[SI-LIST] NETWORK MODELS ADJUSTED TO MEASURED TDR DATA.
- From: "ron@xxxxxxxxxxx" <ron@xxxxxxxxxxx>
- To: si-list@xxxxxxxxxxxxx
- Date: Tue, 14 Oct 2008 14:09:57 -0700
Hi Guys
I was an RF engineer for many years, and was totally committed to the
use of S-Parameters
as an accurate model of a passive network.
When I became an SI Engineer I tried for several years to use
S-Parameters as models of the
source, interconnect and destination of Digital signals. Because I was
using ADS which had a
frequency domain engine together with time domain simulators including
SPICE and harmonic
balance, and the convolution to make the frequency domain models work in
the time domain I was
confident that all would work well, using S-Parameters as models.
However, several problems emerged:
1. Some vendors were supplying S-Parameters without Phase
data(magnitude only)
Such data is unusable.
2. Measured S-Parameters contain noise which result in spikes going
above "1" resulting
in gain for a passive network which causes non-convergence.
The data has to be massaged
to remove the >1 values, which is time consuming.
3. Measurements of differential S-Parameters are difficult to get
because of the single-ended nature
of the Vector Network Analyzers.
4. Calibration of the VNA is very hard to perform out to the end of
the adapters used.
5. Measured S-Parameters must also have a DC line added in order for
the convolution to work.
It was very time consuming, tedious and gave questionable data.
So, in 2006 I was finally successful in adapting frequency domain
transmission models (which generate
S-Parameters durring simulation) to measured TDR data.
THIS WAS VERY SUCCESSFUL FOR SEVERAL REASONS:
1. TDR measurements with differential head SD-24, so the data was in
differential form.
2. TDR data in simple tabular form - time vs magnitude
3. Approximate models are assumed from mechanical data as ADS models.
4. ADS models for transmission lines are simple - trace width,
thicknes, dielectric thickness, dieliectric
constant, length etc.
5. ADS adjustments of length, width and thickness as first cut to
meet approximaely the TDR data.
6. ADS model for frequency dependent (loss and dielectric constant)
uses only 3 adjustable parameters.
7. ADS has Optimization model for final adjustments of network
models to match measured data.
8. Simulation results from adjusting these models in the time domain
is inherently accurate for time domain
simulations such as EYE-Patterns and BER simulation.
In the above I have given away a lot of what took me years to perfect.
However, because I am semi-retired
perhaps some of you can carry on.
Good luck.
Ron Miller
--
Ronald B. Miller, Microwave/SI Engineer RAIL = NOISE + REFLECTIONS
Gigahertz Data Div of MI Corp. \\ // \\ //
7721 Sunset Avenue \\// \\//
Newark, CA 94560 ->JITTER<- EYE ->JITTER<-
tel 510-793-4744, //\\ //\\
fax 510-742-6686 // \\ // \\
www.ghzdata.com RAIL = NOISE + REFLECTIONS
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
