DIY 18650 20 bay Arduino managed Charger

Hi to all that clicked on this thread.
My Name is Mike, I am a 63 Mechanic with a few years in engineering. I have some Electronic experiance but just enough to know what I want to do is possible, even if I am not sure how.LOL

That said on to the topic of my project.

As a half century Tinkerer I have decided to build my own powerwall as have many on the board. I decided on a 20 bay size due to the fact it big enough for a reasonable through put of cells and small enough not to take up much room when the P/Wall is built.

After reading many articles and watching way too many u-tube videos I started to pick the best bits out of others designs. Below is a first draught flowchart of what each cell section will look like.





I am going to use an Arduino Mega Micro to monitor different parts of each cell section, I havent coded since the TRS-80 and commodore64 era, but its like riding a bike, isn't it?

Opionions and comments are more than welcome, I am old school so can take neg comments as easy as positive ones.

Cheers for now.. Mike.

I think i get it - and it looks very interesting!
To make sure that when people comment, they are also trying to meet your aims, can you describe what you are intending?

Hi, your first layout looks grate (as simple as it is) but beware that the INA219 is an ic device that only supports 16 addresses. So either you just make a 16 bay charger or you have to work around that issue (two separate buses or a multiplexer...) I will defiantly follow your progress. There is not just one way to build a charger. Please give us some more information on your goals.
I guess you want to build something, that helps to process used cells. So controlled charging and discharging while monitoring the current and voltage.
One thing that bothers me (personal opinion) is the TP4056 in combination with higher cell counts. "Nerdville" build a TP4056 based charger that processes 66 cells at ones. Be aware that your charger will need a Power-supply that can provide at least 20A @ 5V and your cables must handle that current. After all I think every solution that works safely and fits your needs is a good solution. Feel free to ask any questions and have fun with your project. You will defiantly learn a lot, even when building "just another charger".

Thanks for the feedback gentlemen, its exactly what I am looking for in my current low experience level. But i am a fast learner.
That diagram in my first post is basically how i wish each channel of the 20 bay charger to look. I am trying to keep everything isolated per bay, so a failure will be limited to one "row" only.

Gummiadler made a great point that the INA419' only have 4 addresses, even though the datasheet states 16 only 4 are accessible on the board. I am going to be using 20 of these, which looking at a glance means I will have 5 groups of 4 , and 5 sets of conflicting I2C addresses. You are correct. So below is how I intend to get around that problem.





As you can see, I could expand to more sub-sets of INA419's but think I will save the spare SD/SC channels for a few selectively placed screens, so I can see whats happening at a glace.

All data that I can measure I would like logged onto a spare laptop in spreadsheet form pref Excel, and then write routines to correlate all the info into useful charts. Because of this Data Store wish, I believe I may need to add a second Arduino Mega to help with cataloging data.

Although I am new on this forum I have spent weeks researching all the corners if the web, the more I learnt, the more I realize how stupid I am. Hopefully with the aid of others I can revers that trend and become average.

Raved enough for now.... Oh as for power supply, out in the shed I have 2-3 dozen commercial servers and at least half have good strong power supplies at least the Dells have, Couple that with a capacitor bank to smooth out the supply, we might have dependable power.

Mike

I would go for tp5100 instead of tp4056! It has 12V, 2A charging current and reverse protection...

Regards
Karl

Charly144 said:

I would go for tp5100 instead of tp4056! It has 12V, 2A charging current and reverse protection...

Regards
Karl

Click to expand...

Hi Karl, from what little i have read on these, they charge 2 cells in a series configuration. If that is correct. its not the best config when playing with "unknown" condition Cells. I may be wrong, will need to do a bit more digging to see if I understood it correctly when I read the article.

Mike.

No, the TP5100 can charge in 1s or 2s configurations. In both modes it can charge at 2A max. It does take up a little more pcb space as it requires an external inductor. But for it's pros, I think the space requirement is worth it.

Bugger 25 tp4056's already ordered.. BUT I will still delve into the TP5100, after all DIY powerwalls drink more money than they return dont they. Unless you start at 12yrs old.

BTW my liitoKala 500II bit the dust last night... grrrrrrrrrr.

When you get them, verify that the ICs are legit TP units. There's are LOT of fakes/clones out there. This is one reason why I stopped bothering with the TP4056 (i used to sell them on ebay; i stopped when i realized the ones i had were fakes).

Any give-away on what to look for? I ordered these through Banggood , dont know if thats good or bad not having dealt with them before

Legit ones have the big TP in the upper left corner and the lettering is very clearly visible. If the TP isn't visible and/or the text is very faint (needing a mag lens and light to see it), it's probably a fake (especially if the TP isn't there)




This one is most likely fake




This one is legit. See the TP logo in the upper left corner?

Actually no, but i will look when they arrive, i take it the fake ones dont work or work badly?

I contacted the company I bought the LittoKala 500 II from, they are sending a replacement, no need to return old one.
So because I dont have to send it back, I did what everyone does, I pulled it apart to find out why it died if possible.

2 out of the 4 ss34Schottky diodes are cooked. 0.342 cents each or less than $2 for 10. dunno if its worth the effort, diodes dont die without a reason... meh, might try anyway. It would leave me always wondering What-If.

i read that an Arduino mega will run on 12vdc, however a few posts on various sites say they dont fully recommend it.
So I have ordered a adjustable DC-DC Buck converter, I will set to to give 9VDC and add 1 or 2 1000uF electrolitics just to get the smoothest 9VDC supply.

Gremlin said:

BTW my liitoKala 500II bit the dust last night... grrrrrrrrrr.

Click to expand...

Can you donate it for the greater good/science?

Gremlin said:

i read that an Arduino mega will run on 12vdc, however a few posts on various sites say they dont fully recommend it.
So I have ordered a adjustable DC-DC Buck converter, I will set to to give 9VDC and add 1 or 2 1000uF electrolitics just to get the smoothest 9VDC supply.

Click to expand...

The Arduino has a linear voltage-regulator on board. It reduces the voltage from the "Vin-pin" to 5V.The thing is, the excess energy is waisted as heat. So when feeding in 12V and 500mA the output is 5V and 500mA so 3.5W of heat. That's why it is not recommended to feed 12V dc into the Arduino. The best way is to use the Buck-converter to generate 5V directly and feed it into the 5V pin. This way, you get an efficient power-supply. The additional caps are not necessary but its a good addition to get a more stable voltage.

How bad are the fake's? I think i ordered the fake ones.
Or what they don't do that they actually must do? and vica versa?

100kwh-hunter said:

How bad are the fake's? I think i ordered the fake ones.
Or what they don't do that they actually must do? and vica versa?

Click to expand...

I've Used the "fake" ones. For my small batch (20pcs.) I could not find any dangerous differences to the real ones. The Charging current is not exactly 1000mA but that could be due to low tolerance resistors. The maximum charge-voltage however was always within the Specs of the data-sheet. I use them in some Projects where they are soldered to phone-batteries (they might have build in protection) and so far everything looks good. You should check the end-of-charge-voltage by connecting an electrolytic capacitor to the output of the TP4056 and powering it up. As soon as the chip says the cap is charged, check with a multimeter if the voltage is correct.
I can not say how age or temperature effects the values but so far and for my applications I'm satisfied with them even if they are not genuine.

Hi guys, back with more information and updates on this Charger unit project. As per normal several things have been thought of or suggested by interested readers. This means YAU yet another upgrade.

OK the basic concept still holds true, however my idea for running 2 x 15w 10ohm resistors as a dump load has been thrown out of the window. Several reasons for this, bear with me as I run through the logic behind my thinking.
Here in AU those 2 wire wound ceramic resistors would have cost me $0.35 cents less than a HW586 unit. Now i am not rich and am trying to build this thing for a reasonable cost. My dad once instructed me on the principle of "penny wise and pound stupid" and I applied that here.
I have decided to go the 20 HW586 route for a few reasons (22 actually more on the extra 2 later).

Buying the HW units actually saved me a bit of money,
A) I no longer have to buy Oled displays to get some real time feedback of whats going on.
B) Although I had enough I2C resources to run several Oled screens, I have now freed those prossesing times and resorces for the Arduino Mega processor.
C) The sketch is just so much shorter to write and maintain.
D) the 20 mini Volt meters I ordered are now 15 too many. ( that will teach me not to be over enthusiastic)
E) This one is complicated:-
Using the 2x15w resistors would have worked fine, but would have needed the Mega to keep an eye on the dropping voltage and turn OFF the relay that selects between charge (tp4056) or discharge (Resistors). That would have tied up 20 data lines (or a multiplexer) to monitor this condition. The next problem would have been , as soon as the relay reset from the discharge path, the TP4056 would have seen a flat cell online and launched off to charge it again.
By substituting the HW586 for the resistors, I can now program how low I want it to discharge too, and once that figure has been obtained the HW unit simply switches off the discharge and nothing happens until I instruct the process to repeat.
Bonuses are the HW counts the capacity and displays it at a glance. The Mega is still writing these results to a excel sheet for permanent recording

Hope that was not too boring a read.
As for Vincents Reverse polarity Circuit contribution, I am just waiting for the arrival of he IRL540N mosfets. These circuits will be assembled on a 1" square vero board, then fitted into a 3d printed frame and poured with resin to seal them from harm and damage.

This is still a work in progress, I have a box that is slowly filling up with all the hardware I require. once everything has arrived then the layout will begin. Next post will contain the thinking and reson why I have increased the charger unit from 20 to 22 bays.

Mike
All comments appreciated.

OK new day...

As mentioned I have decided to increase the charg board by 2 additional bays, but these bays are "different" .
These 2 bays are going to be used for 0v cells that we all know and love soooo much.

My classification of a 0v cell , zero volt cell, is a cell that a commercial charger like a LiitoKala or Opus refuses to acknowledge when presented to it. It may actually have more than 0v, but not enough to trigger the charger(s).

Please remember I am a Newbie at this game and am still learning.
Now when I got my first few hundred cells, I sat at the bench slotting then into a Liito charger, if it recognized the insertion the cell went into one bucket, if not it went into another. When all sorted I had 380+ in the good bucket and way way to many in the other. This was initially rather disappointing.
Fast forward a few weeks of reading,video's, chatting and repeat. I grab the "bad" bucket and cobble up a "Dumb" charger system. Taking 20 random cells , one by one I hit them with a lab table top power supply set to 2.8v @ 1amp.
Each cell was timed in this "hard" charger for 10 mins.
Out of the 20 cells I got all 20 to be recognized buy the Litto unit, but, as you may already have guessed with varying results.

Of the 20 cells (red Sanyo's) turned into heaters and were placed in a new DOA bucket
13 others cycled 3 times over 2200mah
3 cycled to 1200-1600mah.
1 just tripped the short circuit protection of he Lab PSU.

Do I think I miraculously recovered a large percentage of 0V cells? Nope, I may be new, I am not dumb.

So *I* think I need to adopt a testing regime for these class of cells. Currently I am leaning towards this:-
3 x dump to 2.8v and recharge to 4.1v then Shelf time of min 14 days to test for self discharge or losses.
Then if they show little or nil voltage loss I can start to look on them as a possible use full cell.

That is the reason for the 2 additional bays on the charge unit, basically to hypothetically slap the cell into paying attention and find if its worth keeping or not. A second chance if you will.
Normal chargers wont do this, the TP4056 module wont do this, so I have designed two bays to play nasty with the cell.

I would love to hear what system others use to sort their cells. Is what I mentioned before a suitable regimen/system?

As always comment appreciated and welcome. Good or bad.

Mike.

Gremlin,

I would love to hear what system others use to sort their cells.

Click to expand...

I sort my cells into 3 different categories. 0V to <2V, 2V to <3V and 3V+
Left to right




0V to <2V are initially resistance checked. If they are 0V andhave infinite IR then the CID has popped and to the recycle bin they go.
If they have an IR of over 100m? they are set aside for further study but they generally don't give any good results other than the green sony GR series which for some reason the IR drops drastically if slow charged.
I gather20 cells of very similar V and put them into my 20p CC/CV charging board at 4.2V and ~50mA per cell.



I fire up the CC/CV power supply and set that to 4.2V and 1A



I watch the voltagerise on my Fluke to 4.19 and the amperage will drop to almost nothing when they are done. This can take a day or two.



Sometimes the V will stall at 3. something V and then I know I have what I call a "Vamp" cell. A vampire cell that is pulling all the mA from the others.
I pull all the cells out and check their V.Inadvertentlythere will be 1 or 2 cells with lower V than the others. I put all the cells that measured good V back in and the meter will show a slow rise again. The LG cell you see there was a "Vamp".I have checked the 20p pack with a thermal camera but so far I have not been able to determine a "Vamp" cell by temp.



The 2V to <3V I mostly treat the same way. IR check first if reasonable and the V is 2.5 and up I will charge them at 300mA on my Zanflairs.
If the V is <2.5V into the 20p charging board they go for a slow rise to 4.2 at 50mA.
3V and up get IR checked again and anything at this point that is over 100m? gets binned. 100m? is a pretty high criteria so there are very few that don't make that grade.Then Icharge them on the TP4056 boards and some of the other of the shelfchargers I have.

Is what I mentioned before a suitable regimen/system?

Click to expand...

I initially did the same thing to "shock the battery into submission" but I have found outthat doesn't really do the cell any favours. It's like hitting them with a defibrillator when they aren't dead yet they are just hibernating and need a slow waking up curve. Kinda like a bear coming out of hibernation.
That doesn't mean you will have a good cell or not but I just had 43 UR18650A Sanyo that where ?1V after the harvest and treatedto the slow wakeup curve come back with very good results. Of course the initial IR showed good promiseand no heaters.