Power-over-Ethernet problem

Topic

New Reply

I have a PoE problem at (only) two wall-sockets, single box, namely that ordinary ethernet works, but there seems to be no sign of Power-over-Ethernet (as shown by a red light appearing initially on the VoIP phone when it is connected). The only recent change has been a replacement of the cables between the PoE switch and the patch panels. All other wall sockets (about 30) work correctly.

Setup (for each one of the wall sockets is):

[*]HP Procurve PoE switch port [call this point A]
[*]Short patch cable from PoE switch port to Patch Panel port
[*]Cable to the wall socket, perhaps 15m long, in ceiling void
[*]Wall socket port
[*]Patch cable, 5m long, from wall socket port to VoIP phone
[*]VoIP phone’s socket [call this point B]

Actions

[*]I have changed the port on the HP Procurve switch – no good
[*]I have changed the short patch cable from the Procurve switch to the patch panel – no good
[*]I have changed the longer patch cable from the wall socket to the VoIP phone – no good
[*]I have changed the VoIP phone for another one – no good
[*]With a cable continuity tester, I have proved that all 8 pins work on the path from point A to point B
[*]I have connected a laptop to the end of the long patch cable from the wall socket, and can connect to the internet without problems

Theory
“Somehow” the power is being lost, presumably somewhere in the ceiling cable (but why do all 8 pins check out on the cable tester?).

Unfortunately I do NOT have a PoE voltage tester, such as this one.

Can anyone offer any suggestions as to why this is not working? Thanks

BATcher

Plethora means a lot to me.

Reply | Quote

Replies

I’d guess that since the “only recent change has been a replacement of the cables between the PoE switch and the patch panels” those are the only points you’d need to check. Wouldn’t you?

In addition to the cables themselves, I’d visually check the jacks to make sure that they haven’t been damaged.

Also, I assume that both ends of each cable are either T568A or T568B? You may not want a single cable with T568A on one end and T568B on the other end (unless a cross-over cable is actually what’s desired).

Without a PoE tester it’s often difficult to quickly determine where you may have a high-resistance point (for instance). The reactance of the line may not be obvious with a simple digital meter even though the resistance would.

Image or Clone often! Backup, backup, backup, backup......
- - - - -
Home Built: Windows 10 Home 64-bit, AMD Athlon II X3 435 CPU, 16GB RAM, ASUSTeK M4A89GTD-PRO/USB3 (AM3) motherboard, 512GB SanDisk SSD, 3 TB WD HDD, 1024MB ATI AMD RADEON HD 6450 video, ASUS VE278 (1920x1080) display, ATAPI iHAS224 Optical Drive, integrated Realtek HD Audio

Reply | Quote

Do you have a punch down tool? If so, re-punch the connections – don’t remove the cable just punch. That’s fixed several issues in the past for me.

cheers, Paul

Reply | Quote

Thanks, both of you!

RockE: I considered that swapping the patch cables from the Procurve switch sockets to the patch panel would be an adequate test, bearing in mind that the whole path from the Procurve end of the patch cable to the far end of the patch cable by the VoIP phone ‘checks out’ with the cable continuity tester.

PaulT: unfortunately I would have problems extracting the patch panel from the comms cabinet to do any ‘punching down’. And the whole cable length checks out, as I said above.

It looks as though I will need to get a PoE voltage tester to take this further…

BATcher

Plethora means a lot to me.

Reply | Quote

You can hire a full tester which will test everything.

cheers, Paul

Reply | Quote

Suggestion: Check resistance of wires and POE power output at the plug. The plug is also a suspect.

Start at the POE power injection point (A), to the losing power connector point (B). The wire resistance between the two points, A and B. POE spec says 20 Ω max.

If high resistance, say, 10Ω and up (each wire from point A to point B), may have wire problem, or wire termination problem (poor punch or poor/old rack contacts, or nearly broken wire at the punch).

But do the obvious first:
1. Use DVM to measure POE power injector no load Voltage, typically 48V at the network wire, ranging 35V-57V. Load the POE at the end of network cable as in usual application. Measure Voltage at POE at the wire again. If it drops significantly, it is the POE power brick losing efficiency, going bad.
2. Sometimes, it is the connector at the POE side that is loose or having intermittent contact. Wiggle the POE injector connector(s) to check.
3. Finally:
Use an Ohmmeter (of a DVM), first measure a roughly same length good cable (shorted at one end), to get some ideas of the wire resistance. (Or calculate it). Then, without POE power connected, short one end of the problem cable, measure Ω on the other side of the net work cable. This measures TWO wires of the cable at the same time. If you have a very long thick copper wire to use, you can use it to measure resistance of a single wire of the network cable.

As for Internet access and networking, you may have slow down, or packet collisions, but network signal may still be good enough for a poor contact or bad wire.

An easier way is measure the termination impedance of the cable (“AC” impedance, not DC Ω). A poor contact/bad wire changes the impedance characteristics.
Note:
For power delivery, it may be a problem with a loosy contact or bad high-resistance wire (10 Ohms or higher), as follows:
W=VxI. Example:
For 5V, delivering 1A = 5W.
If the wire resistance is 5Ω, and at 1A delivery, at the wire end it would have no Voltage left:
5Vsource – (1Ax5Ω) = 0V at the end of wire.

You can say the wire ‘eats’ all the power.

That is why POE must boost up Voltage, to typically 48V.
Same 5-Ohm wire, and 1A delivery, the Voltage at the wire end will be
48V-(1Ax5Ω)=43V
Still lots of Voltage for the load, even after going through the resistive wire.
Then we’ll step down the POE high Voltage to 5V, for normal 5V load, such as USB, etc.
This is not unlike high tension wire for AC power-line that uses kV to 10s of kV for long distance power delivery.

Needless to say, ideally you want zero-DC-Ω wiring. POE spec is 20Ω max.

May need ‘self-made tools’:
1. Two network patch cables, one end cut; wire-end insulation stripped (for measurement).
2. Two Y-adapters so that you can plug the above ‘cable(s)’ in, enabling Voltage measurement.
Plus
some alligator-clip jumper wires.

My way?
1. Jury-rig a 100mA LED circuit (just add R) with 2 alligator clips.
2. A USB cable, the two 5V/0V wires to alligator clips.
How to use?
Need the above two patch cables with open-end stripped-wires.
1. Make sure no POE and network devices connected.
2. The rigged USB cable connects to a USB charger (5V,1A).
3. Alligator clips the USB 5V/0V to the network wire pair under-test
4. The wire pair is now powered by 5V.
5. Alligator clip the LED to wire pair under-test (note polarity). Reversed? No harm. LED can take it.
6. The LED should light up normally.
7. No light = open wire / open contact
8. Dim light = contact problem

Reply | Quote

Thanks, but you credit me with a level of manual dexterity and equipment which I don’t have!
Would you reckon that my suggested PoE Tester will do all you suggest?

BATcher

Plethora means a lot to me.

Reply | Quote

Typically, wire pair (1,2), two wires together, is for power positive, wire pair (3,6) is for negative.
Sometimes it is reversed: (1,2) negative; (3,6) positive.
A DVM will tell.
I hope the load devices, one is not polarity-reversely connected. If it does, it shorts the POE.

Reply | Quote

I forgot:
Check one more obvious thing before the in depth measurement.
If there are more than 1 device load, unplug one device/load at a time. If POE suddenly works, one of the devices is overloading the POE.
If it happens that one extra load is newly added, then it is one too many loads for the POE. May have to get a more powerful POE injector. Or, POE max power spec is reached.

About the POE-Tester, £35, is just that: a Voltmeter + Ammeter + nice plugs/connectors.

I would buy two. £70, ouch!
One to monitor the Injector power source side, the other at the load side.
The Voltage drop (Vsource-Vload) is right there to see.
Current should not exceed 600mA on the network wires.
Mentally calculate (Vs-Vload)/I=R of the wire pair. R should not exceed 20Ω.

I don’t know where the tester measures the Voltage+current.
If it is right at the network wire (high Voltage, about 48V typical), then it is good.

If it only measures right at the power output, and at the load directly, it is partially useful.

As mentioned, the POE high Voltage is stepped down then applies this low Voltage to the load. So, if measured directly at the load, it is only measuring the already stepped down Voltage. This is the output of a step-down Voltage converter unit. One extra device in between.

The measurement should be the [high] Voltage and current at the network wires, not directly at the source, nor directly at the load.
(Obviously my aim is wire connectivity check. Any DVM will be good on Voltage monitor.)

The single big problem using this POE-Tester is that all 4 wires (2 pairs) are tested, all together. (The current is carried by all 4 wires, the two pairs.)
If you found wire resistance/connectivity problem, you still have to further identify the problem pair, or the 1 of 4 wires.

Note:
POE will have new standard spec. At present only 2 pairs of wires are allowed POE.
New spec will allow all 4 pairs, and the network wire handling current will be 1A max.

Reply | Quote

First I must excuse myself because I am a retired Masters Licensed electrician and I have learned some basic rules that sometimes annoys people. You don’t always need fancy electrical testers. Just having fancy tools can lead to false confidence. Keeping things simple helps.

This is the part that sometimes annoys people, a trouble shooter must follow a discipline and not break from it until they have enough experience and recognize a familiar pattern, then they can take a ‘shortcut’.

Power-line gadgets are often blocked because they are used on a power strip. The problem is the filtering or electrical spike blockers that some have. Avoid filters.

Next, a homeowners’ power-line gadgets may be on 2 separate electrical phases. When there are 2 gadgets they need a common link to travel across those phases to communicate. That link is often provided by large electrical appliances and sometimes there are no 2 phase appliances to facilitate that, like a clothes dryer, electric stove, large air conditioner, etc. The phases can be isolated and prevent networking. There are electrical panel ‘bridges’ that will fix the missing link.

The other pain is when there is power-line noise. That is electrical noise, often caused by a microwave or an old electric motor that has worn out brushes (furnace blower, well pump). Noise can be harder to pinpoint, it is a trial of off and on to see if it affects power-line communications.

Lastly, test the units by placing them on the same circuit and see if there is communication. If not, it’s likely a failed unit.

Best of luck.

Reply | Quote

BATcher – mr.wire has said it all. Above all you must plug directly into the wall socket, not via any form of power strip or multi-plug adapter.

As you say you are in the UK, I believe that you can discount mr.wire’s caution about separate electrical phases, as it is standard UK electrical practice to supply only single-phase to domestic premises.

Dell E5570 Latitude, Intel Core i5 6440@2.60 GHz, 8.00 GB - Win 10 Pro

Reply | Quote

Mr Wire and John Fleming: I have made no mention whatever of power strips or powerline adapters (usually known as homeplugs in the UK)!

I am talking about one port of a 24-port HP Procurve Power-over-Ethernet switch and the connection from there to a Voice-over-IP telephone. Both of these are unlikely to be at my home, but are part of the setup of 25-or-so PCs at work.

I was greatly amused by the suggestion that I might be working across two phases of a three-phase electrical supply at between 380V and 415V AC! No, thank you…

BATcher

Plethora means a lot to me.

Reply | Quote

BATcher – OK!

Dell E5570 Latitude, Intel Core i5 6440@2.60 GHz, 8.00 GB - Win 10 Pro

Reply | Quote