My vague recollection of reading about this is that these boxes have to implement a slow “background” erection process – like the vacuum AI does with the pendulous vanes or an electric AI does with some other mechanism.
Aspen went for the airdata solution i.e. constant IAS + constant heading + constant baro alt = straight and level.
Garmin went for the GPS solution i.e. constant GS + constant track + constant GPS alt = straight and level.
The above may be simplistic, however.
So you say they’re using airspeed input only to detect straight and level flight?
The G1000 manual says that it normally uses airspeed, but if that is unavailable it uses GPS. Though I never actually tried to block the pitot tube…
I never understood that flagging business. IMO, it’s almost always better to have less than perfect information than no information at all (even better if there’s a hint that the information might not be fully reliable). So completely blanking the attitude picture sounds like a stupid idea to me.
It’s a bit too fragile for my taste to require airspeed input (which can fail relatively easily by icing the pitot tube if you’re not careful) or GPS (which can be interfered with relatively easily because it’s a weak signal, eg. by an oscillating GPS preamp) for something as vital as attitude information. The old gyros, and most modern (non-certified-aviation) attitude algorithms for MEMS sensors only need the gravity field, which is hard to interfere with :)
only need the gravity field, which is hard to interfere with
Aerobatics? Inadvertant in IMC? I’m assuming the unit needs this in flight, not just for initial set up.
Jesse, can you explain to me why glass avionics (not only Aspen, but G1000 as well) remove attitude information when airspeed and/or groundspeed is unreliable?
These avionics use MEMS sensors to determine the airplanes attitude. These units are often called AHRS (Attitude & Heading Reference System) or ADAHRS (Air Data, Attitude & Heading Reference System).
These systems typically have the following inputs:
These sensors are called IMU (Inertial Measurement Unit)
Additional data is required for the AHRS/ADAHRS to work:
All data is used to compute the attitude. If these sensors give conflicting information the calculated altitude will degrade which can result in "CHECK ATTITUDE or CROSSCHECK indications, or if not corrected over a longer period the display will change to red crosses.
These extra inputs make the signal more reliable AND give a good possibility to determine system errors. This is something a IMU only can’t.
For example, a blocked pitot by ice for example, will result in zero air speed, while there is ground speed and acceleration changes were measured. This data is conflicting, and will lead to a “CHECK PITOT HEAT” indication.
During pitot static testing, it is the other way around, air data is sensed, but no GPS and acceleration is measured, this data conflicts, so it will display “CROSSCHECK ATTITUDE”
My mechanical attitude indicator doesn’t even have an airspeed input, and still displays attitude correctly.
Your mechanical attitude indicator is used a mechical gyro, it doesn’t calculate anything. This is the reason you are required to have a dedicated attitude indicator as backup.
My own algorithm on a tablet computer displays an attitude correctly and doesn’t have an airspeed input as well (and doesn’t use GPS for the attitude data).
That is correct, this is how most IMU’s work, on tablets, phones and for RC aircraft for example. They can not sense its own integrity, which these glass cockpit units can. Additional sensor are needed for this integrity check.
So you say they’re using airspeed input only to detect straight and level flight?
A G1000 will indicate a nose down attitude when pitot pressure is applied slowly and the aircraft itself is standing still / level. If the pitot pressure is applied faster it will indicate red crosses.
On topic:
The Aspen will indicate the CHECK PITOT HEAT message first. When this message occures due to an iced up pitot tube, the pilot hasn’t been paying to much attention, his mechnical air speed indicator would be unuseable as well.
Aerobatics? Inadvertant in IMC? I’m assuming the unit needs this in flight, not just for initial set up.
Tom is the expert on this but AFAIK it is impossible to make an attitude gyro without an erection process… ooops I mean without gravity sensing 
. This is because everything in the universe is merely relative. There is no “up” as such. Sensing gravity is just a cheap and dirty way of getting the “up” (oh dear what have I said?). The proper way of getting the “up” would involve a high precision gyro which detects the earth’s rotation about its axis and then you could work out the “up” vector.
Aerobatics? Inadvertant in IMC? I’m assuming the unit needs this in flight, not just for initial set up.
Well yes, gravity is there even during aerobatics and also in IMC. And you cannot keep accelerating for too long, because you’d eventually exceed VNE in one direction or the other. Therefore, if you average long enough, your acceleration will be zero on average, apart from the g vector.
So all you need is enough bias stability of the gyro to be able to couple it loosely enough the the acceleration sensor. And that is about the only point where a mechanical gyro is still superior to the MEMS devices, IMO.
All data is used to compute the attitude. If these sensors give conflicting information the calculated altitude will degrade which can result in "CHECK ATTITUDE or CROSSCHECK indications,
This seems to imply that the GPS and pitot data is used as an integrity check (I fail to see how such a check could ever be very selective…)
These extra inputs make the signal more reliable AND give a good possibility to determine system errors. This is something a IMU only can’t.
Quite to the contrary. Now what is the failure rate of a MEMS gyro that cannot be detected by other means (i.e. a chip falling off the board by noticing a failure of response on the I2C, SPI or whatever bus is used to attach the sensor)? And what is the probability of the pitot icing closed or the GPS loosing signal for some time? I bet there are several orders of magnitude in between. Now if you use these checks to draw Xes instead of the attitude indicator, then almost always this will not be due to faulty MEMS output, but because of faulty check signals! That is, you’ve reduced the highly reliable MEMS sensor reliability to the lower reliability of the pitot and/or the GPS!
The fact that we see a complaint like this every couple of months while I cannot remember having last seen a complaint about bad output of a mechanical gyro seems to indicate that this design is a bad idea (and IMO completely unnecessary).
Your mechanical attitude indicator is used a mechical gyro, it doesn’t calculate anything.
Of course it does calculate, just implemented mechanically. Many IMU algorithms calculate much the same thing, they’re just implemented very differently. IMO it’s none of the regulator’s business to prescribe implementation details.
This is the reason you are required to have a dedicated attitude indicator as backup.
So that means all the huge amount of money we throw at (software) certification efforts that makes all recent avionics at least 10 times more expensive achieved exactly nothing? That’s what I’ve been saying all the time
When this message occures due to an iced up pitot tube, the pilot hasn’t been paying to much attention, his mechnical air speed indicator would be unuseable as well.
So what. The ASI is about the least necessary instrument. I’ve been taught to fly the aircraft by numbers. I have a table that lists MP settings and expected airspeeds both for enroute and (3°) approach. So if I loose the ASI, I just set engine power according to the table and the airspeed comes out roughly correctly. And if the runway dimensions are slightly larger than Helgoland Düne, that’s really no problem. In fact, my best landing ever (on a 500m runway, no less) was with the ASI covered.
If I loose the attitude information in IMC, I’m dead quite quickly. Now if loosing the ASI means loosing attitude information quite quickly, that transforms a minor nuisance into a much bigger problem, which is a bad idea.
I don’t understand the flagging business. It’s a mathematically provably bad idea since 1948, (it follows from Claude Shannon’s information theory), but aviation seems to be fond of it. (reliability information OTOH is a good thing like the low RPM bar of a mechanical gyro, if it doesn’t interfere with displaying the actual information, like the low RPM bar not preventing the attitude card from moving)
I was recently looking at the accuracy of gyro-compasses and saw it had to be 3 degrees drift in 15 minutes. How good are the MEMS sensors? Is there a reason they can’t simply emulate a mechanical gyro – e.g. higher rates of drift – that means they require additional information for rectification?
ST micro specs around +-0.03°/s/°C offset drift. That means if you change the temperature by 1°C, it will take a completely unaided gyro to drift by 3° within 100s.
While this spec is relevant for a directional gyro without flux gate, I fail to see why it should be directly relevant to an attitude indicator, because you wouldn’t want to run it unaided by the gravity sensor and the magnetometer.
… and nobody will be running a DG without a fluxgate.
Also temp stabilisation is easy to do, even though the current solutions use a relatively high temperature stabilised oven (which is why the EDF1000 and the SG102 run pretty warm).
The answer to this is in the certification requirements somewhere.
