DS125BR800: DS125BR800: IBIS-AMI

Jakob Wicht

Part Number: DS125BR800

Hello,

I am simulating the DS125BR800 redriver IBIS-AMI Model in a 8 Gbps setup using HSpice.

While I got the AMI models running, simply assining them to port-elements in HSpice, I have problems using the IBIS-part of the model for the RX-input in order to calculate a realistic impulse response.
Due to the datasheet we have 50-Ω termination resistors connecting INA_n+/- to VDD and a Signal detection level at 180 mVp-p.
But if I have a look into the "DS125BR800_ramp.ibs" file at "[model] RX_IN" I am missing rising/falling waveforms which could describe its input impedance. Further there are the voltage levels defined as Vinl = 0.8 and Vinh = 2.
Is the model incomplete/wrong regarding the RX part? If I simulate my signal path with just the behavioral AMI, but without the RX-IBIS model, is it sufficient to simply assign two 50 ohms termination resistors?

Kind regards,
Jakob

over 5 years ago

0 Brian Zhou over 5 years ago

TI__Guru** 119995 points

assign to right person

0 Lee Sledjeski over 5 years ago

TI__Mastermind 43795 points

Hi Jakob,

I do not know how to simulate an AMI model with HSPICE. We have done our model work using tools such as ADS to verify proper model operation. The characteristics of the model are embedded within the AMI code. The ramp voltage would not be used for the high speed simulations.

I would add a 2x50 ohm input differential termination if HSPICE does not recognize the termination impedance in the simulation.

Regards,

Lee

0 Jakob Wicht over 5 years ago in reply to Lee Sledjeski

Prodigy 40 points

Hello Lee,

thank you for your fast reply! I replaced the IBIS RX model by differential termination and the results are looking plausible. But there is one point that irritates me:

In the datasheet of the DS125BR800 is written that the differential output voltage VOD is configurable from 0.7 Vp-p to 1.4 Vp-p. But sweeping over VOD in HSpice and Hyperlynx both results show a voltage swing from around 400 mV to 700 mV, which is almost the half of it. I used a very basic setup and the other parameters are kept fix (Limit=1, EQ=0, DE=0), so is there an explanation for this difference I have not in mind?

Thank you and regards,

Jakob

0 Lee Sledjeski over 5 years ago in reply to Jakob Wicht

TI__Mastermind 43795 points

Hi Jakob,

It is likely something in the loading. Can you send a picture of the schematic you are simulating. I get ~ 600mV to 1.3V using ADS.

Regards,

Lee

0 Jakob Wicht over 5 years ago in reply to Lee Sledjeski

Prodigy 40 points

These are the schematic and results (directly on pins 28, 29) in Hyperlynx:

0 Lee Sledjeski over 5 years ago in reply to Jakob Wicht

TI__Mastermind 43795 points

Can you take out the package models and insert a simple 100 ohm differential termination at the receiver to make a comparison?
Thanks and Regards,
Lee