As Yak-52 is not really suitable for any combat simulation and is a very special DCS product, I can show the backstage of the tuning and verification process regarding flight test results.
And this will be a debut unveiling the measured and simulated flight parameters - numerical and coincided at one graph.
The first post will be about its roll rate.
The test was a pure aileron roll starting from a climb to perform 4 full rolls. The rolls started with maximal fast full stick deflection, proceeds with full ailerons and ends with maximal fast stick release (not perfect in both tests though - the real test input was a bit delayed and the DCS test seemes to be overcompensated).
The atmosphere conditions in DCS were set identical to real during the test.
This test allows to check initial roll acceleration (MOIx vs aileron power), steady roll rate (aileron power vs damping) and roll damping (MOIx vs damping). Additionally we obtained roll rate vs speed, and these two parameters (the first divided by the second) can give a ratio that is proportional to so called spiral coefficient (TAS must be used and the wingspan to get it). It is not s necessary to use this spiral coefficient in our case though.
For those who never saw experimental recordings from the real world - the sinusoidal high frequency noise over the curve is a vibrational part. It was not filtered to better preserve phase and time of the original record.
We can clearly see that the model initial angular acceleration is in good accordance with the test results, see
Fig. 4. The deceleration at Fig 2 is in a good accordance to regarding to slightly different control input.
As the initial pitch of both tests were not identical, the speed during the tests is different too, so to compare the roll rate we need to divide roll rate to IAS (Fig. 2).
It is interesting that the real plane has very tiny spiral coefficent slope to high IAS (see the trendline). I can suggest that it could be due to aileron linkage elasticity or wing elasticity or both. As it is barely noticable even a the diagram, I do not think it should be reproduced in the model
And, finally, I’d like to claim that the results represent current stage of model tuning, WIP.
P.S. The uneven (sinusoidal) low frequency at the roll rate is a result of he fact that the X-axis of the sensor was not exactly aligned with the plane axis, so it must be disregarded as a parasite effect.