TDR

I've done both frequency domain and time domain measurements and you can easily use Fourier transforms to get from one to the other after the fact. Invariably, I found that HP8753D VNA measurements followed by an inverse Fourier transform were superior in terms of S/N.

My TDR was a Tektronix S-6 sampler and an S-52 pulser, which have a slight edge in terms of bandwidth. If I averaged >10k measurments, the S/N approached that of the HP8753D, but a lot of patience was needed.

In the end, sometimes one is better, sometimes the other.

Jeroen Belleman

Sure. Several Tek SD-24s and an 11801C mainframe (our 11802 is still sick).

We also license a low cost, miniature, sub-100ps TDR of our own.

TDR gives you good info on the nearest discontinuity and less good info on succeeding ones. (How much less depends on what the nearby one does to the pulse edge.)

Cheers

Phil Hobbs

A long time ago I was helping with the measurement of concert hall acoustics. I made a very powerful portable spark generator which produced peak sound pressure levels of around 140dB spl at 3m range. The impulse response of the hall was recorded with a high quality cassette recorder. The results were disappointing, even with averaging of hundreds of impulses. The problem was that a huge dynamic range was needed with this method. In contrast, playing a pseudo- random noise signal made much better use of the available dynamic range. I think this maps fairly directly to the VNA vs TDR question. John

To this I would add one comment:

VNAs have fairly narrow Receive and Video Bandwidths, often a few MHz, so the smallest discontinuity must be tens of meters long.

By contrast, a TDR has a bandwidth in the GHz, so far shorter discontinuities may be detected.

Joe

Yeah, I wasn't trying to imply that these two pieces of test equipment could be used the 'other way round' as it were. It would never occur to me to ever try using a VNA for DTF type measurments.

No

Okay - now you've got me intrigued. I have a spark generator here which I keep unassembled unless needed. It provides pretty spectacular and intimidating 60kV sparks with an alarmingly loud crackle. But not remotely as noisy as yours. I have to ask - what the hell kind of voltage were you generating??

This is well worth a read, Phil:

About 3kV at over 1kA. It was a large photoflash capacitor about the size of a shoe box with oil impregnated paper dielectric salvaged from a huge ruby laser power supply. It could be completely discharged in about 10us. I built a 3-electrode triggered spark gap with tungsten rod electrodes which eroded fairly quickly. Two electrodes were connected to the terminals of the capacitor with short brass rods and the third was connected to a motorcycle ignition coil with a thyristor capacitor discharge to ionise the main gap. The sound level was measured with a Bruel & Kjaer peak holding sound level meter using a half inch microphone. Thinking back, it was probably only 120dB to 130dB spl pk at 3m. Still VERY loud. John

No, no, it doesn't work like that. You use the VNA to acquire the S11 frequency response over its full bandwidth. You then apply an inverse Fourier transform to the complex S11 frequency data to obtain the impulse response. Then you apply a time integration to finally get the TDR plot with an equivalent bandwidth of the full sweep of the VNA.

Note that the VNA has a directional coupler, so you'll get only the reflection, not the incident *plus* the reflection like a TDR. The nice thing is that with proper VNA calibration, the reflection will start exactly at time zero and many setup parasitics are suppressed.

I could resolve centimeter-sized discontinuities with ease with S11 data from the HP8753D sweeping up to 6GHz.

Jeroen Belleman