Larger DIY BMSs?

Hardware
1

Hi all,
I realise this is not really the correct forum to ask this on - but this is the only place I could find where people are actively discussing real DIY BMSs…

If it is too far off topic, then please delete this thread.

I am contemplating another project, incidentally related to Open Energy Monitor (which I may open another thread on, if that is OK). But for this project, I would need a high voltage battery - at least 320V when empty - so 100s-150s depending on chemistry. Max 10Ah, but probably smaller.

I have been playing around with cell-top BMS designs for a while, pretty much on the same lines as diyBMS. But this is not great for such large packs. Number and size of boards. Slow comms and high current consumption with opto-isolated comms. And cost, which escalates quite quickly when going to 100s+ configurations.

I have played with some optimisations to reduce size and cost.

Capacitive coupled comms:
image

This gives much higher comms rate at virtually no cost - but can not tolerate more than ~0.3V difference between adjacent cells before comms break down.

Active balancing:
image

I have been testing this a bit. Can PWM the PMOS to transfer charge to the previous cell, or the NMOS to transfer charge to the next. Saves a lot of board space and gets rid of thermal management issues. Can run it feed-forward in DCM mode for a really simple design. BUT it results in fairly high current pulses for charging, and I have not been able to find any real research as to the long term effects this might have on the battery.

That takes the cell-top BMS about as far as it can go, but the complexity of wiring 100s of these BMSs up is daunting.

I have been looking at commercial BMICs as well.

L9961TR seems about the cheapest at $2 for 5 cells (plus probably another $1 for an extra uC to handle isolated inter-board comms). But accuracy is poor, and balance current is limited to 70mA - although this could be OK for a well matched low capacity pack.

L9963E is only a little more expensive at $9 for 14 cells with built in inter-board comms and a (fairly) minimal external BOM. 200mA balance current with internal transistors, but can also switch bigger external transistors for higher current balancing. But it requires a minimum 3 layer PCB for its fairly demanding analog layout requirements.

There are a few others out there, but they all seem to have similar capabilities/costs.

Now the questions…
Are there any DIY contenders out there for 100s packs?
Are there any DIY projects out there based on commercial BMICs?
If not, is anyone else interested in investigating one of the above 2 BMICs as part of a modular BMS system?

Any tips advice or other feedback will be greatly appreciated.
Thanks,
Justin

2

Hey Justin,

I’m working on this. trying to run 8 DIYBMS’s in series, to get 400v. Seems it should work, with one master, and 8 monitoring boxes. but, it hasn’t been tested. Have you tried anything yet? how many amps you doing?

3

I haven’t started building anything yet. I did a couple of cell-top BMS variants during lockdown, just for fun. But that is about it.

I am now contemplating building a solar EVSE. But I need a power ‘buffer’ to handle clouds/peak tracking, etc. (It takes 10-30s for the vehicle to reduce charge power after the EVSE signals a different capability. If voltage sags by more than 10%, the vehicle goes into fault mode and disconnects. So you need a buffer to keep up with demand when solar power varies.)

Boost converters are simpler and cheaper for the MPPT. And direct SPWM from a 320-400V battery is by far the simplest and most efficient inverter.

So I need a fairly low capacity 320-400V battery. The smallest I can build which will sustain 35A for 30s.

DIYBMS does not really seem like an option. It may be a fun project, but it will be ridiculously expensive at scale. Can even go commercial for a fraction of the price - just grab a few 24s Daly Smart Active Balancers and hook them up to a DIYBMS controller. Fairly simple protocol to use, and you can pump their data in as though you were 24 DIYBMS modules.

But I would like to stay DIY if I can, which is why is was looking at the bulk monitoring ICs (surprisingly, they are cheaper than high voltage differential ADCs on their own).

Will see how this pans out.

4

you might look at the newer 16s monitor boards that @stuart released recently. For that you would need only the controller and handful of the monitor boards (4-5 maybe?)

But definitely exercise caution as those voltage levels are considerably higher than any diy system would typically use!

5

Ah… Switch from master branch, and there is a whole new world of diyBMS…

1 Like
6

Are you looking at building a high voltage DC EVSE charger?

7

This video might be useful. https://youtu.be/VUp0K2B1ql0

8

Not at the moment. It seems really hard to find decent DIY resources on the DC charging standard.

Far simpler to build a basic inverter and do single phase AC charging.

9

Agreed - so why do you need a high voltage DC battery? Fairly easy to find a 10kW inverter and a suitable 48V LIFEPO4 battery to power a typical 7kW EVSE.

10

It is far sim

A number of reasons…

  1. I have not been able to find a high-C, low capacity 48V battery - need 200A output @ 48V, but don’t require much capacity.
  2. It is far simpler to build an efficient boost MPPT than a buck.
  3. When building an inverter, the boost circuit (to produce the 400V bus) is the most demanding to design/build. If you have a HV battery, you can go directly to 230V AC with an H-bridge and SPWM generator.
  • There are obviously safety issues with such a design. The entire battery is ‘live’, and if you use the simplest MPPT, then the solar panels are ‘live’ too.

Still wrestling with all the design trade-offs.

11

Hi Justin,

I have a diyBMS installation with 80s 230Ah LiFePo4 cells with a goodwe GW25K-ET inverter.
My battery is made out of prismatic cells with a special board layout to match the cell for fast and solder free building
This is what one enclosure looks like (I have 5 of them)

IMG20230922155238
IMG202309221552381920×2553 601 KB

Of course if you pack 100-150s with only 10Ah the cost escalates as the cost per board vs cell is too high. While I like the concept of active balancing, it might be a lot of hassle for just 10Ah.

With 230Ah cells the board cost is small in comparison to the cell cost.

If you are planning to build a battery in the ~3kWh range (10Ah * 320V) I suggest a 16-20s battery with larger cells.

All the best, Patrick

1 Like
12

That is a really nice looking setup. My solar setup is a 16s3p with 120AH cells and a Daly BMS (which at the time worked out cheaper than building 16 diyBMSs). It works, but I would really have liked to build my own BMS.

For the EV charger, the simplest (and probably even cheapest) solution is a 48V battery + 7.2kW Voltronic inverter. But that is again a fairly boring option.

Except for a few LF holdouts, pretty much all modern 48V inverters are HF designs. Which means they start out with a 48V-400V DC-DC converter. And it terms of home-building, that converter is the hardest part to design/build. The actual 400VDC-230VAC inverter is relatively simple in comparison.

But 48V could still be the way to go. Lots of commodity hardware available at good prices. Then it just becomes an integration and software problem. Would actually be similar to my home system, where I have Node Red running the Inverter/BMS/EVSE combination to manage charging.