I received a demo version of a new time domain reflectometer (TDR; model T3SP15D from Teledyne Test Tools). I decided to take it for a test drive by measuring all the cables I had lying around my lab. The results were surprising!
What TDR actually measures
Every interconnect is a transmission line, with no exceptions. If the transmission line is uniform, a signal will see the same instantaneous impedance down the length of the interconnect. The one value of instantaneous impedance that characterizes the transmission line is referred to as the characteristic impedance. This term, and the time delay of the transmission line, are the two most important figures of merit of any transmission line.
The TDR measures instantaneous impedance indirectly based on reflected voltages from a 50 Ohm reference source. As the short-rise time edge travels down the device under test (DUT), reflections occur when the instantaneous impedance changes. These reflected signals make their way back to the source where we measure them with a very fast sampling scope.
At each change in instantaneous impedance, Z2, the reflection coefficient, rho, the ratio of the reflected voltage to the incident voltage, is
If we measure the reflection coefficient at the source location where the sampling scope is connected, we can back out the impedance that caused the reflection as:
We infer what the change in the instantaneous impedance must have been based on the reflection from it. The time at which we see the reflection at the source is a measure of the round-trip delay time from the source to the discontinuity.
The TDR will display the measured reflected voltage, and from this, calculate the spatial variation of the instantaneous impedance, down the length of the transmission line. If we can identify the beginning and end of the line, we can also measure the round trip time delay and from this, the one-way time delay of the transmission line.
Figure 1 shows the raw measured reflection coefficient and the backed out instantaneous impedance as displayed on the TDR screen in the real-time measurement. In this example, I connected the TDR to an 8-inch long transmission line on a circuit board.
The T3SP15D TDR not only measures the instantaneous impedance, displayed on the screen in real time, but will also correct for all the internal reflections and cable launch to provide a calibrated measure of the instantaneous impedance of the DUT. Figure 2 shows the calibrated impedance profile for this PCB trace. This measurement takes about five second to take the averages and do the calibration conversions and display the results.
In this circuit board trace, we see the initial capacitive launch of the SMA into the board and the instantaneous impedance of the line varying by about ±1 Ohm across the line, with an average value of 52 Ohms.
Measuring premium 50 Ohm cables
I use a lot of 50 Ohm cables in my lab to connect scopes and even VNAs to various DUTs. I have always assumed the cables were perfectly uniform transmission lines. I try to use high-end (more expensive) premium cables for the really high bandwidth measurements. I’ve always assumed they were 50 Ohms and were uniform.
With my new TDR, I had the perfect tool to quickly evaluate the impedance profile of my cables. The TDR cable ends in a male 2.9 mm connector. This is compatible with both a 3.5 mm and an SMA connector. To look at a coax cable with a BNC connector, I used a simple SMA to BNC adaptor.
What I found was rather surprising.
I first looked at the premium coax cables. Not too surprising, I saw some very uniform transmission lines, very close to 50 Ohms. Figure 3 shows the typical impedance profiles of two good cables. They are 50 Ohms ±0.5 Ohms. This is very respectable performance.
Using the fast mode, I could literally connect the cable and see the impedance profile on the screen in real time. This allowed me to measure a cable and see its impedance profile in less than 5 seconds. I was able to scan through the dozen premium cables I had lying around in less than 5 minutes.
I was shocked to find some of the cables I was using for high bandwidth scope measurements were not so clean and well behaved. Figure 4 shows two cables that had significant discontinuities.
Of the 12 premium cables I had in my lab, four of them had some sort of discontinuities down their length. Only when the discontinuity was > 5 Ohms was I able to see a physical dent or kink in the cable that was an obvious source of imperfection. Needless to say, I purged my lab of the four less than pristine premium cables.
Low-Cost 50-Ohm Cables
Recently I’ve been using low-cost RG174 coax cables I purchased from Amazon. These have either BNC or SMA connectors. I use them in my scope measurements because they are really flexible. I measured one on a VNA a few years back and measured a 6 GHz -3 dB bandwidth. For most of my applications in the < 8 GHz bandwidth regime, I always thought these cables would be just fine.
After measuring the poor premium cables, I wondered just how bad my stock of a dozen RG174 cables would look. Again, I was surprised, but since my expectations were low, pleasantly surprised.
The spec for RG174 coax cables is 50 Ohm characteristic impedance. Most of the low cost RG174 cables came in very close to this value, and were very uniform. Figure 5 is an example of two representative cables.
In this example, you can see the reflection from the SMA termination inside the cable. This is the biggest source of reflections in the cable. These reflections and the series resistance are what limit the bandwidth of the cable. The poor quality termination of the coax to the SMA connector is partly why these cables are so low cost.
The slight upward creep of the impedance profile could be due to a real impedance variation down the cable, or more likely, due to series resistance. The way to distinguish these two possible root causes is to measure the TDR response from both ends of the cable. If it is a real impedance variation, the profile will look different from one end to the other. If it is series resistance, it will look “uphill” in both directions. This comparison is shown in Figure 6.
While the upward slope is clearly due to series resistance, there is a small impedance discontinuity of about 0.2 Ohms visible at 15.5 nsec from port 1 and at 12.2 nsec at port 2. This is a good consistency check for the sensitivity of how small a real impedance discontinuity can be measured.
I was able to cycle through and do quick measurements on all of my RG174 cables in my lab. Of the more than 12 cables, two of them showed out-of-the-ordinary impedance profiles. These are shown compared to two good cables, in Figure 7.
We take cables for granted. We use them all the time for scope measurements, for VNA measurements, and to connect sources to devices. We always assume the cables are perfect. In many applications, the small variations in impedance in a cable are really not important.
But every now and then, we do a measurement that involves a fast edge or high bandwidth signal and something doesn’t look quite right and we want to explore the root cause. In those situations, you should add a bad cable to your list of “usual suspects.” Unless you have verified the quality of your cables, there is always the possibility that even your premium cables may have some imperfections that might be the root cause of measurement artifacts.
A TDR can answer the question about the quality of a cable in about 5 seconds. I was shocked to find that about 25% of the cables I was routinely using in my lab were defective in some way. In less than 30 minutes of testing, I was able to scan through all of my cables and purge all the out of spec cables which eliminates at least one variable from my list when I am performing high bandwidth measurements and analyzing the results.