Current rating of an RJ45 connection

[Joerg]

Folks,

Have to send 12V/500mA across a connection. So far there is only an
RJ45. Looking at datasheets I saw current ratings of 1.2A and 1.5A which
sounds like a lot.

Is this kosher? I know PoE sends up to 600mA across but I've always felt
uneasy about it when they went up from the (older?) 350mA.

 

[Jamie]

Folks,

Have to send 12V/500mA across a connection. So far there is only an
RJ45. Looking at datasheets I saw current ratings of 1.2A and 1.5A which
sounds like a lot.

Is this kosher? I know PoE sends up to 600mA across but I've always felt
uneasy about it when they went up from the (older?) 350mA.

it's not the wire you need to worry about it's the cheap crimp style
connectors.

 

[[email protected]]

Folks,

Have to send 12V/500mA across a connection. So far there is only an
RJ45. Looking at datasheets I saw current ratings of 1.2A and 1.5A which
sounds like a lot.

Is this kosher? I know PoE sends up to 600mA across but I've always felt
uneasy about it when they went up from the (older?) 350mA.

I think I remember PoE as something like ~400-500mA per pair, but
that's gotta
be including everything from connectors, wire size and how many cables
that are
usually packed together in a large installation

-Lasse

 

[CharlieB]

Have to send 12V/500mA across a connection. So far there is only an RJ45. Looking at datasheets I saw current ratings of 1.2A and 1.5A which sounds like a lot.

Amphenol make a range of connectors rated at 1.5A and I've successfully used them in NRTL listed POE powered products drawing >800mA

Charlie

 

[[email protected]]

I think I remember PoE as something like ~400-500mA per pair, but
that's gotta
be including everything from connectors, wire size and how many cables
that are
usually packed together in a large installation

Don't forget distance. PoE is one of the few times I've had to worry about
"maximum power transfer". BTW, 400-500mA sounds right but my memory is a bit
fuzzy on that, too.

 

[Nico Coesel]

Folks,

Have to send 12V/500mA across a connection. So far there is only an
RJ45. Looking at datasheets I saw current ratings of 1.2A and 1.5A which
sounds like a lot.

Is this kosher? I know PoE sends up to 600mA across but I've always felt
uneasy about it when they went up from the (older?) 350mA.

RJ45 isn't the best connector. In my experience you need to reseat the
connector after a few years for it to make good contact. IMHO sending
a lot of current through it may make things worse. BTW PoE always uses
two parallel wires to carry the supply to the device whether the power
is phantom fed or through unused pairs.

 

[[email protected]]

Don't forget distance.  PoE is one of the few times I've had to worry about
"maximum power transfer".  BTW, 400-500mA sounds right but my memory isa bit
fuzzy on that, too.

sure distance matters when you need to figure out how much power you
have left at the
of the cable, but it doesn't affect current rating

-Lasse

 

[Joerg]

RJ45 isn't the best connector. In my experience you need to reseat the
connector after a few years for it to make good contact. IMHO sending
a lot of current through it may make things worse. BTW PoE always uses
two parallel wires to carry the supply to the device whether the power
is phantom fed or through unused pairs.

Thanks, Nico. Also thanks to all others. One of the things concerning me
is the fact that datasheets derate to zero milliamps at slightly above
70C. Got that info this morning and it was quite the surprise. This is
from the connector we are using:

http://www.te.com/commerce/Document...108-1163EpdfEnglishENG_SS_108-1163_E.pdf

The unit will sit in full sun occasionally and can become as hot as the
inside of a car under that condition.

 

[Joerg]

Note that the maximum recommended current (for power conduction) per
wire for #24AWG CAT5e is 577ma.
<http://www.powerstream.com/Wire_Size.htm>

You can get 22AWG TP for that. Usually only two-pair but that would
suffice here.

I found some references to a TIA recommendation for 360ma per pin
maximum, but never could find the original document.

I found 1.2A as a low value at 25C, so far. But the surprise was a
rather fast derating to almost zero at 70C.

One of the cleanup jobs I dealt with had 12V at about 600ma going
through an RJ45 as in PoE but not using 802.3af power control.
(Actually, it was an RJ50 connector, which is the 10 pin version of
the RJ45). The problem was that after about 20 insertion and removal
cycles, the power connectors were becoming intermittent. Examination
with my microscope shows pits, craters, holes, missing plating, and
evidence of arcing. The problem was that the connector was being
inserted and removed with power applied. The load looked like a BFC
(big fat capacitor) which produced current spikes well above the
recommended 360ma. A peak current limiting resistor in series with
the power connections was the best I could do to reduce the problem to
acceptable levels.

The worst is a BFC on the source side plus a BFC on the other side and
the only inrush limiter is a short cable and an RJ45 connection ...
bzzzt ... *PHUT*

But we'll avoid that here.

While working on the problem, I got diverted into measuring how much
gold was being plated onto the RJ45 pins. The better connectors have
50 micro inches. The junk connectors I bought on eBay barely had 10
micro inches with exposed nickel plating in places. I also had issues
with surface roughness, which reduced the contact area and seemed to
cause some intermittents without arcing:
<http://rj45s.com/Gold_Plating_Problems.html>
(This site is worth reading if you're using RJ45 connectors).

Thanks! I have bookmarked it. In this case we'd be ok because there
would be essentially only two insertions. One at production test and
another at site installation. Both non-powered.

 

[Adrian Jansen]

Thanks, Nico. Also thanks to all others. One of the things concerning me
is the fact that datasheets derate to zero milliamps at slightly above
70C. Got that info this morning and it was quite the surprise. This is
from the connector we are using:

http://www.te.com/commerce/Document...108-1163EpdfEnglishENG_SS_108-1163_E.pdf

The unit will sit in full sun occasionally and can become as hot as the
inside of a car under that condition.

We regularly see plastic boxes supposedly rated for 85 degC badly
deformed after being in a car in Australia.

 

[Joerg]

Careful. That's #22AWG solid, not stranded. Stranded CAT5 is #24AWG
which is what you'll probably see on "portable" equipment.

True, but the 577mA of 24AWG suffice in this case.

That's odd. The only thing I can think of is that either the plastic
connector or the insulation becomes sufficiently soft to cold flow.
That reduces the pressure on the insulation displacement connection,
possibly resulting in a smaller contact area between the stake and the
wire. (This is a guess, this is only a guess).

We just did a test run. Stays under 60C. Whew ...

That's exactly what I had. I don't know how big the cazapitor was
inside the power supply as it would have required explosives to
disassemble. However, the device had about 10,000uF across the PS
input pins:
12V at 0.6A = 20 ohms
t = RC = 20 * 0.01F = 200msec
which was enough to keep the device alive during minor power glitches.
The designer wanted to add more energy storage but ran out PCB real
estate. A supercap would have been better, but at the time, they were
rather pricy.

10,000uF across RJ45 pins? Wow! That is brazen.

Hopefully, it's running on battery, solar, or something not
susceptible to power outages and glitches. If not, run a test to see
how long your device will fly without power. You may want to do some
inrush current limiting.

We'll have a generator with uC triggered starts as needed to replenish a
large battery. So always juice. But the battery will not directly feed
the system, there's a current limiter.

Assumption, the mother of all screwups. The first indication of a
problem showed up in final test, where one power supply and I/O cable
combination was used to connect to the automagic test equipment. After
about an hour of testing, the ATE was rejecting nearly every device
due to intermittent and "noisy" connections. I traced it to the RJ45
connectors and stupidly supplied the test line with a box of 25
replacement cables. That lasted about a week. I knew I was in
trouble when I got a casual email requesting more cables.

That's probably the same feeling people get when a request comes in via
collect call, "Dad, I'd need another $500, soon"

 

[Jasen Betts]

We regularly see plastic boxes supposedly rated for 85 degC badly
deformed after being in a car in Australia.

I've seen 85 degC capacitors (I assume it was the cap) quit in similar circumstances
When I parked I had to unplug the blinker unit and put it under the
seat so that it would stay cool. I never got round to replacing the
cap with a 105 degree part.

 

[SoothSayer]

RJ45 isn't the best connector. In my experience you need to reseat the
connector after a few years for it to make good contact.

All the Gold utilized has changed all that.

Hundreds of billions of normal air connections made every day (all
around the world), and years later still fine. And there is nobody
wandering around flexing all the lines either. Gold does not oxidize at
any rate you would understand. Obviously. Platinum, practically never.

Your experience must have been with tinned 'stingers' on CB radios.

Cat 5 and 6 say for the last 15 years, has been just fine as far as
contact issues go.

And half an amp is trivial.

 

[legg]

Thanks, Nico. Also thanks to all others. One of the things concerning me
is the fact that datasheets derate to zero milliamps at slightly above
70C. Got that info this morning and it was quite the surprise. This is
from the connector we are using:

http://www.te.com/commerce/Document...108-1163EpdfEnglishENG_SS_108-1163_E.pdf

The unit will sit in full sun occasionally and can become as hot as the
inside of a car under that condition.

This could simply mean that a change in colour or surface finish of
the casing is required. Different colours and materials have different
albedo qualities.

White is three times better than grey, order of magnitude+ better than
black.

RL

 

[Joerg]

This could simply mean that a change in colour or surface finish of
the casing is required. Different colours and materials have different
albedo qualities.

White is three times better than grey, order of magnitude+ better than
black.

Not sure if the difference is that stark in real life. I own a white car
and a couple of weeks ago during really hot summer days I had a black
Volkswagen rental. I didn't feel that much of a difference when coming
back to the parked car after severak hours.

 

[Joerg]

What will it do when it's out in the real world? 60C - 25C ambient is
a 35C temp rise. That's not too awful, but with a "commercial"
temperature range of 0 - 70C, that's 105C on the case. That should
boil the tea on your product case.

Most of this civilian computer stuff isn't really geared towards harsh
life. Even garden tools aren't. I had a shovel with plastic handle where
the top plastic layer had softened, discolored and became soft. I didn't
notice until the stuff was stuck to my paws.

Yep. (Disclaimer: It wasn't my design). The designer originally had
a much smaller cazapitor. However, the initial customers experienced
problems with noise on the power line. Rather than fix it in the
power supply (there were several thousand already in stock), he
applied the fix in the device by adding a huge cazapitor. Nobody
noticed a problem until final test started rejecting devices. Better
power supplies were eventually purchased, but the BFC and my
thermistor fix remained.

I am surprised this BFC ever made it through a design review. Or maybe
the ECO was just rubber-stamped and robo-signed

Incidentally, it wasn't my responsibility to fix the RJ45 problem. I
was doing the FCC type acceptance dry run(s), and was asked to "take a
look" at the problem. That somehow morphed into inheriting a
production engineering function. It was a safe thing to do for the
company as I was working on a fixed price contract and would not get
paid extra for fixing their connector problem.

One of the reasons I don't do fixed bid work

 

[[email protected]]

Not sure if the difference is that stark in real life. I own a white car
and a couple of weeks ago during really hot summer days I had a black
Volkswagen rental. I didn't feel that much of a difference when coming
back to the parked car after severak hours.

mythbuster tried it, think they saw a difference of 10'F
between black and white car

-Lasse

 

[legg]

Not sure if the difference is that stark in real life. I own a white car
and a couple of weeks ago during really hot summer days I had a black
Volkswagen rental. I didn't feel that much of a difference when coming
back to the parked car after severak hours.

Really. So then you've got nothing to worry about.

Solar contribution to thermals ia roughly a kilowatt per square meter
in the direct overhead case.

RL

This message has been purged of all expletives or any sentences
containing them.

 

[Joerg]

Really. So then you've got nothing to worry about.

Solar contribution to thermals ia roughly a kilowatt per square meter
in the direct overhead case.

Sure, but there is also re-radiation and reflection. Except maybe for a
Hell Angels Harley or a low-rider rap mobile, vehicles are normally not
painted a dull "primer black". It's more metallic-anthracite plus a
shiny clear coat on top.

RL

This message has been purged of all expletives or any sentences
containing them.

But I didn't use any

 

[legg]

Thanks, Nico. Also thanks to all others. One of the things concerning me
is the fact that datasheets derate to zero milliamps at slightly above
70C. Got that info this morning and it was quite the surprise. This is
from the connector we are using:

http://www.te.com/commerce/Document...108-1163EpdfEnglishENG_SS_108-1163_E.pdf

The unit will sit in full sun occasionally and can become as hot as the
inside of a car under that condition.

Oh, yeah - don't get fahrenheit and centigrade mixed up. This tends to
happen in anecdotal conversation with Americans.

RL