Measurement methodology can play a big role in correlation and accuracy. For example, the
TDR response in
Figure 11.20
illustrates measurement of a 6-in. low-impedance coupon
using a common hand-held type of probe. Notice the initial inductive spike and the amount of
time it takes the response to settle. The purpose of this illustration is to show the possible
variations in the results due to cursor positioning along the response. Measurements that are
taken during the settling time will contain more variability and will therefore be prone to
incorrect readings. To minimize the variability of measurement location along the response,
a good rule of thumb is to measure the mean of the response along the flattest region, as
highlighted in
Figure 11.20
. This works well provided that the structure is long enough to
allow for sufficient settling time.
Figure 11.20: Measurement of low impedance 6 in. microstrip coupon with an inductive
probe.
The second item to note is that trying to measure a short trace would lead to incorrect
impedance values because there is insufficient time for the response to settle. In the case of
Figure 11.20
, the measurement of the 6-in. trace takes several hundreds picoseconds to
settle. If an identical 1-in. trace, for example, was characterized, the ringing from the
inductive probe would not have time to settle before the reflection from the open at the end
of the trace returns to the sample head. Subsequently, there would be no flat areas of the
response to measure, and it would be very difficult to achieve accurate results. This is a
common error made in today's industry. As mentioned previously, proper coupon design is
also critical for accurate impedance measurements. Test coupons should be designed to
replicate the bus design and therefore should take into account trace geometry, copper
density, and location on the PCB.
RULES OF THUMB: Obtaining Accurate Impedance Measurements with a TDR
Complete measurements with the DUT open circuited. This eliminates step amplitude
variation due to dc loading.
Obtain measured data in the flattest region of the TDR response. This should be far
away from the incident step, where aberrations will be most pronounced.
Try to minimize probe discontinuities and cable loss. Use of small, controlled
impedance microprobes along with low-loss cables will minimize these sources of
error.
Utilize test structures that are long enough to allow the TDR signal to settle.
When designing an impedance coupon, try to replicate the topologies that will be used
in the design