The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> Simple RC filter time domain
https://designers-guide.org/forum/YaBB.pl?num=1172548450

Message started by aaron_do on Feb 26^th, 2007, 7:54pm

Title: Simple RC filter time domain
Post by aaron_do on Feb 26^th, 2007, 7:54pm

Hi all,

if i have a simple filter, e.g.

Vin ---vvvvv----------Vout
R |
|
C ==
|
|
GND

and Vin is a ramp with an AC signal on top. Does the filter have the correct characteristics before it reaches the DC solution? i.e. the ramp is still charging up the output node to the DC value, so is the AC response of the filter still 1/(1+RsC). I'm not sure if the filter example I chose is right (maybe too simple). I'm just asking for a general answer.

thanks in advance,
Aaron

Title: Re: Simple RC filter time domain
Post by aaron_do on Feb 26^th, 2007, 8:20pm

Let me just add a different circuit too cos i think the other circuit may be too simple.

Iin
GND----(->) ---------------------------Vout
| |
C1 == >
| > R2
GND >
| node Vx
C2 ==
|
GND

The circuit is such that node Vx takes much longer to charge up than node Vout. So does node Vout retain is steady state filtering characteristics when Vx is charging up?

Oh yeah one more thing...what simulation would i run in spectre to test this? I need the AC response at different time steps.

thanks,
Aaron

Title: Re: Simple RC filter time domain
Post by Andrew Beckett on Feb 27^th, 2007, 3:40pm

Not sure I quite get what you're asking here (perhaps because I couldn't quite make out the ASCII art circuit diagrams - you can upload png files, BTW). However, you can run a transient which runs an ac analysis at specific times. Look at the acnames and actimes parameters for transient. An example would be:

Code:

tran tran stop=100u acnames=[myac] actimes=[10u 20u 30u 40u]
myac ac start=1 stop=1G dec=20

Regards,

Andrew.

Title: Re: Simple RC filter time domain
Post by aaron_do on Feb 27^th, 2007, 8:48pm

Hi Andrew,

thanks for the reply. I've never really tried chaning the netlist for the simulation. I usually just set it up directly using the analog environment interface. Anyway I tried to change the netlist but I couldn't find any results. Could you elaborate on how I should go about this?

thanks,
Aaron

Title: Re: Simple RC filter time domain
Post by mg777 on Feb 27^th, 2007, 9:07pm

I get the sense that you are intuitively reinventing QPSS.

I'm pretty sure there will be no simulation performance gain for linear systems - proving this may require the equivalence between an FFT and inverting the Vandermonde matrix of natural response.

M.G.Rajan
www.eecalc.com

Title: Re: Simple RC filter time domain
Post by Ken Kundert on Feb 27^th, 2007, 10:23pm

None of this is necessary. Your circuit is linear time invariant. The AC analysis will produce the same result regardless of the time at which it is run.

-Ken

Title: Re: Simple RC filter time domain
Post by aaron_do on Feb 28^th, 2007, 2:11am

Hi all,

thanks for the replies. I was asking because my filter was not acting like a filter and I noticed that one of the nodes had not reached the DC value yet.

Just for future reference, how would I actually test the effect I was talking about anyway?

thanks,
Aaron

Title: Re: Simple RC filter time domain
Post by Andrew Beckett on Feb 28^th, 2007, 4:29am

I must admit I thought what Ken did - I couldn't see why you would do this - the response will be independent of the operating point, since it is linear time invariant.

Anyway, there's no need to do the actimes in the netlist. It can be done from the GUI.

If you have a recent enough IC version, setup the ac analysis as usual. Then on the transient options form, go down to the field actimes - and enter space-separated times. The acnames field will open up, and you would specify the names of the ac analyses you want to run at these times. So enter "ac" in that field (the ADE interface will call the instance name of the analysis the same as the name of the analysis itself - so enter ac, xf, noise etc).

If you're using an older version where acnames/actimes doesn't appear on the transient options, scroll down to the bottom of the transient options form, and in the additional options field, type:

Code:

acnames=[ac] actimes=[10u 20u 30u]

If you haven't got the additional parameters field (which appeared in IC5141 USR2), you'd have to go to a netlist-based approach (perhaps with an include file for the analyses). You'll need an MMSIM version of spectre to handle this.

Of course, with your circuit, you'll just see the same frequency response multiple times...

Regards,

Andrew.

Title: Re: Simple RC filter time domain
Post by aaron_do on Feb 28^th, 2007, 5:26am

thanks Andrew

I know it must have taken a while to get that answer down.

I get the point about the circuit being LTI (thanks Ken)... too many hours in front of the computer made my thoughts get a little too wild.

thanks all,
Aaron

Title: Re: Simple RC filter time domain
Post by mg777 on Feb 28^th, 2007, 6:42am

The reasoning in your original question is quite meaningful for a non-linear circuit. For example, add a diode in series with the source.

M.G.Rajan
www.eecalc.com