How to power Mini-PC on robot?

[Kevin Sevcik]

Originally Posted by team-4480

Hi,

We have a mini-pc that we would like to use for some advanced image processing that can only be done on a full-fledged mini-pc. We wrote the code and tested it and it works great. Now the problem is powering it on the robot. On the AC-DC converter, it states that it outputs 19.5V @4.62 Amps. We found a 19V 5A adapter, but unfortunately, when we plug it in, the mini-pc beeps and gives a red light indicating that the power supply isn't good enough, likely due to the lower voltage.

Has anyone ever done anything like this, or does anyone have an idea on how to properly feed 19.5V to the mini-PC? Also, what are the rules regarding use a DC-DC converter?

Thanks a bunch in advance!

The rules that would apply to you are R49, customs circuit may not produce more than 24V, and R10, no individual item may exceed $400 in cost.

I'm told R49 is now in place because someone had a 12V to 120VAC inverter on their robot as their solution to a problem like yours. Kudos to you for not trying to electrocute your inspector!

What brand and model of mini PC are we talking about here? I just googled 19.5V car charger and came up with come likely candidates. I know for a fact that many laptops have an extra pin that transmits data about the charger so the laptop can be certain the charger is "safe"*. This may be a similar issue.

On this I'd be concerned about is how your mini PC is going to deal with riding on a robot. The advantage of raspberry pis, kangaroo pcs, etc. is that they're solid state and generally very compact. There's not much inside to get jarred loose and bounce around. Are you sure your miniPC is going to put up with a headon collision with a defender?

EDIT: Here's an adjustable DC boost converter that should work. Though it suggests you'll need a fan on it to pull 5A. While you're at it, have someone print you a case for it, or it's unlikely to pass inspection.

[team-4480]

Originally Posted by Kevin Sevcik

The rules that would apply to you are R49, customs circuit may not produce more than 24V, and R10, no individual item may exceed $400 in cost.

I'm told R49 is now in place because someone had a 12V to 120VAC inverter on their robot as their solution to a problem like yours. Kudos to you for not trying to electrocute your inspector!

What brand and model of mini PC are we talking about here? I just googled 19.5V car charger and came up with come likely candidates. I know for a fact that many laptops have an extra pin that transmits data about the charger so the laptop can be certain the charger is "safe"*. This may be a similar issue.

On this I'd be concerned about is how your mini PC is going to deal with riding on a robot. The advantage of raspberry pis, kangaroo pcs, etc. is that they're solid state and generally very compact. There's not much inside to get jarred loose and bounce around. Are you sure your miniPC is going to put up with a headon collision with a defender?

Our HP EliteDesk Mini 705 G1 (The first computer on the page) ran 250 dollars. So I would have to find a charger that transmit a signal saying that our charger is safe?

We are really conveniently in the durability of our mini-PC. It will be mounted using the VESA mount on the back and maybe a couple of zip-ties just to be sure. It has an SSD so there are no moving parts inside the computer.

[Adnewhouse]

Something like this should work nicely and is perfectly legal on the robot: http://a.co/7VNpilJ

[Kevin Sevcik]

Meant to put an asterisk on that "safe". "Safe" in this case means both that it puts out enough amps, and that it's the correct brand, etc. Third party chargers don't work on Lenovos, for instance.

But I digress. What does the plug on the end of your AC adapter look like? Is it just the sleeve and barrel, or is there a pin in the center of it as well? Post a picture of it, if that didn't make sense.

EDIT: The rear view image of that PC on the web doesn't bode well. It looks like the power port wants a center pin, which carries that ID information, which, while spoofable, is pretty difficult to spoof.

EDITx2: Since it looks like you need that ID pin, you're back to first or third party laptop chargers. this might work if the OD of that connector is right. I'm afraid this is going to be a little hit or miss to get this working, though. Can you post the model number of the stock AC-DC adapter? If it can be matched up to a corresponding laptop, then you just have to find a car charger that's compatible with that laptop.

[team-4480]

Originally Posted by Kevin Sevcik

EDITx2: Since it looks like you need that ID pin, you're back to first or third party laptop chargers. this might work if the OD of that connector is right. I'm afraid this is going to be a little hit or miss to get this working, though. Can you post the model number of the stock AC-DC adapter? If it can be matched up to a corresponding laptop, then you just have to find a car charger that's compatible with that laptop.

Yes, there is a center pin for the charger.

I will post a picture tomorrow of the official charger brick. Hopefully, we can figure something out. Thanks for the help thus far!

EDIT: Now that I think of it, I believe there was a third white wire when we stripped of the connector to test with our DC-DC adapter. Maybe we just have to apply a certain voltage on that wire and it will work?

[Kevin Sevcik]

Originally Posted by team-4480

Yes, there is a center pin for the charger.

I will post a picture tomorrow of the official charger brick. Hopefully, we can figure something out. Thanks for the help thus far!

Yeah, that's a "smart" charger. Hopefully a pic of the label on the charger will let us cross reference a laptop it works with and then a third part car charger you can power from the PDP.

[rich2202]

IMHO, you should really be thinking laptop with a built in battery.

[marshall]

Originally Posted by rich2202

IMHO, you should really be thinking laptop with a built in battery.

Actually, I'd say this device should be able to run off batteries since most PCs aren't designed to have power yanked from them or drop down randomly so running them from the ROBOT's battery isn't going to work. I'd say that independent battery power is "essential to completeness" to pretty much all COTS computing devices.

However, it is up to all of the LRIs with how they interpret the ambiguously worded rule about COTS device batteries. I know how I'm interpreting it until I'm told otherwise at an event by an LRI.

And based on Al's comments in a prior thread, as long as the COTS computing device's batteries don't interface with the control or drive systems for the robot, you should be following the intent of the rule in his eyes.

EDIT: Or don't test the rules and just use one of these after adjusting the output voltage: http://a.co/gd9yFkM

[rich2202]

Originally Posted by marshall

I'd say that independent battery power is "essential to completeness" to pretty much all COTS computing devices.

"Essential to completeness" is not the standard. Everything on the robot needs something else in order to work.

R37 says: "batteries integral to and part of a COTS computing device". There have been many discussions about batteries and the Raspberry Pi. Until someone sells a Pi COTS with battery, the Pi cannot have its own battery. Team 4480's computer is no different. Hence a laptop (notebook, chromebook, pad, etc.) with a built in battery is the preferred solution.

Just because you have an Operating System on your COTS computing device that does not like a random shutdown is not an excuse. If you had a more fault tolerant OS (and a fast booting one), you would not "need" the second battery.

[marshall]

Originally Posted by rich2202

"Essential to completeness" is not the standard. Everything on the robot needs something else in order to work.

R37 says: "batteries integral to and part of a COTS computing device". There have been many discussions about batteries and the Raspberry Pi. Until someone sells a Pi COTS with battery, the Pi cannot have its own battery. Team 4480's computer is no different. Hence a laptop (notebook, chromebook, pad, etc.) with a built in battery is the preferred solution.

Just because you have an Operating System on your COTS computing device that does not like a random shutdown is not an excuse. If you had a more fault tolerant OS (and a fast booting one), you would not "need" the second battery.

I call shenanigans. Your logic is broken. Laptops do not require a battery to function and can run off of a voltage regulator just like mini PCs and raspberry pis. If we're forced to deal with corrupt operating systems and files then why does a laptop not have to deal with it?

Q&A was asked a direct question about allowing replacement batteries for COTS devices and they chose to answer by explaining that integral means "essential to completeness". You are picking and choosing what that means (Just like I am... see the problem YET?!?!)... and if you're an LRI then you are entitled but no one else is.

[Bkeeneykid]

Originally Posted by marshall

I call shenanigans. Your logic is broken. Laptops do not require a battery to function and can run off of a voltage regulator just like mini PCs and raspberry pis. If we're forced to deal with corrupt operating systems and files then why does a laptop not have to deal with it?

Q&A was asked a direct question about allowing replacement batteries for COTS devices and they chose to answer by explaining that integral means "essential to completeness". You are picking and choosing what that means (Just like I am... see the problem YET?!?!)... and if you're an LRI then you are entitled but no one else is.

Well it seems the Q&A explicitly disagrees: https://frc-qa.firstinspires.org/qa/284

[marshall]

Originally Posted by Bkeeneykid

Well it seems the Q&A explicitly disagrees: https://frc-qa.firstinspires.org/qa/284

So I truly do not understand what FIRST is after here. If they don't want external batteries then specify that.

If they are going to allow batteries for laptops then I truly do not understand why they think laptop batteries are integral. They are not required for a laptop to function. They are not essential to completeness yet that is how they defined it so I'm left to think that a battery must be required for a laptop because if it loses power then it will potentially corrupt the OS or files. Any 1st year EE undergrad can tell you that a power source is a power source yet FIRST doesn't seem to think that. Maybe a cell phone has a battery that is literally required for completeness but that's not strictly true either. All of these devices can be run from an alternative power source like the robot battery.

Now they've said that USB power packs shouldn't be considered integral. Well, ok... what about this: https://www.sparkfun.com/products/13896

IT USES A USB BATTERY PACK!!!

Not to mention the custom circuit rules which allows for other things like this:
Ebay link for a supercap based system that should be able to charge from the robot battery

USB batteries aren't legal but yet this thing seems perfectly legal provided it charges from the robot battery and doesn't power anything on the robot's control system or motors. IT MAKES NO SENSE! It's not a battery after all, it's just a bunch of capacitors in a custom circuit. Bonus, it has a built-in USB port.

I'm committed to kicking and screaming about this for as long as it is ambiguous and believe me, we'll push boundaries on trying stuff.

I really implore FIRST to stop and think about what this rule is trying to accomplish and then explain the intent. At this point they are just being silly.

[techhelpbb]

Originally Posted by marshall

So I truly do not understand what FIRST is after here. If they don't want external batteries then specify that.

If they are going to allow batteries for laptops then I truly do not understand why they think laptop batteries are integral. They are not required for a laptop to function. They are not essential to completeness yet that is how they defined it so I'm left to think that a battery must be required for a laptop because if it loses power then it will potentially corrupt the OS or files. Any 1st year EE undergrad can tell you that a power source is a power source yet FIRST doesn't seem to think that. Maybe a cell phone has a battery that is literally required for completeness but that's not strictly true either. All of these devices can be run from an alternative power source like the robot battery.

Now they've said that USB power packs shouldn't be considered integral. Well, ok... what about this: https://www.sparkfun.com/products/13896

IT USES A USB BATTERY PACK!!!

Not to mention the custom circuit rules which allows for other things like this:
Ebay link for a supercap based system that should be able to charge from the robot battery

USB batteries aren't legal but yet this thing seems perfectly legal provided it charges from the robot battery and doesn't power anything on the robot's control system or motors. IT MAKES NO SENSE! It's not a battery after all, it's just a bunch of capacitors in a custom circuit. Bonus, it has a built-in USB port.

I'm committed to kicking and screaming about this for as long as it is ambiguous and believe me, we'll push boundaries on trying stuff.

I really implore FIRST to stop and think about what this rule is trying to accomplish and then explain the intent. At this point they are just being silly.

I would like to third that an external battery properly engineered - should be just as safe as a in integral battery. Considering that certain products known to smoke have integral batteries that were not properly engineered either.

I believe I understand the spirit of the goal with this rule - but I disagree that integral batteries insure proper engineering and external batteries are more risky.

If it's down the bare legal issue of having someone to sue - consider that most companies DO NOT recommend mounting their systems on a mobile robot so in a Court of Law they'll pull that out as a defense.

Unrelated to the quote above:

I disagree that a small PC that is not a laptop will fair poorly on a robot. I know of several HP Pavilion models (550 for example) that are miniITX motherboards in unusually large cases and take 19VDC power in. With the addition of a cheap SSD these would be no more likely to fail than a laptop and that style board is often mounted in large full size motor vehicles. Granted a full size motor vehicle has a suspension. Years ago someone told me on Chief Delphi that a laptop on a robot would fail if it wasn't as robust as a Panasonic ToughBook so I put a Dell Mini 9 (SSD) and a Gateway netbook (with a HDD replaced with an SSD) on a robot in a box merely padded with foam and drove it over the 6" bumps in the field that year hundreds of times without a single failure of any sort. Could it fail - sure. Did it fail - no. I have watched people drop average HP and Dell laptops down stairs in a padded case in hibernation and they come out in one piece and still work.

As far as how to produce reliable power for such a setup - I strongly recommend that someone boost their input voltage from the battery as high as they are legally allowed then buck it down to whatever voltages they require (use switching power supplies). If 24VDC is the maximum (likely because of the history of the cRIO) then go as high as that. If you don't do this you'll always have risk that the battery will absorb a heavy load and drop output voltage just enough to force a reboot.

If would be better if your PC could run off 16V or less because at 19VDC you only have room for the input voltage to dive about 5-6VDC before the only safety you have to insure a correct output voltage will be the temporary charge of the internal switching power supply capacitors.

Finally I am still game to produce a 3D printed or laser cut case with a Raspberry Pi and battery in it but the requirements for this are more about FIRST rules than some huge business opportunity. If someone wants to put the work in I am will to provide assistance if asked. I figure the goal should be to provide such platforms for a bunch of common FIRST coprocessors because frankly - many times the coprocessors in FIRST are each powerful enough to run a mobile robot outside of FIRST and there could be some opportunity there if that business later provides expanded I/O solutions for those platforms as well. If someone designs the case I have access to many 3D printers and the large format laser cutters at NextFAB in Philadelphia. I can also provide free web services and domain registration to reduce the operation costs. I can finally provide testing and prototyping for the electonic elements. As I have effectively made this offer for this 5 times now on Chief Delphi alone - please be aware that you'd be working on something for next year because this year is cutting it very darn close and if FIRST (understandably) asks for something to examine themselves you are leaving them far too little time to complete that evaluation.

[Jon Stratis]

Marshall, I think you're reading too much into the Q&A ruling. It specifically says that "laptops are designed to have a battery" - even if the battery is swapable, that doesn't mean you're constrained to the one that originally shipped with the laptop, so long as you use a battery intended for use with that laptop by the manufacturer.

But I can't see a way to interpret that ruling to allowing batteries to be added to a computing device that is designed to be run off wall power. Before you try that route, ask a direct question on the Q&A, otherwise you're likely to be disappointed at competition.

[marshall]

Originally Posted by Jon Stratis

Before you try that route, ask a direct question on the Q&A, otherwise you're likely to be disappointed at competition.

I did ask a direct question and got a "go read the dictionary" response. They were being cute with that response and they know it.

I'm already disappointed... nowhere to go but up from here.