A Simple TDR/Test Signal Injector

TDR? Stands for Time Domain Reflectometer. An instrument that sends a signal down a transmission line (like coaxial cable) and then analyzes the reflection. Find a cable fault underground so that you know where to dig! Overhead, in the air, before you take it down! Is it damaged inside a wall? Has a staple penetrated it? Be the hero and go right to the trouble! With a 100 MHz oscilloscope this adapter will allow rudimentary coaxial cable testing with accuracy on cables up to several hundred feet in length. It can enable the user to not only locate a fault but to get a hint about the nature of the failure. There's more! This simple device will also generate a tone that can be used to troubleshoot receivers and audio circuits. Best of all, it can cost little to build and just a few minutes time to do it!

When looking for a fast-risetime generator I found that my test equipment just didn't have what I needed. A search in my junk box provided some 74AC04 hex inverter logic chips that seemed to fit the bill. They can be purchased new for 25 cents each if you need to buy them. Handy parts to have. The first two sections form an oscillator with R1 and C1 setting the frequency to around 1KHz. This feeds two more sections to buffer and isolate the oscillator. Keeping the connections short and the components small allows the output rise times to be under 2 nanoseconds, faster than a 100 MHz oscilloscope can track.

So here's how I built it:

A small piece of printed circuit board, about the size of a large postage stamp, was just a bit larger than the chip with enough room to lay the chip on its back, dead bug style, and bend pin 7 back down and solder it to the ground plane to anchor it. Since you need to ground all unused inputs, I also did the same for pins 11 and 13 as well. This eliminates the need for glue. Remember that with the chip on its back count the pins clockwise. Since pin 7 is on a corner and grounded then use it as a reference : Pin 8 is on the same end across from it, etc.

I bent pins 2 and 3 together and did the same for 4 and 5. I mounted a BNC jack as close to the pin 7 and 8 end of the chip as possible (you need to keep things very short). You can easily install the jack using a 3/8” ring terminal. Remove any sleeving from the terminal, flatten the crimp portion, bend it, as far from the ring as possible, 90 degrees, and solder it to the board. Solder a small 50 ohm resistor (R3) from center pin of the jack to ground. This resistor sets the impedance of the TDR test. If you wish to test 75 ohm cable then you would use a 75 ohm resistor. See? Simple!

Also from the center conductor pin of the BNC you will solder a 300 ohm resistor to unload the inverter section to keep it within its 25 milliamp output capability. Form a loop in the left-over resistor lead at the BNC end so that you have a place to attach your oscilloscope probe. The other end of the resistor just solders to pin 6. Solder a 100 K resistor from pin 1 to the junction of pins 2 and 3. Solder a .005 uF (or thereabouts) capacitor from pin 1 to the junction of pins 4 and 5 and bend the remaining lead over to pin 9 and solder all three connections.

The battery can be just about anything you want to use from 2 to 5 volts. I had a little two-cell AA battery case with integral switch so I used that for mine. I just glued the board to the case. The circuit only draws 6 milliamps or so. Solder the negative of the supply to the ground pin. The positive of the supply goes to pin 14. Perhaps not necessary, but good practice, is to solder a .1 microfarad capacitor from pin 14 to ground as a bypass for the power.

A .01 microfarad capacitor from pin 8 to some kind of probe structure completes the gadget. This was an afterthought and is not used as a time domain reflectometer. The fast risetime of the 1 KHz square wave produces tone from audio through HF. Nice and strong at audio because you need a high level signal there, lower at IF frequencies, and still usable through 180 MHz (S8 signal on AM). I would suggest mounting a probe, insulated except for the tip (I used AWG#12 house wire), on the end of the board opposite the BNC jack. Use a file to sharpen the probe tip. If you use a 1 KV ceramic capacitor it would protect the circuit from the high voltages often found in tube-type equipment. While listening to the tone in a radio's speaker you can walk your way through the stages with the probe to isolate where the problem is. Hand capacity provides enough signal return.

To use as a TDR, connect your 10X oscilloscope probe to the loop at the BNC, ground clip to the nearby ground. With the long-nosed probe and the normal ground clip you will get quite a bit of overshoot on the leading edge of the pulse. You can use the little spring clip ground to reduce this overshoot but there isn't really a reason to do so. That overshoot doesn't affect TDR measurement. You need to use your 10X oscilloscope probe with this. Without it the bandwidth is too low and is un-compensated.

Apply power and you should observe a nice square wave. Increase scan rate to observe the leading edge of the pulse in detail. Attach a 50 ohm cable to the BNC jack and notice the change. If the cable is several feet long and the far end is open then you will see an additional step appear. The time from the initial pulse to this new step (that just appeared) represents the time that it took for the signal to propagate through the cable, bounce off the open end and return to the probe. [1]

If the discontinuity (in this case the open end) represents an impedance higher than that of the cable then the return pulse will be additive and will be higher amplitude than the first part of the display. If you short the far end then this return will be subtractive and will show as a step lower than the prior step amplitude. The amplitude of the return step is affected by both the cable loss and the change in impedance. A pinched cable will show a dip in the waveform since it creates a lower impedance discontinuity.

Terminate the cable with the proper resistance and it will appear to be infinitely long to the TDR. No reflection. [2]

With long cables the details of the reflected signal will diminish due to the high frequency losses of the coax. Still, you can measure several thousand feet with surprising accuracy. Not that you would ever need to.

Since the speed of light in a vacuum is about 186,282 miles per second then it will travel just a bit more than 11.8 inches per nanosecond. If we are measuring the distance to a fault then we need to take into account that the signal must travel to the fault and back. Thus we could measure the delay of the returning signal in nanoseconds and multiply by 5.9 to calculate the distance to the fault (in inches)...that is if the signal were traveling in free space, which it is not. We need to factor in the slower propagation rate of the cable dielectric! This is where Velocity Factor (VF) comes in. It's published for most cable designs but if you have a sample of the cable to experiment with then you can get a more accurate value. Multiply the VF value to get the correct distance.

The cable in figures 3 (Short) and 4 (Open) had a measured delay of 274 nanoseconds. The VF is 0.75 for this type of coax so every nanosecond represents 4.425 inches ( .75*5.9 ). Thus, the distance to fault (in this case the end of the cable) is 274*4.425 = 1,212.45 inches or 101 feet. The reel was supposed to be 100 feet which is close enough considering that the published VF may not be that accurate. 1% is good accuracy.

But wait! There's even more!

By measuring the output risetime of this gadget, you can approximate the actual 3 dB bandwidth of your oscilloscope and probe. Divide .35 by the measured risetime and the result is that bandwidth. Nifty.

Discrete logic chips are so common and cheap that we tend to disregard them. For pennies we get high performance amplifiers and switches that are available in plastic packages that are robust, easy to use, and ideal for battery power. Consider having a few in a drawer, ready for simple projects.

de ND6T


[1] It doesn't matter if you use the beginning or the maximum of the pulse just as long as you use the corresponding portion for both starting and ending measurement. Most DSOs have an automatic cursor feature. Use it if you have it.

[2] Notice that I said “resistance” instead of “impedance”. This is a wide-bandwidth test, from DC to beyond your scope limits, and the termination must be insensitive to frequencies within that range. Antennas appear as opens or shorts depending on their feed methods.

Return to Home