Connect two PHY chips


#1

Hello,
we want to connect the WizFi630 module directly to an Computer On Module (TX53-8030, karo-electronics.com). The computer module has a SMSC 8710A PHY chip.
Both modules (processor and WizFi630 module are on same circuit board.
Is it possible to connect the WizFi630 directly to an other PHY chip without use of a transformer ?
Is capacitive coupling necessary ?
Are there application notes / schematic examples for a PHY <=> PHY connection.

regards
Guenter


#2

Hello guenter

Thank you for interesting our product.

Please refer to the PHY to PHY photos WizFi630.

However, WIZnet has not recommended the PHY to PHY.

We have seen in not be formally tested.

You are recommended to try the test yourself.

Thanks.

WIZnet.



#3

Hi,

Guenter, I’m very interested on this and I’m wondering if you were able to test this and what sort of results you achieved. Is there anything you can share about this with me?

Also, to Wiznet,

Perhaps you’ve heard of any progress made on testing this?

At this point I’m developing a board that will host a WizFi630 and will also host 2 TCP-IP clients connected to 2 LAN ports of the WizFi630. Since all of these components will be mounted on the same PCB I am really trying to save on space and specially avoid using bulky RJ45 connectors with magnetics.

I’m at a point right now were I think I could make things work without using RJ45 connectors by instead using pulse transformers in an IC form-factor but it sounds a bit absurd to do this and have the traces run from one pulse transformer to the next one in a really short distance on the same PCB. If I could do this without the pulse transformers it would be amazing!

Thank you in advance for anything you might share.

Cheers,

Pablo C.


#4

Hi,
I did not use a transformerless configuration in my project. I had no time to test it first. I Think it depends on the two PHYs involved.
It was more safety for me to use a transformer (e.g. wuerth electronics 7490100111A) between the Wiznet module and the processor board. You have to terminate each side of the transformer. You also have to choose the right voltage for the center tab, this depends on the PHY.

Take a look at the AppNotes from MICREL ans SMSC. There are more information regarding transformerless setup.

micrel.com/_PDF/Ethernet/app … an-120.pdf
electronics.stackexchange.com/qu … -magnetics
electronics.stackexchange.com/qu … -magnetics
microchip.com/forums/m830227.aspx

Good luck
Guenter


#5

Hi Guenter,

Thanks for the help, those were some great pointers!

I’ve now gone and looked at the PHY chip on the client side and it’s a TI DP83848J. It looks like it recommends 3.3V for the pull-ups and transformer CT’s in the design guidelines. I also found an Application Note in the TI website discussing transformer-less operation with this exact PHY chip which is a great indication it may be doable (I provide links to all documents below)

TI DP83848J datasheet: ti.com/lit/ds/symlink/dp83848j.pdf
Application note on Transformerless Operation: ti.com/lit/an/snla088a/snla088a.pdf
Ralink Chip datasheet: tracermcc.ru/foto/bender/RT3 … 8_0902.pdf

Now, the issue I’m facing is that the Ralink chip on the WizFi630 apparently has the PHY module embedded in it, but I can’t find any useful information or references in it’s datasheet talking about transformer selection guidelines, transformer CT voltage, transformer-less operation of anything useful. And also, the WizFi630 HW design guide shows the CT’s pulled up to 1.8V which I find concerning.

Do you have any other advise on how to punch my way through this?

Cheers,

Pablo C.


#6

Do you have the datasheet of the RT3050 ?
tracermcc.ru/foto/bender/RT3 … 8_0902.pdf
In chapter 2.5, on page 22.: Current consumption @ 2.0V center tab voltage.

Also the in this thread, in the picture above, there is the information: 2.0V +/+5% for the center tab voltage.
The evaluation board has 1.8V. There is an LDO to get this 1.8V from the 3.3V supply.

Guenter


#7

Thanks again Guenter,

So I think I’m lacking a bit of experience at this point. Now that I know that my 2 PHY chips drive their transformer CT’s at different levels, what does that mean?

Is there still a point trying a transformer-less setup perhaps with capacitive coupling in the lines or not?

And if I do, what should I do about the different voltage levels? Should I maybe add the extra capacitor shown for DC biasing in the MICREL app note you shared previously?

Pablo C.


#8

Give it a try.
It depends on the PHY. If it is current driven it maybe do not work with capacitor setup.
Also the AUTO-MDIX and AUTO-NEGOTIATION may not work properly.

Guenter


#9

Hello guenter

i want to connect two PHY using integrated Magnetics in the same PCB(without Rj45).

in my case:
The first PHY works in 10/100/1000BaseT, so, it have 8 wire MDIO (MDIO1 +/- to MDIO3 +/-)
The second PHY works 100BaseT, so, it have 4 wire (TX+/- and Rx+/-)

please, can you send me a diagram schematics of a possible conection?
Thanks


#10

i have been thinking:







#11

The experience which is needed for the user because the PHY chips are very much important for the drive, though the coupling in the lines are capacitive for the transformer if then also you can take the advice from dell printer support for the correct input.