signal generator output

No, but its part of the job when things are not going right!

;)

--
An ideal transmission line will exhibit only delay and insertion loss,
the insertion loss qualifying it as an attenuator, albeit a tiny one
for a short line at a low frequency.

"John Fields"

** I set up the same conditions as the OP described - ie an RF generator, 1.5 meters of RG58 co-ax and a 60MHz scope with 1 Mohm //20pF input.

AS EXPECTED the input impedance of the cable dips sharply at the quarter wave frequency, 22MHz for my set up - or 28 MHz if the cable is left completely unterminated. This was determined by monitoring the signal at the output connector of the RF gen with a 10:1 probe.

However, at the scope end of the cable there is little sign of any dipping or peaking.

So, the cable acts like a quarter wave stub all right, but the unterminated end has a fairly consistent signal level since peaking and dipping occur simultaneously at opposite ends.

Sue as hell looks like standing waves at play to me.

.... Phil

How could an ideal transmission line have loss? It only includes inductance and capacitance, so it can't dissipate power.

An ideal 50 ohm transmission line, used to extend the output of a 50 ohm signal generator, adds time delay. Nothing else.

Insane. Sorry to say so, but insane.

Nope.

The 200 MHz corner was where I predicted. And I bet the voltage at the scope was flat, no frequency whoopie-doos, from DC up to there. But it is 6 dB/octave after the corner.

If it's source terminated, there are no secondary reflections. The generator gobbles them all up.

Spice seems to have two txline models, the ideal and the lossy. For some reason, the lossy-line model seems to work better, even if you crank the losses down to invisibility.

Gibberish.

Ditto.

John

Yes.

Also yes. That was the point. As seen from the end, the coax adds nothing but time delay to the 50 ohm generator.

Sure, but you don't see them at the scope. The signal at the scope will be flat over frequency until the 160 MHz -3 dB point, and then roll off nicely at -6/octave, unless the scope itself craps out first. Your 60 MHz scope will roll off before the 50r+20 pF corner hits. You shouldn't see any frequency peaks or dips at the scope, regardless of the cable length.

You might see some wobbles if the generator isn't a clean 50 ohm source. In that case, you can get multiple reflections.

Thanks for trying it.

John

--
Poor choice of terms, thanks.

I should've said "real" instead of "ideal".

--
I agree.

Run an AC analysis of:

news:5k49g6dkto6kak9k86v8fp5135kvqpi25q@4ax.com

and take a look at the inputs of the transmission lines for your
Himalayas! :-)

But not at the scope. One normally doesn't open up a coax and probe points internally.

"Standing waves" do not explain the OP's issue. A 50 ohm generator, driving a 50 ohm coax, into a capacitive load, is flat at the load, up to the -3 dB corner, roughly 200 MHz for a typical scope Cin. Then it rolls off nicely at 6 dB/octave. There are no frequency peaks or dips at the load.

Below the corner frequency, the scope sees the same voltage as the unloaded generator would put out at its connector, at all frequencies, no matter how long the cable is.

If you add a 50 ohm feedthru terminator at the scope, the amplitude drops in half and the corner frequency doubles, independent of cable length. Still no whoopy-doos.

(I have described the asymptotic behavior, which is usual in this business. Don't get preachy about a dB or so near the corner frequency.)

John

The first thing you see on that wiki page, in huge print, is the impedance equation. In our case Zin/Z0 = 1, so there is no impedance transformation. The generator is 50 ohms. The cable is 50 ohms. The end of the coax looks like a 50 ohm source to any load, just like the generator looks at its output connector with no external coax in the path. As seen at the load, the whole point of this thread, the coax adds time delay, nothing else.

Invoking "standing waves" to explain the OPs observations is just wrong.

You insist on lecturing about stuff you don't understand.

John

Yes, I placed a 50 ohm terminator at the scope end and now thw rms on the scope is reading same as the generator. Now I just have to figure the best cable length for single pulses that measure 3.5 ns at FWHM, so I can get a good picture on the scope of these pulses. Any thoughts on what frequency/wavelength to assume? A sine approx with period of

7ns?? thanks jk

** Finally.

** Not very likely with a 60MHz scope with a 6nS rise time.

You really are a babe in the woods.

BTW

It is bad manners to use more than one email addy on a newsgroup.

.... Phil

If it's terminated at either end, or at both ends, the cable length won't matter. Very long coax runs can degrade rise/fall times, but a few feet of RG58 won't have any noticable effect at risetimes of, say,

1 ns.

You'll need a fast scope, 200 MHz at the very least, to measure a 3.5 ns wide pulse. 500 MHz would be better.

John

That is why I save my 500 Mhz Old Tek scope for things like that :)

--
Depends.

Ever heard of a slotted line?

--
Perhaps, but if, at the corner frequency, the amplitude of the signal
is down 3 dB and at twice the corner frequency it's down by another
3dB, how is that 6dB per octave?

It rolls off at all possible slopes from udB/octave far below the corner to almost 6 dB/octave far above. I was describing the asymptotic bahavior, which is the way people talk in this business.

Because it makes you look ignorant.

Did you ever have a formal EE education? They teach this stuff, and establish the terminologies that most EEs use.

John

6 dB/octave is the macroscopic slope of the rolloff, the zoomed-out, ultimate slope as you go up and up in frequency. The asymptotic frequency response is what EEs usually describe in words like "flat to 200 MHz, then rolling off at -6 dB/octave." Everybody knows that, if you zoom in on that, it's actually a soft corner, not a sharp change of slope at 200 MHz.

It's like saying "this road is flat for a mile, then slopes up at 5 degrees." The transition isn't knife-sharp, but the description is useful.

It's 6 dB/octave because the reactance of the cap is inverse on frequency. That's all that 6 dB/octave means: above the corner, the voltage goes as 1/f. You've got to get away from the immediate vicinity of the 3 dB point for that to be mainly true.

I didn't make any of this up. It's the way EEs are taught and the way we talk.

John

We have one working 7104 (1 GHz, analog, microchannel plate) scope we use occasionally. All our other analog scopes are retired.

We have a bunch of Tek 11801 sampling scopes, with sampling heads that range from 3 GHz to 40 GHz. They are fairly cheap on ebay these days.

John

not really. the unterminated end of the coax looks like a mirror, asif the coax continued for the same length to an identical source producing an identical signal.

if the end is short-circuited at the end it looks like a mirror (as above) but with the other source producing an exact inverted signal.

yeah, the reflections will be in-phase with the source signal at the unterminated end, so there they will always add constructively.

If it's not a good one the reflections may be partially reflected from it and add constructively or destructively to the source signal.

--
?? 100% natural