DLPC230-Q1: Vcm problem for IBIS model

Hello Wenjun,

Thank you for reaching out and using the E2E forums! An expert from our team will review the thread and get back to you shortly.

Hi Wenjun,

Please allow an additional day. I will be talking more with our system engineer tomorrow, and he has provided initial insights below.

The input you have shown is not differential. Vcm is the common mode voltage of the LVDS differential input--not the Vid (differential voltage) input. Moving the input across the Vcm will not do anything.

The models you are using are single-ended--not differential. The LVDS receiver input IBIS model should be a differential receiver model that represents both P and N pins.

Regards,
Michael Ly

Hi Wenjun,

May we get additional information on what you are attempting to simulate? Were you attempting to simulate an input? We require more information on what your setup is and what you plan to accomplish before we can provide you more help.

Regards,
Michael Ly

Hi Michael,

I am simulating if the LVDS signal quality can meet Vcm(min_0.35V, typ_1.2V, max_1.6V) and VID(100~700mV) requirement which are defined in DLPC230 datashseet.

for the setting, as below figure1

• driver: SN65DSI83 LVDS data driver model
  • driver pattern: 300 ps rising edge, 1.428ns period, pulse pattern
• receiver: DLPC230 LVDS data receiver model
• channel: 5mm 100ohm transmission line

figure2 is differential input waveform, VID close to 800mV, violates 100~700mV requirement

figure3 is common voltage waveform, Vcm at steady state is ~2.4V, violate Vcm requirement. 

so seems receiver model causes the violations, especially for Vcm, so how to solve the model problem? thanks a lot.

figure1

figure2

figure3

Hi Wenjun,

It looks to me like you are still using single-ended inputs for the DLPC230 IBIS model?

Please take a look at the link provided by AMD-Xilinx and see if this helps any. Please let us know your results afterwards.

https://support.xilinx.com/s/article/955152?language=en_US

Regards,
Michael Ly

Hi Micheal,

I changed the receiver DLPC230 model to differential receiver, VID(~800mV pk-pk) and Vcm(2.4V) are still same as SE receiver and violate DLPC230 requirement, so how to solve this problem? is SN65DSI83 driver model not compatible with DLPC230 model? 

thanks

wenjun

Hi Wenjun,

I'll get together with my colleague again and see if he has any input on this. Please allow 2-3 days to get back to you on this.

Regards,
Michael Ly

ok, thanks!

Hi again Wenjun,

My colleague is unavailable until Monday my time. We're sorry for the further delay. I am also attempting to learn more about this from him so that we can better support you. I will continue to keep you updated on the matter and let you know if there are any further changes.

Regards,
Michael Ly.

Ok, looking forward to the update, thanks!

Wenjun,

Quick update: our system engineer will see if they can duplicate your simulation this week. He does not think you need R544 in your schematic, but let's wait to see what results he gets before we decide to remove or keep it.

Regards,
Michael Ly

Thanks! R544 is required by SN65DSI83 IBIS model, please help check.

Hi Wenjun,

Yes, you are correct! R544 is used for simulation accuracy to provide accurate output impedance of the LVDS Driver.

R544 should not be used in an actual board design, however.

Regards,
Michael Ly

Hi Micheal,

We just use R544 in simulation, not in actual board. 

and how about the Vcm problem? for our project, the actual board has not been manufactured, we need do simulation to evaluate if our design meet datasheet requirement, so we need solve Vcm simulation model problem.

Thanks

wenjun

Hi Micheal,

Any update on the Vcm problem?

thanks

wenjun

Hi Michael,

Thanks!

I tried ODT0 of DLPC230 model and add 100ohm external resistor, although the average of Vcm is 1.2V(meet requirement), there is big ringing on DLPC230, as below.

thanks

wenjun 

Hi Wenjun,

I am glad to see this is working for you, but I am sorry that ODT=1 does not seem to work as we expect. Does this help resolve your issue with this, though?

Regards,
Michael Ly