What's Wrong with Microsoft Flight Simulator

By Chuck Forsberg caf@omen.com

I'm working on my practical training for the IFR (Instrument Flight Rules) rating. I own and fly a 1964 Cessna Skylane, N2469R. To fly on the gauges one must be able to hold a course and altitude solely by reference to instruments as a matter of habit, requiring little conscious thought, much as one drives a car without thinking about it. This skill is commonly called "instrument scan". Unless this skill is developed (and maintained), the pilot will not have sufficient "spare time" to deal with navigation, procedures, communications, and other requirements of IFR flight. Since I have a 200 mHz Pentium Pro with all the trimmings, it seemed logical to use Microsoft Flight Simulator 98 to practice my instrument scan.

Some dismiss FS98 as nothing more than a kinder and gentler home computer arcade game. On the contrary, Microsoft advertising insists FS98 is "As Real as it Gets."^TM Microsoft's web page promotes FS98 for IFR training. (The Flight Safety product does not include corrected flight models.) Microsoft claim Cessna approved the flight models for their virtual Skylanes.

Does the reality approach the Microsoft hype??

Simulators are usually cheaper than the real thing, and one can practice maneuvers that would be too dangerous, too tedious, or too expensive to do in real life. The concern is that the techniques and responses you learn in simming may not work in real life. It's a given that a PC based flight simulator is not going to recreate the tactile sense of motion. But that's no excuse for the simulator that proudly proclaims itself "As Real as it Gets"^TM not to accurately model basic flight dynamics and instrument displays of the target aircraft.

I encountered a show stopper defect when I fired up Microsoft FS98 for IFR practice. The Microsoft 182 and 182RG aircraft supplied with FS98 have defective Attitude Indicators. The AI displays a representation of the horizon outside. Controlling the aircraft in IFR flight is difficult if the AI is wrong. Airplane crashes have been caused by faulty attitude indicators. The attitude indicator in the FS98 Cessna singles is way off. The Microsoft Attitude Indicator has a superbly rendered adjustment knob that does absolutely nothing but waste memory and CPU cycles. I have not been able to get Microsoft to address this bug.

To get around this defect, I tried user written aircraft. I found a user written DeHavilland Twin Otter. This plane does not fly well when converted to FS98 but at least the AI was set correctly.

The Twin Otter does not resemble my 1964 Skylane. It has five times the power and grosses several times as much. It climbs at 1600 fpm. The virtue of the Otter is that it was the first plane anything like a Cessna that worked with FS98 and flew straight and level when the stupid Microsoft attitude indicator said it should.

I then found a user written Feissler Storch (storch.zip) that also flew level when the AI said it should. This file has been converted between different versions of MSFS. Apparently something was lost in the translations. It's flight characteristics can best described as strange. The joystick has to be turned up to control this plane. If you enjoy adverse yaw, this is your plane. It takes full rudder input to make a standard rate turn (3 degrees per second, not much of a bank in a bugsmasher).

Landing this thing is a trip in itself. If the landing gear of this WWII tailwheel STOL plane were accurately modeled I would have had nothing but spectacular grounds loops. Sometimes an inaccurate simulation has its advantages.

Despite these faults, the virtual Storch responds to turbulence in a way that reminds me of my real life Skylane.

I also found a freeware IFR control panel for the 182RG (cessnifr.zip). I hacked the layout to resemble N2469R's current control panel. A quick examination of panel.cfg should give you all you plenty of ideas on how to hack it to match a particular plane. The three numbers on each line represent X, Y, and size in pixels. (Hint: do not delete items. If you wish to "delete" an item, just move it off screen.) You are welcome to use this hacked panel.cfg file as you wish. I flew this contraption for several hours, practicing holding course and altitude in light turbulence, and shooting the odd approach. As I got better I threw in more wind and turbulence. My instrument scan was much better in my next real life lesson.

I then found a virtual Cessna Aerobat with 150 HP engine (c152hp.zip). It's a recently written plane that has not suffered so many translations. It is overly sensitive to turbulence, but so is a real 152. The virtual 152 doesn't have a constant speed prop, but that's actually an advantage because Microsoft's simulation of a constant speed prop is so poor.

Below is a list of inaccuracies in FS98 that detract from its usefulness as a training aid for IFR flight.

Defective attitude indicator which indicates a pronounced climb during level flight. Just below the artificial horizon display Microsoft draws a superbly rendered adjusting knob which does absolutely nothing. In a real plane the knob is adjusted after transition between cruise and approach/holding speeds.

The elevator trim in FS98 is quite unlike to the real thing in my Skylane. In the real plane one uses trim to relieve pressure on the controls. Once the desired flight profile is obtained using the yoke, one can set the trim very close to the needed setting in a matter of seconds. The FS98 trim setting mechanism does not give good feedback, requiring a prolonged period of trial and error to get the desired setting. To be fair, I should mention that other sims I've used aren't much better in this respect.

The operation of a constant speed prop is poorly emulated. In a real Skylane, changes in timber and sound give a pilot information about flight conditions. In fact, a real constant speed prop doesn't run at a constant speed at all; it behaves somewhat like an automobile transmission. Most of the cues a real pilot hears are absent in FS98. The nearly illegible manifold pressure gauge compounds the problem.

Navaid signal coverage is grossly inaccurate. This is bad enough that some procedures simply cannot be flown as intended. The Farmington-3 departure from HIO Runway 20 calls for a left turn to a heading of 120, then intercept and proceed via UBG 346 degree radial (course 166) to the Newberg VOR. The Newberg VOR is perched on the summit of Ribbon Ridge overlooking the airport from 11 miles south and 1200 feet higher. Planes on the airport can receive the VOR on the ground.

In the picture at right, delayed reception of the Microsoft VOR results in the path shown in red. The plane is way off to the bottom right of this map before the Microsoft VOR is received. The VOR course cannot be intercepted except by doubling back! The course track in black shows how the Farmington-3 departure from Runway 20 would be flown with real VOR reception.

Beyond textbook basics, Microsoft VORs do not behave like the real thing. The FS98 indicators do not fluctuate like real ones do. Finding a VOR radial in FS98 is little like the real thing.

The FS98 turn indicator twitches as if it were receiving Morse Code, making it nearly useless for controlling the rate of turn or for controlling the airplane in partial panel (vacuum failure) situations.

The turn indicator in my Skylane gives a smooth indication that makes it easy to execute standard rate turns (the basis of many IFR maneuvers). I haven't had that much partial panel time yet, but the turn indicator in my airplane makes it easy to keep the plane right side up. (I'm still working on accurately holding a course with partial panel.)

The stock panel does not include a timer. What appears to be a timer in the Microsoft Skylanes is merely an indicator of the simulation rate. Admittedly a user supplied stopwatch/timer can be used, but some of the considerable expense that went into making a "cool" looking cockpit could have been used to make it correct. In my real airplane, I use the timer function in my KR87 ADF.

I've never heard of anyone getting vertigo using Microsoft Flight Simulator. To be fair, I haven't heard of anyone getting vertigo in any other PC based sim, or even under the hood in a real plane. If you get IFR instruction, make sure some of it is in actual. Otherwise, the first time you fly into actual IMC may be your last, as it has been for some.

The shame of all this is that Microsoft could correct their attitude indicator and turn indicator without straining their resources one whit. Unfortunately Microsoft has not even bothered to respond to my bug reports, let alone fix the bugs. Why can't Microsoft straighten up and fly right? Why such an attitude?

The same technique can be used to diagnose ADF approaches. The picture on left is a plot of the NDB approach to PDX runway 28L "flown" for the first time in a howling crosswind. Notice how the procedure turn (lower right) is mostly obliterated. (Picture edited for clarity)

If you have SIMGPS you can send GPS signals to a second computer running a moving map display such as RMS Technology's VISTA.

To speed up your instrument scan, just do your IFR maneuvers with the simulator running at 2x or 4x.

I have not tried any other mass market flight simulator yet. Their manufacturers are welcome to send me a review copy!

Purpose built IFR simulators such as ASA, Frasca and Jeppesen have accurate flight models but their cost tends to discourage individual pilots.

On Top runs under DOS using a 32 bit memory manager. My system is fitted with a Creative Sound Blaster Live! PCI sound board which at first greeted On Top with silence. The SB Live! DOS emulation was not enabled because all the good IRQs (hardware interrupt lines) had been spoken for. After 8 hours of hacking and rebooting Windows 98 (and nearly deciding to reinstall Windows!) I went into the motherboard BIOS and changed the PCI interrupt status of IRQ 10 and IRQ 15. On Top still does not take input from the throttle on my Sidewinder. Because On Top is not a "well behaved" program, standard screen capture programs won't work with it. If Windows times out and enters screen saver mode while On Top is running, things get really messey. A native 32 bit Windows version would be welcome.

Aside from the choice of operating system, On Top is a credible IFR simulator indeed. An certified instrument instructor (CFII) and $2000 of special hardware make this into a PCATD suitable for maintaining IFR currency. Nobody presets the direction gyro for you, just like on a real plane. The attitude indicator has a working adjustment knob. The turn coordinator really works. Blimey.

On Top has a number of touches that suggest real IFR pilots were involved with the design process. I heartily recommend downloading the simulator and its manual (106 pages in Adobe Acrobat format).

After 8 or 10 frustrating hours of hacking I wasn't in the mood for a bloody checklist, so I grabbed the NDB approach plate for Helena (the only place the demo will fly) and took off. The right button the mouse gives instant access to the throttle. I didn't take the time to familiarize myself with either the airplane or the approach plate. The Helena NDB approach is best studied before the eyes glaze over. Nonetheless I did get the plane up and down without incurring any messages about crashing. Had I done the same thing in a real airplane I should have suffered serious injury at the hands of the CFI or check pilot.
No wonder the vertical speed is drawn in purple.

Many real life pilots have a hard time landing a flight simulator such as MSFS. They complain the sim lacks cues they are used to seeing. Student pilots learning on real aircraft use a variety of visual cues during landing. These cues are different or lacking on a PC based flight simulator.

People who first learn landing MSFS are used to a different set of visual cues. Simmers may learn bad habits that make it hard to accurately land a real airplane.

Practicing stalls with a sim is a mixed bag. To the extent that one learns to keep the nose down at the right times, that's a good thing. But sims may not teach one to use rudder to pick up a wing that drops in slow flight. In slow flight, trying to pick up a wing with the ailerons can make things worse. Slow flight is called the "region of reverse command" because some of the controls don't provide the response they normally so.

It is tempting to practice emergency flight maneuvers on FS98. There is no penalty for misjudgement. But despite being checked by Cessna, the flight models may not be accurate. If you practice an engine failure on takeoff, will the real Skylane do what the Microsoft Skylane did? I've never seen a PC based flight sim get into a spin that looks like what I've seen on King Videos.

Simmers who aspire to flying real aircraft should practice flying and landing without instruments - use the "W" key to make the panel disappear. A common criticism of ex-simmers is that they don't spend enough time looking outside the cockpit. There are too many midair collisions to accept this habit.

A variety of hardware is available to the simmer. A throttle control is essential. A rudder control is important.

I use a Microsoft "Sidewinder 3d Pro" joystick. The rudder is controlled by twisting the grip. I don't know of any plane that works this way, but the joystick is compact compared to a yoke and pedals. Setting it up for a quickie doesn't take an act of Congress. Given the inaccuracies of the flight simulations available, I wouldn't waste time and effort getting anything better for IFR practice.

I also have a CH Products "Virtual Pilot Pro" joystick and a set of pedals. I used this combination for a while to improve my rudder coordination. Unfortunately, the yoke's centering mechanism is so crude that I rather dislike flying with this yoke. I don't use it anymore.

Aside from the "learning disabilities" in FS98, the much vaunted "As Real as it Gets"^TM out-the-window view leaves something to be desired. In my first IFR lesson involving actual Instrument Meteorological Conditions (IMC) one of the delights was the incredible view while flying VFR on top of a low stratus layer. Hills and mountains poked thru the clouds in a fairly tale setting. Microsoft can't get anywhere near the real thing. I pity those who are limited to Microsoft's vision of flight.

Captain Chuck's Excellent flying Experiences