Designer’s Guide Community

topquark

Community Member

Offline

Posts: 61
Thames Valley, UK

Wavescan hysteresis measurement
Dec 07^th, 2006, 3:30am

Hi folks!

I need to measure hysteresis voltage on the Wavescan waveforms obtained using Spectre calculator.
How can this be done?
I've attached a plot where I've manually put markers on the high and low transitions of Vout. The difference gives a measure of Vhys.
I guess this can be done somehow on the calculator to find mutual crossing points of the 2 waveforms.

Any suggestions?

Thanks,
Gautham

Back to top

hysteresis.png

IP Logged

topquark

Community Member

Offline

Posts: 61
Thames Valley, UK

Re: Wavescan hysteresis measurement
Reply #1 - Dec 11^th, 2006, 7:03am

Folks,
I did some work-around for the above problem-
I plot "Vout -Vin" curve, call it Vdiff. Hysteresis is then given by the value of Vout at the zero-crossings of Vdiff.
Now, let Vhi=Vout(at 1st zero-crossing of Vdiff) and Vlo=(at 2nd zero-crossing of Vdiff)
So, Vhys= Vhi-Vlo

This set-up works fine at a given temperature.
My goal is to study Vhys variation as temp. varies. Now, when I run parametric sweep I am unable to get correct calculation of Vhys simply because the calculator calls "Vdiff waveform" by different names like wave_{number}() where number is arbitrary and changes for each run. So, the formula for Vhys is simply not valid for all runs.

How do I resolve this? I'm sure people have a smart way of measuring hysteresis from waveforms. Please enlighten the newbie :)
Highly appreciate any suggestions.

Regards,
Gautham

Back to top

hyst_diff.png

IP Logged

ACWWong

Community Fellow

Offline

Posts: 539
Oxford, UK

Re: Wavescan hysteresis measurement
Reply #2 - Dec 11^th, 2006, 7:17am

You are selecting Vout and Vin as waveform objects (using wave) when you should select them as VT expression (using tran vt).

So your Vdiff waveform should be an expression like:
VT("/Vout")-VT("/Vin")

using the expression for Vdiff in any additional calculator functions (like xval, cross etc.) you need to get the number you want should now work over parametric runs.

cheers
aw

Back to top

IP Logged

topquark

Community Member

Offline

Posts: 61
Thames Valley, UK

Re: Wavescan hysteresis measurement
Reply #3 - Dec 12^th, 2006, 6:06am

Thank you Mr. Wong.
The setup with "Vt" selection of node voltages rather than as wave objects works fine.

Another small hitch-
Say, I define Vhi as said earlier-
value(VT("/Vout2") cross(clip((VT("/Vout2") - VT("/Vin")) 6e-06 8e-06) 0 1 "falling" nil nil))
and similarly, Vlo..
Now, if I define another variable Vhys=Vhi -Vlo
I find that Vhys is not being calculated, but if I give the above expressins explicitly again, it works..
why is this so?

Thanks !
Gautham

Back to top

IP Logged

ACWWong Community Fellow Offline Posts: 539 Oxford, UK	Re: Wavescan hysteresis measurement Reply #4 - Dec 12^th, 2006, 8:00am you need to explicitly give the expression so Vhys should be (and setup in the outputs section of ADE as): value(VT("/Vout2") cross(clip((VT("/Vout2") - VT("/Vin")) 6e-06 8e-06) 0 1 "falling" nil nil)) - value (etc....)
Back to top	IP Logged