Battery Life, 50A load discharge test method/results

Re: Motors/Battery life


Quote:
Originally Posted by sanddrag
Sometimes you really have to question what professionals tell you if it just doesn’t seem right. The battery life probably would not be sufficient for a wheelchair but most definitely would be for an average FIRST robot. Also, the battery is rated as 18 Amp-Hours capacity, not 18 amps supply. Roughly roughly speaking, it means you can continuously pull one amp for 18 hours or 18 amps for one hour, etc. You should look at the spec sheet http://www2.usfirst.org/2005comp/Specs/batex.pdf for a more specific answer.

Plus, twice the motors doesn’t automatically mean twice the current draw. What it does automatically mean, is almost half the loading each, which would mean approximately half the current draw (maybe a little more) each resulting in the total current draw of the drive system being only slightly higher than with only two motors. It is kind of the same concept as walking in snow with stilts or walking in snow with snowshoes. You weigh the same no matter what, but the snowshoes distribute the load better, because they have more area. For a robot, more motors, more load distribution. Less current draw per motor. Only slightly more current draw total.

I’ve performed continuous 50A load discharge FIRST KOP 18AH SLA Battery tests to ‘exhaustion’ (9.72v)

The discharge Vbatt Vs time characteristic for this years is different in shape with less time at a lower Vbatt for 2006 Exides Vs earlier years - perhaps slight different electrolyte implementation ? (requiring P/N change?) Ca Vs Pb?
(2003-5 Ca were 13.30v fully charged Vfloat Vs 13.00 to ~31.10 for 2006
else lot deviation on batts we got though BOTH of this years show this.

Specifically, with a 100% Midtronics charge
(achievable only by overnight charging i.e.
[fast charge to ~80% capacity+several hours capoff for remaining 20% TBD]
(100% not likely achieved during multi-chg-dischg during competition =~80%)

Tests results:
9 min life with Vbatt@term 13.10v noload, @50A load quickly down

11.61v then gradually ending at 9.72v at 9 min,

Vbatt@term drops roughly linear, ~2mV/sec

note: Rload is ~constant so current gradually decreased with time, I=V/R
R=11.61v/50A = .22ohm (note: add .5A @11.61v from fan load)

Equipment:

DMM direct to battery terminals,
Two each Harbor Freight 12v 100A Battery/Load testers in Series (~$15ea)
(50A load for longer test time and match typical 2005/6 robot drain)
Used large 12V .5A KOP fan over vent holes to extend test time over
mfr 10 sec limit
(keeps resistor element loads from burning bright red over 9 min test)

Vbatt@terminals taken ea 10 sec & imported to excel spread sheet & graphed

I don’t have the data & graphs on this computer but will post, if requested.

BTW this method/data was presented at my FIRST Advanced Electronics 2005 Workshop at Calif State Univ Northridge.

Dale(engr294] (TRW / Northrop Grumman)
[email protected] 310 374-8323

more info on battery life:

Per current Exide 18AH ES & EX Battery Spec:
at 20C=78F colder = LESS time
Current Amps Time to discharge End V
0.9 20 hrs 11
4.5 4 hrs 10
10.8 70 min 10
18 40 min 8
36 18 min 8
54 8.5min 8 (last data from graph)
72 4 min 8 following not on graph = extrapolated
90 2 min 8
108 60 sec 8
116 30 sec 8
134 15 sec 8
152 7.5 sec 8
170 3.7 sec 8
188 1.8 sec 8
196 0.9 sec 8

230 Amp for 5 sec = Max discharge current

Measured actuals:
(see also my post on Batt Test @50A load and Excel results attached)
Team 294 robot with 4 CIMs (125A), compressor (11A), Vandoor (13A), 2X FP (10A) + control electronics (2A)
consummed 161 A of average current when pushing against opponent, robot not progressing but with wheels turning slowly on carpet, max throttle.

Note that all motors draw Stalled rotor current at max throttle at start up for around .2sec (with resultant transient droop in battery voltage for that time)

Motor Stall Currents:
4 CIMS ~400A, 2 FP 100A, Vandoor 60A, compressor 34A, (globes 20A)
totalling over 600A for .2 to .3 sec each event (if all Maxed together)!!!

Dale(engr294] since 1998 (TRW / Northrop Grumman)
[email protected] 310 374-8323

RSVP Comments invited, was info useful, clear, or need more ?

also attached: my Slides from Oct 2005 CSUN FIRST Workshop (LA, Calif area)
oops never mind
CD doesn’t allow .ppt

ADMIN note: Lets add extension: .ppt !! to share education data

2006-1_011906 2003-1_110405 BCRbattTst.xls (76 KB)


2006-1_011906 2003-1_110405 BCRbattTst.xls (76 KB)

This is great work. Will be interesting to see the graphs. Usually a battery is ‘dead’ when the slope of the graph takes a turn downwards (the knee).

Im thinking that not many teams will be happy starting a match off with a battery that is down to 9 or 10 volts at the start of a match.

One of the reasons that batteries seem to discharge faster than the A-Hr rating is that batteries go dead by increasing their internal resistance. At any given state of discharge a higher current will cause more voltage drop than a lower current.

Therefore, if you discharge a battery at 50 amps the under-load terminal voltage will drop to 9V sooner (A-Hr sooner) than if you had discharged it at 10 amps. The way you can see this best is to disconnect the load, the voltage will shoot back up.

I noticed one thing about your typical robot-loading numbers. Reading the stall currents on motors separately can be misleading. The current through the wires and connectors on the robot will cause a voltage drop in the system, so having 4 motors in parallel on the battery circuit will cause more voltage drop, therefore the stall current will be less.

also, the motors are on 40A breakers, and the entire robot is on a 120A breaker, so if your calculations are saying the robot is drawing an average of 160A while stalled, you have not taken the voltage drop in the wiring into consideration, or the fact that breakers will start kicking off.

I did some tests several years ago comparing the Exide batteries to the CSA batteries that used to be in the kit to the Yuasa batteries.

I had a set up with a bank of very large (WATTAGE) resistors. I adjusted the resistance to draw 100 Amps at 12V (0.83 Ohm).

I defined a battery as “Dead” when it reached 8V (I think I picked 8V because that was the voltage that the radio at the time would brown out).

I measured the time it took the battery to go from fully charged to dead.

Anyway, “good batteries” would take 450-600 seconds. “bad” batteries would take 250-300 seconds.

Based on my experiments, I believe that unless your robot transmissions or ball handling mechanisms are extremely poorly designed, you will not have to worry about using up a good battery in a 130 second match.

Joe J.

I think it would be helpfull to have a battery tester available thats tests under a specific load while its on the bot.

Turning your control system on and seeing 12.1V on the driver display doesnt tell you much. The battery needs a typical load to drop the voltage to where it will be when the robot starts drawing power.

Measuring the energy left in a battery is not easy or straight forward.

Dale,
I believe you can upload ppt files to the white paper area. You numbers are consistent with our research from several years ago. The first year custom circuits were part of the rules we built a current monitor that had several channels of data recording. That data was ported out to the dashboard at the OI and the battery voltage monitor on the RC also appears at the dashboard. By interfacing the data output, to a palm we were able to display and log conditions we thought would be dangerous to RC life support. (prior to backup battery introduction) It was easy to see the correlation between current draw and battery terminal voltage. With a portable rig we were able to test robots for other teams during practice or in the pits and it showed a tremendous amount of data, even showing one team that had a failed motor in one of their multi motor drive systems.