I was thinking about this idea some more. Why not have both 12V and 48V? We could differentiate between the two during the detection phase with different input impedances in the PD. 802.af uses 25 Kohm, so we could use a different value like 13 Kohm to indicate a low-voltage PD. Once detection is complete, classification would determine the amount of power needed at either 12V or 48V. Obviously the PSE is more complicated now.
Small PDs would be backward compatible with a simple 12V supply, so during the adoption phase of the standard device vendors could still supply cheap AC/DC wall-warts for low-power devices.
For the connector, we should look at designs that are intended for 1000's of cycles and hot disconnection. I don't think either lighter sockets or Molex connectors are suitable. Something similar to USB or Firewire - positive keying, some retention force, wiped contacts. We could follow the USB concept of A, B, and mini-B connectors; or the Firewire concept of large and small connectors. Having A and B connectors would (theroetically) prevent connecting a PSE to a PSE, but the detections phase would keep anything bad happening anyways, so this isn't really needed. So there would just be two connector designs, the large model would be capable of carrying the full 4A, and would be used on all PSE ports and for PDs that need the current. The small connector would be limited to 1A, and would be used for low power devices.
The market for this could dwarf POE. Think of this being applied to every piece of consumer electronics made after ~2009.
I think we're (mostly) in violent agreement. The intent here is to standardize the wall-wart. For the sort of local power distribution that I'm advocating, voltages from 12-50V are reasonable. 12V would be nice because it would require less expensive circuitry for most applications, but it can't deliver much power over long distances. 48V is nice because that's the design point of 802.3af, it still qualifies as low voltage, and it still can power most of my desk-area devices with low current (thin cables and small connectors).
Both of the problems that 802.3af power handshaking solve still apply here. First, power would not be delivered to incompatible devices (wet-ware like children sticking paper clips where they shouldn't). Second, power negotiation would allow the PSE to be a little more user-friendly than blowing a fuse or circuit breaker when you plug in one device too many.
I surveyed the power bricks near my desk. I have 10 providing 20, 18.2, 18, 12, 12 & 5, 5.2, and 5 volts. Max power is 90W (4.5A @ 20V). Total rated output of all wall-warts is approximately 320W. If each wastes 50 mW when the load device is off, then I'm wasting 12 Wh every day. If I had a single PSE that had a minimum power consumption of 200 mW, then I'm only wasting 4.8 Wh every day. Not a big difference, but it adds up over time and people.
I doubt that many (if any) of these devices provides decent power factor correction (PFC). Enough support a wide input range (100-240VAC) that their input stages are clearly a little more advanced than the simple rectifier input stages common on narrow input-range switching power supplies, but I doubt they are well-corrected. Two of my 10 wall-warts are heavy enough that they are almost certainly just a transformer and a regulator - the worst of what we're trying to get rid of. Creating a standard for the low-voltage power interface would allow device manufactures to stop including least-cost wall-warts. This would allow the establishment of reasonable standards for efficiency and PFC for the user-purchased PSE. As mentioned before, adding UPS functionality to the PSE would be trivial.
Less and less electronics actually operate at 12 or 5 volts anymore, nearly everything already has a DC/DC converter inside to provide all the crazy voltages needed (sub-1V processors, 1.8V memory, 20+V LED backlights, and more). Only the smallest (or cheapest) devices still use a linear regulator. Most small rechargeable devices support a wide-range input for recharging. How? A buck/boost DC/DC converter. Increasing the input range to 48V and adding the power negotiation would definitely add cost to these applications. The two high-power devices device in my digital menagerie (laptop and Mac Mini) definitely have DC/DC converters inside so they already do 2 conversion: the AC/DC wall wart (well, floor bricks) and internal multi-output (or multiple) DC/DC supplies.
I see a lot of value here:
* The nastiest of the safety issues are isolated in the PSE as long as the voltage is around 48V (or less). I'm sure small device manufacturers would prefer 5 or 12V because it wouldn't require as much change to their designs as 48V. Small computers and LCD panels would work better with the higher power that 48V could deliver. * The total energy wasted would go down because fewer AC/DC converters would be plugged into the wall. * Power factor would be improved, especially in office environments where every cubicle would have one well designed AC/DC converter instead of 5 cheap ones. * The only difference in the PSE around the world would be the AC cordset - every AC/DC converter I've seen that has PFC also has a wide input-range. * Power negotiation (a-la 802.3af) enhances user safety and may (with proper regulatory approval) allow the PD connector and cord-set to be sized for the power required, not the theoretical maximum power available from the PSE. * If 802.2af compatible voltage and negotiation is used, the devices that n
This is a solution to the OP's problem. I'm looking at all the devices around my desk that have power bricks. Every one uses less than 150W. All but two (laptop and Mac Mini) use less than 15W. In addition I have a 20" flat panel with a integrated AC power supply that is rated at 160W input power, so it could probably use a 150W DC supply also.
I would be thrilled if I had a single power strip with 8-12 SELV outlets capable of delivering a total of ~1000W for everything except my desktop computer and laser printer. It would probably have power factor correction (which I doubt many of the existing bricks provide) and efficiency of 85-92% Enhancing it to act as a UPS would be trivial.
A traveller could take a small power brick with ~4 outlets with ~200-300W output to power a laptop and all the usual digital personal accessories.
I doubt that permanent in-wall installation with a whole-house power controller would every gain much traction, but the saftey advantages of these outlets could eventually encourage their use.
I just see the 802.3af standard as a good starting point for providing an intelligent, high-efficiency power distribution system with enhanced user safety.
I didn't suggest using the physical layer of the 802.3af standard. I would expect a different (hopefully smaller) connector to be used for these DC power outlets. If the negotiation protocol is compatible, then low power devices like a cell phone or MP3 player could recharge from a 802.3af compliant Ethernet jack with an appropriate cable. Higher power devices (small computers, laptops, etc.) would simply not power up.
I would like to propose that we extend the work that has been done for Power over Ethernet (IEEE 802.3af) to higher power levels for consumer electric products.
We can reuse and/or extend the probe/negotiation phase to provide additional power levels, let's say up to 150W (approximately 4A max).
Advantages
1. Enhanced safety because unused outlets provide a high-impedance (power-limited) source
2. Unified power connector for low-voltage/low-power appliances
3. DC outlets could be provided either through centralized power controllers, plug-in power strips, or wall-box mounted controllers
Disadvantages
1. Power controllers will always waste power
2. Centralized power controllers will require point-to-point wiring
3. Significant additional component cost compared to linear regulators
4. Modest additional component cost compared to DC/DC regulators
Some device manufacturers won't want to support this because it would increase the size of the in-device charge controller. For example, I have an iPod, a Treo, and a Bluetooth headset. All support a wide range of input voltages for charging (this implies an internal DC/DC regulator), but none support an input voltage higher than approximately 15V. Designing a DC/DC regulator that supports 48V requires more robust (therefore larger) components. In addition there will be an in-device power controller to communicate with the central power controller. Fortunately this last part is already available for 802.3af applications.
Slashdot Mirror
Small PDs would be backward compatible with a simple 12V supply, so during the adoption phase of the standard device vendors could still supply cheap AC/DC wall-warts for low-power devices.
For the connector, we should look at designs that are intended for 1000's of cycles and hot disconnection. I don't think either lighter sockets or Molex connectors are suitable. Something similar to USB or Firewire - positive keying, some retention force, wiped contacts. We could follow the USB concept of A, B, and mini-B connectors; or the Firewire concept of large and small connectors. Having A and B connectors would (theroetically) prevent connecting a PSE to a PSE, but the detections phase would keep anything bad happening anyways, so this isn't really needed. So there would just be two connector designs, the large model would be capable of carrying the full 4A, and would be used on all PSE ports and for PDs that need the current. The small connector would be limited to 1A, and would be used for low power devices.
The market for this could dwarf POE. Think of this being applied to every piece of consumer electronics made after ~2009.
I think we're (mostly) in violent agreement. The intent here is to standardize the wall-wart. For the sort of local power distribution that I'm advocating, voltages from 12-50V are reasonable. 12V would be nice because it would require less expensive circuitry for most applications, but it can't deliver much power over long distances. 48V is nice because that's the design point of 802.3af, it still qualifies as low voltage, and it still can power most of my desk-area devices with low current (thin cables and small connectors).
Both of the problems that 802.3af power handshaking solve still apply here. First, power would not be delivered to incompatible devices (wet-ware like children sticking paper clips where they shouldn't). Second, power negotiation would allow the PSE to be a little more user-friendly than blowing a fuse or circuit breaker when you plug in one device too many.
I surveyed the power bricks near my desk. I have 10 providing 20, 18.2, 18, 12, 12 & 5, 5.2, and 5 volts. Max power is 90W (4.5A @ 20V). Total rated output of all wall-warts is approximately 320W. If each wastes 50 mW when the load device is off, then I'm wasting 12 Wh every day. If I had a single PSE that had a minimum power consumption of 200 mW, then I'm only wasting 4.8 Wh every day. Not a big difference, but it adds up over time and people.
I doubt that many (if any) of these devices provides decent power factor correction (PFC). Enough support a wide input range (100-240VAC) that their input stages are clearly a little more advanced than the simple rectifier input stages common on narrow input-range switching power supplies, but I doubt they are well-corrected. Two of my 10 wall-warts are heavy enough that they are almost certainly just a transformer and a regulator - the worst of what we're trying to get rid of. Creating a standard for the low-voltage power interface would allow device manufactures to stop including least-cost wall-warts. This would allow the establishment of reasonable standards for efficiency and PFC for the user-purchased PSE. As mentioned before, adding UPS functionality to the PSE would be trivial.
Less and less electronics actually operate at 12 or 5 volts anymore, nearly everything already has a DC/DC converter inside to provide all the crazy voltages needed (sub-1V processors, 1.8V memory, 20+V LED backlights, and more). Only the smallest (or cheapest) devices still use a linear regulator. Most small rechargeable devices support a wide-range input for recharging. How? A buck/boost DC/DC converter. Increasing the input range to 48V and adding the power negotiation would definitely add cost to these applications. The two high-power devices device in my digital menagerie (laptop and Mac Mini) definitely have DC/DC converters inside so they already do 2 conversion: the AC/DC wall wart (well, floor bricks) and internal multi-output (or multiple) DC/DC supplies.
I see a lot of value here:
* The nastiest of the safety issues are isolated in the PSE as long as the voltage is around 48V (or less). I'm sure small device manufacturers would prefer 5 or 12V because it wouldn't require as much change to their designs as 48V. Small computers and LCD panels would work better with the higher power that 48V could deliver.* The total energy wasted would go down because fewer AC/DC converters would be plugged into the wall.
* Power factor would be improved, especially in office environments where every cubicle would have one well designed AC/DC converter instead of 5 cheap ones.
* The only difference in the PSE around the world would be the AC cordset - every AC/DC converter I've seen that has PFC also has a wide input-range.
* Power negotiation (a-la 802.3af) enhances user safety and may (with proper regulatory approval) allow the PD connector and cord-set to be sized for the power required, not the theoretical maximum power available from the PSE.
* If 802.2af compatible voltage and negotiation is used, the devices that n
I would be thrilled if I had a single power strip with 8-12 SELV outlets capable of delivering a total of ~1000W for everything except my desktop computer and laser printer.
It would probably have power factor correction (which I doubt many of the existing bricks provide) and efficiency of 85-92%
Enhancing it to act as a UPS would be trivial.
A traveller could take a small power brick with ~4 outlets with ~200-300W output to power a laptop and all the usual digital personal accessories.
I doubt that permanent in-wall installation with a whole-house power controller would every gain much traction, but the saftey advantages of these outlets could eventually encourage their use.
I just see the 802.3af standard as a good starting point for providing an intelligent, high-efficiency power distribution system with enhanced user safety.
I didn't suggest using the physical layer of the 802.3af standard. I would expect a different (hopefully smaller) connector to be used for these DC power outlets. If the negotiation protocol is compatible, then low power devices like a cell phone or MP3 player could recharge from a 802.3af compliant Ethernet jack with an appropriate cable. Higher power devices (small computers, laptops, etc.) would simply not power up.
As some may know, this standard provides for approximately 15W of power at a nominal 48V. See http://en.wikipedia.org/wiki/Power_over_Ethernet for an introduction.
We can reuse and/or extend the probe/negotiation phase to provide additional power levels, let's say up to 150W (approximately 4A max).
Advantages
1. Enhanced safety because unused outlets provide a high-impedance (power-limited) source2. Unified power connector for low-voltage/low-power appliances
3. DC outlets could be provided either through centralized power controllers, plug-in power strips, or wall-box mounted controllers
Disadvantages
1. Power controllers will always waste power2. Centralized power controllers will require point-to-point wiring
3. Significant additional component cost compared to linear regulators
4. Modest additional component cost compared to DC/DC regulators
Some device manufacturers won't want to support this because it would increase the size of the in-device charge controller. For example, I have an iPod, a Treo, and a Bluetooth headset. All support a wide range of input voltages for charging (this implies an internal DC/DC regulator), but none support an input voltage higher than approximately 15V. Designing a DC/DC regulator that supports 48V requires more robust (therefore larger) components. In addition there will be an in-device power controller to communicate with the central power controller. Fortunately this last part is already available for 802.3af applications.
You heard it here first.