The mail function has been disabled by an administrator.

Deprecated: The each() function is deprecated. This message will be suppressed on further calls in /var/www/rv8/media/foundry/3.1/libraries/cssmin.php on line 2236

I gathered a bunch of airspeed error data on flights 8 and 11, over two weeks ago, and this week I finally did some detailed analysis of the data. The basic test and analysis method is described in my Determining Static System Error document.

The test technique was to fly a four leg box pattern at each test speed, being careful to fly as closely as possible to the target IAS and altitude, at a constant heading. On each leg, I recorded GPS track and ground speed and OAT. My laptop computer was also recording data from the EFIS, GPS, and engine monitor to supplement the hand-recorded data. I flew a range of speeds from 55 to 180 KIAS. The bulk of the testing was done at 3500 or 7500 ft, to be in smooth air. I couldn’t maintain 180 KIAS at altitude, but I managed to find smooth air at 1500 ft early one morning.

Post flight, I used the GPS data to determine the TAS for each test point, using Doug Gray’s method, using the NTPS spreadsheet. I corrected the indicated OAT for the probe’s ram temperature recovery factor (my test results show the recovery factor is somewhere between 0.8 and 0.95, depending on how you look at the data - I am using 0.85 for my analysis here, but the results don't change significantly if I change the assumed recovery factor to 0.8 or 0.95). I used the TAS and corrected OAT to determine CAS. I used the results of my EFIS ASI instrument error ground test to correct the IAS, and the difference between the corrected IAS and the CAS must be due to static source position error.

I was very pleased to see that the static source position error is very low. The error is less than 1 kt at speeds up to 160 kt, and increases to about 1.7 kt at 180 KIAS. The error would make the IAS read too low at low speed, and read too high at high speed.

I am using Van's standard "pop-rivet" static ports, and a AN5812 pitot tube.

The theory says that the error will be a function of angle of attack, and the relationship between angle of attack and speed varies with weight. So, the airspeed error vs IAS curve would change slightly with weight. But, the error is small enough that I am not in a rush to calculate the corrections for other weights.

Note that this chart does not include the EFIS ASI instrument error.


I determined the EFIS and analog ASI instrument errors using a water manometer. I did the test twice, and got similar results. This chart shows the total airspeed system error = instrument error + static source position error for the EFIS.


I was quite surprised to see how closely the hand recorded data matched up to the data recorded on the laptop. I stabilized on each leg of the four leg box for about one minute, recording the data on a knee board when everything seemed to be nicely stabilized. Post-flight, I studied the data on the laptop, looking for the most stable section of each leg of the box, and carefully selected the start and end times to analyze for each leg so the average IAS matched the target within 0.05 kt, and the altitude matched the target altitude within 20 ft. I averaged the GPS track and groundspeed for each of those legs, and used those averages to calculate the TAS.

The results calculated from the hand recorded data matches up extremely closely to the results calculated from the data on the laptop, except for an aberration at 170 KIAS. There is clearly something wrong with the hand-recorded data at 170 kt. It differs from the laptop data by about 2 kt, and is clearly out of line with the rest of the hand recorded data. I probably wrote down one of the numbers wrong. The problem was evident when I looked at the standard deviation in the NTPS spreadsheet - this one showed about 2 kt standard deviation whereas the rest of the hand recorded data had a standard deviation of about 0.5 kt.

Leave your comments

Post comment as a guest

0 / 1000 Character restriction
Your text should be less than 1000 characters
  • No comments found