Low Cost BMS for Home Storage Batteries

[peter_mcc]

1. I haven't seen holes used like that before. I've seen lots of small vias used to try to thermally link the copper on the two sides - if you do that you could try making the copper area on the bottom as big as possible. You might also want to increase the area of the copper on the top side as much as possible. Using the PCB as a heatsink is largely a matter of how big you can make it. Ordering 2oz copper PCBs also helps as bit (normal is 1oz).
A bit more info on the vias to join the sides thermally:
https://www.edn.com/electronics-blogs/t ... ermal-vias

2. I'd put a pull down on the mosfet as a matter of habit. Thinking about it more since my earlier emails to Chris, the failure mode I'd be concerned about is the micro dying and leaving the mosfet gate undriven - it could then float up to the point where the mosfet turns on and flattens the battery. If someone is away or doesn't monitor their battery regularly (eg for a house) they might not notice things have gone wrong until it's too late and the cell is dead. The micro could fail in other ways too (pulling it high or low) so it isn't a guarantee, just risk mitigation.

However... given you have the holes there for a through hole mosfet it would be possible to fit it later on using a through hole resistor.

3. The cap is just there to stabilise the power supply, the specs shouldn't be too critical. Colleagues who know more than me shy away from Tantalum caps because they can fail "short circuit" (bang) - electrolytic caps tend to fail open circuit. Ceramic caps have improved a lot over the years - why not go for a 10uF 10V 0805 ceramic cap like Seeed 302010158 ? You could put a through hole set of pads (2.5mm pitch) next to it for those who can't easily source the SMD part and want to use a through hole electro. Plus because it's so much smaller you could put the mosfet pull down resistor in

This looks like an interesting source of info for the differences:
https://electronics.stackexchange.com/q ... ces-in-use

[4Springs]

Ok, I've placed the order for the modules.
I ended up not putting in an explicit pull-down resistor, although as Peter said, one could be added using the through-hole pads.
I eventually figured out how to properly search the components available through Seeed. The secret is to click on the link to 'Download Library for Eagle and Kincad'. This takes you to SourceForge, where they have the components in a downloadable spreadsheet file. Use your favourite spreadsheet program to search the components.
The result of being able to search well is that found a nest of 10uF aluminium electrolytic capacitors for 3c (US) each. So I'll use them and I've left the C1 pads as-is. A different capacitor could be used here, the pads are a standard 1206 spacing.

I did pour a bit more copper on both sides of the boards. Removed the large holes and replaced with lots of small vias. I left them at 1oz copper as it cost another $50 to go up to 2oz, and I don't intend to run these boards at full power for long.

Total cost was $911 US for 400 boards populated with 8 of the 18 SMD components. Electronics always seems way too cheap.
"I'll have 800 resistors please."
"Certainly Sir, that will be $1.47."

[mikedufty]

Have you got any spares in those 400 boards? I just collected my AVASS batteries and am now thinking about what to do with them.

[4Springs]

Have you got any spares in those 400 boards?

Yes.
Final totals were 352 cell module and 37 master boards ordered by people in the group buy.
There are set order quantities at the board manufacturing place, so I ordered 400 module and 40 master boards.

I've updated the Airtable form: https://airtable.com/shrEdYHvjoLEaMpz6
You can order just the boards, or the boards plus most of the components. Almost all of the components that I ordered had order quantities such that I'll have quite a few spares. But some did not, and I won't have spares at all. So if you order boards+components now, you'll get boards plus all of the components that I have. I'll also give you a list of components with supplier part numbers, so if you are missing some you can see where you can purchase them from.
If you place an order it just sends me an email, it doesn't take any money. So the first thing I'll do is to reply via email to let you know what I have.

By the way I'm very happy to sell all of the spares. I have my own personal spares reserved as part of the 352, so anything above that is just shed clutter as far as I'm concerned.

[4Springs]

Well, after a few working bees over the last several weekends, all the boards in the group buy are now accounted for.

Some were sent to folks (mostly interstate) for them to assemble themselves. We had to remove a resistor from each cell module board (placed there by mistake when I specified the wrong component), and allocate the exact number of each component for each person's order. No two people ordered the same number of boards! Then bundle them all up into padded bags for shipment. I haven't heard back from anyone yet, so presumably everyone has received their order, tallied up their components and found everything in order. Or possibly they've just put the bag into a 'to be done' pile.

Most were purchased by Tasmanians, and most of those people came to the working bees to help assemble them. This was a good fun, if at times rather cold experience, in my garage, in June & July:

IMG_6241.JPG (134.04 KiB) Viewed 1101 times

IMG_20190629_112427050.jpg (286.04 KiB) Viewed 1101 times

IMG_20190706_151912841.jpg (130.61 KiB) Viewed 1101 times

IMG_20190630_090830691.jpg (209.62 KiB) Viewed 1101 times

The bits where we were able to do an assembly line were the easiest. Pick a task and perform it over and over. Eventually you're good at it, and then you run out of boards (we had about 220 to assemble).
The hardest task was fault-finding. About 30% of the boards did not pass their (hopefully comprehensive) testing regime. Most of these were fixed by running over the microprocessor legs with a soldering iron. But some were due to the wrong component placed, or a couple by a fault with the microprocessor. One module board is still in the shed, unfixed. It is a strange fact of electronics that it is sometimes possible to make 20 new boards in the time that it takes to fix one faulty one...

There are a few boards left over by the way, if anyone else wants some. I think one Master board and there should be about 60 Cell Module boards.

Next stop, software. I'll make another post about that.

[4Springs]

Software - what are we going to do with these things??

The current software is available here: https://sourceforge.net/projects/low-co ... ogramming/
The current Cell Module software is version 1.10. I don't expect that this will change.
The latest Master software is H1.00. I've forked to the 'H' version numbers for versions intended for Home Storage systems.

Master code version H1.00 is set up for a 12V battery, intended to be used for Home Storage. It should perform all of the functions in the table in this post here:https://forums.aeva.asn.au/viewtopic.ph ... =25#p72597
I haven't tested H1.00 extensively, but I have used it to charge up my cells. Here is what I did:
I have 32 cells. I arranged them into batteries of 8 in parallel, wiring them together with fairly substantial copper wire. I had 4 of these parallel batteries, and connected these together in series (I put 12V 20A automotive fuses between them as a precaution). At the time I had 4 celltop module boards, so I put one on one of the cells in each parallel battery.
This gave me a 12V battery, so I could charge it with a 12V battery charger. I bought one from Jaycar that charges at 8A.
With the Master board, I connected output 1 to a relay, and put that relay in the 240V supply to the 12V charger.
This meant that I could charge all of the cells as a 12V battery, and the BMS could turn the charger off when full.
It took about 10 days at 8A until the first celltop module reported a voltage approaching 3.6V. The Master turned the charger off at that point. The other cells were not up to charge yet, so I charged them individually using a benchtop power supply. The power supply can only supply about 2A (at 3V). This took about 3 days (a day each for each battery of 8) to charge up the remaining cells.

So H1.00 is somewhat tested, at least for charging. You can download it now and program your board with it, but I'd recommend further testing before you rely on it.
If you'd like to use this for a 48V system there is at least one parameter to change. The file Main.h has a value MaxPackVolts. It is set to 14400 for the 12V system (14.400V), you need to change it to suit the 48V system. Something like 57600 would be appropriate, depending on your charger. This value is intended as a 'backstop' - the battery should not reach this value, since the charger should have turned itself off at a lower voltage.
There are lots of other values that you might like to experiment with, all in that file Main.h

I'm intending to make multiple versions of this software, H1.00 is the first prototype. For my system I'm intending to run an SP Pro inverter, so I'll definitely have software for that. I'd encourage others to upload their versions once working.

[jonescg]

And that right there is the spirit of AEVA writ large
Looks like a fun time despite the cold. I'm about due for another conversion workshop day actually...

[francisco.shi]

Is the oven for reflow?
I bought a similar oven and plan to use it as a reflow but not sure if it will damage the parts.

[4Springs]

Is the oven for reflow?

Indeed it is. Was also used for fruit buns at one stage.
It has a thermostat, we set it to about 200 degrees Celcius. Watch the circuit boards through the glass window to see when the solder melts - then leave it for 30 seconds more to make sure. The whole process takes about 3-8 minutes, although we couldn't see any particular reason why it was sometimes longer.

There was one board that got a little hot:

IMG_20190630_145128709.jpg (131.18 KiB) Viewed 1077 times

The board itself has split, with 'stuff' boiling out from inside it. You can see the stuff at each end - it really stinks and is probably very bad for you. Note also the electrolytic capacitor C1 has jumped from its proper position half way along, hurdled two of the big resistors, then landed neatly, right-side-up, and soldered itself into a new position.
The oven needs to be turned on and pre-heated before use. My guess is that this procedure wasn't followed this time, and the boards were placed in the oven straight away. The temperature inside the oven seems to go way over the thermostatted temperature at first, then settles down after a few minutes. Interestingly only one of the boards in the oven popped (it sounded like popcorn), and it was the one furthest towards the back. The others survived the process.

[brunohill]

For my system I'm intending to run an SP Pro inverter, so I'll definitely have software for that. I'd encourage others to upload their versions once working.

Are you intending to use CANbus comms with your SP Pro or set it up standalone with a customized version of the generic LiFePO4 settings?

[4Springs]

Are you intending to use CANbus comms with your SP Pro or set it up standalone with a customized version of the generic LiFePO4 settings?

The intention at this stage is to use a 48V SP Pro, and let it think that it is a lead-acid battery. I won't be using CANbus, but will be adjusting settings to interact with the BMS. For instance for charging the battery, I'd like to put it into a low-current mode once the first cell reaches a target voltage (perhaps 3.4V). Looks like I can use a relay input (output from the BMS, input into the SP Pro) to indicate that it is time to reduce current. And I can program the value for the reduced current into the SP Pro.
I have no experience with the SP Pro, but the electrician I found said that it is the inverter/charger that he is experienced with. He told me that it won't work straight up, and I'll have to fiddle with lots of settings over a period of time. This sounds perfect!

[brunohill]

I have a SP Pro 48v here that has been using Lead Acid batteries for the last 3 years. I am waiting on a solar upgrade and will then change to the LIFePO4 batteries. I hope to hit it right 1st go.