Battery Capacity Bench tester

How do you store and manage your electricity?
Post Reply
antiscab
Senior Member
Posts: 2700
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Battery Capacity Bench tester

Post by antiscab »

Hi Guys,

im creating a battery capacity tester for use at expected traction currents.

my requirements were:
48v, 2.4kw cont load (50A)
measure Ah and Wh
give a graph of voltage vs Ah drawn

for a load, i was initially thinking in the giant resistor direction.
then i realised alota battery chargers draw around 2.4kw, and then saw this 5kw sine wave inverter

for the data logging, i decided to use this DSO

The idea was to use one channel to measure the voltage of the pack, and the other to measure the voltage across a shunt.

now the problem:
The DSO's original program would only output 520 lines of data to a csv file, when the time/div is set to 1 sec (~10sec data) or more.
when set below 1s/div, the program outputs 10240 lines, but only ~4sec of data.

at 520 lines, and 100mins of data, thats one data point every 11.5sec.
The value measured isn't averaged, its just the value measured over a random 1ns every 11.5s.
For the voltage measurement, thats fine, it doesn't have to be that accurate. also, im not expecting the voltage to change all that much in 11.5sec.

However, for the current, i need it to be rather accurate.
Im also expecting the current to be seeing some ripple.

The DSO's older program version i found would do 5002 lines, or a line of data every ~1.2s.

So heres the question:
Does anyone know of a better data logger i could use?

perhaps a way to do averaging?
i was thinking putting a cap in parrallel with the shunt, so the value measured in the 1ns sample will be more representative of the average value.

as far as i can tell, the lack of data is a flaw in the program, rather than the unit itself.
So a different program can output more data, as both programs i have through away data when operating at the slower sample times.

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells
Mr. Mik
Groupie
Posts: 66
Joined: Fri, 08 May 2009, 04:41
Location: Gold Coast, Aus

Battery Capacity Bench tester

Post by Mr. Mik »

Sorry, cannot help you with your questions, I can barely understand what your problem is! But I really hope you figure it out, because I would also like to have logging capability with my prototype bench tester setup.

What I've built is a kind of "Ghetto" version of what you are trying to create, of course it does not record or draw graphs or calculate anything, but it allows pretty cheap battery testing at high currents.

youtube video here: AMPEASEL
.
.
It's made from an unused easel (but any old furniture will do!), cables, a 100A relay, a few 30A switches, a 1mOhm 200A shunt, 2 voltmeters, cable-ties, a MOM-ON switch and a ON-OFF switch, a battery (to supply power to the relay), and many 12V 50W halogen lights.
.
.
ImageImageImage
ImageImage



It can test at voltages up to 12V (maybe a bit more) and allows to test at currents of approximately 25A, 35A, 45A, 60A, 70A, 80A and 105A. Anything in-between can be arranged by poking more or fewer of the halogen lights into the cables. They have two sharp prongs, just right for the job! (Maybe it could be built for testing at higher than 12V, by using different globes. I'm not sure at what voltage these halogen lamps will blow.)

There is lots of space left for additional lamps, and with more relays it would be endlessly extendable.

The video shows testing of a Prius NHW10 stick.


I thought maybe all it needs is two data logging DMM's and some sort of switch to turn the relay on and off as a function of battery voltage, to control preset cutoff voltage.

Can you think of a way of data logging the voltage vs. current draw on that thing? And to automatically calculate the capacity from the collected data?
Last edited by Mr. Mik on Tue, 08 Sep 2009, 16:05, edited 1 time in total.
There is always a way if there is no other way!
antiscab
Senior Member
Posts: 2700
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Battery Capacity Bench tester

Post by antiscab »

for your setup,

the Hantek DSO

would work quite well.
use one channel to measure the voltage across your shunt, and the other to measure voltage.

the DSO program spits out a csv file, which under excel shows a line of data for every check, which consists of:
time since record start, ch1 v, ch2 v

using some excel wizardry, type in the next column (d) starting at line 3:
=c3*(a3-a2)/3600

then at the bottom of the page type:
=sum(d10240:2)

that will give you Ah.

for Wh, same deal, except at each line you are mutliplying the Ah value and the voltage value.

it should work well for your setup, as your load is purely DC.
the voltage measured over a random 1ns slice of time every 2sec is an accurate representation of the average voltage over the whole 2 sec.

My inverter load is not so kind.
It has lotsa ripple
so for the current measurement, the current varies say between 35 and 65A.
So the random 1ns read is when the current is 65A, so thats what it reports, when the actual average is 50A.

when i get my inverter and set it up, i will measure the actual variance.

more testing to come

if worst comes to worst, i will just use my cycle analyst to measure Ah, and the graph i make will either be in terms of Ah (inaccurately) or in terms of time.

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells
antiscab
Senior Member
Posts: 2700
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Battery Capacity Bench tester

Post by antiscab »

btw,
for general data logging, that DSO may be overkill
i just bought the 60mhz one as i might want to use it for other things aswell, like measuring the output of a bldc controller.

or when i get a decent set of test leads, measuring the harmonics coming out of the pure sine wave inverter i just bought, which the vendors says is less than 2% THD (possible, just not likely at the price i bought it for).
For my application, it doesnt actually matter.

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells
Squiggles
Senior Member
Posts: 742
Joined: Wed, 22 Apr 2009, 03:19
Real Name: Neil
Location: Newcastle NSW

Battery Capacity Bench tester

Post by Squiggles »

Mr. Mik wrote:
It can test at voltages up to 12V (maybe a bit more) and allows to test at currents of approximately 25A, 35A, 45A, 60A, 70A, 80A and 105A. Anything in-between can be arranged by poking more or fewer of the halogen lights into the cables. They have two sharp prongs, just right for the job! (Maybe it could be built for testing at higher than 12V, by using different globes. I'm not sure at what voltage these halogen lamps will blow.)

There is lots of space left for additional lamps, and with more relays it would be endlessly extendable.


You could connect lights in series for higher voltage maybe?
User avatar
Richo
Senior Member
Posts: 3737
Joined: Mon, 16 Jun 2008, 00:19
Real Name: Richard
Location: Perth, WA

Battery Capacity Bench tester

Post by Richo »

Do you need this permanently setup or just a one-off?
There are plenty of loggers around depends on what you want to spend.

You really need to know how many data points you need.
ie 2 analog channels 1 sec log rate and 5minutes 600-1200 bytes.

$900 seems a bit excessive for an inverter to use as a load resistor.

So the short answer is NO but the long answer is YES.
Help prevent road rage - get outta my way!
antiscab
Senior Member
Posts: 2700
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Battery Capacity Bench tester

Post by antiscab »

I need this to be reasonably permanent.
I plan on using this to test Emax traction packs, if the new distributor sells any real quantity.
Most of the parts ive used so far ive either already owned, or will have other uses.
budgets around 1-$200 for a new logger

im using an inverter, as i may need to capacity test ~50 48v60Ah packs in a row (use the capacity test load to charge the next)

the logger has to:
log in a format friendly to excel, or any other spreadsheet program
if reporting average values, report 1 data point per 10s on one channel, for at least 100mins (600 data points)
if the values aren't averaged (as they are with my DSO) then i need alot more

The problem is mainly the current measurement, it needs to be accurate for the Ah value reported to be accurate. (also so the v-Ah graph is accurate).

i will be using my cycleanalyst for recording a total Ah drawn value, and a timer to determine time between start and shutdown.
i could fudge a v-Ah graph with the data from the cycleanalyst and the v-time graph, and pretend the current drawn is stable.

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells
Mr. Mik
Groupie
Posts: 66
Joined: Fri, 08 May 2009, 04:41
Location: Gold Coast, Aus

Battery Capacity Bench tester

Post by Mr. Mik »

Thank you, Matt and Squiggles! Great ideas.
Finally an excuse to buy a scope!

And putting globes in series will allow testing of practically any battery.

My setup is of course too clumsy to change over all the time for testing vastly different battery types, but I have those 160 x 6s NiMH sticks out of the 4 used Prius NHW10 batteries to churn through, so it's worth spending some time on building a setup that allows high current testing.

There is always a way if there is no other way!
Mr. Mik
Groupie
Posts: 66
Joined: Fri, 08 May 2009, 04:41
Location: Gold Coast, Aus

Battery Capacity Bench tester

Post by Mr. Mik »

antiscab wrote:
i was thinking putting a cap in parrallel with the shunt, so the value measured in the 1ns sample will be more representative of the average value.


Matt

How about two caps in parallel, but the second one with a small resistor in series? Or just the one cap with a small resistor? That should cause a time lag in the voltage, but a fairly smooth curve?

I hope this is not totally nonsense...
There is always a way if there is no other way!
User avatar
Electrocycle
Senior Member
Posts: 985
Joined: Sun, 19 Oct 2008, 20:23
Real Name: Andrew
Location: Sydney
MSN: dumhed@dumhed.com
Contact:

Battery Capacity Bench tester

Post by Electrocycle »

If you use one of the later Cycle Analysts you can get it to spit the data out the serial port and log that :)
The Engine Whisperer - fixer of things
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Battery Capacity Bench tester

Post by Tritium_James »

A cap in parallel with the shunt won't be practical, the shunt is really low impedance (duh!) and so to do any useful filtering the cap will be huge, especially for a 0.5Hz or lower rolloff that you need for your 2s sampling interval. A quick calc says with a 1mOhm shunt you'll need a 79 Farad filter cap to get a -3dB point at 0.5Hz!

Maybe a 2-stage RC filter or even something with an opamp and an active filter?
Gow864
Groupie
Posts: 202
Joined: Sun, 27 Jul 2008, 04:37
Real Name: (John) Gary
Location: Bellingen NSW

Battery Capacity Bench tester

Post by Gow864 »

If you are not part of the solution, you are part of the problem.
antiscab
Senior Member
Posts: 2700
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Battery Capacity Bench tester

Post by antiscab »

Gow864 - yeh i already own one of those eagle tree data loggers (or rather did, until i blew it up)

I doesn't have any more useful functionality than the cycleanalyst.

Electrocycle - i didnt know the later cycle analysts could output to a serial port. i will have to investigate further.

TJ - a active low pass filter might be feasible, i haven't actually made one before. I might have to come back to this, after looking at opamps and such.

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells
Mr. Mik
Groupie
Posts: 66
Joined: Fri, 08 May 2009, 04:41
Location: Gold Coast, Aus

Battery Capacity Bench tester

Post by Mr. Mik »

Mr. Mik wrote: How about two caps in parallel, but the second one with a small resistor in series? Or just the one cap with a small resistor? That should cause a time lag in the voltage, but a fairly smooth curve?

I hope this is not totally nonsense...


Ahem, ahem, cough, cough, .... that was nonsense!

There is a good reason why I use nothing more complicated than a relay and an egg-timer for my measurements!   Image
There is always a way if there is no other way!
BjBlaster
Groupie
Posts: 65
Joined: Mon, 07 May 2007, 06:44
Real Name: Beau Walker
Location: Australia
Contact:

Battery Capacity Bench tester

Post by BjBlaster »

I asked my self this same question when building my mains power logger. I decided that even a 1 second intervals, it was accurate enough to work out what my house is using a year. You need to scale the project according to time. If you want to smash a cell in 5 minutes, then you need more points to get an accurate representation of the measurement. I love the egg timer :) Sometimes that is all you need to test something!

In saying that, I like the fact you want to use an inverter to actually "run" something useful whilst you are testing. It seems a waste of power otherwise just to heat up some ceramic.
If it's stuck, force it. If it breaks it needed replacing anyway - bj's shed
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Battery Capacity Bench tester

Post by Nevilleh »

Jaycar sell a "Digitech" digital multimeter that has an RS232 interface plus software that you can run on a Windows computer that will log data all day. From memory its only about $79. I've used a couple to log voltage and current when doing discharge tests on PbA batteries.
Also, Tamura sell Hall Effect current sensors, L01Z***S05 series that go from 100A to 600A and cost less that $US20 each. DigiKey might have them, or you can try Tamura Corp in the US. Better than current shunts!
Edit
Jaycar don't stock this meter any more, but you can get an equivalent from here:
http://users.tpg.com.au/users/p8king/qm1538.htm
Last edited by Nevilleh on Fri, 11 Sep 2009, 10:23, edited 1 time in total.
Mr. Mik
Groupie
Posts: 66
Joined: Fri, 08 May 2009, 04:41
Location: Gold Coast, Aus

Battery Capacity Bench tester

Post by Mr. Mik »

BjBlaster wrote:...
I love the egg timer :) Sometimes that is all you need to test something!

In saying that, I like the fact you want to use an inverter to actually "run" something useful whilst you are testing. It seems a waste of power otherwise just to heat up some ceramic.


If we combine the two we can boil eggs to perfection with otherwise wasted energy! Image

But I was exaggerating a bit, the egg-timer is actually a multi-function electronic timer which I often use to boil eggs. It counts up, down, or round and round...
There is always a way if there is no other way!
drowe67
Groupie
Posts: 147
Joined: Fri, 18 Apr 2008, 20:14
Real Name: David Rowe
Location: Adelaide
Contact:

Battery Capacity Bench tester

Post by drowe67 »

I built a battery tester using a Wifi router and a $2 PIC microcontroller, have blogged about it here:

http://www.rowetel.com/blog/?p=119

It samples voltage and current every few seconds and stores them in a text file on the router. A later version uploads the data to the Web. The uC has a watchdog timer than kills the test if the router goes down (to prevent accidental over-discharge) and also stops the test if the voltage drops beneath a threshold.

As it's Wifi I can monitor a running test from anywhere in the house.
Driving an EV every day
http://www.rowetel.com/ev.html
antiscab
Senior Member
Posts: 2700
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Battery Capacity Bench tester

Post by antiscab »

thanks for the responses guys :)

Neville - the uni-T DMM you linked to doesn't appear to be true RMS, and will have the same problem my DSO setup will have. reported current will be wrong.

Having said that, i hadn't thought of using a DMM. ill see if i can find one that is both true RMS, and support data logging.

drowe67 - your setup is rather impressive. perhaps i could feed the output of a DMM directly to a PIC or similar. I think this project is starting to suffer from scope creep :p

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Battery Capacity Bench tester

Post by Nevilleh »

antiscab wrote: thanks for the responses guys :)

Neville - the uni-T DMM you linked to doesn't appear to be true RMS, and will have the same problem my DSO setup will have. reported current will be wrong.

Having said that, i hadn't thought of using a DMM. ill see if i can find one that is both true RMS, and support data logging.

Matt


Matt, I use a dc load for battery testing, so a true RMS meter is not relevant. (My load just heats the jug for coffee!)
The Tamura current sensors go to 10 KHz, BTW.
Mr. Mik
Groupie
Posts: 66
Joined: Fri, 08 May 2009, 04:41
Location: Gold Coast, Aus

Battery Capacity Bench tester

Post by Mr. Mik »

antiscab wrote: for your setup,

the Hantek DSO

would work quite well.
use one channel to measure the voltage across your shunt, and the other to measure voltage.

the DSO program spits out a csv file, which under excel shows a line of data for every check, which consists of:
time since record start, ch1 v, ch2 v

using some excel wizardry, type in the next column (d) starting at line 3:
=c3*(a3-a2)/3600

then at the bottom of the page type:
=sum(d10240:2)

that will give you Ah.

for Wh, same deal, except at each line you are mutliplying the Ah value and the voltage value.

it should work well for your setup, as your load is purely DC.
the voltage measured over a random 1ns slice of time every 2sec is an accurate representation of the average voltage over the whole 2 sec.

My inverter load is not so kind.
It has lotsa ripple
so for the current measurement, the current varies say between 35 and 65A.
So the random 1ns read is when the current is 65A, so thats what it reports, when the actual average is 50A.

when i get my inverter and set it up, i will measure the actual variance.

more testing to come

if worst comes to worst, i will just use my cycle analyst to measure Ah, and the graph i make will either be in terms of Ah (inaccurately) or in terms of time.

Matt


Thanks again, Matt!

I finally got around to do it, including the spreadsheet calculations.

Now I can do discharges at up to 100A and calculate Ah and Wh, despite the non-linear resistance characteristics of the halogen lamps.

And show graphs for voltage, current and power:

Image

The graph shows the overlay of discharge curves for the best and the worst stick out of a Prius NHW10 battery with 40 sticks (of 6 cells each). They had rested 2 weeks to include self-discharge in the test.
Test starts at 30A, then 100A, 60A, then 30A to the cutoff voltage of 5.5V

Stick 17 discharged 5.34Ah and 34.55Wh; Stick 35 discharged 4.96Ah and 30.42Wh.

Interesting how the relationship between Ah and Wh is not the same:

4.96Ah/5.34Ah = 0.93
30.42Wh/34.55Wh = 0.88

How come???

Is that a special representation of the sticks different internal resistance?

Or maybe rather due to the dropping resistance of the halogen lamps as the voltage drops; that keeps the current going almost horizontally = constant, while the power drops ever more steeply towards the end.
There is always a way if there is no other way!
antiscab
Senior Member
Posts: 2700
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Battery Capacity Bench tester

Post by antiscab »

Mr. Mik wrote:
Interesting how the relationship between Ah and Wh is not the same:

4.96Ah/5.34Ah = 0.93
30.42Wh/34.55Wh = 0.88

How come???


The Ah (coulombic) efficiency shows you how much the secondary reactions have consumed between start of charge and end of discharge (over one cycle in other words).

the Wh effiency also takes into account the voltage drop due to internal efficiency.

thus Ah efficiency > Wh efficiency.

other chemistry efficiencies:
lead acid: 0.9 Ah, 0.7Wh
lifepor4: 0.99Ah, 0.9 Wh.

the nimh effiencies fall if there is a significant amount of time between the charge and discharge parts of the cycle due to the self discharge.

all interesting stuff.

thanks for showing us what hardware you use :)

my new and improved tester is/will be based around a DC motor to ACIM DOL.

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells
Post Reply