Naturally, I was excited to hear that a pro developer was creating a native FSX P-38, and I was delighted to find their principle 3D modeler was Gibbage -- a renowned 3D modeler and fellow P-38 nut.  Seeing as I was never satisfied with my FS2004 models in FSX, I figured this was an opportunity I just couldn't pass up.

My final versions of both the sound set and flight model were made for FS2004 and I knew there would be some tweaking involved to get them to behave as they should in FSX.  I just didn't feel comfortable letting anyone tweak my babies, so I consented to come out of FS retirement to rework the sounds and flight model that kept me occupied for about a decade.

I am now in the process of revisiting each and every sound file, every line in the sound configuration and every aspect of the flight model to tune them all for FSX.   Many people are familiar with the visual aspects of FS development; it's relatively easy for developers to show the visual progress of their model.  Not many simmers know as much about goes on behind the scenes in developing the sounds and flight model.   So I decided to make this little "blog"  to share some of my progress.

Flight testing: home base and fun

For each model I work on I choose an airport to work from.  All test flights are based out of that particular airport during the entire development, unless some special circumstances warrant otherwise.  For this FSX rework of the sounds and flight model, I have chosen Fairchild International in Port Angeles, Washington (KCLM).  I grew up in the Seattle area and often pick Washington airports for my base of operations.  In this case, where I need to revisit each flight model parameter I needed an airport at sea level and near a large body of water (for safe dive and stall testing).  As a bonus I've got the Olympic Mountains nearby and one of my favorite cities, Victoria B.C. Canada just across the way.

I like to have fun flight testing, so I when practical, I set up a flight test plan and go through all the motions as if I were testing a real a/c.  In other words, I'm not just putting the plane a set altitude and speed and tweaking parameters.  I take my planes through full-length flight tests.  For example, when I test high altitude performance I don't just slew up, or place my a/c at a set altitude, I start at the ground and fly up to altitiude.  As such, by the time I'm done with a flight model, I've spent a lot of time with it. (I have literally logged several thousand hours in virtual P-38s over my "career")

Below is a shot I snapped this evening as I did some stall testing.  I spent 45 minutes flying back and forth across the ~10 mile wide Strait of Juan de Fuca and tested 1g clean and dirty stall characteristics with various load conditions, making minor adjustments to lift parameters. 

BIG NOTE: the visual model (exterior and interior) is my P-38J-15-LO and NOT the Milviz/FSD P-38!!!

Flight testing: migrating from the old to the new

I spent tonight working with the milviz P-38 visual model and my model, noting differences that may affect the migration from mine to theirs, such as model center point relative to CoG, contact points, thrust lines, exhaust port locations, empty static angle, full-power angle, gear compression ratio, light points, etc.  For the most part, things matched up pretty well, but several things need tweaks here and there.

After making some of those tweks, I took the plane out for a spin afterwards and did a few touch-n-go's. 

Sound -- my life's work and in FS

I thought it might be of interest to tell a little about myself related to the sound part of this P-38 project, and offer a few general comments about sound in FS.

My Background

By profession, I hold a BS in Physics and a MS in Acoustical Physics, I am INCE Board Certified (akin to a professional engineer, but in noise control) with nearly 20 years of experience in acoustical physics working with vibro-acoustic and aero-acoustic systems. I  currently work for a major industrial company that makes big yellow tractors, where I work a lot with engine noise and fan noise, among other things.  I am a university-trained musician, with 30+ years of experience playing, recording and making music.  I have my own recording studio, lots of instruments, equipment, etc. and have released three CD.  As an avid warbird fan (mostly a certain twin!) I carry a binaural portable digital recording system with me to airshows and record pretty much everything I can get. So I have accumulated a fair number of original recordings from many different kinds of planes. I also record many other industrial and mechanical sounds for my library.  I have contacts in most areas of industry and have obtained great recordings from them as well.

A Few General Comments about Sound in Flight Simulator

Sound design in FS is not as well-defined as say making the 3D model or the texture set.  It is not as black-and-white or “right” or “wrong.” As such, the FS designer must use quite a bit of art to recreate the essence of the plane’s characteristic sound. And where’s there’s art, there is interpretation.

The sound playback engine is probably the least sophisticated portion of the simulator. If we feel that the flight model is indeed capable of making “1%” flight models, I would say the sound playback system can probably at within 20%-30%. Basically, we have sound files which can be looped (played over and over smoothly) that we mix and match per the engine / prop settings. (Plus all the other miscellaneous sounds, such as doors, gear, etc.)

Even if one has a good recording of the actual plane, putting a loop from that recording into the sim does not guarantee perfect results. The game does affect the sound such that the wave file often needs significant modification to get it to sound good in the game.

Matching up the low, medium and high rpm sounds is another difficult task. In a real physical system with rotating components, there are several characteristics of sound: tonal, broadband and impulsive noise. Cyclical sounds, such as the propeller or engine combustion will have well-defined tonal components that increase in frequency with rpm at some multiple of the primary rotational frequency. These are called orders. As these orders increase they may excite system resonances or antiresonances, augmenting or attenuating the tone respectively.

The orders can be simulated fairly well with the simple FS playback system. However, system resonances, which are a characteristics of the system remain stationary and, due to the simplistic nature of the sim, can not be represented well in the sim. The very nature of the sim prevents it.  Broadband sounds can be simulated reasonably well, as with impact noises, depending on the type.

With FSX, the directivity of a sound source can be crudely modeled.  Prior to FSX all sound sources radiated uniformly in all directions.  This is called monopole radiation, or spherical divergence.  FSX introduced the idea of a sound "cone" which can be use to approximate dipole and, if the sound designer is clever, quadrupole radiation, as often occurs with high speed propeller / fans such as the propeller of the P-38. 

Just Struttin' Around

After working through (but still not quite resolving) some technical difficulties, as well as an unplanned trip out of town, I spent tonight working on the gear suspension.   The image below shows the model in profile with some of the key lines and angles to which I compare Lockheed prints and specifications and the many hundreds of photos in my books and private collection. 

There was quite a variety in how the '38 sat statically.  Some pictures show it at quite an angle, with the front oleo nearly fully extended, while others show it nearly fully compressed.  Perhaps the variations were due to different standards of maintenance, or because of "tuning" the suspension for a particular type of field, or even the amount of ammunition (or photo equipment) in the nose, or a combination of these and more. 

Lockheed specs indicate a nominal ~5.6-degree nose-up static angle, but I actually prefer something closer to 6.5 to give it a bit more personality.  The image above shows an angle closer to 8.5 degrees with full ammo, still a bit too steep. 

Along with the static angle, one has to also be mindful of each gear's compression ratio, static deflection, damping ratio and such.  So even though the plane may sit right, it may handle horribly during taxi, or not reach the proper angle during run up to take off.  All the gear parameters have to adjusted in concert, all the while testing, testing and more, testing.

The image below shows a tried-and-true trick for working with contact points -- putting nav lights at the contact points of interest.  In this case, I am working on the extended gear points.

Up and Down

Tonight's task was to verify time to climb from sea level to 20,000 ft, some high altitude handling and compressibility tuck. 

It is well known that the '38 was one of the first planes to encounter difficulties with compressibility so this is an important feature to implement in the flight model.   The image below was taken a few seconds into a power dive from 30,000 ft as I sped past 400 mph indicated and began to feel the loss of elevator authority. 

Not long after, the nose began to tuck under and I had lost most control.  Despite this being the L visual model, I was actually testing the J flight model so I did not have the dive recovery flaps. I had to get of this one the old fashioned way: cut the throttles, wait for thicker air and slowly pull out of the dive just before the overspeed caused airframe failure.  I barely made it -- leveled off at about 1,500 ft, going 450mph.