Measuring S - Parameters in Spice

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
Hi all,

I flicked through Google last night and saw some quite simple set-up
whereby, given an accurate enough transistor model, one might measure
S - parmaters for any device under consideration purely via
simulation. Is this feasible? Or is it really only possible to
reliably do this empirically with real components?

Thanks,

p.
--

"What is now proved was once only imagin'd." - William Blake, 1793.


Re: Measuring S - Parameters in Spice
> Hi all,
>
> I flicked through Google last night and saw some quite simple set-up
> whereby, given an accurate enough transistor model, one might measure
> S - parmaters for any device under consideration purely via
> simulation. Is this feasible? Or is it really only possible to
> reliably do this empirically with real components?
>

I dont follow what you are asking for. A bit of confusion between
"measurement" and "simulation" here.

*All* the parameter sets (h, s, abcd etc) are mathematicly *identically*
equivalent. A far as simulation goes, it don't care a toss what set you
use.

S - parameters came into use for R.F, because it is easier to *measure*
S parameters in the real world at high frequencies. By and large, S
parameters themselves, in my opinion, are a pain in the arse. If you can
get a good spice model at h.f, you are better using it, especially,
because this will allow one to do transient simulation as well.

Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.




Re: Measuring S - Parameters in Spice
On Fri, 06 Aug 2004 14:38:22 GMT, "Kevin Aylward"

>> I flicked through Google last night and saw some quite simple set-up
>> whereby, given an accurate enough transistor model, one might measure
>> S - parmaters for any device under consideration purely via
>> simulation. Is this feasible? Or is it really only possible to
>> reliably do this empirically with real components?

>I dont follow what you are asking for. A bit of confusion between
>"measurement" and "simulation" here.

I think you know. I've since Googled a bit more and found that
"gwhite" (can't recall his first name off hand, but he's the author of
the VHF/UHF DX Handbook) posted to one of the radio amateur groups
that he'd had more accurate results 'measuring' S-parameters from
Spice simulations than the manufacturers provided in their datasheets.
He was able to 'adjudicate' between the sheets and Spice by virtue of
having access to a decent VNA.

>*All* the parameter sets (h, s, abcd etc) are mathematicly *identically*
>equivalent. A far as simulation goes, it don't care a toss what set you
>use.

True enough.

>S - parameters came into use for R.F, because it is easier to *measure*
>S parameters in the real world at high frequencies. By and large, S
>parameters themselves, in my opinion, are a pain in the arse.

Indeed? I'm surprised. Measuring them might be a pain in the arse, but
once you have them, they do make it much easier to visualise any
mis-match when subsequently plotted on Mr. Smith's chart.

>If you can
>get a good spice model at h.f, you are better using it, especially,
>because this will allow one to do transient simulation as well.

Whatever, but what I'm trying to arrive at is a circuit simulation
which will enable the modulus and phase-angle components of reflected
waves to be displayed via Spice. If you or anyone else can come up
with an accurate way of doing so, then I'd be interested to hear about
it.

p.
--

"What is now proved was once only imagin'd." - William Blake, 1793.


Re: Measuring S - Parameters in Spice

>
> Whatever, but what I'm trying to arrive at is a circuit simulation
> which will enable the modulus and phase-angle components of reflected
> waves to be displayed via Spice. If you or anyone else can come up
> with an accurate way of doing so, then I'd be interested to hear about
> it.

Easily done.

The biggest problem is the S Parameters vary with bias on the Transistor.
But you should get reasonable results at different bias points off a good
Spice Model.

I have an old 2 page App Note from Wes Hayard (W7ZOI) in DOC format that
describes a setup that will work in all the various Spices. If you provide
an address (suitably camouflaged) I'll send it.

But essentially you drive the device you want to measure the S Parameters
with a 2 volt AC source through a 50 ohm series resistor. The other end of
the source is tied to ground. That is, if the impedance of your system is 50
ohms.

In shunt with this connection to the device is one side of another 1V AC
source, set so it subtracts 1V of the applied 2V signal on the side not
connected to the device input.

That side of the 1V AC source is tied to ground through a high value (1Meg)
resistor.

Then the voltage at the top of that high value resistor will be S11.

If you tie the output connection of the device you're trying to measure the
S Parameters to 50 ohms to ground (use a coupling cap if necessary) then the
voltage at the top of that 50 ohm load to ground will be S21.

Turn this around to measure S22 and S12.

Robert




Re: Measuring S - Parameters in Spice
Robert, Robert, Robert.... did you really write this article
(sci.electronics.cad) at the date of (Sat, 07 Aug 2004 22:39:02 GMT)?

[S Parameters in SPICE]
> I have an old 2 page App Note from Wes Hayard (W7ZOI) in DOC format that
> describes a setup that will work in all the various Spices. If you provide
> an address (suitably camouflaged) I'll send it.

I am interested in this app note, can you send it to me?

renan_tdb AT yahoo DOT com DOT br

[]s
--
Chaos Master®, posting from Brazil. REPLY TO GROUP!
MSN: wizard_of_yendor[@]hotmail[.]com http://marreka.no-ip.com
We've slightly trimmed the long signature. Click to see the full one.
Re: Measuring S - Parameters in Spice

>I have an old 2 page App Note from Wes Hayard (W7ZOI) in DOC format that
>describes a setup that will work in all the various Spices. If you provide
>an address (suitably camouflaged) I'll send it.

Sounds just what I'm looking for. Thanks, Robert. Address is
pb@osiris1.co.uk with no mods needed to reach me. I just don't care!

Best regards,

paul
--

"What is now proved was once only imagin'd." - William Blake, 1793.


Re: Measuring S - Parameters in Spice
> On Fri, 06 Aug 2004 14:38:22 GMT, "Kevin Aylward"
>
>>> I flicked through Google last night and saw some quite simple set-up
>>> whereby, given an accurate enough transistor model, one might
>>> measure S - parmaters for any device under consideration purely via
>>> simulation. Is this feasible? Or is it really only possible to
>>> reliably do this empirically with real components?
>
>> I dont follow what you are asking for. A bit of confusion between
>> "measurement" and "simulation" here.
>
> I think you know. I've since Googled a bit more and found that
> "gwhite" (can't recall his first name off hand, but he's the author of
> the VHF/UHF DX Handbook) posted to one of the radio amateur groups
> that he'd had more accurate results 'measuring' S-parameters from
> Spice simulations than the manufacturers provided in their datasheets.
> He was able to 'adjudicate' between the sheets and Spice by virtue of
> having access to a decent VNA.
>
>> *All* the parameter sets (h, s, abcd etc) are mathematicly
>> *identically* equivalent. A far as simulation goes, it don't care a
>> toss what set you use.
>
> True enough.
>
>> S - parameters came into use for R.F, because it is easier to
>> *measure* S parameters in the real world at high frequencies. By and
>> large, S parameters themselves, in my opinion, are a pain in the
>> arse.
>
> Indeed? I'm surprised. Measuring them might be a pain in the arse, but
> once you have them, they do make it much easier to visualise any
> mis-match when subsequently plotted on Mr. Smith's chart.

I don't use Mr. Smith, that's an even bigger pain in the arse.

>
>> If you can
>> get a good spice model at h.f, you are better using it, especially,
>> because this will allow one to do transient simulation as well.
>
> Whatever, but what I'm trying to arrive at is a circuit simulation
> which will enable the modulus and phase-angle components of reflected
> waves to be displayed via Spice.

Why? These "waves" do not "exist", i.e they are imaginary, in a lumped
component model such as a spice one.



Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.




Re: Measuring S - Parameters in Spice
On Sun, 08 Aug 2004 06:28:15 GMT, "Kevin Aylward"

>Why? These "waves" do not "exist", i.e they are imaginary, in a lumped
>component model such as a spice one.

I know what you mean, Kev, but this is the very reason I'm posting the
question. It *is* possible to contrive a simulation to establish a
device's S -parameters in Spice, notwithstanding real-world
distributed reactances. I simply would like to know *how* it's done!
--

"What is now proved was once only imagin'd." - William Blake, 1793.


Re: Measuring S - Parameters in Spice
The article below shows a good method to measure the S-parameters
using Micro-Cap. It should work with any simulation programme.

http://www.spectrum-soft.com/news/winter2001/sparameters.shtm

mehran



> On Sun, 08 Aug 2004 06:28:15 GMT, "Kevin Aylward"
>
> >Why? These "waves" do not "exist", i.e they are imaginary, in a lumped
> >component model such as a spice one.
>
> I know what you mean, Kev, but this is the very reason I'm posting the
> question. It *is* possible to contrive a simulation to establish a
> device's S -parameters in Spice, notwithstanding real-world
> distributed reactances. I simply would like to know *how* it's done!


Re: Measuring S - Parameters in Spice
On 12 Aug 2004 08:39:20 -0700, mehran_gupta@yahoo.com (mehran gupta)

>The article below shows a good method to measure the S-parameters
>using Micro-Cap. It should work with any simulation programme.
>
>http://www.spectrum-soft.com/news/winter2001/sparameters.shtm
>
>mehran

Many thanks, Mehran. That's just the kind of info I'd be trawling the
'net for and been unable to find.

Regards,  Paul
--

"What is now proved was once only imagin'd." - William Blake, 1793.


Re: Measuring S - Parameters in Spice
On Fri, 06 Aug 2004 14:06:52 +0100, Paul Burridge

>Hi all,
>
>I flicked through Google last night and saw some quite simple set-up
>whereby, given an accurate enough transistor model, one might measure
>S - parmaters for any device under consideration purely via
>simulation. Is this feasible? Or is it really only possible to
>reliably do this empirically with real components?
>
>Thanks,
>
>p.

It's in this list of app notes:

http://www.orcadpcb.com/pspice/applicationnotes.asp?bc=F

                                        ...Jim Thompson
--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
We've slightly trimmed the long signature. Click to see the full one.

Site Timeline