The "Eierlegendewollmilchsau"

Hello everybody,
first: I hope my english is understandable. I'm from Germany and not a native english speaker.

To the title:
I want to build a Battery-Charger-Discharger-Terster-Logger-Unit. An all in one thing, os as we call something like this in Germany " eine Eierlegendewollmilchsau" (egg-laying - wool - milk - pig)!
As far as I am at constructing my idea, I've ordered some parts and have already made some "proof of concepts".
The goal is a device to:

• charge up to 8 cells at max. 3A
• discharge up to 8 cells at max. 3A
• measure the charge- and discharge-capacity
• measure the cell-temperatures while charging/discharging
• get the charge-/discharge-curves and store them to a memory-card for further use
• and is compatible with a wide rage of cell-chemistry and physical sizes

My primary goal is to learn more about IC's, batteries and microcontrollers on the way. After all its all about the fun of doing it right?
Secondary i want to build a device with open source hard- and software, that others can build and modify for there own use.

I know this sounds like a lot of functionality that not everybody would need or use but that is kind of the idea .
So in the next days I would like to let you now in more detail what I've done so far and what components I would like to use.

Many users are going this route. Have you looked at other projects? This seems to be one of the major projects, other than the power packs themselves, that most users are pursuing. Really neat overall

Many people have done this and one of the better you can find here:
https://secondlifestorage.com/t-Brett-s-Arduino-8x-Smart-Charger-Discharger

I have another version of above ongoing that will be shown in a week or 2. Kind of samy type but more slave/master layout for 1-300 cells in first batch.

Thanks for the feedback. I've already seen many of these chargers and they were kind of my motivation. However something I did not find was a real CC-Discharging mode or what I would like to also implement, a constant power discharge. Also the ability to log the process to evaluate the aging of a cell is a thing I want to try out.

Did you mean 3A charge into each 1x 18650? This is too high & may cause a lot of heating, especially for bad cells.
If 3A = shared by 8x cells OK, but 0.375A maybe slower vs 0.5A, eg takes longer to charge/discharge.

Hi Redpacket,
I mean it as it is written, up to 3A per cell and for a variety of cell-types. There are cells beyond 18650 with a capacity of more than 3500mAh.
To charge these cells with a tp4056 at 1A would take ages. I got the number because the DC-DC converters I wanted to use are capable of currents up to 3A. But that does not mean I want to charge every time with a current that high. It is often a good thing to stay under speck with these things to get a longer servicetime out of these converters.

For higher currents its generally cheaper to get a proper RC charger.

Doing proper CC is not hard to implement. Its just a matter of cost. Doing CC with OP amp feedback and Nmosfet and with a shunt feeding it you can easily set current you want to discharge of. Google have many different rather simple designs. In the ends its about how much it can cost per cell

I will follow your progress!

Hi Gummiadler,
willkommen im Forum!

With every respect to your plan (which is for sure a universal Instrument for every secondlifer) i would like to point out some issues.

The problem of an Eierlegende Wollmilchsau is, than it can be the ultimate tool for a work which can be perfectly defined beforehand.

So, every critique does not go to the plan you have, but to the sum of specifications you include in your "Sau".

Is CC discharge neccessary ? You included it.
Is IR Measurement necessary? Hardly anybody seems to be using it in the test. You didnt include it.

I hope you understand what i mean.

The direct opponent of an ElWMS concept is "modular and flexible", which is obvious in its terms.

I wish you good luck for your project, and i will for sure read every line you publish, and i am sure you can get every necessary support from every knowledgeable DIYer here.

Ok, so here is a little update / more information on my plans:
First I want to talk a little bit about the Hardware I intend to use. I want the charging and discharging happenautonomously and not totally dependent on a microcontroller.
So for the charging this module would do it for me. The biggest advantage is the higher inputvoltage, so I can use some random powersupply later on. In the meantime I have done some tests and the module can provide up to 3.56A but gets really hot while doing that. Maybe I'll be changing it for this module because it can provide up to 5A (I will stay at my 3A limit as the rest of the circuits is designed based on that current) and therefor will never reach its maximum rated current.
The discharging will happen with a IRLZ44N N-channel mosfed, driven by a LM 358 OpAmp mainly because I had the parts already on hand and the circuit is easy to build with only a view parts. By using a mosfet, the tester can also discharge Ni-XX type cells with a low voltage at a higher current. And by controlling it with a microcontroller it can simulate a resistive load as well.
For current measurement I like to use the 5A ACS712 current sensor. I know, some of you don't like this type of sensor for totally legitimate reasons. I could use a sens resistor but I had some of these laying around, gathering dust and I wand to put them to a good use. Also the output-voltage makes it easier to read with an ADC.
Speaking of which, as an analog to digital converter, there will be an ADS1118 per cell als this chip does not only feature a differential measurement for getting the cell voltage but also has a build in temperature-sensor to monitor the cell. The resolution is 16-bit and higher as many ADC's in microcontrollers.
The "brain" will be an Arduino mega 2560. Thanks to brettwatty and his project, I've already ordered some mega pro boards from robotdyn. At first I wanted to use an Arduino nano but soon I had discovered that the program-memory and the pins wouldn't be sufficient.
To control the charge- and discharge-circuits I'll be using one AD5206 for two cells. This way I have 3 pots per cell ( charge-voltage and -current, discharge-current).
For the user interface I have to admit to make it a bit overkill. It will be a 2.4'' touch TFT with a SD-card for storing the data. Also I want to put some indicator-LEDs on each cell to give an additional optical feedback depending on the cell-status. There will also be a Button on each cell to get a fast access to the settings of it. (maybe a longer press will stop the current program...)




I have already soldered a test circuit to replicate one cell stage and will test it this week. Maybe it will work as I intended, so I'll leave you with a picture of it before it has the chance of exploding .

Nice!

Just for ref im working on a slave testing station design where you can run 1 cell or 300 cells at once. As simple as it can be but at same time accuracy good enough. Its not CC but I do meassure the current for being able to calculate the capacity worked with. I think that in software i can tweak it close enough so the numbers are consistend and ALOT better than Li or opus testers.
The goal of my design is that you can easily scale up and keep cost down and at same time get the data to wherever you want.

(Hijack of the thread )

Gummiadler said:

Ok, so here is a little update / more information on my plans:
First I want to talk a little bit about the Hardware I intend to use. I want the charging and discharging happenautonomously and not totally dependent on a microcontroller.
So for the charging this module would do it for me. The biggest advantage is the higher inputvoltage, so I can use some random powersupply later on. In the meantime I have done some tests and the module can provide up to 3.56A but gets really hot while doing that. Maybe I'll be changing it for this module because it can provide up to 5A (I will stay at my 3A limit as the rest of the circuits is designed based on that current) and therefor will never reach its maximum rated current.
The discharging will happen with a IRLZ44N N-channel mosfed, driven by a LM 358 OpAmp mainly because I had the parts already on hand and the circuit is easy to build with only a view parts. By using a mosfet, the tester can also discharge Ni-XX type cells with a low voltage at a higher current. And by controlling it with a microcontroller it can simulate a resistive load as well.
For current measurement I like to use the 5A ACS712 current sensor. I know, some of you don't like this type of sensor for totally legitimate reasons. I could use a sens resistor but I had some of these laying around, gathering dust and I wand to put them to a good use. Also the output-voltage makes it easier to read with an ADC.
Speaking of which, as an analog to digital converter, there will be an ADS1118 per cell als this chip does not only feature a differential measurement for getting the cell voltage but also has a build in temperature-sensor to monitor the cell. The resolution is 16-bit and higher as many ADC's in microcontrollers.
The "brain" will be an Arduino mega 2560. Thanks to brettwatty and his project, I've already ordered some mega pro boards from robotdyn. At first I wanted to use an Arduino nano but soon I had discovered that the program-memory and the pins wouldn't be sufficient.
To control the charge- and discharge-circuits I'll be using one AD5206 for two cells. This way I have 3 pots per cell ( charge-voltage and -current, discharge-current).
For the user interface I have to admit to make it a bit overkill. It will be a 2.4'' touch TFT with a SD-card for storing the data. Also I want to put some indicator-LEDs on each cell to give an additional optical feedback depending on the cell-status. There will also be a Button on each cell to get a fast access to the settings of it. (maybe a longer press will stop the current program...)




I have already soldered a test circuit to replicate one cell stage and will test it this week. Maybe it will work as I intended, so I'll leave you with a picture of it before it has the chance of exploding .

Click to expand...

Gummiadler, pls chnage your name.
To you US guys, a gummiadler is translated a "Rubber Eagle", with the german meaning of a fried chicken having 10000 flight hours - if you understand what that means .

To your modules, you should at least use the second module, you will not be happy about what happens with the weaker one if its happens do deliver 3 A for several hours... . I have similar modules here, but older and without CC control. On higher currents check the thermal stability, long term, in the range of your needs.

Give The IRF a small resistor, say 0.2 Ohm, to limit the current if your firmware happens to full open the IRF. Vishay has btw wonderful resistors in the TO220 case with 50 Watt nominal. On a cooler, of course.
See example https://www.vishay.com/docs/50005/rto20.pdf

I have the ACS 712 by myself, i have no experience atm why it is rejected from the folks. Try to mount it away from other high current wires. For dedicated "Measurement "Task of course do something else.

ADS 1118 is very interesting, didnt know that, will see if i can implement this in my Active balancing project.

The AD5206 Potentiometer - take care because the Potentiometers are only operating within the supply voltage of the AD. Since at least the Voltage pot may be bound to the output voltage of the Step down converter, it might get interesting if the software happens to to try outputting a greater voltage. Check the CC pot in the converter of its level as well.
If the 256 Steps are sufficient i cannot say, but because the Voltage range of the converter might be 10V-something you might get into trouble or make a modified divider concept preventing this.

And a user interface is never overkilled....

@ daromer
Your project looks very interesting. I will defiantly have an eye on that one!
@Cherry67
->username: ok, I will consider that.
->DC-converters: You are Probably right, I have already ordered the other modules and they should turn up in the next weeks.
-> Mosfets: they will be driven by the op-amp with a frequency at about 150kHz. At least that's what I've been told. Sadly I do not own an oscilloscope jet to confirm that. In this time, the current has no chance to rise to a critical level before the mosfet is turned of. Maybe i will use a gate-resistor to operate the fet a little bit slower if needed.
->Potentiometer: I use the Pot and the second channel of the opamp to generate an analog voltage for the DC-DC converter which is than fed through a 1K resistor to the feedback pin of the lm2596. The original Pot is still on there and trimmed so that the voltage goes from 0.8V (5V from the pot) to 5.0V (0V from the pot). I think I should make a dedicated post to explain it in more detail along with a schematic. Basically, this way the pot can never get more than the supply 5V and if you want to raise the output-voltage you could do that by changing the value on the trimmer-pot on the module. Also the 255 steps of the pot will be from 0.8V to 5V so ca. 0,016V/step. The current pot is in line with a 5V regulator on the DC-Dc board itself and another high-ohm resistor to make a voltagedevider for the same voltage as the shunt-resistor. So this should be save as well. Testing will reveal the truth.
Here is a schematic of the converter!

Gummiadler said:

-> Mosfets: they will be driven by the op-amp with a frequency at about 150kHz. At least that's what I've been told. Sadly I do not own an oscilloscope jet to confirm that. In this time, the current has no chance to rise to a critical level before the mosfet is turned of. Maybe i will use a gate-resistor to operate the fet a little bit slower if needed.

Click to expand...

All looks well thought through except the above.
Of course its depends from the reason why the op-amp drives so high frequent, but why do you think the curent will not rise that quick ? Capacity lowpass of the gate or Speed of the mosfet ?

why do you think the curent will not rise that quick ? Capacity lowpass of the gate or Speed of the mosfet ?

Click to expand...

I'm sorry, but I do not fully understand your question.
If it is why I am sure, the current will not rise above my set value?
Because the current-sensor will provide a feedback for the opamp.
Or why I doubt the speed?
Because the ACS712 has just a bandwidth of 80kHz. As far as I can measure it, the Current seams to stay stable but I don't have the ability to measure its ripple. if I get a switching-frequency of 80kHz I'll be happy.
I think I will make myself a Christmas-present this year.

Gummiadler said:

I think I will make myself a Christmas-present this year.

Click to expand...

Ok, i am not sure i do understand your reasoning. Maybe its the problem of the meaning of "switching frequency", the bandwith of the regulating circuit.

Other issue, when you say you have no oszilloscope, there is but one solution - get one .

If you cant afford something good, then google DSO 138. ist a 10$ issue, with plastic casing. Yes, it is just a plaything with severe limitations against someting professional, but it is far better than nothing.
I have been working for years with professional equipment in my profession, but could not afford the real thing for myself.
Then my son (!) pointed the DSO out to me..... and every now and then i use it. For instance the Mosfet for the Kelvin Thingy, i could esily check if the On-Voltage was really down to zero with the driving voltage i had. And it shows as well the AC of a cell with intermitant load....

And you could see with it if your regulating circuit is oszillating....

It has been a busy week for me and I did not get much done on my project. But there was something in my mail, that I think would be interesting for you guys.
Background:
At first I planed to use the standard-18650-holder for my project. But as they arrived and I test fitted a cell, I discovered, that it was to much afford to get the cell back out without damaging anything. Also 18659 cells are the only ones that would fit in there. So back to the drawing-board. I searched online for something better and found some ridiculously cheap Lipo-Chargers for less than 3$ for a 4-slot unit. everything in me said "This has to be dangerous garbage!"
So guess what I had in my mail:



The miracles of "chinesium" make this devise have a 16.4W peak outputpower and a 10W peak inputpower.



The internal circuit has some unbranded Chips with a TP4056 compatible pinout. The traces are a bit ... unconventional ...



but the power-cable has a surprisingly large conductor. Maybe I will keep it for another project.



The circuitboard is reduced tu the bare minimum not even a inputfiltercap .



After disassembling and reassembling I checked the poor thing for performance. Now that I know what is inside.
Charging is 0.6A per cell so 2.4A on the USB cable. Witch is about the maximum that is save to transfer via a USB connector. That is why the connector gets pretty toasty after a while. The cable itself stays cool. The end of charge voltage is around 4.21V to 4.23V.
After all I am really surprised how much you get for 3$. Don't get me wrong I will never use this thing as a charger. All I wanted was the sliding mechanism, witch looks like it is up to its task. Please don't buy these things as a Charger, this could be just another cheap way to burn your house down.

Gummiadler said:

It has been a busy week for me and I did not get much done on my project. But there was something in my mail, that I think would be interesting for you guys.
Background:
At first I planed to use the standard-18650-holder for my project. But as they arrived and I test fitted a cell, I discovered, that it was to much afford to get the cell back out without damaging anything. Also 18659 cells are the only ones that would fit in there. So back to the drawing-board. I searched online for something better and found some ridiculously cheap Lipo-Chargers for less than 3$ for a 4-slot unit. everything in me said "This has to be dangerous garbage!"
So guess what I had in my mail:



The miracles of "chinesium" make this devise have a 16.4W peak outputpower and a 10W peak inputpower.



The internal circuit has some unbranded Chips with a TP4056 compatible pinout. The traces are a bit ... unconventional ...



but the power-cable has a surprisingly large conductor. Maybe I will keep it for another project.



The circuitboard is reduced tu the bare minimum not even a inputfiltercap .



After disassembling and reassembling I checked the poor thing for performance. Now that I know what is inside.
Charging is 0.6A per cell so 2.4A on the USB cable. Witch is about the maximum that is save to transfer via a USB connector. That is why the connector gets pretty toasty after a while. The cable itself stays cool. The end of charge voltage is around 4.21V to 4.23V.
After all I am really surprised how much you get for 3$. Don't get me wrong I will never use this thing as a charger. All I wanted was the sliding mechanism, witch looks like it is up to its task. Please don't buy these things as a Charger, this could be just another cheap way to burn your house down.

Click to expand...

second life slider!

I think I owe you guys an update on my project. Unfortunately my laptop died on me, having blue-screen after blue-screen. So I had to recover it and my data first. Sorry!
In the meantime, the Micro-controller and LCD got here and I made a little Prototype. I will post a video of it doing what I could implement so far later. There is so much important stuff to do for the holidays, that this project got a little bit slow...

Dont worry. All good things need their time.

Just a note on those CC/CV buck modules, the "CC" part is a bit loose, it will start dropping current early, for example in my case using a XL4015 module (5A rating) at 1.5A CC, it starts dropping current off at 4.07V (with 4.2V CV voltage). I've also been working on a custom cell tester with graphing, for example one prototype is running cycle tests here using the above mentioned XL4015 module: https://thingspeak.com/channels/563140 and another here using a custom boost/buck converter for discharge/charge: https://thingspeak.com/channels/570759 I'm working on a v2 using a standard MOSFET+Op-Amp circuit for constant current discharge instead of discharging into a buffer pack. My tester is only 2 cells though, I'm already near the processing power limit of the STM32 microcontroller I'm using for it.