Spartan6 programmable impedance termination accuracy

[nagaraj.s01@gmail.com]

Hi,

I learnt from Spartan6 ug381.pdf >>

"The Spartan-6 FPGA programmable output driver impedance feature is similar to the DCI feature available in many other Xilinx FPGA families. However, an important difference is that the programmable output driver impedance is not calibrated, and therefore some variation to the target output impedance value will exist across process, temperature, and voltage variations."

I'm curious to know the actual variation for UNTUNED_25 / UNTUNED_50 / UNTUNED_75. Where can I find the range?

Thanks.

[austin]

n,

In the IBIS models, the SLOW/WEAK and FAST/STRONG corners are the nominal (expected) one sigma variations.

[nagaraj.s01@gmail.com]

Hi @austin,

Sorry, it is not clear to me how large is that one-sigma variation? For UNTUNED_50, is 30ohms one sigma? or 49ohms one sigma? How about for UNTUNED_25 and UNTUNED_75?

--

Regards,

nagaraj.s01@gmail.com

[nagaraj.s01@gmail.com]

Hi @austin

I came across this spec from ds162.pdf for ROUT_TERM:

"

Description Min Typ Max Units

Thevenin equivalent resistance of programmable output termination (UNTUNED_25) 11 25 52 Ω

Thevenin equivalent resistance of programmable output termination (UNTUNED_50) 21 50 96 Ω

Thevenin equivalent resistance of programmable output termination (UNTUNED_75) 29 75 145 Ω

"

Are these Min and Max values 1-sigma numbers or are they 3-sigma numbers? Fyi, I am not going to stress the fpga under any Voltage or Temperature conditions. I'll be using it under nominal conditions.

--

Thanks.

[austin]

Yes,

I did not realize it was also in the data sheet.  Those numbers are likely worst case.

The IBIS models would be more reasonable I believe.

Generally, in a digital system a match of 1:2 TO 2:1 is acceptable.

[nagaraj.s01@gmail.com]

@austin

Thanks for the reply.

Could you please explain how I should launch IBIS models and how do I find the actual value in the IBIS models? I am using Xilinx ISE webpack 14.7.

Regards,

-Nagaraj

[nagaraj.s01@gmail.com]

Again, it is not clear to me if the spec values are worst case one-sigma or 2-sigma or 3-sigma numbers and what is the standard deviation (sigma)?

[austin · Accepted solution]

Data Sheet numbers are 3-sigma,

If it is more or less, you get a free part.

The IBIS models are here:

https://www.xilinx.com/support/download/index.html/content/xilinx/en/downloadNav/device-models/ibis-models/spartan-series-fpgas.html