P440 Data Sheet / User Guide
DRAFT
53
reasonable (e.g., operation at minimum range and maximum transmit power). It is believed to be a
conservative estimate of the system’s ranging Precision and Accuracy.
The precision of measurements can be improved with averaging. Measuring 6 readings will
normally improve the accuracy by a factor of 2. Averaging more than 6 readings will have only
marginal improvements.
The measurement campaign referenced above was conducted in 2011. Since that time several
significant improvements have been made to the range measurement algorithm. The overall
measurement accuracy (precision and bias) has been field-verified using a survey-grade laser. While
the test was limited in scope, the results suggest that the overall measurement accuracy is
substantially better than the value quoted in this specification. These tests were conducted in an open
field and indicate overall accuracy closer to 1 cm. While the tests’ size is too limited to warrant a
change in the published specification, the user should expect similar results, especially when
operating in a similar environment.
By constraining operation to a narrower field of application, users have achieved higher performance.
One user has reported accuracies of 0.5 cm while a second reports 2 mm accuracy.
6.4.2 Precision and Accuracy in NLOS Conditions
Time Domain does not have a specification for accuracy in Non-Line of Sight (NLOS) environments.
This is because of the wide variety of conditions that can be encountered. For example, if one is
measuring range inside a building that is constructed of wood frame and drywall (also known as
sheetrock or gypsum board), then one will experience a level of performance that is less than but
close to LOS conditions. This is because wood and drywall do not significantly attenuate or disperse
RF signals at the P440’s operating frequency. At the opposite end of the propagation spectrum would
be operation inside a metal ship. Because metal blocks radio frequencies, the operating range would
be limited to the size of the room. Operation in NLOS conditions must be determined empirically.
Having said that, we routinely measure range from one side of our office space to the other (a
distance of 30 meters through an environment that is not only NLOS but is also occluded by large
amounts of metal) with an accuracy of better than +/- 1 meter.
6.4.3 Bias and Calibration
The Precision (Standard Deviation) and Accuracy (Bias error) of the TW-TOF range measurement
are called out in
Section 6.1, Table 3
and discussed in
Section 6.4.1
.
It should be noted that Bias is measured using the standard Broadspec antenna with a 90-degree SMA
connector.
When a P440 leaves the factory its bias will be zeroed to within a few centimeters. This ensures that
when a user takes a range measurement, the result should compare favorably with a measurement
taken with a laser. This assumes that the measurement is clear line of sight, not in compression and
not affected by Fresnel cancellation or enhancement. If that is the case, then most users will find the
result satisfactory. Furthermore, this result will not change materially over the operating temperature
range.
However, sometimes it is necessary to use a different cable, connector or antenna. In this case the
Bias will change. Furthermore, some applications may require a more accurate zeroing of the Bias
than what is normally provided from the factory. In either of these cases it will be necessary to
characterized and adjust individual P440s. Application note
320-0327 Distributed Calibration of
Time Domain UWB Radios
describes such a process. Once the user has determined by how many cm
the calibration needs to be adjusted, then this number should be converted to picoseconds and entered
through the RCM_Set_Config_Request API message or through the Configuration Tab on the
