DP83825I: Question about the network coupling circuit

Frank Van Hooft1

Part Number: DP83825I

We are looking at using the DP83825I to update an older existing design. As part of this update we need to remain compatible with the existing network coupling circuit that is already present. Predictably, the old network circuit is different to the circuit shown in the DP83825I's datasheet.

This is our existing circuit we need to use:

As can be seen, the existing circuit contains 49.9 ohm resistors, and it biases the transformer center taps to 3.3V

Will this work, at all, for the DP82825i? We're quite willing to accept shorter cable distance (within reason), or is this circuit going to be a disaster for this PHY?

Thanks very much.

over 4 years ago

0 Vikram Sharma over 4 years ago

TI__Genius 16030 points

Hi Frank,

825 has integrated termination and hence if you will populate these extra resistors of 49.9 : signal level transmitted and received by phy will become half + power consumption will increase. I believe it should still transmit fine with lower cable lengths, but depopulating the components is also a constraint ?

But center tap biasing is a complete no. You will have to depopulate L203. Phy biases the center tap itself and it will create a short.

Also pin 1 and pin 6 can't be connected together and will need separate cap. (this one is not depopulation but addition of cap, for rest depopulating will do)

Regards,

Vikram

0 Frank Van Hooft1 over 4 years ago in reply to Vikram Sharma

Expert 1120 points

This is a great answer - I really appreciate it. Thanks.

OK, so clearly this PHY is a terrible choice for that old coupling circuit. I will look for a different 10/100 PHY. Do you have any suggestions?

Thanks again.

0 Vikram Sharma over 4 years ago in reply to Frank Van Hooft1

TI__Genius 16030 points

Hi Frank,

DP83822 is a fit.

Regards,

Vikram

0 Frank Van Hooft1 over 4 years ago in reply to Vikram Sharma

Expert 1120 points

Thanks for the suggestion. The coupling circuit shown in the DP83822 datasheet shows the 49.9 ohm resistors being tied to 3.3V. Our coupling circuit doesn't have that. Is that a concern?

Thanks again.

0 Frank Van Hooft1 over 4 years ago in reply to Vikram Sharma

Expert 1120 points

See post above

0 Vikram Sharma over 4 years ago in reply to Frank Van Hooft1

TI__Genius 16030 points

Hi Frank,

I think you are referring to wrong diagram. Please check the diagram with magnetics and not for fiber. Kindly recheck and let me know if you find anomaly.

Regards,

Vikram

0 Frank Van Hooft1 over 4 years ago in reply to Vikram Sharma

Expert 1120 points

I've copied the diagram from the DP83822 datasheet below, and circled the 3.3V connection. Our existing circuit (shown in the first post) doesn't have a 3.3V connection at that point. Is that a concern?

Thanks.

0 Frank Van Hooft1 over 4 years ago in reply to Frank Van Hooft1

Expert 1120 points

Hi Vikram,

What did you think of the post above? Thanks!

0 Vikram Sharma over 4 years ago in reply to Frank Van Hooft1

TI__Genius 16030 points

Hi Frank,

Ah! Now I see what you are talking about. Yes we will need the connections as mentioned in the datasheet. But I will double check the impact once again without center of resistor divider connected to supply. Will reply back tomorrow.

Regards,

Vikram

0 Vikram Sharma over 4 years ago in reply to Vikram Sharma

TI__Genius 16030 points

Hi Frank,

I checked with the team here and we still have doubt about the disconnection. Team wants to do a bit more analysis to confirm this. Do I have time till middle of next week?

Regards,

Vikram

0 Frank Van Hooft1 over 4 years ago in reply to Vikram Sharma

Expert 1120 points

Of course. Thanks very much!

0 Vikram Sharma over 4 years ago in reply to Frank Van Hooft1

TI__Genius 16030 points

Hi Frank,

We will need that extra connection to supply.

Regards,

Vikram

0 Frank Van Hooft1 over 4 years ago in reply to Vikram Sharma

Expert 1120 points

Thanks very much for checking Vikram. OK, so that PHY is not a winner. That's a shame. I've been looking at a few PHYs, it appears, unless you think otherwise, that TI doesn't have a PHY that can drive that circuit. Ah well. Thanks again for all the help.

0 Vikram Sharma over 4 years ago in reply to Frank Van Hooft1

TI__Genius 16030 points

Hi Frank,

Do you know from where these requirements of exact circuit coming from? Each phy may have different requirements.Even depopulation of components not allowed?

It is sad that for our phy to not be considered because of this as each of our 100mbps phys have unique advantages over competition.

Regards,

Vikram

0 Frank Van Hooft1 over 4 years ago in reply to Vikram Sharma

Expert 1120 points

The processor & ethernet PHY are on a plug-in board, that plugs into the base board which contains the ethernet magnetics etc. So the upgrade path is simply to unplug the processor / PHY board and plug in the new one. The base board is not changed. Hence the new processor / PHY board needs to be able to drive the existing magnetics circuit.

I agree, I would like to move to a newer PHY. But it's not looking possible, not only with TI PHYs but also not with other manufacturers PHYs as well. It looks like we'll actually have to reuse the old PHY on the new processor / PHY board. Crazy, but that seems to be the way.

Interface

Interface forum

DP83825I: Question about the network coupling circuit