5 port gigabit switch powered over POE via one of the cables

[Nominal Animal]

Yes; all true. Can be done for less.

No; it came from Amazon and there is no "support" or "customer service". You send the box back.

Already did. Netgear GS108Tv3

Sorry; missed that. Good price too.

Are you sure you looked at GS108Tv3 and not GS108Ev3?  The GS108Tv3 I can find here are 80+€.

[Nominal Animal]

Oh, and apologies to peter-f, bingo600, and Monkeh: I was very wrong.  :-[
I've done quite a bit of looking at product datasheets, devices on sale all over, and completely changed my stance on PoE terminology.  A big thank you for Monkeh to not let my error and bad advice stand uncontested.

My current understanding and suggestion to anyone getting PoE stuff:

• Terms 'Active PoE' and 'Passive PoE' are useless, do not trust them.  They are undefined terms, and people use them to mean whatever they want.
  Even 'PoE' itself does not refer to any standard, just to the idea of using Ethernet cabling, somehow, for power delivery.
• Look at the exact product datasheet or manual, and check for explicit mention of compliance with
  • IEEE 802.3af (15.4 W)
  • IEEE 802.3at (25.5 W)
  • IEEE 802.3bt (71.3 W)
  to find out if the device.  A compliant supply (PSE) will not destroy any non-PoE device connected to a powered port.
• 10Base-T and 100Base-T use only two out of the four twisted pairs in an Ethernet cable for data transfers: pins 1 (Tx+), 2 (Tx-), 3, (Rx+), and 6 (Rx-).  The two other pairs (4, 5, 7, 8)  are unused.
• 1000Base-T and faster uses all four twisted pairs in an Ethernet cable dor data.
• IEEE 802.3af and 802.3at have three different modes:
  • Mode A: uses two twisted pairs for power, pins 1, 2, 3, 6
  • Mode B: uses two twisted pairs for power, pins 4, 5, 7, 8
  • Mode C: uses all four twisted pairs for power; often called '4PPoE'
  Supplies (PSEs) detect the support of devices (PD), and only provide 48V if the device has the signature of a proper 802.3af/at device.  The polarity (between pairs) may vary.
• The signature for a powered IEEE 802.3af or 802.3at device (PD) is a 24.9k kΩ 1% between the twisted pair (1,2) and (3,6) for mode A, between twisted pair (4,5) and (7,8) for mode B, and both for mode C.
• IEEE 802.3af and 802.3at 'negotiation' is basically just the detection of the presence of this resistance at low enough voltages and currents to not destroy common devices with just 150Ω of resistance between pairs due to 75Ω termination to ground.
  There are chips like LTC4267 (802.3af/at) and LTC4295 (802.3af/at/bt) that handle that detection; such datasheets describe how it is done.
• It is stuffing 48V without checking, to ordinary Ethernet devices that causes the connector or circuitry to blow by trying to dissipate several watts of heat within the tiny connector or a small termination resistors, that breaks things.
  Standards-compliant PSEs do not break ordinary Ethernet devices; and PDs used on ordinary Ethernet behave just like ordinary Ethernet devices.
• Ethernet cables themselves should have less than 100Ω DC resistance, but non-copper cabling (CCA, copper-clad aluminium, and similar) may have much higher resistance and fail for power delivery, even though they seem to carry data mostly fine.
  Expected DC resistances for proper <10m copper Ethernet cables, measured using an ordinary cheap multimeter, per wire, are in the 1Ω to 5Ω.
• I am not exactly sure of the IEEE 802.3bt 'negotiation', but it is backwards compatible with 802.3af and 802.3at.  It adds new power classes and 'stuff' for the higher power support.
• Manufacturers do not seem to agree whether a 802.3af or 802.3at supply equipment (PSE) or powered device (PD) should support more than one mode or not.  Some do, some do not.  So, just because both devices are compliant with 802.3af/at, does not mean they will for sure work with together, if it happens one is mode A only and the other is mode B only.
• There exist supplies ('injectors') that work with 10BaseT and 100BaseT Ethernet cabling, using the unused two pairs for power delivery using various voltages, often 24V or 48V.  These do not do any kind of detection or negotiation, and usually just treat the unused pairs as DC voltage lines.
  They are not compatible with IEEE 802.3af/at/bt supplies (PSEs) and usually not even devices (PDs), and will easily destroy standard Ethernet ports.
• There exist supplies ('injectors') that work with 10/100/1000 Ethernet, using magnetics but simply biasing the pairs at different voltages for power delivery.
  These do not do any kind of detection or negotiation either, and usually just treat the unused pairs as DC voltage lines.
  They are not compatible with IEEE 802.3af/at/bt supplies (PSEs) and usually not even devices (PDs), and will easily destroy standard Ethernet ports.
  MikroTik devices often support being powered via such, but since there is no standard, be sure to check the compatibility thoroughly, and label such cabling so that nobody confuses them for normal Ethernet cabling.

[peter-h]

Are you sure you looked at GS108Tv3 and not GS108Ev3?  The GS108Tv3 I can find here are 80+€.

Similar here: 53 quid plus VAT here: https://www.broadbandbuyer.com/products/39344-netgear-gs108t-300uks/

But not 150 quid like that Ubiquity one. Their boxes are generally good but I have one of their WIFI APs which does 2.4 and 5 GHz and the 5 GHz is basically useless. It has a fantastic amount of config which like on a lot of chinese boxes runs extremely slowly - probably contains a whole Linux dist on a CPU running at 16MHz and with 16k RAM

will easily destroy standard Ethernet ports.

In milliseconds. 15 watts in two 0603 or 0402 resistors. Hanrun and some of its counterfeiters make integrated RJ45s with capacitors in series with the terminators, which prevents the damage - it is an old RS485 trick because it gives you AC termination without the DC loading - but these cost quite a bit more. I am sticking with the standard variant made by a number of vendors, including the counterfeiters, and all these work fine, priced from 50p for the fake Hanrun to 3 quid for German (chinese) ones. We had another thread

https://www.eevblog.com/forum/microcontrollers/rj45-with-integrated-magnetics-different-configurations/msg4127149/#msg4127149



Curiously, I've just noticed, if a cheap injector which just does the 2nd two pairs, blows up the resistors, the product will still work if it is only 10/100, and most embedded boxes are just 10/100. It may emit a "pop" and some smell. I wonder if the cheap injectors drive all four pairs?

[Ranayna]

• Look at the exact product datasheet or manual, and check for explicit mention of compliance with
  • IEEE 802.3af (15.4 W)
  • IEEE 802.3at (25.5 W)
  • IEEE 802.3bt (71.3 W)
  to find out if the device.  A compliant supply (PSE) will not destroy any non-PoE device connected to a powered port.

Great post
The quoted is the key here: Do not buy crap that does not mention the official PoE Standards in it's datasheet or manual. Mikrotik for example, for as much as i like their stuff, is not using standard compliant PoE.
The TL-RP108GE does not mention either 802.3af or at, so you have to assume it's not using these. So to put it mildly, it is trash.

The Netgear GS108Tv3 seems to be a decent switch. But be careful, it is marketed as "Smart Managed", and on some Netgear devices that means you need a Netgear account to manage the device.
D-Link DGS-1100-05PD might also be an alternative.

[MarginallyStable]

The Netgear GS108Tv3 seems to be a decent switch. But be careful, it is marketed as "Smart Managed", and on some Netgear devices that means you need a Netgear account to manage the device.

I'll admit. When I first bought a few of these switches. Netgear attempted to limit features with a firmware update unless you registered the product. Myself and several others complained to netgear about this. I stated I would never recommend a netgear product again if they stick to this. How is it right to remove features that were working on the original firmware to require a registration. Well, I reverse engineered the product activation key and self registered my product. Since then netgear removed the registration requirement. These switches can me managed completely through the local web interface, or can we switched to "cloud mode" to use a netgear account to manage. In fact my switches have never been allowed internet access to call home. These are great switches, I'm glad netgear fixed their craziness so I can keep using them (and now recommend them)

[peter-h]

It is working perfectly now. Thank you for all your help

The Tenda PoE30G-AT is driving the TP-Link TL-PoE10R splitter.

The forum has changed to HTML; how do I set it so it drops formatting?

[peter-h]

This is rather interesting. From above

• Mode A: uses two twisted pairs for power, pins 1, 2, 3, 6

That should mean that my Linksys POE switch should deliver power over the first two pairs (of the cable which has only those connected) and there is no need for the Tenda injector.


Linksys data sheet: IEEE 802.3, 802.3u, 802.3x, 802.3ab, 802.3az, 802.3af, and 802.3at (PoE+)

So I tried it, and it works! The TP-Link TL-PoE10R splitter is now powered from the other end of the system!

[Nominal Animal]

So I tried it, and it works! The TP-Link TL-PoE10R splitter is now powered from the other end of the system!

Nice!  I just wish PoE router datasheets/manuals were explicit on the modes they support.

But.. does this mean you bought both the Tenda PoE30G-AT and the TP-Link TL-RP108GE for nothing? Can you get your money back?

[peter-h]

Correct.

I will keep the Tenda injector; could be handy one day.

The RP108 is going back. The vendor is dragging their heels over the return though.