Frequency compensation (I think)

Thread Starter

transconductance

Joined Jun 29, 2019
48
I am trying to instrument 100BASE-T Ethernet. Sounds like it's not an RF question but I think it is.

I want to tap into an ethernet pair without loading it any more than necessary for purposes of testing the signal. I don't want to demodulate it. I just want to do a basic envelope detection on it - an AM radio. It's a differential signal, so I need to turn it into a single-ended signal. I need a BALUN.

My first thought was to use a normal Ethernet isolation transformer without any termination resistors, but there's still the magnetizing inductance which is fairly high at 350uH. So then I thought about putting resistors between the transformer inputs and the network pair. Solves the loading problem and I can deal with the attenuation but results in a lot of distortion. So I feel like I need to shunt the resistors with small capacitors, like an oscilloscope probe. Then, I thought, well maybe I don't even really need the resistors.

At this point I'm in over my head mathematically and have no idea how to figure it out. I think I can safely assume I'm dealing with a 31.25 MHz signal since that's the whole point of the MLT-3 modulation scheme. So somehow I want to resonate out the transformer inductance and make it look resistive at 31.25 MHz but at the same time minimize the loading.

Maybe what I really need is to start from first principals and design a balun instead of trying to shoehorn in an Ethernet transformer.

Now you see why I think this is an RF problem.
 

DickCappels

Joined Aug 21, 2008
8,719
You could use a differential scope probe if you have the money to buy or rent one. You can also look at the signals differentially with a two channel scope. In the case of using the scope you might want to use X100 probes or FET probes to minimize capacitive loading.
 

Thread Starter

transconductance

Joined Jun 29, 2019
48
Commercial baluns for 100BaseT are made. It used to be common to take thinnet (10Base2) 50Ω coax and use a balun to run it on CAT3 (telephone twisted pair). Now, its 10 or 100baseT over 75Ω coax.

Random example: https://www.amazon.com/Dualcomm-Ethernet-over-Adapters-DECA-100/dp/B009AGCLVG
But those are designed to terminate the transmission line. And what I want is a non-invasive tap. The line will be terminated on both ends by the normal Ethernet devices. Maybe I should explain that this is an automation system.
 

Thread Starter

transconductance

Joined Jun 29, 2019
48
You could use a differential scope probe if you have the money to buy or rent one. You can also look at the signals differentially with a two channel scope. In the cause of using the scope you might want to use X100 probes or FET probes to minimize capacitive loading.
TL;DR I suppose what I'm trying to do is build a differential probe...

In fact, I have looked at the signals on a nice expensive Tek 'scope. I should have made it clear that this isn't a one-time measurement. I'm trying to make a compact instrument that people in the plant can use. A go/no-go red-light/green-light box that measures the amplitude. We've had problems where an Ethernet device works fine on the bench and passes all testing but, once it gets out in the plant, it has the kinds of problems that indicate the signal is marginal.

I was hoping not to have to put together a differential amplifer with a high-impedance input but that may be the correct path.
 
Top