Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
oh...just choose options after clickig on the tran setting and there u will see max step and minstep..if u set them both equal then the step size is always same..other wise the step size used by spectre can vary from minstep to maxstep..
I am very thanks for your reply. Actually, I find there are another option: strobeperiod. Is it the same function could let me change the step size? However, it could not work after I set it.
Do you mean that I should set the maxstep, minstep and strobeperiod in the same time in order to simulate the exact time step value and give out this reult ouput?
Thanks
wccheng
Added after 11 minutes:
Dear v_naren,
I have tried to go to the windows you mentioned. I just could find out step and maxstep. I could not find out minstep. Where is the minstep box?
Do you really necessary to set a fixed time-step? It is actually not recommend to fixed the time-step used in simulation, as the required time-step is depends on many factors (such as the signal changing speed, convergence difficulties, etc.). Fixing time-step will probably leads to either inaccuaries, or unnecssary long simultion time, or even non-convergence.
The way how to set the max step is changing based on what version of Spectre you have.
I would recommend to go to "Analysis setup" -> "Transient"-> "Transient Options"
there are 2 "TIme step parameters" - step - which I believe is the starting step size for simulation and "maxstep" maximum step size.
In case you set the maxtimestep your simulation will run much longer and will create more data on disk.
In most times you can avoid it by changing abstol/reltol or other parameters.
Also in later versions of spectre there are preset precisions for the simulation - check those.
Regarding putting the right time step, sometimes it is important to put the fixed time step. Convergence can be obtained by increased RELTOL or by following DPTRAN as homotopy.
My friend argued that if I set the strobeperiod = 1ns, it will calculate the value at 1ns and then give out the result. Is it correct? Or, it just gives out the result at 1ns without calculate it at 1ns before?
For example: I set the step = maxstep = 2ns. Then it will calculate the exact value at 0s, 2ns, 4ns, 6ns,......
At the same time, the strobeperiod = 1ns. It will give out the value at 0s, 1ns, 2ns, 3ns, 4ns,.......
My question is arisen. In the output result at 1ns, 3ns, 5ns......, does this result will calculate first and then give it out OR
it just estimate the result based on the result caluated between 0s and 2ns (I set in the step = maxstep = 2ns)? Which one is correct?
When you specify the strobe period, the simulator will specifically save all the points as per the period you have specified. For example, if you have given the strobe period to be 2ns, then the simulator will save the data for every 2ns.
But when you run the tran simulation, your tstep should be less than the strobe period. So, strobe period = N* tstep (where N is an integer). Or else the simulator might not give accurate values.
If I need to get the accurate value of every 1ns, I need to set the step = maxstep = strobeperiod =1ns? Then it will calculate and print it out the value in 1ns period?
OR
I just set the strobeperiod=1ns and leave step and maxstep box in empty. Then it will calculate the value at every 1ns period and save the output automatically?
You can give a very conservative value in step box and leave the max step box empty. These things are telling the simulator where the next time step should be.
While strobe will tell the simulator to save the values as per the strobe period given. This will help in data reduction and is particulary useful in mixed signal circuits where we can eliminate all the data points which are having a high frequency component. We particularly use it in Data converters where we want to run a fourier analysis or an FFT. Using strobe we remove the unnecessary points and only punch(save) points where the signal has settled. This gives better results.
You can look into the earlier post for the values of strobe and tstep you should give. I hope this helps.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
