BatteryBeak: Utility of making internal resistance and state of charge measurements

[WinDnDusT]

Originally Posted by Joe Ross

What method does your team currently use for determining the condition of batteries?

My team's method isn't really controlled. It's more like 'This battery looks new, it's probably good' or 'Oops, that battery died during the match, let's not use it again'.

We don't have a person who has the domain knowledge to create battery management processes and policies for us. And we really felt the consequences of that this past year. We started off the 2013 season with 6 new batteries and by the end of the season, not one inspires confidence for use in matches. We took all of our batteries to a neighbor team's workshop and borrowed a machine that put a load on the batteries. We put a 200 A load on the batteries and not one of our batteries made it 5 seconds before dropping to unusable voltage. This did not contrast nicely with the neighbor team's batteries which lasted much longer (around 35-45 seconds I think).

So I'm trying to gain the requisite knowledge to be able to step into this role.

[Al Skierkiewicz]

Win,
The cells with reduced capacity that are shown in Ether's post and my discussion are cells that for one reason or another have some of the plates within the cell damaged or detached. The change in internal resistance is very small but the cell has less available power density and so it dies out sooner than the other cells. I recommend that teams do not try to pull 200 amps from a battery for extended lengths of time. Yes, robots do pull more than that but it is for much shorter periods of time. (except in extended pushing with another robot) The effects of this high current, extended test, is to warp the internal structures of the cells and raise the internal temperature to the point where cell relief valves vent the higher pressure (and water vapor) to atmosphere. I also believe that the separator (absorbed glass mat) may become damaged as well producing another type of internal resistance that discharges the cell constantly. In these batteries, the terminal voltage will fall in multiples of 2 volts after sometime (several hours to a few days) from being removed from the charger. All of the test methods including a voltmeter will show this defect when the cell has become completely depleted. In worse case events, the discharge is extreme enough to cause heat, leakage, case swelling and steam.
Above all, remember that our batteries have a stated life of only 400 charge/discharge cycles under optimal conditions. Our severe use cause this life to be reduced to perhaps 300 cycles or less.

[WinDnDusT]

Al, thanks again!

I now have a much better understanding of what the Battery Beak is capable of telling me in terms of internal resistance.

Thankfully, pulling the 200A from the batteries was the first and only time we've done that (although I realize the risk for damage isn't mitigated because of this). I'll advise my team against any tests of that sort in the future.

I am also looking into the CBA IV from West Mountain Radio that you mentioned. Out of curiosity, would the CBA IV allow you to find the Peukert constant of a battery? I know the (approximate) value of the constant when the battery is new, but I also know that it changes as the battery ages.

Also, I'm interested in keeping to the 50% percent rule for our batteries. Assuming that rule applies to our robot batteries, will the Battery Beak assist me here? If I take a state of charge measurement before using a freshly charged battery, do some work with the battery, and then take another state of charge measurement from that battery, will I know the true amount of capacity removed from the battery (In units of % rather than Amp hours)?

[Al Skierkiewicz]

Win,
I am guessing you are trying to make a judgement based on usage to predict battery life. There are some methods to do that but none are very accurate since the load varies with each match. The CBA will allow you to make a calculation to get Pleukert's constant but it will take several tests at different loads to do that. Unfortunately, the life of the battery and the temperature are constants you cannot readily measure with any accuracy. I am not sure what your 50% rule is, can you elaborate? I do have some rules of thumb I consider in battery use. These are merely observations reduced to writing...
1. If you deplete the battery in one two minute match, you have an inefficient drive system. If your robot suddenly starts to do this, you have a broken drive train.
2. Don't use 6 motor drives simply because you have they are the best. I can show you more teams that eat a battery than get a benefit from the multi-motor drive.
3. If you are only using 4 motors and succumb to rule #1 above, you have chosen the wrong final gear ratio for the game. More important if you are only using two motors.
4. A battery should last long enough to drive in two or three matches minimum before running down.
5. Never trust any of the above rules, always use a freshly charged battery.
6. Even batteries that come right off the charger can be bad and fail in a match.
7. The robot is talking to you, listen to what it is telling you.

In troubleshooting a robot electrical issue, I watch the battery voltage monitor on the DS dashboard. If you are driving and the voltage is falling to 6 or 7 volts you are going to have a problem soon. At sustained voltages of less than 6 volts, the regulators in the digital sidecar start to drop out, killing drive to your robot. At less that 4.5 volts, the power supplies on the PD cease causing the Crio and/or the radio to reboot. There are a number of reasons that could cause this. Too much load current, dead battery, too long of a #6 wire run, improper crimp or loose connections on the battery, the main breaker or the PD, will all contribute to low voltage.

[Ether]

Originally Posted by Al Skierkiewicz

I am not sure what your 50% rule is, can you elaborate?

I think he means do not exceed 50% depth of discharge of the battery.

[DonRotolo]

Originally Posted by WinDnDusT

(Don, did you know that your team was sitting behind mine at Championships during Einstein matches? We were thrilled because our team numbers are sequential)

Yeah, we've kinda fallen for that great UPS brown. I have a 1675 keychain from some regional (Cleveland?) where we were neighbors, that I treasure. (We also met 1677 somewhere!)

Originally Posted by WinDnDusT

Does this imply that a battery can have reduced capacity without having an increased/abnormal internal resistance?

Internal resistance is more than a hypothetical number. If a battery has 0.1 Ohm internal resistance (for example), it becomes a limiting factor on how many Amperes it can deliver. 12 volts/0.1=120A (assuming zero resistace across the terminals). Now double the resistance to 0.2 Ohm, and the absolute max current drops to 60 A.

Now in the real world, external resistance is significant, but this theoretical exercise helps explain how internal resistance is a good indicator of battery health (and capacity). So, your answer is No.

[WinDnDusT]

Yes, sorry.
I am referring to discharging a battery to only 50% depth of discharge.
I am wondering if the state of charge measurement after two minutes of activity (real match or simulated match) is accurate enough to allow promoting a 50% discharge policy with my team or to use as a troubleshooting technique to indicate that the robot uses too much power. I suspect that due to surface charge, it may not be.

[Al Skierkiewicz]

Win,
Your biggest issue with 50% rule is determining the 50% SOC point. If you were to accurately determine the current output of the battery and added a fudge factor for battery life, you might be able to predict the actual charge state. Under current robot rules, you cannot add anything between the battery and the PD except the main breaker and associated wire. Due to the nature of the DC source, loop current monitoring really doesn't work. You can however, measure the voltage drop across a known length of #6 wire. This resistance will change very little with temperature but will give you an indication of the current via Ohm's Law. Maxim makes a chip set to monitor battery charging in laptops and other devices made especially for this. The only issue is getting enough resolution over the current swing you would expect. Start current on a four CIM drive could easily exceed 400 amps and with the robot resting, current is likely less than an amp or two. As Don pointed out, battery current is affected by the internal resistance but more important, that resistance has a voltage drop. At nominal internal resistance of 0.011 ohm, 100 amps will drop 1.1 volts. 400 amps will drop 4.4 volts giving you an output of only 7.6 volts. When you add in the drop in voltage for a depleted battery, you are now approaching the point where the power supplies start dropping out. It is this effect that gets most robots into trouble. That is why the field monitors robot voltage in addition to other things (like packet loss, connection, etc.). If we see loss of connection or packet loss at the same time we see a drop in robot voltage, we know what is at issue.
To put this in perspective, loose connections (and bad crimps) on the main wiring, long cable runs and the main breaker can all sum to the same resistance as the internal battery resistance. So at 100 amps you have a 2.2 volt drop at the input to the PD instead of a 1.1v. Remember that includes the red and the black wire together since all robot current flows through both wires.
If we come to Milwaukee next month we can talk further.

[Michael Hill]

Originally Posted by Al Skierkiewicz

Win,
Your biggest issue with 50% rule is determining the 50% SOC point. If you were to accurately determine the current output of the battery and added a fudge factor for battery life, you might be able to predict the actual charge state. Under current robot rules, you cannot add anything between the battery and the PD except the main breaker and associated wire. Due to the nature of the DC source, loop current monitoring really doesn't work. You can however, measure the voltage drop across a known length of #6 wire. This resistance will change very little with temperature but will give you an indication of the current via Ohm's Law. Maxim makes a chip set to monitor battery charging in laptops and other devices made especially for this. The only issue is getting enough resolution over the current swing you would expect. Start current on a four CIM drive could easily exceed 400 amps and with the robot resting, current is likely less than an amp or two. As Don pointed out, battery current is affected by the internal resistance but more important, that resistance has a voltage drop. At nominal internal resistance of 0.011 ohm, 100 amps will drop 1.1 volts. 400 amps will drop 4.4 volts giving you an output of only 7.6 volts. When you add in the drop in voltage for a depleted battery, you are now approaching the point where the power supplies start dropping out. It is this effect that gets most robots into trouble. That is why the field monitors robot voltage in addition to other things (like packet loss, connection, etc.). If we see loss of connection or packet loss at the same time we see a drop in robot voltage, we know what is at issue.
To put this in perspective, loose connections (and bad crimps) on the main wiring, long cable runs and the main breaker can all sum to the same resistance as the internal battery resistance. So at 100 amps you have a 2.2 volt drop at the input to the PD instead of a 1.1v. Remember that includes the red and the black wire together since all robot current flows through both wires.
If we come to Milwaukee next month we can talk further.

My biggest issue with ChiefDelphi...

[Ether]

Originally Posted by WinDnDusT

I am wondering if the state of charge measurement after two minutes of activity (real match or simulated match) is accurate enough to allow promoting a 50% discharge policy with my team or to use as a troubleshooting technique to indicate that the robot uses too much power.

It's an interesting question. I haven't seen this specific question answered here on CD.

Are there any teams out there who regularly measured and recorded internal resistance and state-of-charge using battery beak, before and after each match? Would you be willing to post your data (and procedure^*) here?


^*for example, how long did you wait after charging (to let surface charge dissipate) to measure the before-match values, and how long did you wait after the match (to let the battery stabilize) to measure the values

[Ether]

Originally Posted by WinDnDusT

Does this imply that a battery can have reduced capacity without having an increased/abnormal internal resistance?

It is certainly possible for 2 different battery designs (call them A and B) to be such that A has less capacity than B while having the same internal resistance.

I suspect that it's also possible for a (damaged, old, or abused) FRC battery to have reduced capacity without showing a definitive increase in internal resistance.

Tom Line and/or Hugh Meyer,

If you are following this thread, do you by any chance have internal resistance measurements for the bad-cell batteries you tested?

[Hugh Meyer]

WinDnDustT,

I would highly recommend the west mountain CBA tester. It gives a good solid indication of the state of the battery under test. We purchased one several years ago and it has given us data that takes the guess work out of knowing the state of the battery.

We test our batteries at least once per year during build season, so we have fresh data on our complete inventory before competitions start. I compare the results to previous years to decide if a battery will be used for competition. By looking at the discharge curves as they change from year to year we can tell when a battery is nearing end of life. Bad batteries go to the recycler and marginal batteries are marked “Not for Competition”. Only good ones are used for competition.

Throughout each event we log the state of charge before the match using the battery beak tester. Attached is a copy of what we do. I have several years of log data and so far have not found the internal resistance to be much of an indicator of anything useful.

Since we got the battery beak the first year it came out we have not had a battery fail in a significant way. We do have some that are getting higher than normal. I have been told by the people that make the battery beak that an internal resistance in the range of .035 is a trashed battery. Haven’t seen that yet, but we have some getting up to .028. In an odd sort of way I am hoping to see a failure this season just so I can see what the internal resistance is when they fail. Most of our batteries measure in the range 0.015 - 0.025. These batteries perform fine.

One thing that is important, but not mentioned so far in this thread is the state of your connections. In our FRC world by far a much more significant issue is the connections that people make. It is extremely important that you make good tight clean solid connections throughout the circuit. When you test a battery your measurement is only as good as your weakest connection. Since you are not a rookie team I assume you know this but I wanted to mention it just to be sure others reading understand the importance of good connections.

Attachment 1 is our log template that we complete at competitions.

Attachment 2 is our inventory data sheet that shows battery history. You can see when we got our beak tester where that data starts to populate.

Attachment 3 is a log sheet completed.

Seeing a team go dead in a match because of a dead battery is really sad. More than once I have heard a team declare they unintentionally went out with a dead battery. ALWAYS know the state of your batteries!! The beak tester and CBA tester are great tools!

-Hugh

[Mr V]

Originally Posted by Al Skierkiewicz

Battery Testers of this type http://www.mcmaster.com/#battery-testers/=o19fra use high current loads, 60 and 125 amps or simply 125 amps and produces a meter deflection to show good or bad. A car battery (.003 ohm) has an internal resistance typically 1/4 of our battery (.011 ohm) so this tester will not give you a good idea of the state of the smaller battery. I cringe at teams that use this battery tester as it will continue to load the battery as long as the switch is held "on" which under certain conditions can damage the battery.

As long as the switch is not held in the "on" position for too long there is no risk to damaging the battery since it does not put any more of a load on it than does use in a high current demand FRC robot.

I wouldn't recommend using it on a daily basis but at the beginning of the season or before heading to competition is not a bad idea.

[Tom Line]

Originally Posted by Ether

If you are following this thread, do you by any chance have internal resistance measurements for the bad-cell batteries you tested?

Unfortunately, no. We don't record any data from the battery beak - it's our 'fast' test during competition.

We test our batteries using the west mountain device in the first two weeks of the season to weed out any batteries we killed during demonstrations, then rank the rest based on Amp-Hours.

[evanperryg]

Originally Posted by Al Skierkiewicz

Win,
1. If you deplete the battery in one two minute match, you have an inefficient drive system. If your robot suddenly starts to do this, you have a broken drive train.
2. Don't use 6 motor drives simply because you have they are the best. I can show you more teams that eat a battery than get a benefit from the multi-motor drive.
3. If you are only using 4 motors and succumb to rule #1 above, you have chosen the wrong final gear ratio for the game. More important if you are only using two motors.
4. A battery should last long enough to drive in two or three matches minimum before running down.
5. Never trust any of the above rules, always use a freshly charged battery.
6. Even batteries that come right off the charger can be bad and fail in a match.
7. The robot is talking to you, listen to what it is telling you.

making this into a poster. We actually had an interesting occurrence related to rule 6 at champs. One of our 2012 batteries came directly off the charger, 13.7v. After the match, the battery was around 8v. We figured it was a fluke and threw it on the charger. about 5 minutes later, I go and check this battery to see if it is ok. The thing is hot(as in burn-your-hands hot) and still at 8v. Long story short, nothing is a fluke with a battery.