2 AWG Battery Cables


#40

Yes, and it’s fine to argue that teams ought to look at larger than 6-gauge wire (we use 4-gauge, ourselves, and might look into SB120s). But it’s important to use a realistic operating point when making the decision.


#41

Would you please post a short “executive summary” of the method you used to do this?


#42

*(https://github.com/blair-robot-project/449-central-repo/blob/master/RoboRIO/src/main/java/org/usfirst/frc/team449/robot/components/RunningLinRegComponent.java)

Essentially, we just kept a running linear regression of battery voltage (as measured by the PDP) versus total current draw (also as measured by the PDP) over a window of a few seconds. The intercept of this should be the battery’s no-load voltage, and the negation of the slope should be the total resistance in the battery-PDP circuit.

Of course, if the current draw remains roughly constant, this method won’t work - but in actual robot operation, it’s usually moving around enough to work quite well, we found.*


#43

We had a rough go because of power connections with the main breaker and battery before. So it’s pretty important for sure.

However, we have purchased this crimper.

It works great for anderson connectors. We buy our andersons from them as well.


#44

I would absolutely advocate for the SB120 and at least 4AWG wire for the Battery to PDP circuit for most teams. I don’t believe the power supply circuit spec has paced the motor rules for FRC, and at this point the SB50 and 6awg is at the very edge of its use case.

We used 2AWG welding cable and SB120’s for our supply side circuit this year and were extremely pleased with the results. One of our major concerns and decidedly the failure mode of high powered 700 series eight motor drive trains is tripping the main breaker. Peak discharge current is divided across eight motor channels providing highly effective protection for the motors, which pushes the failure mode up stream to the Main Breaker. While a few of the benefits have been indicated here, one that may have been missed is the extra thermal mass in that circuit. The best route for heat to leave the main breaker is through the conductor. If you put more copper in the circuit, directly adjacent to the breaker, it serves to pull more thermal energy out of the main breaker in contrast to a comparable 6AWG setup.

Ultimately, we ran our robots (competition and practice chassis) all season with eight motor 775 drives with no programmatic protection (current or voltage limiting) through 85 matches and dozens of hours of autonomous and driver practice and never replaced a motor or blew a main breaker. Further, as finally configured, we were running 15 775’s on the competition robot with no power use issues to speak of and I firmly believe our decision to use the SB120 and 2AWG cable was a factor in that performance.


#45

I want to point out that the 600% up to 3 sec is the upper bound of the range not an absolute. The graph also shows that the breaker could trip at that level in as little as 1 sec. This is of course due to the manufacturing tolerances of a product that is built to a price point and not individually certified to a narrow spec.

The other important take away from the data sheet is the temp derating curve. The lower bound crosses the 100% line at 77 degrees so I’m going to assume that is the ambient and device temp used for the time vs % graph. I’m going to further assume that the test is started with the device at that temp and subject to the listed load immediately. They are thermal breakers and the heat does build up with time especially since we use one of the sealed versions. So if it has been operating at say 50% of rating for a period of time before being subject to say a 300% load it will trip quicker than the same breaker that was subject to a 300% load immediately.

A good example of this are the circuit breakers that used to be incorporated in many automobile headlight switches. As the started to fail the common scenario was that you could drive around for 20, 30, 40, minutes without any issue. However drive for a longer period, the breaker would trip, the lights would go out and since it was an auto reset come back on a second or two later. However once it tripped it would trip again soon there after and your headlights would turn into flashers until the switch was allowed to cool w/o any load for a significant period of time. Some of that was due to the switch to wire connection gaining resistance over time and putting heat into the breaker. So as we all say good connections are imperative and the heat a poor connection generates can affect other things. This is also a point as to why a larger wire is a good idea. The studs are the most effective heat path out of the breaker and the greater the thermal mass of what is attached to them is pretty much the only way to affect in-use heat dissipation. So potential breaker performance improvement is one of the benefits of the larger gauge wiring that hasn’t been mentioned.

12 milliohms is the lower bound of the range of specifications of one of the legal batteries, while 15 milliohms is upper bound of that specific battery.

Unless your set up and code is designed to measure the voltage of the battery at the battery terminal then you are reading the resistance of the entire circuit which includes all of the connections, wire and breaker. The Battery Beak on the other hand takes its readings through fewer connections and less wire. Note it is possible that the Battery Beak’s code is designed to account for the expected resistance of a the wire, terminals and connections that are properly done. However the lowest reading I’ve seen with recently acquired batteries from the mfg that states the 12-15 milliohm internal resistance is 17 milliohms.

Readings taken internal to the PDP will include potentially more than double the wire and certainly more than double the terminals and connections. So while it gives the big picture that does not mean it is necessarily more accurate than the Battery Beak just representative of the total supply system resistance.

I fully agree that you should stay away from the upper bounds of the breakers rating, especially due to the build up of heat over time detailed above since a tripped main breaker means the robot is dead for the rest of the match.

I would also like to see this. Since this discussion is about the potential benefits of 2 vs 4 vs 6 ga the code could be used to determine the real world benefits if someone was willing to take the time and swap between the different gauges using the same battery with each set up.

Then we could actually put numbers to the benefit side to balance against the costs and also find the operating point at which the benefits actually are seen.

It is also important to note that you can to a point compensate for the lower available voltage by increasing the motor control output. In other words the improvement of larger supply side wiring doesn’t provide any benefit at lower demand operating conditions.


#46

I’m well aware of this, which is why I mentioned the gauge wire we use and the length of the run. But, as we’re interested in operating conditions on an actual robot, this resistance (including wiring to the PDP) is what matters, not the “actual” internal resistance of the battery.

I would also like to see this. Since this discussion is about the potential benefits of 2 vs 4 vs 6 ga the code could be used to determine the real world benefits if someone was willing to take the time and swap between the different gauges using the same battery with each set up.

I’ve already posted the code and an explanation above.


#47

Hi all,
There was an interesting talk at worlds in Detroit a few weeks ago about testing batteries.
Their tester actually uses big power resistors (max current I think is 125A) turned on and off with mosfets, and they have software to simulate the rigors of an actual match, as well as just a quick test. Read their paper.
One of the things they came up with is the internal resistance.
In short, the mid teens in mOhms is about the best, and their testing shows how batteries definitely get worse (internal resistance wise) as they age, get dropped, etc…
Here’s their web site, no affiliation, just an electrical guy who is interested in these things.

Also, I’ll throw in there that we ran some numbers last fall with our electrical team and while I don’t remember the exact numbers, we came to the loose conclusion that one is only able to deliver about 60 amps to a stalled CIM, give or take 10 amps, in the drive train.
This would indicate to me that the combined system resistance, PDB, SRX, wire, connectors, breakers, etc., is about the same as (again, loosely) as the resistance of our 4 drive motors.

I like the idea of 2 AWG, and may try it out next fall with the team.
As far as sharing batteries, etc, just make an adapter. I know we have several of these to go from the SB50 size to the regular 45A power poles we use for testing purposes.
Although, now that I said this, there could be rule problems? 120>>50?

Thanks for an interesting thread everyone,
Mike


#48

The current rule set, specifically R46, allows for only one SB type connector pair on the robot, so adapters aren’t legal.

It’s probably something that one could lobby FIRST for a rule change on, since the wording of the rule strikes me as being something created to eliminate ambiguity on the high current pathways of the robot, and not recognizing some inherent issue with multiple SB type connectors in series.

If you really want to use both SB120s and loan batteries, just bring your old SB50 wire assemblies and swap them onto the loaned out batteries. It’s not all that much effort, really.

We do have a SB120 to SB50 adapter, but it’s really just for adapting the battery beak these days.


#49

As a note for those not familiar with the history, prior to 2015, teams were required to use the SB50. The wording of the rule was changed in 2015 to allow other SB connectors, as teams were expressing a desire to use the SB120 on their robots.

I would have sworn there was a question that year about using a 50-120 converter with borrowed batteries, but I searched through the Q&A archive since then and couldn’t find one. The closest I could come up with was Q739 in 2016, which was asking something a little different but got the answer:

No, additional connectors attached to the battery would be a violation of R36. R36 is an exclusive list of what may be connected between the ROBOT Battery and the Power Distribution Board, as shown in Figure 4-10.

There was also Q254 in 2017, which asked about using a 50-120 converter between the battery and a battery charger, and that was deemed legal.

I agree with your interpretation, that this helps to remove ambiguity. As for safety… I personally don’t see a problem with it, but I’m not an expert and wouldn’t want to make that ruling. I think the 2017 Q&A helps point towards a pass on safety, although that’s for a circuit limited to 6A, which is much less than we routinely see on a robot. But the rules are the rules, and without asking on the Q&A next year to get clarification (or a change to the rule next year), I would have to rule it illegal at my events.

That said, I personally wouldn’t want to use such an adapter - I’d rather come prepared to swap the connector out completely. The increased resistance of having both a pair of SB50’s and SB120’s, along with the increased opportunity for one of them to not be fully connected/come unplugged is enough to make it a no-go. Replacing the connector is a quick enough process if you have a replacement handy.


#50

Through out this discussion the common thread is how to over power a couple of milliohms, of resistance. That is so much worry about so little.

Your concern should be about friction, in all of the moving parts. Friction will consume far more power than a couple of milliohms, in a foot of wire. Seriously, there is little more than one foot of wire, in any of the branch circuits.

Once you guys graduate, and if you study electricity, you will find that publications (Nat’l. Elect. Code ) look at power differently. It will be about conduit fill, balanced or unbalance load, distance, amperage draw, and so on. When you spec out a wire pull for a machine pulling say 125 Amps at 100 - 150 feet away, then wire size means something. Your boss won’t be happy if you throw away a couple of hundred dollars oversizing the wire, just because.

Every pound of wire, you add to your robot, is another pound of wire you have to make power to carry, and lift, in the end game.

Make sure your wire connections are correct, and tight; And, your mechanical parts work smoothly and easily. Okay I’ll get off my soapbox now.


#51

I feel like bringing up the electrical code is a little ironic. There whole thing is over sizing wires (for good reason). The loads are very different.

I’m not sure this discussion is about the amount of energy consumed by the additional resistance but about the reduction in performance caused by the voltage drop.


#52

Okay, the wire resistance for 10 gauge wire is 0.001 Ohms. For 12 gauge it is 0.002 Ohms. Twice the resistance, that’s a lot or is it.
Amps = Volts / resistance. Sample: 1 V / (1 ohm load + 0.002 ohm for wire) = 0.998003992 amps or a 1.996 mv drop for the 12 ga wire. The 10 ga wire is 0.999 mv per foot. So it all works out to a difference of ~1mv drop per foot.

My argument is a chain too tight will cost more power than the wire difference.


#53

I can’t think of a reason why I would purposefully increase friction. Typically I don’t try to design things with excess friction, and I teach 1072 not to do the same. There’s no reason not to reduce friction and change battery wires, unless changing b attery wires isn’t worth it independently; the two are unrelated concepts and need different amounts of attention.
Changing battery wires is more to prevent brownouts due to the drivetrain than anything else; very little else on the robot is usually the source of a problem.


#54

For what it’s worth, I’d argue in-code current limiting is far-and-away the best solution for brownout-prevention in modern FRC, especially given how easy it is to implement with the resources currently available to teams (specifically, the Talon SRX).

Bigger battery wires can allow you to use a more-aggressive current limit, but you shouldn’t be relying on them alone to prevent brownouts; not when such a simple, reliable solution is currently available.


#55

12 AWG ~ 24 lbs / 1000 ft
10 AWG ~ 38 lbs / 1000 ft
8 AWG ~ 63 lbs / 1000 ft

6 AWG ~ 96 lbs / 1000 ft
4 AWG ~ 153 lbs / 1000 ft
2 AWG ~ 234 lbs / 1000 ft

Suppose your drivetrain is 6 CIMs, the (one-way) wire length from battery to PDB is 18 inches, and (one-way) wire length from PDB to speed controller to motors is 15 inches.

Upgrading your wire from 12 AWG to 8 AWG and 6 AWG to 2 AWG will add just under a pound (.648 lbs to 1.647 lbs), but will increase the voltage your 6 CIMs will see at 40 amps each by about .3 volts.

An extra 72 watts of power for an added 1 pound of wire seems like a pretty easy upgrade.


#56

Need numbers…


#57

We use one of the West Mountain Radio CBA / Battery testers to test all our batteries yearly for health. I just went and checked our known good batteries (14 of them) and almost every one of them is checking with a battery beak in the range of 19-23 milliohms.

Anyone else have some actual measurements from their batteries? It only takes a minute to test all your batteries. I’m sure the battery beak isn’t testing the way the manufacturers do, but I’d like to believe it isn’t off by 25%.


#58

Friction is a portion (or in some cases, all of) the D gain in a PID loop. There are several reasons to purposefully increase friction, for example eliminating backdrive after climbing when the motors are disabled (think 2016 and 2013). We have intentionally built friction into components to help regulate their velocity or position control (in particular when closed loop may not have been an option – eg in a relay circuit).
In 2017 we adjusted friction on belt drives to help eliminate slip when in contact with Fuel. Sometimes simply reducing friction isn’t a viable option, you need more power applied from the motor. I would argue thicker wires help with that.


#59

OK, I think it is time to talk about how manufacturers test their parts and make up the specs, motor curves and discharge curves. The test systems that are used have very large cables so there is very little resistance between battery and load. When a battery is speced at 600 amps, that is with as little resistance as possible and a very accurate load resistance while monitoring the current. Given that test criteria, 600 amps will drop 6.6 volts across the battery internal resistance at 0.011 ohms. The CIM motor is likewise tested with large wire so that when the curve is labeled for 12 volts, that is measured at the motor terminals.

Now for real world robot systems. The load in our robots is not accurately maintained and is very dynamic. Most of the well designed robots, I have been able to investigate, will have a large CIM motor at stall draw between 100 amps and 120 amps if only one motor is running. Typical four motor drives will draw more like 90 amps for each motor. Many robots will introduce significant resistance in the wiring, connectors and wire such that large CIM motors are more likely to be drawing 80 amps maximum with only one motor running. All robot current flows through the primary circuit (the battery wiring, the internal resistance of the battery, the main breaker and the PDP). So lowering the resistance in that path will result in better performance but your mileage will vary. You can’t change the breaker, the SB connector or the PDP so that only leaves the wiring. The discussion thus far has shown that #4 wiring would be 0.0008 ohms compared to 0.002 ohms for #6.

For this discussion, a typical robot will have at least .005-.010 ohms if one were to account for series resistance of the main breaker and all connections. You can see that as you total up the resistance, the available voltage at the motor is vastly reduced. Using the CIM motor specs, the internal resistance of a stalled motor is 0.091 ohms. So if we run the calculations on that (with no other load), the battery is only really delivering 275 amps with a four motor drive and #6 wiring. Moving to #4 would make that 282 amps. Knowing that there is a voltage drop across the primary circuit (Battery, wire, breaker, PDP) 275 amps will drop 5.75 in the primary circuit, or will produce 6.25 volts at the output of the PDP. Moving to #4 will make the 6.2 volts rise to 6.4 volts. For a six motor drive, the output of the PDP would be 5 volts for #6 and 5.4 volts for #4. For reference the RoboRio disables output (true brownout) between 4.5 and 6.3 volts at the input to the RoboRio. So using #4 wire could protect brownout (by a whole 0.1 volts) if everything in your system was perfect.