TLK105L with the SAME PHY side termination for TX and FX

Christos,

If I understand your intent correctly, I think it would be better to use stuff options to provide both networks on the version of the board with copper and fiber options. Check out the DP83640T-EVK board schematics at:

http://www.ti.com/lit/pdf/snlr007

You could do something similar with your board.

Patrick

Thank you, Patrick.

Once pc board with PHY will be a "standard" one, it is intended to have a unique mounting process and BOM. This is the reason I wish to use the SAME termination at PHY-side circuit. Do you know if some of of my two proposed circuits can work suitably? If not, do you have any solution?

Christos

I am not sure that I understand how the board will be constructed or used.  Do you intend to have the transformer / RJ-45 connector and the fiber optic transceiver both be populated on the board at the same time and both be connected to the TD+/- and RD+/- device pins at the same time?

Patrick

No.

Trying to clarify my first text...

* In the “standard” board I’ll have MAC, TLK105L and “PHY side” termination components;
* this “standard” board will be connected to a “custom” board, which will have two mounting options: RJ45 and fiber.

Each “custom” board will be populated with proper “RJ45 side” or “optical transceiver side” components according the mounting option. However, the “standard” board needs to be always the same one – and, so, the “PHY side” components also need to be the same, including the termination ones.

Is it clearer now?

Christos

OK.  It is becoming clearer.  There are some constraints to consider here. 

In fiber mode, the PHY will operate in MDI mode, i.e. the PHY will always transmit via the TD+/- pins and will always receive via the RD+/- pins.  In copper mode, most applications use Auto-MDIX to swap the pins which transmit and receive in order to accommodate straight or crossed cables. In your application, when operating in copper mode, would Auto-MDIX be disabled so that the PHY will operate in MDI mode? 

Patrick

Ok, Patrick.

You detected a good issue! I'm almost right that Auto-MDIX can be disabled in my application. (I'll confirm!) If so, can I use in the "standard board" (for copper as well as for fiber):
- my proposed circuit for TD pair (50 Ohm to 3V3 bias)?
- anyone of the two proposed circuits for RD pair (130 / 80 Ohm bias or 50 Ohm to 3V3 bias with capacitor decoupling)?

P. S.: Anyway, I think that, even if Auto-MDIX will not be disabled, 50 Ohm to 3V3 bias with capacitor decoupling can be ok for both pairs. What do you think?

Christos

Hi, Patrick.

I'm assuming that my proposed solution will work and I'll continue the project according that.

If you see another limitation or problem, please inform.

Regards.

Christos

Christos,

Sorry for the delay.  I am still not sure this solution will work.  My two main questions are:

1. Have you confirmed that Auto-MDIX can be disabled in the application?
2. What will be the datapath for the RJ-45 board?  Will there be a transformer or will you AC couple directly to the RJ-45?

Patrick

Ok, Patrick.

Thinking better, a unique circuit in my “Standard PHY Board” may not work properly with MDIX. However, it seems ok with Auto-MDIX disabled. (I’ve confirmed that it’s not necessary in my applications.)

The attached drawing shows the solution I was referring to. It seems that all components will work in their normal conditions with anyone of the two arrangements: “Standard Phy Board + Customized RJ-45 Board” and “Standard Phy Board + Customized Fiber Board”. Only one “specific” feature is required from TLK105L: to conveniently accept 130/80 Ω biasing signals (PECL) in its RD pins regardless of TX or FX operation. Is it right? Can I hope this from TLK105L? (I suppose that its differential input circuit doesn’t change with operation mode.)

Regards.Christos

6177.Proposed_solution.pdf