The story of a little powerwall

Artlav

New member
Joined
Oct 18, 2018
Messages
27
It all started 4 years ago, when i was on an Ebay spree (this happens sometimes...) and bought some cute solar cells.
As luck would have it, these turned out to be genuine Maxeon Sunpower cells, pretty much top of the line.

[img=640x480]
After some looking around, i managed to find a place to get them professionally soldered and laminated and ended up with a pair of solar panels.

[img=640x480]
And there the luck ran out - the panels added up to about 9 V max power point, and were neither here nor there.
Eventually they went into the attic and haven't seen the light of day for years.

A year later i slowly started accumulating 18650 cells, mostly from used laptop batteries.
I always wanted to make a big battery based power system, a KWh of storage at least, and often salivated over industry sized batteries, but so far sanity always prevailed.
As the cells accumulated, i started to wonder how to sort them and figure out their state and capacity.
A regular li-ion charger i had kept giving random values, so eventually i just gave up and made my own battery exercise unit.

[img=640x480]

It was fairly simple - a linear constant current source to charge the cells at 1A, a big resistor to discharge them, multiply by 4, add a microcontroller to monitor and record everything to a microsd card.
It would cycle the cells 4 times automatically, over the course of a day or so, charge them up to 50% and beep.

With this i went through the whole collection and managed to get them characterized, with a charge performance, estimated ESR and precise capacity.

[img=640x320]
The results weren't that good - about half the cells were dead, with capacity under 500mAh. The rest were evenly distributed across the 1000-2000mAh range.
I used the best of the crop to make phone powerbanks and stuff, but by and large they sat in a box unused for a few years.

Eventually i had a lapse of sanity and bought a couple of large LiFePO4 cells on Ebay.


[img=640x853]

On a quick check they looked fine, but i had no real use for them... So they sat in a box for some time.

Fast forward to present days.
Having recently moved, i've been unloading all the crap i accumulated from my previous garage sale and ebay sprees.
I recently made a bunch of autonomous cameras, and one of them sat on the second floor of the garden house, staring out of the window.
It was powered by an extension cord running all over the place, and people kept forgetting to plug it in.

So one day i figured out i can solve this problem and unload a bunch of junk at the same time - slap the solar panels onto the roof, add batteries, power the cameras from that.
Easy, right?
I started composing the system and testing things.

[img=640x667]https://i.imgur.com/GHpEj6Y.jpg[/img]

Rather quickly it became obvious that i've been had with the LiFePO4 cells - they barely held any charge and instead of 40Ah they were more like 4Ah.
While contemplating the fragility of human planning, i noticed the box of 18650 cells.
Cells that were sorted and characterized years ago, and added up to quite a noticeable volume.

So i took the best of them, made a 2s6p pack totalling 10Ah, built everything up and tried it out.

[img=640x1055]https://i.imgur.com/IJgt8Zp.jpg[/img]

It worked quite well, even if a bit underpowered - the panels put out 40W, and with only 70 Wh of storage it wasn't a good match.
But now it was time to get serious.

First, instead of duct taping the panels to the roof, how about a proper frame.

[img=640x1138]https://i.imgur.com/4r9y4wu.jpg[/img]

Then, instead of a bunch of spaghetti, how about a proper cutting board.

[img=640x718]https://i.imgur.com/HoJEj45.jpg[/img]

16 more cells were added, from the middle of the list.
That's another 10 Ah, bringing the total to 120Wh.
Also in frame an MPP controller/charger, a reverse protection diode, a low voltage cut-out and a step-up converter with a fuse.
This all went onto a wall, neatly mounted this time.


[img=640x1137]https://i.imgur.com/07cbKj6.jpg[/img]

All this thing powers is a couple of cameras and a 4G/Wifi converter. It can also dual-power a few LED lights around the house.
Maximum power output of 60W, as limited by the step up converter.

Not the scale you folks are used to, but it is power and it is on a wall. :)

Since then i had a couple of additions.
First, i found 12 more cells. These appear to have been bought new and are protected flashlight types.
No idea where they came from, but they added another 12Ah to the system.
So the total is 32 Ah or 230Wh and 40 cells in 2s20p configuration.

[img=640x1138]https://i.imgur.com/9QkNXB7.jpg[/img]

I also added an innovative monitoring system consisting of a spare camera with an LED taped to it, pointed at the instruments.

[img=640x1138]https://i.imgur.com/PceOB6V.jpg[/img]

This lets me see how things are going at any time.
I might make some sort of an ESP based board to actually track everything later, but i'm too lazy at the moment to do it.



image_ifokmd.jpg


Odds are i'll be expanding the system over the years.
But for now, that's all folks. :)
 
Really Nice! :cool:

Great job
 
Could you provide some details of that cell tester you are using?
 
Geek said:
Could you provide some details of that cell tester you are using?

I call ita battery exerciser.
Wrote about it on EEVBlog forums - https://www.eevblog.com/forum/projects/diagnosing-lithium-cells/

Summary:


image_skrwjz.jpg


Approximate layout of this version (don't have the original schematic):

[img=640x513]http://orbides.org/img/181023-charger_sch.png[/img]

The cells got got 4 linear 1A CC-CV chargers (sense resistor between cell and ground) with switches above them, and 4 load resistors with switches in line with them attached.
There are measuring lines on each terminal of each cell (4 wire sensing with calibration), allowing it to measure voltage (top-bottom), current (bottom-ground) and ESR (charge/discharge off for a millisecond, observe voltage difference).

The rest of it is a microcontroller with microsd card, buttons and a screen.
It automatically does 4 cycles of charge-discharge, while logging all the above mentioned data to a file on the card, one per each direction and step.
The whole ordeal takes about 24 hours.

It's probably a bit of an overkill for just measuring cells, but my goal was to actually get a good idea of how the cells behave, age, react to being charged at -30*C, differ in capacity between 4.0, 4.1 and 4.2 top voltage and so on.


I suspect that if come across more than a few weeks worth of cells i'll be making more of these, which means i'll have to clean it up and order a bunch of real PCBs, at which point it might become good enough to not to be too ashamed to put it on Github.
So if anyone is interested in making one, do let me know.
 
Definitely would like to make one. Is it Arduino based? What micro-controller are you using?

Brett has posed designs for one, but it is more complicated than what I would need.

I struggle with multi pin SMD soldering though. I would have toadapt the design for through hole soldering.
 
Geek said:
Is it Arduino based? What micro-controller are you using?
The MCU is ATMEGA324p, so it can probably be shoved into the Arduino compatible shape. No idea what it would take, however. Can Arduino bootload from a microsd card, or do you have to have a USB->UART converter and a computer connection?

I'm basically using avr-gcc and my own file-on-microsd-card bootloader, i find that to be much cleaner than Arduino stuff.

Geek said:
I struggle with multi pin SMD soldering though. I would have toadapt the design for through hole soldering.
Hm. I'm the other way around - prefer surface mount stuff and compact layouts.
I do have the tools to work with it all, however...

The original board is already mostly through hole so i guess it's possible to replace the SMD ATMEGA with a DIP one, but even that would require considerable re-arrangement of all the stuff.
Given that FT232RL it needs to be Arduino-able is SMD only, i'm not sure if there is much point to doing that.
 
Artlav said:
Geek said:
I struggle with multi pin SMD soldering though. I would have toadapt the design for through hole soldering.
Hm. I'm the other way around - prefer surface mount stuff and compact layouts.
I do have the tools to work with it all, however...

The original board is already mostly through hole so i guess it's possible to replace the SMD ATMEGA with a DIP one, but even that would require considerable re-arrangement of all the stuff.
Given that FT232RL it needs to be Arduino-able is SMD only, i'm not sure if there is much point to doing that.

Agreed using a 44 pin DIP would really mess with the size of that PCB. However you can get an FT232RL soldered to a PCB for through hole mounting.

I would love the skills and equipment to do my own SMD work.

Artlav said:
Geek said:
Is it Arduino based? What micro-controller are you using?
The MCU is ATMEGA324p, so it can probably be shoved into the Arduino compatible shape. No idea what it would take, however. Can Arduino bootload from a microsd card, or do you have to have a USB->UART converter and a computer connection?

I'm basically using avr-gcc and my own file-on-microsd-card bootloader, i find that to be much cleaner than Arduino stuff.

That complicates things. I believeArduino can load code (native hex) from SD. I don't think they can bootload from SD.
 
image_kzeuxi.jpg


It's a mix of a bit of scary and a bit of exited.

One pack done, made with 32 cells.
216 Wh of energy. One mile of an electric car's range.
The coldness of it, the weight, the size, the danger and the unseen potency all add to the mix.
That i'll be wrapping it in heater wire does not help either.


A few days ago i came across another used laptop battery seller, and made a mistake of asking how many he got.
Turns out he got a whole box...
So now i have a box full of batteries, and am slowly going through them.

I've ordered PCBs for more cell exercisers, and for the month it'll get them to get here i'll be designing and making frames, heaters, controllers and the rest of the things while painfully attaining 8 cells a day...

The plan is to make a 7s32p system, which will probably be a separate project from the existing chaotically grown 2s20p one.
 
And they are assembled.
And they are working.

Well, not all of them, some are only looking nice.
9 out of 24 slots are dead, 3 of which were due to an assembly miscalculation, 4 due to a mysterious bug that turned out to be flux residue under the resistors, 1 due to a crappy part, and one due to me not wanting to test a single cell on it's own.

The whole assembly also features a UPS made out of the "protected" cells (which had to be amputated once i realized they trip from the just the capacitor charge surge), mostly for the ability to move it from socket to socket without having to wait for the cycles to finish.

...can i sleep now?

[img=800x800]http://orbides.org/img/181112-chargers.jpg[/img]
 
Wow, need some close up dude - great work
 
hbpowerwall said:
Wow, need some close up dude - great work

Like that? :)

[img=800x1073]http://orbides.org/img/re_20181115_025302.jpg[/img]

I've been contemplating making this thing open source, but the design was quite a slap-it-together so i'm a bit ashamed of it.


Meanwhile, after re-testing all the cells from before (3 weeks worth in 3 days!) and debugging the setup i'm about to start increasing the cell count again.
Sadly, the good cells box is on the right and the bad ones are on the left.
Out of 114 cells in there only 56 that are decent, of which28are good.
It's going to be a long trip... :/

[img=800x450]http://orbides.org/img/re_20181115_030412.jpg[/img]
 
Yep, that works lol well done - looks neat enough to a noob ake me
 
Despite the lack of snow, it's well below freezing outside here.
And Li-ion does not like being charged in the cold.

Let's make some winter-grade batteries.
-Get the battery pack assembled
-Apply thermometer
-Wrap it in fiberglass tape
-Apply nichrome heater wire and a power out for BMS
-Wrap it in insulation material, secure with duct tape
Done.

Should work down to -20*C based on my earlier tests...
Hopefully.
The BMS logic would be "if below +5*C and sun is on, heater on. If above +5*C or no sun, heater off. If Tb<0*C, charge off".


image_ktymok.jpg
 
Hmm, not sure I wanna comment about pack #4 on the right :p
 
Nice work on the gear you made :)

Seems like a very high rate of "bad" cells? Are you sure they are bad? Maybe worth a 2nd test?

re heating, is there a way you can have the battery system inside a warmer area of the house?
 
Korishan said:
Hmm, not sure I wanna comment about pack #4 on the right :p

In that case, here is the full battery. :)

image_rjajqm.jpg


Redpacket said:
Seems like a very high rate of "bad" cells? Are you sure they are bad? Maybe worth a 2nd test?

Bad is relative...

77 laptop packs, out of which 61 contained non-dead cells, where dead is anything leaked, corroded or showing under <1V.
That gives 297 cells.

Out of them, 41 ended up dead in testing, as in either heaters or have a capacity under 1000mAh.
That leaves 256 usable cells.

In further sorting, 100 of them were sorted into the "bad box" meaning they are high self discharge, high variance (capacity drops between discharges), high ESR, high degradation (under 60% from rated if known) or have a capacity below 1500mAh.
That left 156 cells in the "good box".

These are further sorted by the same parameters into 85 good, 59 mediocre, 8 low and 4 missing.

Finally, 128 of the best of the good and mediocre were selected to form the 4 32p packs shown here.


So, how much of a bad ratio this is?

Redpacket said:
re heating, is there a way you can have the battery system inside a warmer area of the house?

No, they will be sitting in a place that is rarely occupied during winter, so no heating to be expected most of the time.
 
Back
Top