Measuring the orientation sensitivity of a UHF RFID transponder, where do the deviations come from?

I measured the orientation sensitivity of a UHF RFID tag using a Voyantic Tagformance Lite RFID test device. I rotated the tag along the 3 axes using a stepper motor and a rotating table.

The device measures the path losses from the device to the tag location using a reference tag with known properties. I positioned the reference tag to the measurement location and started the calibration process. The device does a frequency sweep from 800-1000 MHz ands sends a read command at varying power levels until the tag responds. Once the tag responds, the path loss is calculated using the known properties of the reference tag.

Path loss = Transmitted power - Power on Tag

• Power on tag is a known property of the reference tag
• Transmitted power is the power level at which the tag responds

After the path loss has been measured for every frequency, the reference tag can be replaced with the tag under test. Using the same measurement as above, the device calculates the read range with following formula:

Where $$\P_{tag}\$$ = transmitted power (at which tag responds) - path loss (measured using the reference tag)

and $$\P_{max,EIRP}\$$ = 3.28 W

This is the first measurement orientation (I measured 4 copies of the same tag)

and here is the result as a polar diagram. Depicted is the read range as a function of orientation angle

Next rotation

and the result

Last measurement

Result

My questions:

1. The last measurement shows large deviations between the tags, what could be the reason for this? Could it be reflections from the rotating table? I could not properly reproduce the measurements, I always get more or less different results for this measurement.

Edit: What causes the fluctuations from angle 0 - angle 180? 180° - 270° is exactly what I expected. The material I put the tag on is styrodur, it is pretty much electromagnetically neutral.

2. Why is the overall shape of measurement 1 and measurement 2 different? I would have expected a very close result between those two measurements.

Overall, the results are more or less what I expected but there are some discrepancies that I can't explain. Can anyone shed some light on this? My guess would be that the rotating table is somehow interfering with the measurement in the last two experiments.