Using PoE to power devices

[cglaeser 0]

CAI Networks sell an inexpensive Ethernet I/O device called WebController. See http://www.redhillnetworks.com/products/webcontrol/webcontrol-faq.html

 

The device requires 6 to 9 VDC. The device would be easier to deploy if it used PoE for power. I called the company yesterday and suggested they add PoE in a future rev, and he appreciated the feedback, but indicated that using PoE was not really a good option for them. I don't think he had considered PoE before, so perhaps he was shooting from the hip. What are the design considerations when converting 44 VDC to, say, 5 VDC in a PoE application? In particular, what is the part cost, and what is the efficiency? Are there evaluation breadboards available? Is it possible to easily add PoE to power an existing non-PoE device?

 

Best,

Christopher

[thewireguys 3]

http://www.trendnet.com/langen/products/proddetail.asp?prod=130_TPE-112GS&cat=59 Edited December 15, 2010 by Guest

[cglaeser 0]

http://www.trendnet.com/langen/products/proddetail.asp?prod=130_TPE-112GS&cat=59

 

Thanks!!! - Christopher

[Soundy 1]

I have a customer who uses these with Linksys WAPs on his sites: http://homesupport.cisco.com/en-us/wireless/lbc/WAPPOE

 

The injector component includes a 48VDC power adapter and is fully 802.3af compliant. The receiver "extracts" that and regulates it down to 12VDC for the AP. I haven't taken the receiver apart (I just use the injector for a PoE camera now, and had to open it up to repair a flaky power connector), but I imagine it's a pretty simple design, extracting the 48V from pairs 4/5 and 7/8 and feeding it through a voltage divider and basic 2-3A regulator IC (I'm guessing there's a voltage divider in there since most off-the-shelf regulator ICs accept max 35V input).

[thewireguys 3]

This one you can select the output voltage.

 

 

http://www.trendnet.com/langen/products/proddetail.asp?prod=130_TPE-112GS&cat=59

[ak357 0]

I have a customer who uses these with Linksys WAPs on his sites: http://homesupport.cisco.com/en-us/wireless/lbc/WAPPOE

 

The injector component includes a 48VDC power adapter and is fully 802.3af compliant. The receiver "extracts" that and regulates it down to 12VDC for the AP. I haven't taken the receiver apart (I just use the injector for a PoE camera now, and had to open it up to repair a flaky power connector), but I imagine it's a pretty simple design, extracting the 48V from pairs 4/5 and 7/8 and feeding it through a voltage divider and basic 2-3A regulator IC (I'm guessing there's a voltage divider in there since most off-the-shelf regulator ICs accept max 35V input).

 

little more involve then just "power regulator and voltage divider"

think if u want to drop in your example 10-13 v ( 48-35)

with current around 1 A then then divider has to withstand about 10 Watt

and so on

[Soundy 1]

I have a customer who uses these with Linksys WAPs on his sites: http://homesupport.cisco.com/en-us/wireless/lbc/WAPPOE

 

The injector component includes a 48VDC power adapter and is fully 802.3af compliant. The receiver "extracts" that and regulates it down to 12VDC for the AP. I haven't taken the receiver apart (I just use the injector for a PoE camera now, and had to open it up to repair a flaky power connector), but I imagine it's a pretty simple design, extracting the 48V from pairs 4/5 and 7/8 and feeding it through a voltage divider and basic 2-3A regulator IC (I'm guessing there's a voltage divider in there since most off-the-shelf regulator ICs accept max 35V input).

 

little more involve then just "power regulator and voltage divider"

think if u want to drop in your example 10-13 v ( 48-35)

with current around 1 A then then divider has to withstand about 10 Watt

and so on

Granted, but nobody says it has to ONLY drop to 35V; that's just the maximum for most off-the-shelf regulator ICs. You could also use a divider to drop it BY 30V, down to 18, and then regulate it from there... or use a regulator with a higher input tolerance.

 

Anyway, like I said, that's just a guess, because I haven't actually opened the receiver up to see what's in it. If you REALLY want to know the specifics, I can crack it open and take some pictures...

 

BTW, 802.3af-2003 specs up to 15.4W per channel (minimum 44V @ 350mA), so a basic PoE supply wouldn't be giving you anywhere near 1A anyway.

[cglaeser 0]

BTW, 802.3af-2003 specs up to 15.4W per channel (minimum 44V @ 350mA), so a basic PoE supply wouldn't be giving you anywhere near 1A anyway.

 

Right, it's roughly a third of an amp at 44v. Some splitters deliver 10W at various voltages (e.g. 2A @ 5V).

 

According to the tech at CAI, he was concerned about the reliability of low-cost dc-dc power devices. FWIW.

 

Best,

Christopher

[Soundy 1]

BTW, 802.3af-2003 specs up to 15.4W per channel (minimum 44V @ 350mA), so a basic PoE supply wouldn't be giving you anywhere near 1A anyway.

 

Right, it's roughly a third of an amp at 44v. Some splitters deliver 10W at various voltages (e.g. 2A @ 5V).

 

According to the tech at CAI, he was concerned about the reliability of low-cost dc-dc power devices. FWIW.

 

Best,

Christopher

I'd be concerned about the reliability of low-cost DC-DC *step-up* devices... not so much about step-down, since it's so easy to do: the venerable 7812 1A 12VDC regulator IC can be had at a local retailer for $1.25 each, or in a 10-pack at 78c each ($7.80 for the 10-pack). I don't think anyone would consider the decades-old design "unreliable"

[cglaeser 0]

the venerable 7812 1A 12VDC regulator IC can be had at a local retailer for $1.25 each, or in a 10-pack at 78c each ($7.80 for the 10-pack). I don't think anyone would consider the decades-old design "unreliable"

 

This is my understanding of the conversation. I don't think he was questioning the reliability of all regulators, and I believe they use them in their designs. I think he was comparing the reliability of regulators that reduce the voltage by a modest margin versus those that reduce the voltage by a factor of eight and increase the current by a factor of six (my words, not his, but that was my understanding). Is there a difference in reliability of regulating a 7.5V@2A to 5V@2A versus a PoE splitter converting 44V@0.3A to 5V@2A?

 

Best,

Christopher

[Soundy 1]

the venerable 7812 1A 12VDC regulator IC can be had at a local retailer for $1.25 each, or in a 10-pack at 78c each ($7.80 for the 10-pack). I don't think anyone would consider the decades-old design "unreliable"

 

This is my understanding of the conversation. I don't think he was questioning the reliability of all regulators, and I believe they use them in their designs. I think he was comparing the reliability of regulators that reduce the voltage by a modest margin versus those that reduce the voltage by a factor of eight and increase the current by a factor of six (my words, not his, but that was my understanding). Is there a difference in reliability of regulating a 7.5V@2A to 5V@2A versus a PoE splitter converting 44V@0.3A to 5V@2A?

 

Put it this way: there SHOULDN'T be. As with anything else, it depends on the actual circuit design and construction, but there's no THEORETICAL reason one should be any more reliable than the other. A step-down circuit starts with something as simple as a voltage divider (two resistors); can be made more accurate by the simple addition of a zener diode, can be a complex IC like a 78xx, or even a fancy PWM switching supply.

 

Now for example, the 78xx-series regulators are limited to 1A, and at the upper end of that, require a heat sink; running them full-out without a heat sink, or applying an extended overload without fuse protection, will lead to the chip burning out... this is a case of reliability being wholly a function of circuit design and component choice (if you're going to require >1A, use a different unit with the proper rating).

[mike_va 0]

Don't use a 7812 at 1A, input voltage is too high at 48V (35V max) and power dissipation will be too great (48-12)*1A=36W not gonna happen. Even a fairly large heatsink at 1C/W would put you at a 36C rise, too hot to touch. You could make a more complicated linear circuit but this is not the way to go.

 

Most designs do this with switchers, that way you can also increase the output current to greater than the input at lower voltages. Look at it as power in and out.

[cglaeser 0]

Don't use a 7812 at 1A, input voltage is too high ...

 

Thanks, this is helpful, and echos the comments of the CAI designer. He said that using a) the more common 7812 type of approach would produce too much heat and insufficient power, and b) he also did not want to use a switching approach due to reliability. Comment a) makes perfect sense. I don't know if the approach in b) is less reliable, but he is confident in the reliability of his existing design of regulating an input with lower voltage and higher current.

 

Best,

Christopher

[Soundy 1]

Don't use a 7812 at 1A, input voltage is too high ...

 

Thanks, this is helpful, and echos the comments of the CAI designer. He said that using a) the more common 7812 type of approach would produce too much heat and insufficient power, and b) he also did not want to use a switching approach due to reliability. Comment a) makes perfect sense. I don't know if the approach in b) is less reliable, but he is confident in the reliability of his existing design of regulating an input with lower voltage and higher current.

 

Best,

Christopher

Just for the record, I didn't recommend using a 78xx for this purpose, I only used those as an example of common, low-cost, reliable IC regulators. There are also 3A and higher-output ICs available - like this: http://www.national.com/mpf/LM/LM350.html#Overview

[mike_va 0]

...did not want to use a switching approach due to reliability.

 

It is a little more difficult to design a good switcher, but they can actually have a longer lifetime if done correctly due to the lower losses. Heat directly impacts lifetime, to get 15 years for example in a space application we usually derate to 105C max junction temp etc.

 

Perhaps when he said not a good option he meant there was no in house expertise or money to allocate to development.

 

Most people do not understand magnetics and control loops well enough to cobble a switcher together, and the comment about low cost might refer to using certain parts that will not have enough derating to have good reliability, low EMI etc. There are definitely more subtleties to designing a switcher than using a standard 3 terminal linear reg (which is pretty resilient and tolerant of a wide range of design practices).

 

The other inherent advantage is that you are not limited to the input current. For example to make the math easy 44V 0.1A (current limit) available, linear reg you can get 12V 0.1A and must dissipate 3.6W. Switcher you can get 12V 0.367A assuming no losses, or if we say 95% then losses are 0.22W.

 

Having said that all these cameras that run off 44V POE must have a switcher inside, to step it down to voltages that logic level parts like to see.

 

Even my Bosch Dinion (non-POE) is spec'd to run off something like 10.8-39VDC (in addition to 28 VAC) or something like that. I have it hooked up to a HP variable lab supply. As the voltage is increased the current goes down as expected with a switcher (i.e. roughly constant power). The other advantage switchers can give (not all though) is isolation, which has bearing on both grounding and various fault and environmental situations.

 

Lastly someone could try to argue that a switcher (especially based around POE) is slightly less reliable just based on the increased part count compared to a linear reg.

 

Nearly everything has a switcher these days however and they can be designed to be very reliable (although perhaps less so if someone tries to save the last few cents). The major players can sometimes save the last few cents but to do this successfully there is usually a pretty rigorous qualification program behind it all.