Battery capacity tester, 0-60V 0-20A

[Mad Professor]

Good day all.
 
I am looking to be starting a new project within the next few months. So I am now in the planing stage of things before I try jumping in with both lead feet.

Before I go into things to far I will say now that I have been working with electronics for about 8 years, but only at a very basic hobby level.

Over the last 6 months I have been helping a local e-bike shop with testing e-bikes electronics and the battery packs.
Quite often I am given battery packs to test as they don't seem to hold there charge, or the run time is not as expected.

As I am in the UK, 90% of the battery packs I have had dealings with are 36volt, but I have also had 24volt and 48volt packs, ranging from as little as 5Ah upto 20Ah.

I am now looking at trying to build my own Battery capacity tester rig, for testing the whole back and individual battery cells.

And this is where I am going to need help and advice.

As said above at this point in time I am purely in the planing stage of things, and will more then likely be an evolving project.

Voltage range: 0-60V (Readout to 3 decimal places).
Current range: 0-20A (Readout to 2 decimal places).
Max power rating: 1200W.
Basic LCD Interface / Menu (For Now)
Constant current / Constant power modes.
USB and or Serial interface for external control and or data plotting.

If I get of the terminology of things wrong please do correct me.

Thanks for your time and advice.

Best Regards.

[Mad Professor]

I guess my 1st question should be on the subject of heat dissipation.
As stated above the max power rating is: 1200W
I am going to need a large heat sink and probably forced cooling to be able to dissipate 1200W.
Can anyone please advice me on how I go about calculating the size of heat sink I will need.

I do already own a quite large heatsink: 300x100x32mm Datasheet

The above datasheet quotes: Thermal Resistance 0.04C/W with two papst type 3312 fans.

Thanks again for your time.
 
Best Regards.

[nowlan]

Your voltages are not very common. Other wise I would have suggested you look at the new battery analysers. They go beyond shorting the battery out. 12v however.
http://www.ctek.com/au/en/chargers/Battery%20Analyzer

I am wondering if something like a Maynuo electronic load would be appropriate. Some diagrams to inspire you.
http://gerrysweeney.com/maynuo-m9711-dc-electronic-load-unboxing-teardown-and-quick-test/

Of course if you want the diy/fun factor, see who else chimes in.

[turbo!]

Lead acid batteries are rated in 20hour rate. A 12v 12Ah battery means that it will support 0.6A load for 20 hours until cut off point. For most things these batteries are used for, this is far from real life representation. At 12A draw, I think you'll get about half that capacity. In electric toys, you're pushing 20-30A, so even less that that. Consult battery manufacturer datasheet and it should tell you the constant resistance, constant current, etc load.

They can fail two ways. Inability to provide the current level needed, or not providing the capacity needed. 

[max_torque]

Pretty much this project can be split into three parts:

1) a suitable adjustable load (and a big one at ~1200w)

2) A suitable measurement system that can measure and integrate current wrt time to get battery capacity (and probably measure voltage so you can draw a discharge characteristic)

3) A method for controlling all that^^^^ stuff (micro controller etc

Each of those parts, whilst not difficult, is potentially non trivial and time consuming, especially if you want to be able to  ensure you have a repeatable and accurate system (repeatability is probably more important than absolute accuracy i guess)

For the load, i suggest water cooling.  1200w to air is either  a high temperature or a lot of air to move, 1200w to water is a lot less trouble (you have relatively "low" peak voltage (60v) so i think you could just about get away with simple isolation methods (assuming no 'public' or untrained access to your device.  A total loss cooling system, running cold tap water from the mains would do (1.2kW with a 60degC upheat only needs 300cc/min (P = Mdot.Cp.DeltaT)

The actual load is more difficult. An Active load (pass transistors etc) that can accommodate any input voltage and any set current is getting a little big and pricey at 60v and 20A, but not too bad.  Some chunky TO-3 NPNs would do the job, if cooled properly (the interface between the transistor and the water is going to require a bit of thinking about).  Alternatively, as a starter, you might just decide to simplify with some basic power resistors, that can be switched in series or parallel to give some basic "ranges" for the discharge etc

Measurement of the battery voltage is less of an issue, but you need to establish what your system resolution is going to be like, potentially you might need some "range" switches again for your input dividers to maximise your resolution (depending on how many bits your ADC is etc

Current measurement is probably the biggest issue, and for a one off device, it's probably cheaper and more accurate to simply buy an off-the-shelf current transducer from LEM (or who ever) giving you a relatively high accuracy and isolated current proportional voltage that can be fed almost directly into your micro's ADC.  Again, potentially two (selectable) ranges might be needed to maximise current resolution.

The control part is probably the easiest, pick your fav micro (PIC/AVR etc) and get coding!  You'll want to add suitable safety monitoring for voltages (min and max) Current (min / max) possibly an additional battery isolator (large relay etc), battery temp measurement, load temp measurement etc.

I'd also suggest that the DUT is housed in a steel, fireproof box, kept somewhere that a fire will not lead to significant damage!

(LIPO's etc, can bite you, loosing your test device would be annoying, loosing your house, rather more so.......  ;-)

[max_torque]

EXTRA:

Assuming you're going to use an "unpressurised" water cooling system (for simplicity / safety) i would suggest a max system outlet temp of say 80degC to avoid boiling (boiling persay isn't bad, as it helps cool the system, but unless you want your device to double as a "teasmaid" ;-) it probably is best avoid for safety reasons (scaulding etc!). To minimise water consumption, use the micro to monitor the temp of the heat sink, and perhaps use am RC servo (easily commanded by your micro) to open/close a tap to modify the water flow rate through the system.

[NiHaoMike]

If you test enough batteries (probably not the case?), maybe it would make sense to hack a few cheap grid tie inverters?

[Mad Professor]

Thanks for everyone's advice and input, I have a lot to take in and think about.

So with the advice given above, I think I will start low powered unit and work my way up to the high powered unit.

For now I think I will start with designing and building a unit to test single cells, up to a discharge rate of 20A.

3.2v nom / 3.6v max
3.6v nom / 4.1v max
3.7v nom / 4.2v max

So worse case should be: 4.2V * 20A = 84W

I have some new (old stock) computer heatsink & fans that are suitable for up to 115w TDP, And I was thinking about using one of them and bolting the power fets directly to the bottom of the heatsink.

Regarding the safety concerns around the lithium pack(s) / cell(s). When ever myself or anyone else is test / charging / discharging a pack or cell(s), the pack or cell(s), are always placed within a closed fire resistant enclosure. To date we have not had any pack(s) or cell(s) catch fire.

[max_torque]

The issue will be keeping your power silicons junction cool enough so you don't have to spend mucho £££ on loads of parallel devices!
  If you look at the SOA graph for your chosen devices, you will see the "DC" curve is power limited (ie, as voltage dropped across the device climbs, the max permissible current falls) and if you check the "power derating" curve you will see that the devices power rating is reduced with case temperature (in order to keep the devices junction below the max permissible temperature.

So, you need two things:

1) a very low thermal transfer co-efficient between your devices and your heatsink

and

2) a very low max heat sink temperature (lots of fast cold air, or water cooling etc)

Of course, it is relatively easy to prototype a single power device to establish a sensible maximum rating for your layout, and then to just parallel up extra devices to meet the eventual max current (and hence power) requirement!

[Mad Professor]

Thanks again for your advice, I am just watching over some of the old eevblog videos.

EEVblog #105 - Electronics Thermal Heatsink Design Tutorial


EEVblog #102 – DIY Constant Current Dummy Load for Power Supply and Battery Testing


EEVblog #281 – BK Precision 8500 Electronic Load Teardown

[cosmos]

I have used a Maynou M9712C for exactly what you describe (I converted our ebikes non balancing cell monitoring to a balancing type)
I also have had brief contact with an ebike shop that had lots of batteries (non returnable full container) with capacity problems (seems a result of non balancing monitoring and spread in cells)
The Maynou (and probably others) will count Ah for you and you can set a discharge stop voltage.
Connect it, start and come back an hour later to see the result ... simple, I would not think twice to purchase this over building my own if it was for work.

The Maynou has high resolution readout (mA and mV LSB) and can be controlled and read from a PC, it appears to be designed for battery applications.

[peter.mitchell]

Whilst discharge capacity is nice, it is also nice to know hows much charge capacity is needed to achieve the said discharge capacity. Eg, if it takes 20ah in to get 10ah out and for it to read charged, but only 10ah in to get 9ah out but not read charged.