AFE5832: IBIS Model Validity

Hi Charl,

Thanks for reaching out.

The device's IBIS model is used by many of our customers but so far we havent heard anything about this issues.

Can you elaborate as to what exactly is the issue.

By the way, the AFE5818 and AFE5828 also has the same output interface. Can you try with AFE5818 model and let me know if you face similar issues.

You can find the model here:AFE5818 data sheet, product information and support | TI.com

Thanks & regards,

Abhishek

Hi,

When I drive one of the LVDS outputs to a 50ohm testload I get a 70mv Differential swing,

See below my setup:

Hi,

I tried simulating it. I think the mistake is the way load is connected to the buffer. Can you connect differential 100 ohms? Device doesnt support 50 ohms to ground.

Results:

Regards,

Shabbir

Thanks Shabbir,

Now I am driving this model into a LVDS18G Receiver on a Polarfire FPGA, 100ohm terminated using an ideal transmission line and get the following:

Any Idea why the non-monotonic signal response comes from?

Link to IBIS model for Polarfire:

https://www.microsemi.com/document-portal/doc_download/1243538-polarfire-gpio-industrial-model

Can you show your circuit used for simulation?

Regards,

Shabbir

Hi,

I tried simulating using ADS and I observed ringing. I feel it might be due to package parasitic of Polarfire LVDS input buffer. In the model i couldnt find an option to place termination inside the buffer. That might help. Can you check with Polarfire team about it more? Since placing 100 ohms termination alone is not showing an issue, it is not due to AFE IBIS model.

Regards,

Shabbir

HI,

There is a model with ODT-100 see the below model selector:

Then about the ringing. I am not so sure it is on the PolarFire Side.

If I drive the Same buffer wusing an Virtex 4 Output buffer I get no ringing:

See below my result comparison. Orange Waveform is for Vertex as driver and red waveform is for AFE as Driver. Both cases the PolarFire is receiving with LVDS18G_ODT-OFF_HYST-OFF.

Hi,

I tried simulating 100 ohms model too and saw same result. Can you check the below cases at your end? Simulate without any termination and then simulate with PolarFire LVDS input buffer. If there is internal termination in Polarfire buffer then  I must have seen reduction in swing.

Below is simulation result: Blue is without connecting the LVDS buffer and red is with buffer. Same swing is observed. can you check same at your end?

Regards,

Shabbir

Here are my results:

Simulated at 100Mhz 50% Duty.

See alot of ringing when driving the Polarfire Buffer.

So here if you see, the voltage swing is same for two cases. Which means the internal termination is not available. Whenever device is offered a termination, swing will become half. This ringing might be due to parasitic passives of package and input network.

Thanks!

Regards,

Shabbir

Seems like you are right. I now changed the parasitic values for the polarfire Ibis model to "0" and getting a far better result: