devop
Joined: 05 Dec 2006
Posts: 87
|
 30 Oct 2009 7:26 how to calculate the EM |
|
|
|
|
Hi :
how does the ICC calculate the EM problems??
go to lef to find the max current density limite,
then ....how to calculate a max density of a certain path??
thanks a lot
Added after 42 minutes:
| devop wrote: |
Hi :
how does the ICC calculate the EM problems??
go to lef to find the max current density limite,
then ....how to calculate a max density of a certain path??
thanks a lot |
To calculate current density, Astro first needs to run timing analysis to determine the transition time for each net. Once the current density is calculated, it is checked against the constraint to determine whether it is a violation. The constraint depends on the width of each metal segment and the temperature specified.
I think it's easy to calculate the current flow through VDD to GND(based on the lib),but how to calculate the current flow through a signal net, calculate base on the extracted RC model of net?I think it's a huge data.
can anyone help?
|
|
aliputa
Joined: 01 Nov 2009
Posts: 3
Location: China
|
01 Nov 2009 16:34 Re: how to calculate the EM |
|
|
|
|
as i known, there is hardly a EM violation on the signal nets. It more likely happens on the P/G nets, so, if we discussed some EM violation, it is often on the P/G grid.
As others friends said, the EM violation threshold will be determined by the PVT(Process, Voltage, temperature), and it just like a look-up table, and its Axis is the PVT.
if you want to calculate the signal EM with some information, there is a typical threshold value--"1mA/1um", but it is maybe old, you should know the information about the latest advanced process, it's a pity that i don't know these data.
so, i wish these could help you.
maybe there is some error, it‘s pleasure to be correct.
Added after 10 minutes:
| devop wrote: |
I think it's easy to calculate the current flow through VDD to GND(based on the lib),but how to calculate the current flow through a signal net, calculate base on the extracted RC model of net?I think it's a huge data.
can anyone help? |
sorry , i noted these words just now.
i have heard some information from a engineer from a EDA software company that "EM could be calculated by the metal width, switch activities, voltage(on this net)", you said the RC, yes ,if there is a constant voltage, the current will be determined by the R.
After calculate ,the value which have been computed will be compared with the EM violation Threshold(on some PVT).
and you say that is a huge work, i think you should know your chip clearly, you should know where is the path which could have a EM violation most likely. and you choose these path to be checked instead of all those path.(just like everyone should check the P/G nets for EM violation ,because we know the PG could have a EM violation most likely)
|
|
