RMS and AVERAGE Calculation with respect to power.

Hi,

I am interested to know how spectre calculates the dc(avg), peak and rms currents , and how the average current can be negative ?

I have just used the keyword "rms" and " em_avg" in the .meas statement. Please provide some insights on this.

Thanks,

This has very little to do with the original question, which is one reason why the forum guidelines ask you not to post on the end of old threads.

It also would have helped if you'd given the actual .measure statements you're using. Note that we primarily support .meas/.measure statements for compatibility with other SPICE simulators; the normal way to do this with Spectre would be to use Spectre MDL or ADE to compute the quantity. I'm assuming you have something like this simple spectre netlist:

//

I1 (n1 0) isource dc=1m type=sine freq=1M ampl=2m
R1 (n1 0) resistor r=1k

tran tran stop=2u

simulator lang=spice
**.option em_recovery=1k
.meas tran av1 rms I(R1)
.meas tran av2 avg I(R1)
.meas tran min1 min I(R1)
.meas tran max1 max I(R1)
.meas tran av3 em_avg I(R1) em_rec=1

If I do that, I get:

Measurement Name : tran
Analysis Type : tran
av1 = 0.00173168
av2 = -0.001
av3 = -0.001
max1 = 0.000997474
min1 = -0.00299432

In my results. The RMS is obviously always positive, but the avg isn't because it calculates the area under the curve in question divided by the interval (in this case the whole simulation since I didn't specify from and to in the .meas statements), and if you have more negative regions than positive the total area may end up being negative. In a definition of em_avg I found in a request for support for this .meas syntax, it says:

• Measure the average of positive part of the waveform (when I(t) > 0): "pos"
• Measure the average of negative part of the waveform: "neg".
• final em_avg current is: max(pos,neg) - EM_REC * min(pos,neg)
• the polarity of em_avg current is same as the polarity of the max(pos,neg), i.e. Positive, if abs(pos) is more than abs(neg) and otherwise.

Regards,

Andrew.

Thanks, Andrew for the information but still I have one query.

How spectre calculates average current for improper waveforms, like if the waveform don't have any standard structure(triangular,square...) , and it has many negative and positive peaks.

Does PSTEP value is used calculate , and also correct me if I am wrong, suppose " PSTEP = 1e-10" it takes the value of that required parameter in this instants of time.

Thanks,

Spectre doesn't really use "PSTEP". It does support it as an option, but it only affects the "print" or .print statement. Spectre always computes time steps based on the local truncation error, which means that it dynamically adjusts the time steps to ensure it follows the curves with sufficient accuracy. The average would be computed by integrating the area under the curve (using a trapezoidal integration, most likely) and then dividing by the interval. 

So I don't think PSTEP is relevant (and I'm not sure why you'd set it, presumably because you've got a netlist from another SPICE simulator?)

Andrew.