Maker Pro
Maker Pro

Simple transmission line question

F

Fred Abse

Jan 1, 1970
0
Try a diode termination some time >:-}

I have. Measuring trr. Switching from an amp forward. Source termination
on the pulser, and a 1 ohm load. straight into a 50 ohm, 1GHz vertical
amplifier. Took a week to get the source termination good enough to be
sure that what we saw was the diode, not the test rig, with some quite
expensive resistors.

Datasheets can tell big lies;-)
 
F

Fred Abse

Jan 1, 1970
0
That's an excellent solution when the input capacitance of the scope is
zero...or inconsequential at the frequencies of interest.

50*22pF= 1.1ns.

Will degrade a 100MHz scope to 95.4MHz
 
F

Fred Abse

Jan 1, 1970
0
OK, Thanks Fred. I don't really need a new design. I'm just trying to
understand what I already did :^) Is the above "tee" circuit any better
than 450 followed by 56 ohms to ground? It certainly attenuates more.

It looks more like 50 ohms. Not perfect, since you don't know an accurate
source impedance. Attenuation is less important than termination accuracy,
here, provided you can still see what you want to see.

500 series, and 56 shunt from a 500 source looks like 45.2 ohms, or 1.1
VSWR. OK for RF, but maybe not for pulse reflections.

What I'd do is load your source with a known resistor, and calculate the
source resistance from the voltage drop, then calculate a suitable pad.

However:

Instead of source termination, use a 450 ohm resistance in series with the
coax, at the sending end, and terminate the scope end with 50 ohms. That's
effectively a 10:1 probe without reflection. I do that all the time, with
50 ohm wideband vertical amplifiers. If it loads too much, try a 100:1
4950/50 version instead.

The main application of source termination is where you want to get pulse
power down a line into a load that can't be matched, such as measuring trr
of a diode.
 
F

Fred Abse

Jan 1, 1970
0
Appendix A on "how to drive a scope" in Art of Electronics, 2nd edition,
by Horowitz and Hill looks like a reasonable student guide to me. I must
warn you that I have never tried it on, or seen it tried on, real live
physics students. :) Because of when it was written, it's for analog
scopes.

Standard disclaimers apply; I don't get money or other consideration from
any companies mentioned.

That assumes that students still read *books*.
 
F

Fred Abse

Jan 1, 1970
0
I like fat cable because it has the least attenuation at UHF.

Depends how UHF you go. High enough, and fat enough, it starts to
think it's waveguide, and you get all sorts of modes happening.
 
F

Fred Abse

Jan 1, 1970
0
Yeah but then you have to tell people to terminate the line. This way
it's 'mindless'.
(Instruments for students.)

That might actually leave an impression on the student that won't be
forgotten. Two lessons in one. That's teaching.

I hadn't spotted that Jim had already suggested this, so I did, too.
 
F

Fred Abse

Jan 1, 1970
0
Exactly, 56R is a better match, but even your resistor precision gets a 5%
tolerance too. Just pretend you don't see the 10mV overshoot, it's just
monitor front end parasitics ....

I think George can afford 1%, or better, resistors. Hell, the difference
in price is so small that it only matters if you're making millions of
units.

I don't know of a development lab that even stocks >1% on 0805 and below.
For cheeseparing clients, there is always Monte Carlo (costs extra).
 
J

Jasen Betts

Jan 1, 1970
0
That doesn't work. Coaxial cables have basically no parasitic inductance,
there's nothing to charge up unless perhaps the cable was grounded at
both ends. or some other wierdness with the powersupply.

oops colour me clueless.
 
J

josephkk

Jan 1, 1970
0
Yeah but then you have to tell people to terminate the line. This way
it's 'mindless'.
(Instruments for students.)

George H.

It may well be better to leave the problem in the test setup. Most people
learn faster from failures than their successes. Think about it.

?-)
 
F

Fred Abse

Jan 1, 1970
0
Construction Specifications

Inner conductor : Solid BCCAI 2.74 mm
Dielectric : Foam polyethylene 7.24 mm
Outer conductor : Aluminum tape 7.39 mm
Overall braid : Tinned copper 8.13 mm
Standard jacket : Black polyethylene 10.29 mm

That's not fat. Similar to RG213, or LDF4-50.

Andrew make much fatter cables than that.
 
D

DecadentLinuxUserNumeroUno

Jan 1, 1970
0
That's not fat. Similar to RG213, or LDF4-50.

Andrew make much fatter cables than that.

Just get LMR400. Not exactly the same...
 
W

whit3rd

Jan 1, 1970
0
If you put DC into a long unterminated coaxial cable it will store a
charge, just like any other capacitor.

It's much more wonderful than 'any other capacitor'. The fault scenario had
a cable as part of the tuned resonator on a spark-gap, just like old-time radio
transmitters; that was because a transmission line terminated with too-low
impedance is an inductor. An open line is a transmission line terminated with two-high
impedance, and (as stated) is a capacitor. But in all cases, there's a time
constant in addition to the inductance or capacitance, which is what makes
gizmos like Blumlein pulsers possible.

One cannot dismiss as 'negligible' either the inductance, or the capacitance, of a transmission
line. If you do, the characteristic impedance loses both its parents.
 
J

josephkk

Jan 1, 1970
0
Air core coax (with occasional disc spacers) can be bought at 50 ohms and is
close to c. GR used to sell true precision air core 50 ohm lines; probably
someone still does.

You can buy foam coax that's 0.89 c.

And both of them have issues when and how tightly you bend them. There
are also spiral dielectric (still about 2/3 to 3/4 air) coaxes available
with intermediate velocity and bending properties.

?-)
 
J

josephkk

Jan 1, 1970
0
On Mon, 22 Apr 2013 10:03:54 -0700, George Herold wrote:
I stick in a series combo of 450 ohms and 50ohms to ground,
and take the monitor output from the 'top' of the 50 ohms. (and eat
the factor of ten in signal level.)

That will present less than 50 ohms to the line, do this:

T-pad:- 500 ohms in, 50 ohms out, -20dB, theoretical values:
[deleted ....]

OK, Thanks Fred. I don't really need a new design. I'm just trying
to understand what I already did :^) Is the above "tee" circuit any
better than 450 followed by 56 ohms to ground? It certainly
attenuates more.

George H.

It's -30dB (20log(Uo/Ui))! That's a waste!

A minimum loss matching L-pad between 500 and 50 ohms would
have 474.3 ohms in series and 52.7 ohms to ground at the 50
Ohm side. That's about 4dB better and the best that can be
done with resistors.

Jeroen --Every dB counts-- Belleman

1 percent resistors are pretty standard now. Looking it up...
475 ohms and either 52.3 or 53.6 ohms will be rather close.

See:

http://www.logwell.com/tech/components/resistor_values.html

?-)
 
J

josephkk

Jan 1, 1970
0
Normally, source termination uses a SERIES resistor at the source.
So, if the driving source has a 5 Ohm impedance, you would use a
45 Ohm resistor in series between source and the cable. You WOULD
get a reflection from the open-circuit end, but it would be fully
absorbed at the source end, so the output end would see no reflection.

For a parallel resistor to properly terminate the source end, the
source would need an infinite output resistance, which prevents it
from creating a signal.

Jon

Maybe, i have seen plenty of pulse generators that were current mode and
thus had 50 ohm shunt termination.

?-)
 
F

Fred Abse

Jan 1, 1970
0
One cannot dismiss as 'negligible' either the inductance, or the capacitance, of a transmission
line. If you do, the characteristic impedance loses both its parents.

"To lose one parent is unfortunate. To lose both sounds like carelessness"
(Oscar Wilde, "The Importance of Being Earnest", more or less.)

;-)
 
W

WoolyBully

Jan 1, 1970
0
"To lose one parent is unfortunate. To lose both sounds like carelessness"
(Oscar Wilde, "The Importance of Being Earnest", more or less.)

;-)

Silly savage.
 
Top