It depends on the frequency, fan out and slope of the transitions. You should simulate (all transitions ideally or random values maybe) and calculate the average energy per transition. You can calculate using the same values and methodology of the cells in the .lef of your library.

You can use the Calculator to obtain the average power or rms power for a mutliple number of cycles of the clock frequency. But you will have to simulate the circuit at two or more operating frequencies to get the static and dynamic power consumptions. In the simplest case, from just two operating frequencies, you get static and dynamic from solving the two equations and two variables problem: pt = ps + pd*freq (pt = total power, ps = static power, pd = dynamic power, req = operating frequency). In general, you will use the collected data for a least squares fitting (linear approximation method)   

I would rather prefer to include the average and rms power measurement in the simulation scripts (hspice), as accuracy of simulator is better than calculator. Note that calculator uses the simulation results aftet taking account of simulation accuracy parameters (simulation time step/resolution).  

Good luck,

There's no (easy) way to separate the dynamic power contribution in Spectre simulations. I assume you are simulating a digital CMOS circuit. In this case, if you can determine the average power and the static power separately, one possibility is to subtract the static contribution from the average power. This difference is the dynamic power contribution.

As static power is very less, the subtraction from avg. power will always give avg. power (Avg. Power is almost same as dynamic power). But can this validate the results calculated from the normal power expressions ?

for e.g. In the dynamic power calculation Load capacitor is one of the parameter. This is the capacitive load present due to the circuits present after the test circuit or a physical capacitor connected at the output ?

Thank you all for giving your valuable suggestions.

@Víctor Navarro-Botello, I'll follow the steps provided for the calculation. I have one more query (How to calculate the total power and How it is different from avg. power ?).

It depends on how complex your circuit is. If you only have an inverter loaded with Cload, let's say, then your power formula (e.g. 1/2 * alpha * Cload * Vdd ^ 2 * Frequency)  will be very close to what you get as average power (assuming static power is negligible, as you wrote). If it is more complex, e.g. a set of cascaded inverters with a single Cload at the output of the last one, then you cannot use the same formula, as you will be ignoring the contributions of the internal capacitances (e.g. gate capacitances of all the intermediate inverters).

For full adder two outputs are there (Sum and Carry). I have connected two capacitors(one for each). But after the layout considering Parasitics more than 200 internal capacitances are present. Now it is very difficult to calculate the Cload and power expression theoretically. Also most of the calculated parameters got doubled from that of schematic value.

Subtract static power from total power in order to compute dynamic power. i.e Dynamic Power = Total Power - Static Power.

Exactly Sandeep, this is why you need simulations and cannot use power expression (at least not that easily).

Sandeep, the total power is the measured average or rms power from your simulations. Note that for a good static power estimation, you must simulate the circuit at very low frequency of operation. And to avoid high errors, the highest frequency you must use for the dynamic power estimation must assure all the transitions have time enough (this is, is far enough from maximum operating freq). If you don't take this assumptions, it could lead to variable instead of constant dynamic power consumptions vs freq, which are not suited for linear approximation. 

Note also you can work with both, average or rms power values, and are not the same: avg power is the power consumption, while rms power is more related to power dissipation (heat, charge movement, or energy dissipation). Capacitive loads will exhibit near zero average power consumption, because it consumes reactive power but not active power. But in terms of rms power, it will exhibit power consumption. So, avg power is a measure of the active power consumption (real part), while rms power better suits a metric for both active and reactive power consumption.

Each transistor on a chip dissipates a small amount of power when it is switched, and transistors that are switched rapidly dissipate more power than transistors that are switched slowly. The total amount of power dissipated per unit area due to switching of a chip's transistors is called dynamic power density. There are two factors that work together to cause an increase in dynamic power density: clockspeed and transistor density. IEEE Journal on Low Power has so many papers related to the topic. For dynamic power it is important to device the conditions under which you want to measure it. In particular you need to decide on a data/clock rate. Then you need to be careful to ensure that the time period you average over is sensible. For example, if you clock at 1MHz you could measure over 1.5us and 2.5us and you will get a different answer. You will get a different answer if you arrange there to be one or two transitions within the measuring period even if the measuring period is the same.

Sandeep, simulate your work with zero activity factor (i.e. no switching signals). The power (current) dissipation you observe is static power. Then you can subtract it from your actual measurement to observe dynamic power.

If you need dynamic power in real time, then export csv files, subtract values manually and plot the resultant values either manually or by import (never did this but it should work).

Edit (Credit: Victor-Navarro-Botello): Counting on the probabity factor of each input combination at a steady state, you can extract an 'average' static power dissipation that would provide a more accurate value.

Furkan, just a comment about what you said: when simulating with zero activity factor, do not forget to simulate each state / static condition (for example, low and high output conditions). Each state will typically have a, associated probability factor. The average static power will be the weighted value from all observations, taking into account the probabilities.

You can also refer as static power consumption associated to each state. 

@ Furkan Ercan...I already did exactly you said (Manual Subtraction) considering all the possible states....I think the result from this method is correct but could not validate it as I have no method for the validation. One way is to compare it with other simulators.

In the first method static power can be calculated but as actual measurement is not giving the total power, just plots it at certain steps, It is up to us how to measure it. 

Use power meter simulation as quoted in Kang, Weste and Rabaey books 

1

2