Apparently there is a bad interaction between the mains power adapter, the POE midspan injector, the POE power extractor, the wires, and the EL inverter.
You've told us which midspan injector and POE power extractor you are using, so that's already 40% of the information we need to solve this problem -- 40% better than many Stack Exchange questions. :-)
I'm going to assume you're using the
EL Inverter - 12v
mentioned at the bottom of the
EL Escudo page you linked.
I'm also going to assume that, with the inverter disconnected,
everything else you're trying to power off the same PoE link
(the EtherTen and anything else you have connected to it) uses less than 0.25 A at 12 V.
You are going to check these assumptions, right?
According to that EL inverter datasheet,
that EL driver is only specified to run over a fairly narrow range of 11 V to 13 V,
and it pulls a normal operating power of 2 W to 5 W (typical) to 9 W.
That implies that at 12 V, the normal operating current is 0.166 A to 0.417 (typical) to 0.750 A.
Some forms of PoE can send more than 13 W of power to the powered device, which is more than enough to supply the worst-case 8 W steady-state power required by your EL driver.
Your "plug the device directly into the power adapter which was running the PoE" test indicates that you have plenty of power coming out of the mains adapter.
As Toybuilder mentioned, some midspan injectors have a current limiter.
It is possible your EL driver simply pulls more power (steady-state) than your particular POE setup can handle -- in that case, you are forced to use a power supply that can give your EL the power they need.
However, the particular midspan injector you linked has no such current limiter.
Yes, the datasheet says "maximum current: 1.5 A total across all ports".
But that is the absolute maximum that the rectifier diode bridge is guaranteed to work before it overheats and dies; it's not the point where some current limiter cuts the power.
(It's more like a car that can go 100 miles per hour, but at 101 mph the wheels fall off; and less like a car that can go 100 miles per hour, but at 101 mph the governor kicks in and smoothly reduces the speed back to 100 mph).
Since your current is less than 1.0 A, you should be good here.
I think this is most likely your problem.
With your current setup,
there is 4 diode drops between the power supply and the EL inverter
(2 in the PoE midspan injector, 2 in the PoE power extractor on the EtherTen).
Typically a diode drop is around 0.6 V, so these four have a total voltage drop of roughly 2.4 V.
If your power supply happens to supply exactly 12 V DC,
then the voltage at the EL inverter will be roughly 9.6 V --
less than the "11 V" minimum spec.
(The voltage drop over a long Ethernet cable will only make this worse).
If you have a bench power supply, it would be pretty simple to use it to power your midspan injector at 12 V and then ramp it up until your multimeter shows you are actually getting 12 V at the EL driver input -- you'll have about 14.4 V at the bench power supply output.
If that bench test successfully lights up the EL light, perhaps the simplest way to fix this "under voltage" problem (and free up the bench supply for other things) is to use one of the alternatives mentioned on the EtherTen page:
- first: install the "Power-over-Ethernet Regulator 14-24V" on the EtherTen. Second: plug a mains power adapter that puts out 24 V (really, anything in the 15 V to 24 V range, with at least 1 A) into your midspan injector. OR
- first: install the "Freetronics Power-over-Ethernet Regulator 802.3af" on the EtherTen. Second: use some standard 802.3af power sourcing equipment. (It will supply at least 44 V to each device).
An EL inverter has some similarities to a switching voltage regulator.
Many switching power supplies have difficulty starting up with a current-limited source.
Jerrold Foutz describes this problem in more detail in
"Latchup of Constant-Power Load With Current-Limited Source".
Some possible things to try:
Hook the EL inverter directly to the EL wire (without the EL Escudo shield between them). Does the same problem happen? If so, the main problem is somewhere other than the EL Escudo shield.
Try connecting a very short cable (~6 inches?) between the midspan injector and your EtherTen controller. If it starts working, you know one problem is the inductance and resistance of the ethernet cable.
Add a big capacitor directly across the power wires just before they go into the EL lamp driver. This might be enough to supply that start-up burst of energy.
Add that capacitor, then somehow let the capacitor charge up first, and then later turn on the EL lamp driver.