Hello,

I'm confused about the difference between Time Domain Reflectometry (option 010 on our N5242A) and simply setting the sweep type to CW freq. I can see that a whole lot more numerical horsepower goes into a TDR, because it's taking a broadband frequency sweep, and calculating the inverse Fourier transform to put us in the time domain. What would be the difference in interpretation?

Also, I've noticed that the starting level of Re(S11) in the transform mode seems to vary as we make changes to capacitances on the MCM module. Is this a mathematical artifact? nothing about my edge launch connector or the transmission line leading up to the MCM has changed.

Thanks so much!

yours,

Raj

I'm confused about the difference between Time Domain Reflectometry (option 010 on our N5242A) and simply setting the sweep type to CW freq. I can see that a whole lot more numerical horsepower goes into a TDR, because it's taking a broadband frequency sweep, and calculating the inverse Fourier transform to put us in the time domain. What would be the difference in interpretation?

Also, I've noticed that the starting level of Re(S11) in the transform mode seems to vary as we make changes to capacitances on the MCM module. Is this a mathematical artifact? nothing about my edge launch connector or the transmission line leading up to the MCM has changed.

Thanks so much!

yours,

Raj

In CW time, we measure the mag and phase of a response (say, S11) at a single frequency and we measure it over and over again and plot the response as a function of time. It is really only a single frequency measurement, and if the DUT is constant over time, it doesn't show anything interesting. If the DUT is pulsed, for example, you can see the pulse envelope in the CW time sweep.

Now, S11 in the low pass mode will be pure-real, because the time response of any network is real. We actually enforce this by our math, so the imaginary part in low pass is always computed as zero, because we assume the negative frequency response is identical to the positive.

The starting, and ending values are computed as a result of the DC point of the circuit. We can't measure DC directly. but we infer it from the first three frequency response data points. The troube comes when the length to the first response is so long, the S11 phase changes too much, and we don't get a good extrapolation of DC. This causes the DC level to jump. Often, adding more points will improve this response. Now, if you are capacitively coupled, the capacitor at the input can have a lot of frequency response at low freuqency, so the phase may shift a lot and not allow the S-parameter (in frequency domain) to be extrapolated the same. Interestingly, if we did do a perfect job of extrapolating, we would see DC=infinite (or at least big) because of the input blocking capacitor.