 2.jpg
    2.jpg
    199.4 KB · Views: 14
  • 1.jpg
    1.jpg
    104.3 KB · Views: 13
Wiring completed. And it works :) !
 

Attachments

  • m.jpg
    m.jpg
    153.1 KB · Views: 8
  • q.jpg
    q.jpg
    193.7 KB · Views: 8
And it's done!

I might disassemble it once again to get one more transparent coat on top of the wood box, but that's all.

A bit disappointed with the SCC. There is a 0.2V drop on it with a 6A load. But I'll live with it. In and out wires are 4mm2 and I don't think it is a wiring issue.
 

Attachments

  • 2.jpg
    2.jpg
    91.7 KB · Views: 14
  • 1.jpg
    1.jpg
    104.8 KB · Views: 13
Nice clean job.

The more and more I see you guys with 3d printed parts it’s wanting me to snag one
 
The boost voltage was increased to 13.9V. The SCC reads the cell voltage with 0.1V lower than it is. There is a minor voltage drop across wires, fuse, switch, and connectors (0.05-0.06V at 8A) also. With the current settings, the SCC gets the cells at ~3.45V, keeps them there for around 10 minutes, and then goes in float mode. The cell voltage after 10 more minutes drops to 3.398-3.399V. This is the disbalance between cells - 0.001V at that point.

The busbars produce a higher voltage drop on connection points than the lugs. The low-quality lugs I have are giving ~0.5mV voltage difference between the cell terminal and the lug. The busbars are producing 2-3 times more (consistent on all connections). This is at ~8A load. The terminals were cleaned from oxidation and the busbars were covered with silicone grease. The busbars themself can't be sanded at all - the copper starts to show up.

I'm leaving it as it is. The voltage drop on the busbars is not going to be an issue for my usage conditions. I have ordered high quality lugs though and I may reconsider it.
 
Back
Top