Fast Pulse Generator

And that's why people need to state their problem as well as their guess at a solution to the problem.

Tektronix managed to measure that with a 0.4V 50ps step back in 1976, so 1.5V ought to be more than sufficient. It also has a resolution of a couple of cm, and worked in driving rain and could be stored under a foot or water.

Your application is so undemanding that all you need is a 100kHz or so square wave driving just about any logic gate. 74AC is the traditional starting point, but many others would do.

Depending on the accuracy you require, you may need to verify the relevant speed of light in the cable.

I would always take the velocity factor into account anyway, even if I were not planning to sell these on to others who might actually expect to get what they've paid for!

To go the capacitance route I'd need to idenfity it and look it up. Many of these reels and cables don't have helpful markings. Weigh them? There are so many different reel dias and widths, so that won't help. Can you expand a little on your resonance idea?

Not even if I want to know the lengths to the nearest inch? If not this time, maybe next time I might....

Well, as ideas go it's got simplicity on its side, that's for sure.

It's just my subtle way of hinting that I don't seek replies from trolls, that's all, Jeff. Not that I see them unless someone else quotes them, anyway.

once again you prove yourself to be the Link King of s.e.d Jeff! Some great stuff there; many thanks.

Not in this universe, no. But I'm not looking for the dV of the rise times, only the time interval between the source and the 'echo' so to speak.

There's a catch: The slower the risetime, the more uncertain the interval. The electronics version of the uncertainty principle.

Jeroen Belleman

Most run-of-the-mill 50 Ohm coax has 100pF/m capacitance, give or take 5% or so.

An open-ended perfect coax cable looks like a short circuit from the other end if it's a quarter wavelength long. You could apply a sine signal and tune to the lowest frequency where you see an impedance null.

Either way, you'll need to measure a known length to get the capacitance/m or the velocity factor and find the real length.

Jeroen Belleman

Note that you /don't/ need a pulse that is

So you're saying a one-off fast transition from low to high would suffice? That may be right for all I know. I need to get up to speed on this stuff. I'm really more of a frequency-domain person.

Yes. Or high to low, of course.

Think back to basic Fourier Transform theory, the type you learned in maths before leaving school to go to university.

The only difference between a step and a pulse is that a step has more energy in the low frequencies.

The transition time determines the highest frequency that contains energy.

The resolution is determined by the highest frequency; the energy in the lower frequencies is irrelevant.

So, to get the best range resolution, have a fast transition. To get a good range, put more energy into the cable.

For pictures, look at the Tek 1702 manual, or elsewhere.

Better to cut off a small piece and measure it.

If you can access both ends, measure resistance.

Many

Sweep it with a sine wave generator and find the amplitude nulls. With an open far end, the first null is when the reflection is out of phase with the generator, 1/2 wave round-trip.

You could probably design a simple oscillator too.

Measure inches with a ruler!

TDR is great, but there are usually easier ways to measure a cable length. It's really useful for evaluating PCB traces, and that needs picosecond resolution.

TDR is used to find opens and shorts and things on inaccessable coax and twisted pair and fiber cables. It will tell you the cable impedance too.

My 11801/SD24 applies a 10 mA negative current, steered with fast diodes. That makes 250 mV into the local termination and a 50 ohm coax.

When Tek pulled out of the Channel Islands, they allowed the locals to buy the 1502 boards and start Polar Instruments.

Thanks, I had wondered about the origin of Polar Instruments.

piglet

Yup, I use that guy's gizmos for test jigs--they're cheap enough that you can just leave it in there.

Cheers

Phil Hobbs

If it's single-shield coax such as RG-58, there will probably be enough crosstalk between turns to mess up your measurement. +1 on using a capacitance meter.

Cheers

Phil Hobbs

Yeah Jeroen said that too. I remember having great fun for a day or two as a grad student with a (semi-fast) pulse generator, 'scope, and spool of coax, with both ends available.

So in the frequency domain... (frequency domain reflectometry) I've got reflection peaks and nulls... And then, if I think of pulsing it in the time domain... it's somehow all those reflection peaks that get summed to give me a return pulse.. Sorry, just having fun thinking in two domains. :^) One thing I tend to forget, thinking in the frequency domain, is that you have to let the time go to infinity (sum many cycles).

George H. (who would cut off a 1' section and measure C and the mass. and then C and mass for spool... tape the cut-off section and your measurements to top of spool, for later reference and resale.) Are there no numbers on the coax?

Tek assembled scopes there to avoid european import duties.

Oh I just need to add, that I have no good image, for what a pulse traveling down a transmission line look like in freq. domain.... OK I can take a picture of the pulse at each small time step, and take the Fourier transform (FT) of that.. In which case the FT looks like a bunch of amplitudes and phases that change with time... But since in taking a FT I'm assuming infinite time.. which I can't. A lot depends on the time 'window' in ways I don't fully understand.

George H.