Stupid question?

[scadaguy]

To locate cable fault by TDR method, a pulse is injetted and the reflected pulse is heard so that distance could be calculated. Is it possible to use current method by coupling instead of voltage method? Would current have the propagation delay and wave property of reflecting upon a high resistance?

 

[itsmoked]

Well think about it.... How are you going to get a current thru a cable to "reflect" if the cable is broken?

If you knew precisely the capacitance of the cable you could charge the cable up and see how long it takes to reach a certain voltage via current flow. In other words measure the actual "charge" injected. But you'd probably not be real accurate. maybe to withing 30's of feet?

Keith Cress
Flamin Systems, Inc.- <

http://www.flaminsystems.com>

- kcress@<solve this puzzle>

 

[VEBill]

"...current method by coupling...?"

I think that I understand your question.

It is possble (but why bother?).

The reason it is possible is that you could couple a short pulse into the wire ('wire' is a better word than 'cable' in this discussion), and it would generate a conducted pulse in accordance with the characteristic impedance of the wire (even if the wire is open circuit further down the line).

The V/I ratio of any TDR pulse is always set by the characteristic impedance of the wire. The pulse doesn't know if it's voltage (connected), or current (coupled), derived.

The pulse would 'bounce' off the fault just like any TDR.

Then you need couple it out again.

'Coupling' instead of connecting would be less efficient in terms of coupling loss (many dB). Difficult to make a big pulse and difficult to get the return signal out again.

Another problem with 'coupling' is that your pulse would probably head out in two directions (instead of one).

 

[scadaguy]

Thank you all for useful information.

I just tried with a CT as coupler and a long copper wire loop as the cable. 100kHz pulses were injetted. Expected reflected pulse and sending pulse were monitored using digital oscilloscope. Since it's lab test, the fault impedance was just simulated with decade box and the earth with real coppy conductor.

I hardly detect any reflected pulse of propagation delay which can be easily demonstrated with voltage method.

I always suspect that in real application what would be the effects of true earth to the result.

 

[Warpspeed]

Really, for a TDR to work properly it needs to be a proper transmission line of some sort, and you usually try to match the pulse generator into the characteristic impedance of the line.

That way you get a reasonable amount of pulsed energy transmitted into the line, and the reflected pulse is totally absorbed by the generator. That is important so you don't get another unwanted reflection from the source.

The reflection you see will be the amount of energy returned, which will be almost all of it, if the far end is shorted or open circuit. How much less depends on transmission line attenuation both ways.

It certainly does work, and I once caught a government agency tapping my home telephone line, believe it or not.

That was many years ago, and it is all now a bit of a well appreciated joke, because I now have personal friends that used to work for that very same agency as electronics engineers back in those days. They remember that incident very well, it actually caused quite a stir, and I caught them red handed back then, ha-ha-ha.

The infamous Warpspeed strikes again.

 

[scadaguy]

Yes really appreciate sharing of real applications and cases.

I see what your meant and I just tried again and again.

And I am quite sure that for lab test the key problem is by coupling method a low impedance loop is needed for a strong enough pulse to flow, but this would not enable you to see the reflection. With impedance in the loep (simulate cable fault or open conductor), back to basic and analgous to "open jaw" clamp-meter, you would not see any pulse.

And i think of applications of power line communication (eg, homeplug) that they may not suffer from interference from reflection even no termination of grounding is considered. Any opinions?

 

[Skogsgurra]

I must question the use of a CT. Unless it is a Fischer broadband CT (50 - hundreds MHz), you will not see anything. And even with such a monster, I doubt if it will work. You need rise and fall times in the 10 nanoseconds range if you want any resolution.

You can experiment cheaply and efficiently with TDR "voltage mode" by arranging a small relaxation oscillator with drivers, all on one cheap chip (the 74HCT14 for instance). There is a simple design in

http://www.elecdesign.com/Articles/Index.cfm?AD=1&ArticleID=6260

and it even has a picture showing outgoing and reflected pulse.

Gunnar Englund

http://www.gke.org

 

[scadaguy]

Sorry for not mentioning details of CT before. I am using a CT from PLC (powerline communication) modules which should be high frequency response. But thank you for your reminding about the frequency of injetted signal. I will try later.

I have read your suggested article before. Actually I have tried voltage method successfully and see good results. I want to set up and test by coupling current method and compare the results. But up til now I still cannot see the reflection.

 

[Skogsgurra]

If it is PLC, you cannot expect more BW than around 100 kHz. That's very far from the band-widths needed to see any reflections. I would say about 20 MHz and above. So you have to increase the BW around 100 times - not easily done.

Gunnar Englund

http://www.gke.org

 

[Comcokid]

OK, I'm a little late in comming into this discussion.

If you are using a CT to inject a pulse into a coax , then you are going to inject the pulse into both the center conductor and the outside shield. Normally on a coax a pulse is applied to the center conductor with the ground (shield) as reference. But if you inject the same pulse onto the center conductor and the ground all you might do is turn you whole cable in a Goubau-line type transmission line (a single wire transmission line). The velocity of propagation is high in a G-line.

Now, if you injected on one side of a twisted pair, or opened you coax to inject only on the center conductor then you would have a pulse. However, a pulse with amplitude V on a transmission with impedance Z is the same whether it originated by coupling a current or conducting a voltage.

In order to do any useful TDR work, you need a fast risetime (i.e. bandwidth) in you transmitted pulse.

 

[Warpspeed]

Agree with the above comments. A suitably fast edge needs to be injected, and it needs to be injected so there is both voltage and current in a ratio to match the characteristic impedance of the line you are measuring.

Energy needs to be launched down the transmission line, and much of it will be lost at the point of transmission if there is a serious impedance mismatch at the source.

Also, the "mode" of transmission, differential mode, or common mode, needs to suit the normal operating characteristics of the particular transmission line. Injecting signals in unusual ways may not give meaningful results.

I rather suspect that just injecting a current from a very low source impedance that has almost zero signal voltage, is not going to launch enough energy to be useful.

The rise time of the pulse needs to be appropriately fast to suit the distance and resolution requirements of the measurement. Extreme speed may not be a prime requirement if the required total range is miles, or tens of miles. But if it is only a few feet, some serious bandwidth will be required to see anything at all.

 

[fsmyth]

Sure. Done all the time. But if you are using inductive
(or capacitive coupling, for that matter), you will be
working with RF equipment - preferably a network analyzer,
but it can be done with a signal generator, a coupler, a
suitable detector, and the appropriate stubs, connectors,
cable lengths, and knowledge.
For what is normally (now) call TDR, inductive coupling
is inappropriate; the rise time will never be sharp enough
without some VERY expensive gear. Current or voltage can
be used, but voltage is easier to measure with conventional
test equipment.
All this is sorta beside the point - TDR needs a reference
point which is usually the point of connection (one only).
Otherwise, you will be seeing reflections from two separate
ends.
<als>