Forums » General Electronics Discussion

Is there really no transmission line model? It would be quite helpful!

For transmission line simulation I suggest you to use a new tool that will be soon available in the Web: https://picasaweb.google.com/118285010585894969594/DWSIMAGES

Andy, more docs regarding DWS are available here:

https://docs.google.com/#folders/0Bx-ZqV10CSiNcVZRU3drdHpGcmc

DWS will be soon available in the market on PCs (Windows,Linux,Mac)

You can also become a member of my SWAN Linkedin group where the news regarding DWS are published:

http://www.linkedin.com/groups?home=&gid=2673743&trk=anet_ug_hm&goback=%2Enppvan_%2Fpierobelforte%2Egmr_2673743

Piero

This series of articles describes a way to implement a general purpose single ended transmission line:

http://i.cmpnet.com/rfdesignline/2010/06/C0580Pt1edited.pdf

http://i.cmpnet.com/rfdesignline/2010/06/C0580Pt2edited.pdf

http://i.cmpnet.com/rfdesignline/2010/06/C0580Pt3edited.pdf

(Note that part 3 is a bit confused about which G sources are Laplace functions: Fig11 is right and the listing at the end is right but the two differ.)

However, whilst it is possible to implement this transmission line subcircuit in CircuitLab, until CircuitLab includes the ability to parameterise values in expressions, it would be very difficult and likely to be highly error prone to try to enter and edit the expressions.

This is an example of where implementing feature requests:

https://www.circuitlab.com/forums/support/topic/n98tb8z5/feature-request-parameters-please/

and, to build a component or hierarchical block to simplify re-use in other circuits;

https://www.circuitlab.com/forums/support/topic/f3yqcvkf/help-or-feature-request-attachment-of-user-circuits-to/

would be a big help.

:)

Sorted.

Managed to make the LC lumped element model converge properly.

and a slight modification ....

Here you will find an Application Note I've just written to compare the effects of lumped circuit approximation for modeling an ideal transmission line: https://docs.google.com/file/d/0Bx-ZqV10CSiNS1JwTnc1NVIyTWs/edit

An evaluation of LC-cells lumped TL approximation in terms of WCEDs (Worst Case Eye Diagrams) in the case of 1,000 cells is given here: https://picasaweb.google.com/118285010585894969594/DWSIMAGES#5733799510606247170 I carried out this evalution up to 5,000 (five thousand) cells, I'll publish the results soon.

Here you can find a Facebook group where the difference of SWAN with respect conventional simulators are discussed: http://www.facebook.com/SpicySchematics

This video shows an actual comparison between a commercial simulator (MC10) and SpicySWAN during the simultaneous simulation of the same circuit 10_LC_TL.This circuit is a simple LC cell circuit that some simulators use to model a TL. To get similar accuracy levels, both sims are carried out the same time step of 10 femtoseconds on a time window of 0-50ns. MC10 runs locally on a quad core I7 PC. SpicySWAN sim is carried out by SWAN/DWS engine running on a SpicySWAN server in the Web The difference between the two sims is less than 1mV.The same sim carried out locally on the same PC using SWAN requires 8sec with the time step set at 10 femtoseconds.

http://www.youtube.com/watch?v=pTMBnvUChog

Here my last paper on TL model comparative analysis using V-I plots:

https://docs.google.com/file/d/0Bx-ZqV10CSiNYlZENzdfOEk4aEk/edit

https://docs.google.com/file/d/0Bx-ZqV10CSiNMk1SNC1zSDNlVkk/edit

Here a GALLERY of SWAN/DWS sims screenshots: https://picasaweb.google.com/118285010585894969594/DWSIMAGES#5729896428676687522

Here a screenshot related to SWAN/DWS sim of the time-domain STEP RESPONSE OF A CHAIN of TEN 5th-ORDER LOW-PASS CHEBYSHEV FILTERS ( wc = 10MEGRAD, fc = 1.59MHz ). The filters transfer functions are described by their poles in the S-PLANE.

https://picasaweb.google.com/118285010585894969594/DWSIMAGES#5727141692576492418

SWAN/DWS max error between circuital implementation and H(s) in the case of unit step response of a simple LPF (R 100ohm, C=1nf, H(s) single real pole at -10 MEG rad/s) is 4.03 NANOVOLTS (Tstep : 1ps to 1ns).

Microcap 10 ( http://www.spectrum-soft.com/index.shtm ) is a version of Spice with a relatevely good model of Transmission Line and the ability to carry out the simulation at fixed time step as required by circuits with TLs inside. This benchmark shows the accuracy problem of MC10 in a relatively simple situation of a RL TL circuit used to model skin effect losses. DWS resulted 500X faster than MC10 in the same situations with a maximum error of 1uV while MC10 waveforms show several accuracy problems. http://www.slideshare.net/PieroBelforte1/2013-pb-dws-vs-microcap-10-rl

I too am looking for the delay line function. Although my application is not RF but rather acoustic/mechanical. It would be very helpful for trying simulate some piezo and other mechanical devices to check out control approaches.