typical values for IR drop and ground bounce simulation

[allennlowaton]

Hello EDAboard fellows,

I would like to ask for your help regarding the typical values of metal wire resistance and bonding wire inductor to be placed into the VDD or ground for the IR drop and ground bounce simulation.

Thank you for spending time reading this one.

 

[erikl]

Depends on quite a lot of things like current consumption, power routing strategy, no. of layers and power pins used, overall die size, a.s.o., you'll know probably. For preliminary simulations on schematic we used to use ±50mV IR drop at the power supply, and 1nH/mm for bonding wires.

R(L)C extraction will give you more trustworthy values. And from packaging companies you can get package and wire bond models.

 

[allennlowaton]

thank you for the reply erikl, would you mind if I will ask you
how to execute the ±50mV IR drop on HSPICE?
My supply runs from 2.7V to 5.5V.
And also the conservative length for the
bonding wire so that I can obtain an initial
value of the inductance for bonding wire.

 

[erikl]

There are several methods to consider the IR drop: the simplest (and least accurate) is to subtract 100mV from the power supply value (2.6 instead of 2.7V). A bit better is to use a series resistor (both with GND & VDD) which drops e.g. 50mV each for mean DC current. This allows you to consider noise resulting from peak currents.

Bond wire lengths differ depending on die mounting, package type and pad-to-pin distance, of course. For top-up mounted dice and DIL or Quad type packages, a medium bond wire length is around 3..5mm. I used 4nH in simulations, AFAIR. For top-down mounted dice this is less (≈ 1nH).

 

[Teddy]

Bond wire characteristics depend on material, length and diameter of the wire.
fo example 1.3mil Au wire with cca 5mm length will give about: L~0.4nH and Resistance~18mOhm
Those are measured values I usualy use. I do not have any Cu wire data.

 

[dgnani]

Bond wire characteristics depend on material, length and diameter of the wire.
fo example 1.3mil Au wire with cca 5mm length will give about: L~0.4nH and Resistance~18mOhm
Those are measured values I usualy use. I do not have any Cu wire data.

Your value for inductance seems off by an order of magnitude (rule of thumb is 1nH/mm) and also the estimated resistance is way too small

e.g. see **broken link removed**

 

[erikl]

... the estimated resistance is way too small

e.g. see **broken link removed**

These curves consider the RF resistance of the bond wire, i.e. they include their frequency-dependent L & C contributions.
Pure Ohms' (DC) resistance of a 5mm 1.3mil diameter Au wire is ≈130mΩ.

 

[allennlowaton]

thanks for the reply, erikl, teddy, dgnani...
My application is in power, specifically LED driver.
My frequency of operation is 1MHz..

 

[dgnani]

These curves consider the RF resistance of the bond wire, i.e. they include their frequency-dependent L & C contributions.
Pure Ohms' (DC) resistance of a 5mm 1.3mil diameter Au wire is ≈130mΩ.

Consider that the curve at 100MHz gives 120mOhm when you adjust for the increased cross section (data are for 1mil diameter so you divide by 1.3^2)

 

[leo_o2]

Cu wire will decrease dc resistance by 20%~30%.