Is the Viggen navigation system inertial or not?

[renhanxue]

"is the Viggen navigation system inertial or not" - the greatest debate in the history of forums, locked by a moderator after 12,239 pages of heated debate,

 

dril references aside, this question has come up a couple of times before, and earlier today @Rudel_chw asked this in a direct message. This post is mostly a reaction to that.

My answer to the question is "not really", or "not primarily". It does use dead reckoning just like an inertial navigation system (INS) does, and it does use accelerometers and attitude gyros just like an INS does, but it does not primarily rely on them, so it's not really inertial. Pseudo-inertial? Partially-inertial? Whatever you might call it.

An INS by the usual definition of the term only (or at least primarily) relies on accelerometers and attitude gyroscopes to calculate how it's being moved. The accelerometers measure the acceleration of whatever they're mounted to, and the attitude gyros measure the attitude relative to the earth, and if you combine the two you can figure out how the entire system is moving relative to the Earth. That means an INS is by definition ground-referenced (or rather referenced to Earth's field of gravity, I guess).

The Viggen system on the other hand is primarily atmosphere-referenced. I've translated a block diagram from the SFI:

As you can see, it starts not from inertial measurements but rather from air data measurements. First it uses temperature and air data to calculate the true airspeed. Then the angle of attack is used to calculate an estimate of the vertical and longitudinal speed separately (transverse speed is assumed to be negligible here). This is then combined with attitude information from the FLI-37 ADI and the autopilot's moment gyros to calculate the aircraft's true velocity vector relative to the air. To get the movement relative to the ground (which is what we actually want), the nav program then compensates for wind speed, which is either measured using a doppler radar (ground speed measurement), or forecast wind as programmed into the computer (input with VIND/RUTA/MÅL). The ground-referenced velocity vector is then calculated and integrated every 103 milliseconds to continually keep the aircraft's position updated.

This isn't quite good enough though. There are two problems: first, transverse speed (side to side) can't be calculated from air data, and second, the air data system reacts too slowly to really make this precise enough. To mitigate these problems, there's an accelerometer unit that measures acceleration in the longitudinal and vertical axes; acceleration in the transverse axis is taken from the transverse accelerometer that's used for the autopilot's sideslip correction system (y'know that RENFLYGN knob nobody ever touches). These accelerometers react much faster than the air data system, and so they're used to correct the velocity vector calculated from air data. This taken as a whole is good enough.

So, in the transverse axis the system actually is a "real" inertial navigation system - it relies purely on attitude gyros and accelerometers to determine movement in that axis. This, the SFI says, is acceptable because the velocity in the transverse axis is usually more or less net zero (the aircraft doesn't typically fly sideways, after all), with only small variations back and forth, so relying purely on acceleration gives sufficient precision.

You might ask though, if the aircraft has attitude gyros and accelerometers anyway, why not just go purely inertial? The main answer is that to do that you'd need much more expensive equipment. You need more precision than the Viggen's gyros and accelerometers actually have to make it work well, and probably you'd need a dedicated computer for it too. INS systems were absolute state of the art in the early 1960's when the Viggen was designed, with the Minuteman and Apollo guidance systems being some of the first really practical examples. You couldn't just buy an INS off-the-shelf in the 60's; Saab would've had to design one in-house. INS became a lot cheaper and more reliable in the 80's with better computers and the commercialization of the ring laser gyro.

The Viggen system isn't as accurate as a real INS, but at the time it had a significant cost advantage. It's noteworthy that almost the entire system is just a software program that runs in the CK 37. The only hardware it uses that the aircraft wouldn't otherwise have are the two accelerometers for the longitudinal and vertical axes. It also has the side benefit that it's capable of cold starting in two minutes flat because the gyros just aren't that sensitive. True INS systems of the era could take tens of minutes to align from a cold start. It was an engineering compromise: good enough to fulfill the requirements the air force had, but not more complex and expensive than it had to be.

Edited November 25, 2023 by renhanxue

[Rudel_chw]

9 minutes ago, renhanxue said:

It's noteworthy that almost the entire system is just a software program that runs in the CK 37. The only hardware it uses that the aircraft didn't already have are the two accelerometers for the longitudinal and vertical axes.

Thank you so much for clearing-up this matter, as I understood that the Viggen didn’t had an INS system, but didn't knew the real details 

Edited November 21, 2023 by Rudel_chw

[widen76]

@renhanxue - Nice! Thanks for the easy-to-understand explanation! 

[renhanxue]

important follow-up questions:

• do we need an INS alignment chart (ref: the sandwich alignment chart)?
• does anyone have a precise technical definition of the Swedish term "momentgyro"? I get the gist of what they do (I think) but I want all of the details

[QuiGon]

On 11/21/2023 at 9:00 PM, renhanxue said:

"is the Viggen navigation system inertial or not" - the greatest debate in the history of forums, locked by a moderator after 12,239 pages of heated debate,

 

dril references aside, this question has come up a couple of times before, and earlier today @Rudel_chw asked this in a direct message. This post is mostly a reaction to that.

My answer to the question is "not really", or "not primarily". It does use dead reckoning just like an inertial navigation system (INS) does, and it does use accelerometers and attitude gyros just like an INS does, but it does not primarily rely on them, so it's not really inertial. Pseudo-inertial? Partially-inertial? Whatever you might call it.

An INS by the usual definition of the term only (or at least primarily) relies on accelerometers and attitude gyroscopes to calculate how it's being moved. The accelerometers measure the acceleration of whatever they're mounted to, and the attitude gyros measure the attitude relative to the earth, and if you combine the two you can figure out how the entire system is moving relative to the Earth. That means an INS is by definition ground-referenced (or rather referenced to Earth's field of gravity, I guess).

The Viggen system on the other hand is primarily atmosphere-referenced. I've translated a block diagram from the SFI:

As you can see, it starts not from inertial measurements but rather from air data measurements. First it uses temperature and air data to calculate the true airspeed. Then the angle of attack is used to calculate an estimate of the vertical and longitudinal speed separately (transverse speed is assumed to be negligible here). This is then combined with attitude information from the FLI-37 ADI and the autopilot's moment gyros to calculate the aircraft's true velocity vector relative to the air. To get the movement relative to the ground (which is what we actually want), the nav program then compensates for wind speed, which is either measured using a doppler radar (ground speed measurement), or forecast wind as programmed into the computer (input with VIND/RUTA/MÅL). The ground-referenced velocity vector is then calculated and integrated every 103 milliseconds to continually keep the aircraft's position updated.

This isn't quite good enough though. There are two problems: first, transverse speed (side to side) can't be calculated from air data, and second, the air data system reacts too slowly to really make this precise enough. To mitigate these problems, there's an accelerometer unit that measures acceleration in the longitudinal and vertical axes; acceleration in the transverse axis is taken from the transverse accelerometer that's used for the autopilot's sideslip correction system (y'know that RENFLYGN knob nobody ever touches). These accelerometers react much faster than the air data system, and so they're used to correct the velocity vector calculated from air data. This taken as a whole is good enough.

So, in the transverse axis the system actually is a "real" inertial navigation system - it relies purely on attitude gyros and accelerometers to determine movement in that axis. This, the SFI says, is acceptable because the velocity in the transverse axis is usually more or less net zero (the aircraft doesn't typically fly sideways, after all), with only small variations back and forth, so relying purely on acceleration gives sufficient precision.

You might ask though, if the aircraft has attitude gyros and accelerometers anyway, why not just go purely inertial? The main answer is that to do that you'd need much more expensive equipment. You need more precision than the Viggen's gyros and accelerometers actually have to make it work well, and probably you'd need a dedicated computer for it too. INS systems were absolute state of the art in the early 1960's when the Viggen was designed, with the Minuteman and Apollo guidance systems being some of the first really practical examples. INS became a lot cheaper and more reliable in the 80's with better computers and the commercialization of the ring laser gyro.

The Viggen system isn't as accurate as a real INS, but at the time it had a significant cost advantage. It's noteworthy that almost the entire system is just a software program that runs in the CK 37. The only hardware it uses that the aircraft wouldn't otherwise have are the two accelerometers for the longitudinal and vertical axes. It also has the side benefit that it's capable of cold starting in two minutes flat because the gyros just aren't that sensitive. True INS systems of the era could take tens of minutes to align from a cold start.

Awesome write-up!  It gave me some answeres to some detailsq that I was not aware of. Thank you!

[MAXsenna]

@renhanxue

Excellent writeup! Thank you!

Momentgyro. Well, we all know what "in a/a moment" means, and in Norwegian moment can be short for dreimoment (nor.) /(vridmoment swe.) which means torque. Results I find most likely are "control moment gyroscope" or "quick gyro stabiliser". Not sure if they are the same though.

 

https://en.wikipedia.org/wiki/Control_moment_gyroscope?wprov=sfla1

[Machalot]

On 11/21/2023 at 12:00 PM, renhanxue said:

The accelerometers measure the acceleration of whatever they're mounted to, and the attitude gyros measure the attitude relative to the earth, and if you combine the two you can figure out how the entire system is moving relative to the Earth. That means an INS is by definition ground-referenced (or rather referenced to Earth's field of gravity, I guess).

I think it's more accurate to say an INS is by definition inertially-referenced, not Earth-referenced.  The accels and gyros measure motion relative to an inertial reference frame, and the Earth is decidedly not an inertial reference frame by virtue of its rotation rate of 15 deg/hr.  It's easy enough to convert to Earth-relative motion for navigation purposes using a few equations in the software.

 

On 11/21/2023 at 12:22 PM, renhanxue said:

does anyone have a precise technical definition of the Swedish term "momentgyro"? I get the gist of what they do (I think) but I want all of the details

My guess would be "rate gyro" since the outputs of the "moment gyro" in the block diagram are angular rates.  Update: On the other hand In pretty sure the Viggen has a stable platform gyro system, so ignore this. 

 

On 11/21/2023 at 12:00 PM, renhanxue said:

It also has the side benefit that it's capable of cold starting in two minutes flat because the gyros just aren't that sensitive. True INS systems of the era could take tens of minutes to align from a cold start.

(This is my opinion as an engineer, and I'm prepared to defer to you if you know otherwise.)  My understanding is that the rapid start is because no alignment is performed at all.  The runway heading is just taken as the initial heading of the dead reckoning process.  The relatively poor accuracy of this method is tolerable because of the use of visual/radar nav fixes and ternav fixes to keep the drift under control.  Perhaps the Viggen's gyros aren't sensitive enough to perform a gyrocompass alignment, which requires that they be able to measure the Earth's rotation rate significantly above their own measurement errors over time in order to find north.

In principle any INS could be initialized rapidly like the Viggen's, by just supplying the initial attitude (azimuth, elevation, roll).  In fact that's how the Stored Heading Alignment mode is initialized in modern jets, although they typically follow it with a gyrocompass alignment for 90 seconds or so to make it more accurate.  But you could skip the gyrocompass step if the nav drift is tolerable or you can do nav fixes.

Edited November 27, 2023 by Machalot
Not a rate gyro

[Spirale]

Very intructive :THX!

[LastRifleRound]

Good stuff my dude