Trying to Measure Ethernet Current

[Electro Fan]

I used a couple terminal blocks to to connect two halves of an ethernet cable.  Trying to measure the current on the various leads.  When I insert either a Fluke 179 or an Aneng 8008 into the circuit with any of the ethernet wires the signal crashes.  Sometimes if I just jumper the two DMM leads the signal returns.  I think maybe the "burden voltage" is a tad too much for what I am guessing is sub milliAmp current.  Any suggestions on a meter that might measure tens or single digit micro current with relatively low burden voltage?   Maybe this is a good reason to have Gold or regular (non-gold?) U Current?  Are they still being made/sold?  Thanks

[Cerebus]

Ethernet on twisted pairs is, from memory, 1V differential into 100 ohms. So you need to be looking at something that's going to put way less than 100 ohms in series with the signal. The current shunt resistor on any typical meter is going to be way too high for the kind of range you're interested in (milliamps basically). A high current range on a 6 1/2 digit meter, perhaps even 5 1/2 digit,  might give you a ball-park reading without clobbering the Ethernet at the same time.

[Jeroen3]

What are you intending to get out of it? POE device current?
Normally ethernet is isolated on both sides using signal transformers, there won't be much current. And if there is, it's in the MHz range.

[AndyC_772]

You can't just split a differential pair carrying 125 Mbit/s and add a massive long loop into just one half of it, then expect it to work.

Ethernet is transformer isolated with 50 Ohm single ended / 100 Ohm differential impedance. If you want to know the current involved in the actual data signalling, measure the voltage differentially with a scope, and calculate it from there.

If you're trying to measure PoE current, I suggest making up an adapter with a couple of back-to-back PoE capable transformers, which will allow you to separate out the supply current from the signalling.

It would be helpful to understand what you're really trying to measure, and why.

[Fungus]

I used a couple terminal blocks to to connect two halves of an ethernet cable.  Trying to measure the current on the various leads.

There's no way this will work:
a) Those currents switch in the MHz range so a multimeter will be totally unable to see them
b) You'll probably cause errors so Ethernet will shut down and you'll see nothing

Hint for getting good answers in forums: Tell us what you're trying to achieve, not what you're doing.

[Electro Fan]

"There's no way this will work" - Roger that.

"a) Those currents switch in the MHz range so a multimeter will be totally unable to see them" - Yep, I should have understood/expected that.

"b) You'll probably cause errors so Ethernet will shut down and you'll see nothing" - In some cases it did shut down the Ehernet, others it didn't - I think it was teetering near the edge.

"Hint for getting good answers in forums: Tell us what you're trying to achieve, not what you're doing"- Well, I kind of did. I made a "breakout box" (more or less) and I measured to see what current was present.  That's what I was trying to do, just measure, no other objective than to see if I could measure the current and how much was present.  I expected the current to be relatively low (uA?) - and I'm still curious to know what the current might actually be; and I'd love to figure out what techniques and tools would enable a reasonably accurate measurement.

Among other things that went wrong, I didn't take into consideration the frequency but I think mostly what I don't have a grasp on is differential.

To be clear, I'm not interested in looking at PoE, just regular Ethernet data signal traffic.  And I'm just measuring because that's how I learn - lame as that might be.

So, what I think I need to understand better is Ethernet and especially the principles of differential signaling.

"You can't just split a differential pair carrying 125 Mbit/s and add a massive long loop into just one half of it, then expect it to work."

I sort of get this, but why?  Here are some guesses:  Because when you split the pair a) the differential is no longer at work? and b) even if it was, the DMM burden voltage is too high?, in addition to c) the frequency is way above what a regular DMM will handle?  Feel free to correct me on any and all of this.  Thx

"Ethernet is transformer isolated with 50 Ohm single ended / 100 Ohm differential impedance. If you want to know the current involved in the actual data signaling, measure the voltage differentially with a scope, and calculate it from there."

- The transformer isolation takes place where?  In the router or switch?  What about in a PC port?  Are the ports (in a router or switch) facing the user at 50 Ohm single ended? and 100 Ohm differential facing the interior of the Ethernet network, ie between routers or switches?  That doesn't seem to make sense because the same Ethernet principle has to work all throughout a network (including PC to router/switch), so where is the 50 Ohm single ended / 100 Ohm differential impedance transformation taking place?  And what drives the 50 Ohm / 100 Ohm relationship?  (I'm guessing some use of Ohm's law like voltage dividers or something...?  Maybe it's as simple as instead of one circuit at 50 Ohms it's 2 circuits x 50 Ohms = 100 Ohms?)  As you can see, I'm kinda/very confused about the transformer isolation and where it takes place and why.

My objective is to figure this stuff out conceptually (and mathematically) and then be able to actually measure it.  If anyone has a link to a primer or wants to post some explanations that would be great.  Thanks!

PS, I've read this:
https://en.wikipedia.org/wiki/Differential_signaling
 - might just need to get it to sink in better and fill in some blank spots but any and all guidance welcome and appreciated 

[Fungus]

Are the ports (in a router or switch) facing the user at 50 Ohm single ended? and 100 Ohm differential facing the interior of the Ethernet network, ie between routers or switches?

Both.

It depends on the direction of data in that particular pair of lines.

In the old days you had to know which way everything went and use a cable that either crossed or uncrossed the data lines, eg. You needed a different cable to connect two PCs than to connect a PC and a router.

https://en.wikipedia.org/wiki/Ethernet_crossover_cable

These days the chips are smarter and mostly auto-configure themselves.

My objective is to figure this stuff out conceptually (and mathematically) and then be able to actually measure it.  If anyone has a link to a primer or wants to post some explanations that would be great.  Thanks!

Just imagine a pair of wires with a 50 Ohm resistor across one end.

You apply a voltage to the other end, current flows, that's it.

The secret is that the receiver senses movement of electrons, not a particular voltage. This is much more immune to noise.

[jahonen]

Every interface port into twisted pair ethernet should be functionally isolated (isolation ratings are not sufficient for safety) with a purpose made transformer. So this isolation is in practically every ethernet-connected device. The transformer is integrated into the connector itself (magjack), or it is on the PCB separately near the connector.

Regards,
Janne

[james_s]

The signals on an ethernet cable can be considered RF, think of it as radio waves travelling down a transmission line. If you split open a twisted pair and route one side of it through something else then that part is not a twisted pair anymore, you're introducing all kinds of noise and reflections. When dealing with frequencies this high you have a situation where the wavelength is significantly shorter than the length of the cable.

Generally both ends of the cable are transformer isolated, you'll find a transformer at each port in a switch/hub and in a network interface in a PC. Occasionally the transformer is built into the jack itself.

[AndyC_772]

My objective is to figure this stuff out conceptually (and mathematically) and then be able to actually measure it.  If anyone has a link to a primer or wants to post some explanations that would be great.  Thanks!

OK, let's back up a little.

Ordinarily I'd completely agree that learning by measuring a real, working system is a good thing to do. In this case, though, I think you'd benefit considerably from reading up on how Ethernet signalling works.

Assuming we're talking about 100BaseTX, there's one transmit pair and one receive pair, and the other two are unused. Modern equipment is capable of handling the cross-over from Tx to Rx whichever way round the cable is wired, so you can't say for certain (without interrogating the PHY) which pair carries data in which direction.

There can't be any DC current, because there's a transformer at each end of the link, and transformers don't pass DC. (There's your answer on the current measurement!)

A differential pair is a great way to carry a high speed signal, but only if it's implemented correctly - by which, I mean the two halves are the same length, and they are physically routed together so they pick up the same interference. If you split the pair, and especially if you add some delay or other degradation to one half only, then the voltage difference between the two at the receiver is no longer a nice, 'clean' version of the transmitted signal. I guarantee the reason you're seeing a drop-out of the connection is nothing to do with burden voltage.

[Electro Fan]

Yes, we're talking about 100BaseTx

Thanks for the posts - I'd be happy to read anything more anyone thinks might be instructive.

I found this:
https://electronics.stackexchange.com/questions/27756/why-are-ethernet-rj45-sockets-magnetically-coupled

But I think I need to get my head much better around transformers and differential signaling in addition to going deeper on the Ethernet architecture.

Back to transformers, saw this:

https://van.physics.illinois.edu/qa/listing.php?id=2354

Q: from your understanding or research of electromagnetic induction, can u explain why transformers can be used in AC power systems and not DC power systems for power distrubution

- Anonymous

A:  That's a really good question.

For a transformer to work, the current in one coil has to somehow make current flow in the other coil and the circuit it's connected to. A DC current in one coil will make a magnetic field on the other coil, but a magnetic field by itself won't drive any electrons around. A CHANGING magnetic field, however, does create an electric force which will accelerate those electrons in the other coil into carrying a current. This process is described by Faraday's law of induction. You get a changing field from an AC current, since the current which makes the field is changing.

- so if no DC what exactly (AC?) is on the wires between the transformers?

[james_s]

Yes, it's AC, very high frequency AC, in other words RF.

[Electro Fan]

Yes, it's AC, very high frequency AC, in other words RF.

Thanks!

[AndyC_772]

Have a look at:

http://www.ti.com/product/DP83848C/datasheet/application_implementation_and_layout#SNLS2662067

See section 7.2.3 "application curves" - the scope trace on the left is 100BaseTX. The top pair of traces show each of the wires individually (ie. TX+ and TX-, or RX+ and RX-, there's no way to tell), and the bottom trace is the difference between the two, which is the signal actually used by the receiving device.

[Electro Fan]

What are the chances I have a scope trace of an Ethernet signal from one of the pairs, maybe with a ground connection that is too long?  Thx

[AndyC_772]

Better than average, I'd say.

[Electro Fan]

Here is another showing the frequency of 125 MHz (not too bad for a 70 MHz Rigol).  I think 125 MHz is the specified symbol rate for 100BASE-TX.

[Electro Fan]

Or maybe to be more specific and correct, that previous image was the symbol rate but the cursors in this image show the frequency of 62.5 MHz (which yields the symbol rate).  Feel free to correct.  Thx