Rules Change I Would Like to See - Batteries

Under the current rules, all systems must be powered from the single 12 volt battery, with the exception of sensor/computing systems that have their own built in battery.

I would like to have the “built in battery” portion removed, allowing separate batteries for coprocessors such as the Raspberry Pi.

The reasoning behind it is that there are sensors and processor boards that are meant to run on USB power, and may not be easy to hook up with anything else. The USB port of the RoboRio can’t supply enough power for these systems. It would make it easier for teams to make use of more advanced sensing and control strategies for which the RoboRio is not well suited.

The exact wording could be problematic. I’m not sure I could come up with a perfect wording of the rule, but if someone could, then it could work. I would want a rule that allows “portable phone charger” type power supplies to be used to power those auxiliary boards that are designed to run vision coprocessors or similar algorithms.

Arguments for:

We are trying to encourage young people to use cutting edge technology. This makes it easier.

Since it is only used for computing, there are no safety issues with supplying power to motors and actuators.

The current rules encourage the use of tools with built in power devices, which might not be the best for the job.

Arguments against:

It makes the inspection process more complicated. An inspector might miss some sort of “sneak circuit” that could allow an actuator to be activated when the robot is supposed to be disabled.

Although it allows use by students of cutting edge technology, it also makes it easier for mentors to put in systems that are far beyond student capabilities, allowing an even greater advantage to “mentor built” bots.

“Making things easier” is not always a worthy goal for a First program. Working within constraints might be more important, so the current constraint is simply an engineering challenge that teams have to work within.

Although the intent is straightforward, the actual wording of the rules might be extremely complicated. If the rule is too simple, it might be easy to exploit loopholes for unintended capability. If it is too complicated, it might result in accidental violations due to misunderstanding the rule, or misinterpretations by inspectors resulting in confusion and controversy at competitions.

And now, fill in your own advantages and disadvantages. This is just something I have in mind. I am not adamant that we need to do this right away. I simply find myself wanting to do some Raspberry Pi work with some students, and I find that I will have to do some rather odd things to power the board, that might not be the safest for the equipment. (There’s no human safety issue that I can see.) It would be easier if I could just velcro on one of those battery power packs and power the Pi directly. It may be that there is an obvious problem with this idea that I haven’t thought about, that might turn it into a very bad idea.

An alternative would simply be to modify the next generation RoboRio or control system boards to provide more USB power. I don’t know how far in advance these things are designed, but my guess is that there are already modifications in the works. I just don’t know if we are likely to get new or modified control systems in 2, 3, or 4 years.

Why not use a 12V to 5V adapter? Wouldn’t that count as a custom circuit (and already be legal)?

It is already possible to power a Raspberry Pi off of a robot. You just need to wire the power cable to the 5V/2A port on the VRM. My team used it this year for our vision system and it worked just fine.


[strike]Allowing 12v to 5v regulators would be a lot more useful in my opinion.[/strike] There’s not need to have a separate battery when you can just step down the PDP voltage.
Sometimes the VRM doesn’t have enough power to give. Even the 5v 2a slot isn’t enough for some applications, and I dislike the extra loose wires that it requires. Ultimately though, there’s not enough need for more 5v power to make it worthwhile imo.

DC-DC converters are legal.

We have also used automotive USB power converters to power USB and other 5V devices on both our competition and non-competition robots (e.g. t-shirt cannon); you can get these at almost any local convenience store. (Example). It is usually a simple matter to remove the power port connectors and expose the wires to provide the 12V input. There are also units with wire pigtails or spade connectors for inputs, designed for installing in a dash board; a bit more effort to locate, but neater-looking.

I would suggest against a USB converter because you’re really only guaranteed 500 mA. Odds are it can actually do more than that, but something to watch out for, especially if you’re doing anything computationally intensive on the Pi. I’d also watch out for some of the cheaper automotive USB converters. The power is often really ugly (lot of switching noise), which is generally fine for charging your phone, but for a Pi, it may be a little more sensitive to switching noise.

Thanks for illustrating the issue very well.

Yes, there are ways to do it, and lots of teams do them. However, there are issues with them. Gotchas. Things to look out for.

If the students were just allowed to use devices that were actually designed specifically for providing power to a device that was specifically designed to use that power, those gotchas wouldn’t exist.

Again, some people might look at that as a good thing. Providing an easy solution is not necessarily the best thing for a FIRST team. The question that the rule makers have to ask themselves is whether they want to encourage people to make use of the fancier processing available through the use of coprocessor boards. If so, do they want to make a little bit of electrical engineering a prerequisite, or do they want to let them cut straight to the processing part, assuming they are willing to pay a small weight penalty. (I would assume that, unlike the main battery, any batteries used to power auxiliary computers or sensors would count against the robot weight allowance.)

I don’t think there’s a “right” answer to the question. I have an opinion on the subject, obviously, but I won’t say that there is an obvious answer.

So far, though, the people providing information haven’t really discussed advantages and disadvantages, just that the rule change isn’t necessary because there are engineering solutions available for teams under the current rules. I suppose the disadvantage is that it introduces an unnecessary change to solve a problem that, in many people’s opinion, already has an adequate solution.

What would you define as a dedicated USB power source? The powerbanks for phones are one example, but those have a non-negligible power storage compared to the robot battery itself. One can simply use a larger DC-DC regulator to get bigger 5v power.

What I have in mind is the powerbanks for phones. The exact wording is something I would leave for the rule makers.

Sure. You could use a DC-DC regulator. It’s not that hard, but it requires either a student or mentor who understands the requirements, and a little bit or rewiring and cable cutting to hook the output of that to whatever the coprocessor board requires.

It just means that a programmer who wants to do some cool programming has to team up with someone who understands how to get an adequate amount of 5 volt current out of the available power sources. A student whose focus is programming, and who knows nothing about electricity and power supplies now depends on having someone else on his team that can take him past the “easy” step of finding the appropriate regulator.

Not that that’s a huge deal, but on smaller teams that might be the thing that prevents the budding programmer from making it happen during competition season.

One thing that I have seen, both in my professional career and in FIRST robotics is programs stalled because the person assigned to the task knew how to do the “hard part”, but not the “easy part”. Just last week at the office, I had to coach an expert Java programmer through the arduous task of reading a voltage from a device. We have been working on automotive diagnostics via Canbus. She knew everything possible about networks, communications protocols, and data structures, but we had a new task to mix in voltage data read from a data acquisition device. She knew everything about how to write the program, but she was utterly perplexed about where to connect the wires, or even what to connect them to.

I’ve seen it happen in FIRST as well. The electrical people can’t do anything because they can’t write a two line program. The software people can’t do anything because they can read the sensor, but they can’t connect it to a power supply. This proposal just makes it easy to get over that hurdle.

Whether or not that is a good thing is a matter of opinion.

I would guess that $5 noname chinese usb battery packs off of Amazon are exactly the type of thing the GDC has created so many electrical rules to ban. Unless they only approved a few specific models, regulating something like this would be a nightmare. Not to mention the 5V/2A isn’t even enough for many computers. Odroids (5V/4A) and Jetsons (12 or 19V ??) come to mind.

I think a lot of what you’re describing is just the reality of FRC, and robotics in general. The mechanical people need somebody to program PID for them, the electrical people need the programmers to make the lights blink, and the programmers need the other two to do everything they need to.
I agree that finding people who know electronics is hard (of the entire “Electrical” division on our team, maybe 2 of them know electronics; the rest are programmers only). But the knowledge that it takes to wire a 5v circuit shouldn’t be something beyond the capabilities of a few mentors to tackle.
Personally, I’ve found my knowledge both of electronics and of programming to be very useful, especially when it comes to using microcontrollers. Also in helping the programmers figure out electrical problems. :stuck_out_tongue:

Exactly. There are safety concerns as well given the prevalence of lithium ion chemistry batteries. You do NOT want a large one of those punctured or shorted. Cell phones and Kangaroos have pretty small batteries, and laptops are typically above the $400 cost limit (or considered too heavy). Punctures in even small LiPo batteries can be quite destructive, and a large lithium ion battery on a FRC robot sounds like a really bad idea when DC to DC converters (or doing vision processing on the driver station) are perfectly viable alternatives.

While we’re at it, I think it’s worth a mention that you are allowed up to 3 Additional VRM units on your robot according to R48 of the 2016 Manual. If you do have a processing unit on your robot that requires more than 2A (but less than 20 due to the fuse on the PDP), that is a possible option. As many others have mentioned, DC/DC converters may also be of interest.

Honestly, if you have a processing unit on your robot that is constantly drawing more than 2A, it might be a good idea to ask yourself “is this worth the trouble?”. Something to keep in mind that electrical engineering, especially on a mobile robot, usually has the challenge of working with the power you’re given. FRC Robots have a pretty big supply of power going to them as is, do we really need more?

Many of the 12V converters feature a 2.1A port and a 1.0A port, which is enough for a pi. You can always add a couple of capacitors (it’s a custom circuit) to clean it of if necessary. Our preferred method is still the VRM or other robust, regulated DC-to-DC converter. Jaci’s bottom line is right on the money:

Remember, in addition to the battery, we can (in most recent years, anyway) have pre-charged pneumatics tanks, pre-wound springs, and gravitational potential energy. (R35 A, C, and B respectively in 2016.)

Finally, If you really want that separate USB battery-power, charge up an old laptop, and power your pi off its USB port. There’s no rule that says you have to turn the laptop on, much less do any processing on it. If you can use a classmate from a KoP, you won’t even have to list any dollar cost in the BoM.

As I see it. The problem the second battery solves is not losing your sensors and CPU from brown outs. The old (pre 2009) control system had a battery for this purpose because a brown out would cause a loss of control without disabling the robot. A second control power battery Powering the Rio and radio along with other custom circuits would solve a lot of issues. It would also add cost and complexity to the robot.

Would another possibility for the auxiliary processing be the android system used in FTC? The come with batteries and meet the COTs definition.

A cool idea, but I think the VRM (or a second or third VRM as needed) would do the trick.

We powered a Pi and an IP camera (and its LED ring light) off the VRM on our competition robot. On our practice bot, where we were a little less, uh, picky, we plugged the Pi into the USB port on the RoboRio.

Well, that’s a pretty compelling argument against my proposal. If there’s a genuine safety concern, that pretty much trumps everything else.

The DC-DC converter plan seems like a better idea.

The safety aspect can be dealt with fairly easily. Cost and complexity still remain. The battery powered tool industry has pretty much gone to Li ion for everything except the bottom line tools. They get badly abused and have a low risk of fire/explosions. First could pick one or two commercial solutions and restrict maximum current draw and require specific chargers.

Another option for Arduino and PIs is somebody to package it with a battery and case. As long a it a company and they make it available for sale, it becomes a legal COTs device. (insert the usual future rules disclaimer here). Make it nice enough and maybe AndyMark would stock it for you.