08-25-2016 04:18 AM
I am using Kintex 7 FPGA (XC7K160T-1FFG676C) along with its IBIS model downloaded from Xilinx website (kintex7_ibis_v2_0).
I'm using this model in Hyperlynx. Generally, with higher drive strength, we generally have a shorter rise time and a lesser output impedance. I can see this for LVCMOS18, slow HP.
For the particular case of (LVCMOS18, slow HR), I am getting identical information for drive strengths for 8 mA as well as 12 mA.
Rise Time (20-80) = 569 ps
Output Impedance = 30 ohms
Is this an error of the model?
08-25-2016 04:23 AM
08-25-2016 07:17 AM
No. All models are from hSpice(TM Synopsys) of the actual IO circuits converted to IBIS models, using TSMC models for process. They are verified by the characterization team before production.
It is likely that once 'strong enough,' you see no more 'improvement' in the metrics mentioned. Not at all unusual.
08-25-2016 08:18 PM
I think I have got my engineering intuition off by a little! Hope you will guide me on this!
I thought that having a bigger drive strength would decrease the rise time. I did some simulations. I just took some drivers and tested their rise times without connecting them to any circuits. I then connected them to a simple resistor circuit to get the output impedance. I did it for virtex 6 (XC6VLX75T-1, FFG484) as well as kintex 7(XC7K160T-1, FFG676). All the rise times (RT) are (10-90)%.
I am sharing this information here:
1) LVCMOS18_SLOW - 8mA => RT = 570 ps; Zout = 42 ohms
2) LVCMOS18_SLOW - 12mA => RT = 388 ps; Zout = 29 ohms
3) LVCMOS25_SLOW - 8mA => RT = 580 ps; Zout = 43 ohms
4) LVCMOS25_SLOW - 12mA => RT = 396 ps; Zout = 30 ohms
5) LVCMOS18_SLOW_HP - 8mA => RT= 762 ps; Zout = 46 ohms
6) LVCMOS18_SLOW_HP - 12mA => RT= 533 ps; Zout = 29 ohms
7) LVCMOS18_SLOW_HR - 8mA => RT= 881 ps; Zout = 30 ohms
8) LVCMOS18_SLOW_HR - 12mA => RT= 881 ps; Zout = 30 ohms
Except the case in #8, if we see the cases, drive strength of 8 mA is giving shorter rise time and smaller output impedance.
Hence I got confused for scenario #8 of LVCMOS18_SLOW_HR for the Kintex device IBIS model. Have I done something wrong here?
08-25-2016 09:02 PM
Sorry for the error in the last post.
I wanted to say that the in all cases except in the scenarios of #7 and #8, drive strength of 12 mA is giving shorter rise time and smaller output impedance compared to those with the drive strength of 8 mA.