The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> High-Speed I/O Design >> LVDS TX with pre-emphasis
https://designers-guide.org/forum/YaBB.pl?num=1214974938

Message started by nowicq on Jul 1^st, 2008, 10:02pm

Title: LVDS TX with pre-emphasis
Post by nowicq on Jul 1^st, 2008, 10:02pm

Hi all.

I have a question about the pre-emphasis of LVDS TX.

If the data rate is about 2Gbps, is it needed to use the pre-emphasis?

How about using large current (more than 4mA) without pre-emphasis?

I searched some papers about the pre-emphasis but I could not find any good examples.

If you have some advice on the pre-emphasis, please let me know.

Thank you.

Title: Re: LVDS TX with pre-emphasis
Post by loose-electron on Jul 3^rd, 2008, 11:56am

The intention of a "pre emphasis" system is to get, at the receiver, something that approximates a square wave.

Model the system with a backplane/cable/interconnect model and you will get degenration of the transient edges.

If you do a system without the PE, and just throw more current into the cable, the signal overall gets bigger but waveshape remains the same.

Title: Re: LVDS TX with pre-emphasis
Post by email_gz on Aug 27^th, 2008, 11:29pm

Hello:

"If the data rate is about 2Gbps, is it needed to use the pre-emphasis?"
The decision of using equalizer or not lies on the channel conditions around the communication bandwidth. If the channel loss is much more great than 3dB or there is significant ISI, then the channel equalizer is necessary.

"How about using large current (more than 4mA) without pre-emphasis?"

Big transmitter signal can also generate big ISI, so you still need equalizers if the channel isn't good enough.

"I searched some papers about the pre-emphasis but I could not find any good examples."
paper:
W. J. Dally and J. Poulton, “Transmitter equalization for 4Gb/s. signaling,”
JSSC and ISSCC, many,many......
theory book:
W. J. Dally, J. W. Poulton, Digital Systems Engineering, Cambridge University Press, 1998.
J. G. Proakis, Digital Communications, 4th Edition, McGraw-Hill, 2001.
"If you have some advice on the pre-emphasis, please let me know."
1: careful measure or model you physical channel.
2: make you pre-emphasis programable.

Title: Re: LVDS TX with pre-emphasis
Post by smarty on Sep 4^th, 2008, 10:04am

Hi,
The need for using pre-emphasis in driver depends on the signal loss which is going to occur in a transmission media, as pointed out by loose-electron

Ideally you want the signal transmitted to be recieved at the reciever end and not die out, in that case receiver will fail to sense that signal.

Increasing just current is not just going to improve the signal loss, but it is going to increase the area and power.

Efffectively to check your LVDS performance u should model all the components in the path like the package, board trace, connector and so to get the real feel of your driver design.

Also, these techniques are added in design to improve the ISI effect so just by doing a 1010 or a 1100 pattern will not show the effect. You might need to simulate using a PRBS pattern ( example: long sequence of 1's followed by a single 0 (111110111010000010000) to test the effect. (correct me if my understanding is different)

Thanks,
SBR

Title: Re: LVDS TX with pre-emphasis
Post by loose-electron on Sep 11^th, 2008, 12:03pm

In addition to the above -

The PCB trace needs to be a distibuted network with multiple RLC elements. Do not lump them all togther for the response will be different.

Data encoding for these devices can be in a run length limited (RLL)fashion (8B/10B is commonly used in these systems) or NRZ data (n run length limitations)

With the above, the patterns to test the system need to include a bunch of different possible patterns including isolated 1's in long strings of 0's and a number of other possible test cases.