Will FIRST ever allow 48V PoE+?

It would be very convenient if FIRST finally lifted (or increased) the limitation on max voltage on the robot, which is currently 24V. Teams could finally use industry standard PoE switches to power the radio and other coprocessors. And these PoE switches aren’t even expensive anymore.
Only $65 for a GbE model Amazon.com
$40 if you’re willing to step down to a 10/100Mb model https://a.co/d/cOL9gcQ

I’m not even sure I understand why the voltage limitation exists in the first place.

Do you not trust the DC redundant power inputs, on those switches?

It is not about powering the switch itself, but the stuff downstream from it, like the radio and coprocessors. I do trust the redundant power inputs on these switches

Industry convention has long considered 30V the threshold of safety from electrical shock. I would not ever expect to see 48V allowed.

(Yes, OSHA defines it as 50V. But we’re not talking about trained electrical professionals here.)

5 Likes

There wasn’t much of a point in having anything above 24V in the past, and if what I’ve heard is correct it exists specifically to prohibit teams from having mains voltage on their robots using inverters.

I think allowing proper active PoE would be nice. It might need to come with a control system upgrade so there’s a FIRST-approved switch to go with it and the roboRIO (or other main robot controller) can at least tolerate 48v on the Ethernet port if it doesn’t support PoE.

No change to the RoboRIO should be required. Active PoE isn’t destructive to non-PoE devices. The PoE negotiation process doesn’t damage devices that don’t support the 802.3/at/af/bt standard.
If anything, the current use of passive PoE is quite dangerous, as its very easy to accidentally plug in an Ethernet cable constantly carrying power + data into a device that isn’t expecting it, and kill a NIC.

5 Likes

This ought to involve some cost-benefit analysis, including some feedback about what risk postures are desirable. (Obviously, FIRST’s risk posture will dominate the discussion, but others may have relevant perspectives.)

Newer cordless power tools can frequently output this voltage. Devices connected to household mains voltages up to 240 V are commonly operated by unskilled people. Are these adequately safe for the same kinds of users, in the same kinds of circumstances? Is it because of features inherent to those applications, which are absent in PoE? We should be specific about what deficiencies exist.

Are we protecting against all shocks, or just the ones that cause trauma (rather than mere transitory pain or discomfort)? We presumably are fine with the existing level of risk of the kind of shocks that a 12 V battery can supply (e.g. if an electrode penetrates the skin). With clean, dry hands, most people won’t feel a 57 V shock1 any more than they would feel a 12 V one. With penetrating electrodes, even small voltages can interfere with motor nerves controlling muscle, or can apply hazardous currents internally to tissues with something on the order of 300 Ω. (That link is a very accessible discussion of the theory and the risks.)

Are there mitigations that we could use, like specifying the kind of connector to one that is adequately touch-safe (maybe 8P8C Ethernet cable ends aren’t technically touch-safe, but they are decent…but homebuilt cable ends to interface with arbitrary devices might not be), or requiring devices to autonegotiate with the power supply unit so that they don’t energize at full voltage until a compatible device is connected (that seems to be in the PoE design).

And of course, to be fair, is this a solution in search of a problem? Do the current rules adequately provide for the majority of needs without unreasonably inconveniencing anyone?


1 I mentioned 57 V because apparently that is the upper bound for a PoE device. Our hypothetical rule change would presumably permit devices designed for 57 V or less.

1 Like

The reason PoE devices are going up to 48V or 57V is because they need power, and Ethernet cables have thin, 22 or 24awg conductors.

What would you accomplish with 48V PoE?

Do we need more power to any devices? Is 12w not enough? Or is it more of a plug-n-play compatibility thing?

I’m all for doing 48v PoE, especially since the conductors and algorithms are finger-safe. But adding complexity should come with advantages.

I personally would like to see a faster booting radio. Or maybe one with a supercapacitor to help with brown outs or loose wires

2 Likes

Back in 2012 I think, the rules were a little less defined and we did run 48V POE with off-the-shelf hardware to power our IR network camera. We had a looooong chat with the head inspector about it, but in the end were able to use it. The explicit language restricting maximum on-bot voltage would come later. (EDIT: incorrect information) I suspect the only reason we’re still allowed 24V today is because its what the old cRIO’s needed to run and this wasn’t taken away.

Some solenoids use 24V and the different pneumatics modules have jumpers or other ways to switch between 12V and 24V depending on what a team is using. I think this is generally the only area on a bot that might use 24V.

2 Likes

Ah, you’re right; I totally forgot about solenoids. 24 VDC is a pretty common industrial controls operating voltage, so it does make sense we allow it.

My guess (though I similarly haven’t done any kind of risk analysis) is that is isn’t so much about features inherent to those applications that make them safe, but the risks of having these things on robots that are exposed to much more abuse (like contact with other robots, getting knocked over, and so forth) than the average item “commonly operated by unskilled people”, leading to there being more likelihood of shorts and other cases where components aren’t operating “within spec”. I could see there being higher risk with a 48V short than a 24V short (to other components at least, and maybe even to nearby people).

1 Like

I could maybe see a world where 48V PoE is made legal but not 48V in general. It could go a step further and limit that only certain switches be used (e.g. that have 12V supply voltage, step it up internally, and have been tested for failure modes), e.g. something like this: https://www.amazon.com/LINOVISION-Industrial-Gigabit-DC12V-48V-IEEE802-3af/dp/B09HGWLZSD/

3 Likes

Tangential to this subject, has anyone experimented with Ubiquiti 24V passive PoE equipment on a robot? I’m planning to pick up one of their NanoSwitch switches and play with it over this summer – it has PoE pass-through, so an injector supplied with 24V could power it and a radio. It wouldn’t be a legal way to power the radio at this time, but maybe that could change. It’s unfortunate that a Limelight can’t take 24V PoE…

It would be useful for the powering of coprocessors, like a Raspberry Pi via PoE. And even the radio uses more than 12W. For example, the REV Radio Power Module outputs 18V @ 1A max, allowing for a total of 18W of power for the radio alone. And it would simplify wiring a good amount. Simply provide power to your switch, then plug in all your Ethernet devices. No need to run separate power wires, or have a separate PoE injector.
Also as stated earlier, switching to active PoE would also prevent costly mistakes, such as plugging in an Ethernet cable carrying power into a device that is not passive PoE compatible, like a laptop, or the RoboRIO.

Why wouldn’t it be legal? I powered the radio via a passive quad port PoE injector that was powered by a 12V-24V buck boost converter. The middle port on the radio can accept 24V passive PoE, and 24V is a legal voltage to use on a FRC robot.

Because of R616:
*Power radio as specified – Part 1. The wireless bridge (radio) power must be supplied by either:
A. the 12V 2A output of a CTR Electronics Voltage Regulator Module (VRM) (P/N am-2857,
217-4245), as shown in Figure 9-13, and must be the only load connected to those
terminals or
B. using an Ethernet cable between a REV Radio Power Module (RPM) (P/N REV-11-1856)
and the “18-24v POE” Ethernet port on the wireless bridge
The clause about “only load connected” plus the blue box only specifically mentioning PASSIVE CONDUCTORS makes it pretty clear that a 12-24V buck-boost converter should not be allowed under the current rules.

The RPM itself is a buck boost, 12V-18V

It is, but it’s specifically allowed by (B). The rule says the radio must be powered by either one of those two devices, and nothing else is allowed.

1 Like

Like @Peter_Johnson said, that doesn’t make a random 12-24V buck-boost legal. The rules are very specific on what is and isn’t allowed here.