tdr pulse generator

TDR is just a pulse generator and osciloscope in one instrument.
The idea is to use a pulse that travels along the DUT and observe
the pulse itself and the reflection. Any discontinuities along the
way will appears as pulse positives or negatives depending of the
value of impedance. respect the value of impedance of the
pulse generator.

Agilent has many very nice free appliction notes. Here's one that describes some of the math in TDR measurements.

Time Domain Reflectometry Theory (AN 1304-2)
http://cp.literature.agilent.com/lit...5966-4855E.pdf

Hi,

Allow me to share my comment here.
Guess by now you should have some understanding on theTDR.
By learning the TDR measurement, it is very useful to identify the bad location and discontinueties.
SOMEHOW, I have learned that not much theory/comments on the correction part, like how to particularly modify/reduce/rework the discontinueties.

Pls look at this information as well.
I hope you can share it to me if you have it.

Thanks

Thank you for the comments. Can I use a directional coupler to identify the reflected pulse? I need to design a TDR for aother application which is not a fault locating in a transmission line.

sure you can...but if you use a high impedance oscilloscope
then you can use a regular "T" to connect the pulse generator,
the oscope and the DUT. Just as is explained in:
http://www.tkk.fi/Misc/Electronics/circuits/tdr.html.
One small advice is: the reference plane is the output
just after the pulse generator. if you use delay in your oscope
your reference plane can be put wherever you want..
One more advice is: depending of your applicaton you
migth need a faster pulse gen(shorter rise time). If the distance
of the transitions are too short in comparison with the rise time
you will see them as one big transition.

I was upload this article from " circuit cellar magazine"

see this link

https://www.edaboard.com/viewtopic.p...tometer+cellar


SOKRAT

Hi boy,

What is your application ? Perhaps we can advice better if we know what you need.

Need information about TDR application in soil moisture measurement.

I thinnk for soil moisture they use a TDT rather than a
TDR:

http://www.sowacs.com/feature/automata/index2001.html

http://www.surechem.com.my/download/.../P1/P1-64e.pdf

http://aas.bf.uni-lj.si/oktober2005/19zupanc.pdf

Thanks for the link.

It looks like TDR sensor measures time from pulse to receptioon of reflected wave from probe ends.

I am trying to find what pulse is to be send - unique or frequency , requirements for pulse rise and fall time , pulse amplitude and reflected signal level and conditioning requirements .
The paper you mentioned says freq of pulse is about 15 to 45 MHz .

My interrest to TDR is just hobby - i have own about acre and wish to make autoirrigation system for personal use.

Reflectometry measurements can be TDR or TDT.
TDR means Time Domain Reflectometry and
TDT means Time Domain Transmissionmetry(some thing
like that). What you see is two probes that is why
I said they use TDT, since they inject a pulse and
measure the pulse in the other probe. With only one
probe you will not be able to measure "transmission
time". For TDT the repetition frequency is irrelevant
as long is not too short to discriminate on pulses.
What is important is the rise time and power(level).
Big rise time will not be able to discriminate between
pulses, low power level will not be able to discrimina
te in the second probe from noise.

I see what you mean . Have a look to this document (figure 23):
http://www.cprl.ars.usda.gov/programs/MAKE_PRB.PDF

There are 3 wires - middle is connected to inner of coaxial and 2 outers to outer of coax respectively . That make transmission line somehow assymmetric and close to coax structure . Of course there are also 2 wire based probes - i guess it is symmetric line the tdr signal is injected into .

Well , now the question is to find a way to digitize such signal without Tek equipment . Just an idea
(though may be stupid enough):

If we will measure integrated signal on TDR probe starting from time the pulse is injected till next pulse in way that each time integration starts little bit more delayed to pulse - by calculating difference between adjacent integration (find a derivative) samples it could be in theory possible to get a real sampled signal for very short time (manageable delay time between adjacent integrations) . Making multiple measurements we can cut out spontaneous errors in measurement (suppose it has a Gaussian distribution).

That is needed because for probe length 10 cm resolution should be in picosecond interval . I suppose this is the only way to construct probe for which most probably costs no more than 100 USD in practical implementations .

Delay could be managed, let say, via varicap based voltage controlled principles or something similar.

I think in the pictures they do is just transform the coaxial cable in a balance
trasnmission lines. The first transition: coaxial to tree wires, produce a dip in
the display, that is an indication of place or location. The second transition is
when the signal reach the tip of the three wire structure and then reflect. The
length(or time) of the reflection will depend on the transmission line itself and
the soil. So if you use the same three wire structure you can map this distance
to moisture contain. BUT, if you change the separaton distance between electrodes
then you need to calibrate again.
There is not need to integrate, the distance information is already in the display.

Added after 1 hours 10 minutes:

Lets see some numbers:
For a distance of 20 cm the time separation between rods would be about
22.22 psec. If you allow 12 psec for rise time, the bandwidth required for
oscilloscope would be 3GHz, a fairly common oscope nowdays.

The frequency of pulses is irrelevant. And regards to the question if use sine
or a pulse, is a little more complicated to explain but pulse is the way to do it.

That is ONE CRUDE soil probe! You will have all sorts of inductive ringing at the feed point where the coax make an ABRUPT transistion to 3 conductors!

As a start, one could improve the two outer conductors by bending them toward the center at the feed point, and using some metal foil to make a ground plane "cone" where the braid transitions into the two outer conductors.

Hello everyone!

Could someone tell me why do I use a very small rise time for the step signal? I understand that having a small tr will lead to a very wide bandwidth signal. My question is why do I need such a wide BW? Other question is, assuming that I m working with that wide BW, what I see in time domain, is it the reflected spectrum of the incident signal with some modification on it depending of the load?

I have read several application notes and all of them say that a very small tr is needed but there is no explanation of the reason...

Thanks in advance for your help!

In a short, the bandwidth respectively rise time will determine the TDR resolution. Slow risetime = low resolution.

Thanks FvM, now I see it a little bit better!

Now, If we try to understand it from the side of the frequency domain ? How does affect the fact of having a small tr?

Agilent's application note introduce the TDR technique numerating differents methods. One of those methods is using a single wave to see how is the response for that unique frequency. But then, it says that we would need to repeat the measurement with different frequencies in order to have a better understading of what is going on our transmission line. Then it introduces the TDR technique, and here is where I get lost... I suppose that having a little tr is like testing with a wide range of frequencies at once... as long as we reduce the tr, we are covering a wider Bw... is this concept correct?

TDR spells time domain reflectometry, thus I think, it's obvious to use a time domain perspective to explain it's behaviour. In my opion, transmission line reflections can be understood more intuitively in time domain. But it's a matter of your region of interest. If you're familiar with small band RF applications and use to think about transmission lines in terms of S-parameters or VSWR, you may prefer frequency domain. Both representations are however related by a fourier transformation and can be converted, if you like to.