5kWh Lifepo4 battery pack and associated electronics

eranrund

Member
Hi!

Wanted to share some photos of my build of a 5kWh LiFePo4 battery packs that utilizes 16 100aH CALB cells and hopefully answer any questions people might have regarding such a setup. In addition to building the enclosure and installing the pack, I redid a bunch of the electronics. The stuff I currently have is:
* Electrodacus SBMS40 solar charge controller/BMS
* Reliable Inverter 3KW - Some cheap Chinese inverter that so far seems to do the job. I've tested my microwave, kettle, angle grinder, heat gun and AC with it just to see if it's willing to run all these types of loads and it did that without issues. At the peak I was drawing around 100A from the battery to power two heat guns with nothing acting funny for over an hour.
* GL.iNet GL-AR750 Travel AC Router which is connected to a permanently installed Nexus6 phone. This serves three purposes:
1) GPS tracking of the van using google maps location sharing.
2) Provide me with a WiFi access point. It uses a separate carrier than my primary phone so I get better coverage.
3) Provide internet to a Raspberry Pi
* Raspberry pi - currently the only thing this does is pull data out of the Electrodacus to give me remote monitoring/graphing of amount of energy produced/consumed.
* Small reverse camera mounted on the side of the van, since I have no windows and might want to see who's knocking at my door
* A custom wireless LED dimmer controller. I have two transmitters - one by the door and one by my bed, that control a receiver mounted where all the other electronics are and allow separate dimming of two zones in my ceiling LED lights (front and rear)
* A similar wireless controller that is controlled by a fob on my keychain for turning on and off two large LED light bars I have mounted on the roof at the front and rear of the van. I also have wired buttons for those indoor, again using custom hardware/software I designed
* A 500W charger to allow for shore charging, to supplement the 600W solar array mounted on the roof
* I also run all the standard loads (fridge, water pump, diesel heater, etc).

There might be pieces I'm forgetting since this project has many parts in it, but that's a rough overview of what I've done with the electronics. Most of it is probably unnecessary but I enjoy this kind of work, and this is my main hobby.

The full album of some pictures I've taken is at https://imgur.com/a/nkE8J0d but I also attached a few select photos to this post.

Cheers! And thanks again for being such a wonderful and helpful community, my Sprinter adventure would've been a failure if it wasn't for the people here taking the time to answer questions and help others.
 

Attachments

autostaretx

Erratic Member
Gotta love the "one size fits everything" wall outlet...

PowerWallDetail.png

(except South African plugs)

--dick
 

eranrund

Member
Yeah, it gets the job done :) Its powerful enough to run my Macbook pro charger which is the only thing I use that mini inverter for.
 

daws314

New member
Nice work, good to see unistrut in use, it's so versatile.

I'm planning on using a Raspberry Pi in my build to control the lights via touchscreen, I decided to use Pi so I could have a dimming function which I plan to use a 16 channel PWM board but also to create lighting zones where I can have predetermined locations and colours, I'm very impressed with your wireless dimmers.
 

eranrund

Member
I initially purchased 8 of the cells from Craigslist in used-in-new condition buy a guy that decided to go to a bigger 48v pack for his house for around $900. That honestly provided enough juice 99% of the time, before adding a microware and a kettle. But I wanted to not have to think about running out of juice so I added another 8 cells later on.
The other 8 cells cost me ~$1300 from a US supplier after a bunch of time wasted on a different second hand lifepo4 back that ended up measuring ~3kWh for $450 - so a good deal but it was too big and heavy (came out of an electric vehicle of some sort). I believe those ones are still being sold so if you're happy with 3kWh of usable capacity for around $450 + shipping and ~160lbs, I can find the link.

Anyway, $2200 for the 16 cells, and the BMS/charge controller was around $250 I believe. Those were the expensive pieces. The steel enclosure was cheap, I had a bunch of the 1" square tube around in the shop so I just had to weld it together and get some unistrut to mount it.
 

altide8

Member
What are you using for the 500w charger. I have the sbms 40 as well. Also, what are you using to get the information form the sbms to the pi? Just the web browser and html file from electrodacus?


Sent from my iPhone using Tapatalk
 

eranrund

Member
I'm using a Meanwell RSP-500-27 power supply (it support current limiting which is what makes it work in this setup).
To get the data from the SBMS I connected the SBMS40 to the RPi using a USB to serial converter (https://www.amazon.com/gp/product/B076LNBB7P/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1). I am running https://github.com/eranrund/van-pi/tree/master/sbms_exporter on the pi. It is a small Go utility written by someone on the Electrodacus forum. It pushes the data to a Prometheus instance that is then graphed using Grafana. I have a GSM access point in the van, so the pi has constant internet access.
 

RedBlogger

New member
Thanks for sharing your setup! I'm thinking about a very similar setup of 16x100Ah CALB cells.

A couple of questions that I'm just going to fire at you speed-round style if that's alright :)

1) What do your high and low voltage limits look like in your BMS? Are these based off top balancing your cells? It looks like you only have ~5mV spread between your cells.
2) Are you choosing to only stay in the "flat" part the voltage-vs-SOC curve? I noticed in one of your pictures that your one cell is "red" (does that mean balancing is active?) at 3.350V
3) What do your charge profiles/voltages look like in your solar and shore power controllers? I'm not too familiar with the Electrodacus BMS so not sure what kind of charge profile is available for solar, but from flipping through some of the manual I would assume it gives you a lot of room for control.
4) In the course of your research, did you come across any BMS's that can do individual cell monitoring? I'm planning on a 12V ~5kWhr system, but am not thrilled about 4 CALB cells per "logical cell" in the BMS. Do you know if the Electrodacus would allow you to do parallel strings for a 12V system e.g. a 3P4S configuration if one had 12x100Ah cells?
5) How accurate do you find the Coulomb counting for the SOC calculation?
6) Curious what kind of real-world solar output you see from your panels (and I guess what part of the country you are in)? I'm guessing you're not always making 70W like in your picture :) Is this something you track over time with the RasPi?

Thanks!
 

eranrund

Member
Hi!

1) I don't remember the exact values I have it set up for, I'll need to check. I think the low is at around 2.8? If I recall correctly, high is currently at 3.4v or 3.45v. I bump it to 3.5v if I am going somewhere I don't expect sufficient sun (the 600W array goes a long way in refilling what is usually a pretty modest use of the battery). I could go higher, but I don't need the extra capacity for my day-to-day (read: a weekend adventure every few weeks/months) so I figured no need to get the batteries to 100%. The cells were originally bottom-balanced. I discharged them all individually to around 2.6v (I don't recall the exact number), and then connected them all in parallel for 24 hours. The 5mV spread seen in the album could be deceiving since you're not gonna see much of a difference when the cells are at the flat part of the discharge curve. Here's a photo of where they are right now, showing a slightly bigger delta:
1597434900409.png
2) The red color is just a configuration option in Grafana (the dashboard software). It should've been set to a higher threshold, thats about it. I am mostly staying in the flat part part of the curve, but am allowing for it to go a bit above that to detect >80% SOC.
3) The Electrodacus is a PWM controller, so whatever voltage comes in from the solar array goes out to the battery. Since I am using 24V panels this setup works well and is not significantly less efficient than an MPPT charger (to my understanding - the Electrodacus guy has a long video on the topic). For shore charging, my charger is current limited to about 18A if I remember correctly, so I am delivering a constant ~500W to the battery until the BMS detects a cell that crosses the 3.45v (or whatever I set it to). At that point the relay connecting the charger to the battery is disconnected and shore charging stops.
4) I haven't seen a reasonably priced solution for a per-cell BMS. I am not a huge fan of the 2P configuration I have, but I came to terms with it assuming it is likely sufficiently safe with good quality cells that are properly matched and from (hopefully) the same batch. I think for parallel strings you could use an SBMS0 (the "entry level" Electrodacus) per string, but the costs and complexity builds up fast. I would suggest buying bigger cells and avoiding this mess. I would've done that had I now already had 8 cells when I upgraded to 16.
5) Seems accurate but hard to tell since I have never ran out of juice. The current draw I am seeing matches what I expect based on what I have powered on. I have done some measurements over the 120V draw when running my microwave using a Killawatt and the numbers added up as expected. It's probably not 100% accurate but close enough.
6) I am mostly parked in San Francisco, CA. The highest output I've seen was around 440W, which I believe is what is expected from a flat array in this part of the world. I do track it over time (but apparently only 2 weeks - I need to see why so short, storage is cheap!), here's a graph from the last 30 days at my new place in Oakland:
1597435489067.png
Note that this graph is watts per hour. The reason it's only peaking at 200W is that usually the battery starts pretty full, so by the time the sun is up at the angle that yields >400W the battery fully charges very quickly and the controller disconnects the solar array. The ramp leading to the tall peak is a combination of filling up whatever was drawn during the night, and the sun rising up and providing and increasing amount of power.

Here's a single day:
1597435853409.png

Looking at actual current coming from the array, you can see it performing nicely:
1597435710283.png

and a single day:
1597435863223.png

I hope this answers your questions, and happy to answer more! Cheers!
 

RedBlogger

New member
Thanks for all the data! The solar tracking looks fantastic.

I was wondering about Coulomb counting accuracy if you were using some kind of low-SOC disconnect, but it seems like you'll never get down that low. Probably also not a concern if you are using bottom-balanced cells.

I have seen one system/package for individual cell monitoring from a company called Starlight Solar (link: http://www.lithiumrvbattery.com/Lithium_RV_Battery/GBS_100AH_Cells.html)
They seem to sell a package of 'X' number of 4x100Ah cells along with a BMS+display+monitoring boards. The BMS they sell appears to come with a board for monitoring 4 cells in series - so a 5kWh setup would require 4 parallel strings of 4 series cells to have individual monitoring of the 16 cells. Their 400Ah setup goes for ~$3600 + shipping (from Arizona).
Going with ~8x200Ah cells in a 24V system would be a tidy solution, but preclude alternator charging. Everything's a tradeoff I guess...
 

eranrund

Member
No problem!

I paid $1040 + shipping + tax for my more recent purchase of 8x100Ah CALB cells (200Ah 12V pack). I don't have a record of how much the previous 8x100Ah were as I bought them 2nd hand on Craigslist. So this would be $2080 for 400Ah. But that excludes BMS/Shunt/Solenoid/etc...
The Costco deal seems pretty sweet. I wonder what does the lifetime warranty covers.
 

RVBarry

Active member
12V 560aH under $1000
 

Top Bottom