[Prev][Next][Index][Thread]

Lumped vs. T-line - You be the judge...



Hi All,

	I did Bob's experiment!  It seemed to work very well.

1.	I made a ground plane out of foil and split it so as not to cause a
shorted loop.

2.	I used a 100 ohm resistor is series with a 10 ohm resistor as an
isolator.  The 1 ohm resistor gave too small and noisy of a signal for
practical purposes but I think it gives just the same results as the 10 ohm
which gives good signals.  This is a great idea!  It totally eliminates
generator loading while providing a very low impedance drive signal.

3.	I just brought the generator (<1 ohm out Z) right up to the coil base
since it is small and used very short wires so all the feed coax problems
are eliminated.  I grounded everything at the coil base.

4.	I did have to hook the 16.6pF/10Meg scope probe to the output of the
coil but I think I have justified that.

5.	If I model my secondary as a transmission line.  MicroSim says that it
has an intrinsic impedance of 31000 ohms.  It seems to work out so I guess
the program just knows...

Ok.  So I think I did everything right ;-)

I made a MicroSim model of the whole thing using both lumped and T-line
models to compare the output waveforms.  I took the input stuff and the
scope probe loading and all that into account.  I also took the scope's
60MHz bandwidth into account which did make a little difference.  The drive
signal was a 1kHz square wave.  My generator has a 30nS rise time which is
also in the models.  The MicroSim models are pictured below:

http://users.better-dot-org/tfritz/site/misc/Bob03.gif

The resulting output are here:

http://users.better-dot-org/tfritz/site/misc/Bob04.gif

With a square wave step function, The lumped model gives a nice sine
response.  The T-line model gives a 725nS delay and then a distorted sine
response with lots of little jerks from the reflections and stuff.

So there is a big and easy to measure difference between the two.  the
results should be "definite" Ha ha ha 

And here is the actual scope waveform:

http://users.better-dot-org/tfritz/site/misc/Bob05.gif

Here is a sharper data pull to Excel:

http://users.better-dot-org/tfritz/site/misc/Bob06.gif

No doubt about it, the waveform is distorted much like the T-line model
would predict but far from exact.  Hardly a clean sine wave.  However,
there is definitely a rise in the wave before the 720nS a T-line model
would predict which the lumped model does predict.

So we see the fast output rise of the lumped model but distortions similar
to the T-line model...  So the fast rise is definte proof of the lumped
model and the distortions are proof of the T-line model...

I guess this should make everybody happy... or nuts!! ;-))

Cheers,

	Terry