[sdiy] RC filter questions + SPICE RC filter simulation questions

[cheater cheater]

On Fri, Apr 16, 2010 at 18:38, Neil Johnson <neil.johnson97 at ntlworld.com> wrote:
> Hi,
>
> cheater cheater <cheater00 at gmail.com> wrote:
>> Hi guys,
>> I have been playing around with pspice today, simulating and analyzing
>> some very simple circuits.
>>
>> I analyzed a simple RC filter: AC voltage source goes through R=1kR
>> and C=1uF to ground. I analyzed it and got the familiar curve.
>>
>> Then I stuck some more stages like that on it: there's a series of
>> four 1kR resistors and after each one there's a 1uF cap to ground. I
>> analyzed it and got the expected curve at the last stage, but the
>> first stage has a distinctive bump. Additionally the cutoffs of all
>> the stages are much lower. Where does the bump come from? Is there a
>> specific law that defines this?
>
> You're analysing intermediate stages in the whole circuit so expect the responses to be different from single circuits.  In your extended circuit =
>
> R1 + (C1 || (R2 + (C2 || (R3 + (C3 || (R4 + C4))))))
>
> (where '||' denotes parallel connection and '+' denotes series connection).

yep, I understand in full that the two circuits are not supposed to be
the same. I'm just wondering why a bump, and not something else... say
a ripple, or two bumps :-)

> As for the bump... on this website:
>
> http://www.logbook.freeserve.co.uk/seekrets/index.html
>
> look at the picture of the back cover and have a go at simulating that circuit.
>
> That should entertain you :)

I see a low-pass filter. There's indeed a bump on one side of the top
capacitor and not on the other. What more can you tell me about this?

I'll check out the book, thanks a lot for the link, looks cool.

>> I am trying to find a way to filter a .wav through the filter and look
>> at the voltage output. Is this possible to do in pspice? If not, do
>> any other spice packages allow that?
>
> LTSpice.  Can do both wav in and wav out.

will check it out. I have installed it yesterday already, but I have
found pspice easy to use so hadn't bothered with ltspice yet. In the
meantime I've found a 'piecewise linear voltage source' which allows
loading stuff, but it needs to be an ASCII file of some stupid format
(see http://www.orcad.com/forums/ShowPost.aspx?PostID=13872) rather
than a normal wav file, so I guess ltspice wins this one by far.

Scott,
> If you're expecting a filter with 4 poles at 1/(2*pi*RC), that's not what
> you've built.  It's not 4 independent stages, it's all one circuit, and a
> measurement taken at any point will be affected by the entire rest of the
> circuit.  You've got 4 poles, but they're all at different frequencies,
> resulting in a rather flaccid transfer function.

Hmm, that's what I expected to some degree, but rather by guessing
than making an informed prediction - after I saw the different
transfer function I thought, yes, of course, the further sources are
sucking current away. Thanks for reassuring that. But what surprised
me is the shape of the transfer function, which is funny. I expected a
curve the slope of which changed monotonically - in this instance, I
expected the transfer function to become steeper and steeper the
higher above cutoff we go. In fact, I expected at least the first few
derivatives to be concave, to use mathematical wording (maybe the
first 2-3).

Aaron,
> I worked out the transfer function of this sort of circuit a few years ago; it was a huge pain and took multiple pages. If I recall correctly, Tim Stichcombe did the same analysis, and fortunately my result matched his:
>
> http://www.timstinchcombe.co.uk/synth/Moog_ladder_tf.pdf
>
> See p. 34

Great answer - thanks! BTW, do you still have your own analysis up
somewhere? It would be cool to have two sources comparison :-)

Cheers,
D.