What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Synthetic Vision Failure Incident - Safety Report

rjtjrt

Well Known Member
http://www.atsb.gov.au/media/5772247/ao-2016-064-final.pdf

This is a report from Australian Safety Bureau, relating to synthetic vision misleading indication on take-off.

Whilst this is due to misleading input from rad alt causing the sythetic vision to malfunction, it is still of note that Synthetic Vision is fallible, and highly experienced pilots had enormous difficulty ignoring the false display even when they realised it was misleading, as well as the usefulness of a standby instrument that did not have synthetic vision.
 
Last edited:
Great post. I'm about to install an EFIS with Synthetic Vision and start my instrument training. Really good to know about failure modes and where to look! Thanks.
 
Very interesting indeed. Thank you for pointing out this report.

When reading what Honeywell was considering as corrective actions, I was left wondering what level of cross-check/comparison is being done by the multiple sensors on the aircraft; GPS x 2, Radalt x 2, AHARS x2, and who knows what else. Having the synthetic vision so heavily reliant on RADALT seems highly unwise. Somebody failed to do a full FMECA on this system, or so it would seem.
 
Also makes a strong case for an independent standby ESI, in this case an L3 Trilogy vs the Honeywell primary.
 
Imagine if it was IMC at 50ft AGL

If there had of been a low overcast or fog layer in this case , say at 50ft, and you have both large primary displays showing and telling you to pull up how would you catch on quick enough to trust or even look at the small back up?
 
If there had of been a low overcast or fog layer in this case , say at 50ft, and you have both large primary displays showing and telling you to pull up how would you catch on quick enough to trust or even look at the small back up?

I think this is a great question that should be asked by more pilots who launch off into the flag. My guess is that the question exists, but isn't verbalized.

I have SV and love it. However, what you're asking is interesting and speaks to actually using SV.

First off, there is the certified aspect of needing an independent backup within the pilot's view (this is has a defined angle and lateral separation requirement in the reg's). If the primary display failed and you had a properly installed backup, it should be fine - assuming you're trained and proficient with your instruments.

Second is the requirement for backup in an experimental airplane, as in - there isn't one.

Now, for the common sense approach. I never take off of an airport that I can't land at. A 50 foot ceiling is well below any legal instrument approaches for standard GA aircraft, which 200' AGL is lowest precision approach minimum I know of, and you'd certainly have very little time to react to any changes, let alone failures.

My direct response to your question is that, no, I doubt you'd have much time to react to anything with a 50' foot ceiling. Considering a 500 fpm descent rate for landing, that's only 6 seconds from the time breaking out to impact; not much time for any reaction - let alone instrument failure determination.

Now, a question I have is "do you trust SV for landing in non-visual conditions?". It's not legal to use SV in lieu of having the runway environment visually, but if you have an emergency - you can use any means necessary, which would include SV, to safety land. For the record, I've used a safety pilot and landed under the hood on SV - right down the centerline.
 
As Canadian Joy pointed out, failure of the radar alt causing the primary EFIS SV to go "wonky" is a FMEA case that seems to have not been addressed. If the RA is such a significant contributor to that function, failure of the RA should clearly be annunciated and probably shut off SV. I would surmise that the RA is only used so significantly for low altitude phases, and may have been missed in the FMEA for that particular phase of flight. Just a guess.
 
Great post. I'm about to install an EFIS with Synthetic Vision and start my instrument training. Really good to know about failure modes and where to look! Thanks.

I highly recommend that you turn the SV off until you earn your rating. SV can become a distraction. IMO your scan needs to be rock solid without having to rely on augmented presentations like SV or even HITS/Pathways.
 
Last edited:
i understand that having the SV show bad information will be distracting, but it sounds like this crew was almost using it as primary information. i assume that the aircraft had dual AHRS dual FD, and FMS. so pitch and roll should have been accurate on both screens. if the cross hairs were in the box and the IVSI showed climb on both screens that should have been primary information. automation dependency is rapidly becoming a very major issue in avaition today.

bob burns
rv-4 N82RB
 
I have to agree with Bob
There are two methods taught at Flight Safety
in regards to takeoff Pitch information.

One is to arm the Go Around mode which will give you a preset
pitch of roughly 10 degrees. the other is to go into Pitch mode and dial in a Vertical mode to 10 degrees.
I personally like the Go Around and after lift off have the co pilot bring up Nav or heading and FLC to 250 knots

I utilize the roughly the same procedure on My RV7 with the Dynon SkyView. I preset the Altitude that is my first assigned altitude, and arm Heading mode. After take off I hold the Autopilot disconnect which brings up a display REL to ENG, release the disconnect button and the autopilot engages in the attitude and Vertical mode and heaing mode. Then you can select NAV or Heading and off you Go on your Trip
able to concentrate on flying navigating and communicating.

I never refer to synethic vision on takoff as you should have good knowledge of
the terrain and possible emergency contingenties prior to takeoff.
likewise. I do not takeoff if I can not get back in with in 50 miles in multi eng jets and never in Single engine (Now)not below 1000' ceiling. I use to do Night IFR, Over Mountains, taking off in fog on one motor.
But That was when I was young, stupid and flying for whom ever I could to build hours.
Lucky? or Blessed Stupidity?


You can get Radar altitude alerts on takeoff,if your taking off and a hill is in your departure path but again You should have knowledge about that hill before
takeoff. A prime example is a Runway 25 departure out of Eagle,Co. or out of Aspen.
But the RA,radar altimeter,tied into the Synethic Vision as in the story I have never heard of before. As the findings stated the pilot should have relied upon
his PDF for primary pitch information and not the Synethic Vision over riding his sensory inputs. It is aweful easy to sit in the living room and evaluate but it is another story when it happens to you.
 
Last edited:
i understand that having the SV show bad information will be distracting, but it sounds like this crew was almost using it as primary information. i assume that the aircraft had dual AHRS dual FD, and FMS. so pitch and roll should have been accurate on both screens. if the cross hairs were in the box and the IVSI showed climb on both screens that should have been primary information. automation dependency is rapidly becoming a very major issue in avaition today.

bob burns
rv-4 N82RB
I disagree with this analysis, slightly. Yes, following the FD command bars and a positive climb being displayed should result in satisfactory performance, I cannot argue that point at all. But when you have sources of (normally) infallible information giving conflicting data, the potential for confusion exists, and until you can figure out what data is correct you run the risk of putting the aircraft in, or permitting it to remain in, a hazardous situation through no real fault of your own.

In this case, take away the SV aspect, and consider a traditional RadAlt instrument. You're passing 200' on climb, and you notice the RadAlt winding back to 0, even though you have TO power set, climb attitude established and a positive rate. In these few seconds (time slows down...) all sorts of thigns go through your mind...I've never heard of a RA failing like this, so is it right? But I've got Power & Attitude, so that must equal performance, surely?!?, but I'm sinking, or am I? Has the #1 ADAHRS shat itself, and is the RA really correct? Where's that bloody standby and what's the tiny VSI on it say?

This bears some similarities to the CRJ that went down in Sweden last year, inasmuch as a bad sensor fed believable data to the PF's PFD, upon which he acted. In this case, though there was an on-the-ball P2 who stepped in. As the report says, it could have been a different story if the operation was (as is normally conducted) Single-Pilot.
 
This shows how systems training has changed over the years. It use to be that during systems training, you had to build the plane. Now, systems is a whole different game. With all the computer integration you need to know the basics of where the data comes from. Sv is nothing more than a computer image based on position.what give the system that image? That is really a need to know item. I have had plenty of ra failures, I've only had one ahrs failure and the system worked as designed and I knew immediately what was failed and what was good. Amazingly, it always comes back to the basics. Power,pitch and airspeed = performance.

Bob burns
 
Back
Top