(Long) A failed AI does not mean a failed vacuum pump

tawood

En-Route
Joined
Sep 22, 2015
Messages
2,558
Location
SE Michigan
Display Name

Display name:
Tim
So this is my first winter with an IR, and I've found staying current is difficult...well, its difficult to stay current if you are looking to fly in actual conditions here in Michigan while avoiding ice. But I've been patient, watched the weather daily, and even took a few different days off of work when the weatherman has predicted some instrument weather, only to be kept on the ground by un-forecast fog or extreme winds. My last 4 approaches were all done in October, with several more in August, so my time was running out for doing this in actual.
Finally, this weekend's predictions for semi-low ceilings, 50+ degrees on the ground, freezing levels well above the MSA, and mild surface winds held true, and I went to my hangar early this morning at about 7am, trying to get up and fly before the afternoon predictions of heavy turbulence kicked in. When I got to the airport, the airports all around were at about 5000 foot unpredicted ceilings, but all the TAFs were saying that through the morning those ceilings should come down to about 1000 agl. So, after killing an hour+ tidying up the hangar, I took a look on foreflight on my phone, and saw that several airport ceilings about 40 miles to my north were all down to around 900-1200 feet, and were predicted to stay that way for the day. There were no airports between me and them, and my field, as well as the other fields to the E/S/W of me, were all still 5000 feet. So, I took off without a clearance, just planning to get a pop up clearance as I got close to the low clouds. (I've also never gotten a pop-up clearance before, so I did this to "scratch it off my list".)
First surprise was that the low clouds were literally just a couple miles north of my home field, so that right after turning north from takeoff, I had to turn/circle and call someone for the pop-up. Once done, I climbed and headed to my first airport for a non-precision approach. Within moments of leveling from the climb, I entered the clouds and started flying on instruments (no AP in my bird). I flew along, feeling very good, remembering how much I like actual (but hate the hood). ATC gave me vectors to my first approach (VOR), then I was cleared, and flew it no problem, ending with a low approach/missed. On to the next one.
I called missed, climbed to 3000 and was vectored to the next one, an ILS. After awhile ATC gave me a turn, and also a descent to 2300...which I did and did fine, until...

Just as I was leveling at 2300, I hit some turbulence, chop really...not bad turbulence at all. In fact, during the past summer when I practiced/obtained my IR, I flew in MUCH MUCH worse summer air current stuff...but as I hit the turbulence, I got the weirdest sensation, like none of my instruments made any sense: my initial impression was that some were showing a climb, some were showing a dive, some a bank, some not, etc. For just a few seconds, I had this sensation, but then the air smoothed out, the feeling went away, everything seemed fine, and I flew on. My first thought was, "Wow, I'm rustier than I thought!" I crossed checked the gauges: now I was sure that everything was in agreement. I took a look at my vacuum: 5", right where it should be...ok, all good.
I flew the ILS to another low approach, and called missed again, to go fly my third/last approach. As I was being vectored and descending to the third and final approach, again I hit some light turbulence, and again I got the same sensation that NOTHING made sense. Because I wasn't so "caught off guard" this time, I thought to myself, "This is not me, this is a problem with the gauges." My first reaction was to suspect the AI, but I looked again at the vacuum, and it again showed 5 inches. It sounds silly to me now, but I just didn't consider at the time that the AI would fail on it's own...I really expected the AI to fail ONLY due to a failed vacuum pump. But just like before, after just a few seconds, everything again looked fine: plane flying level, cross check of gauges showed everything in agreement. I now dismissed the thought that it was an instrument problem and thought, "Maybe it was me again."
I finished the approach with a last low approach, and asked for vectors back to my home drone. The better weather had been heading my way, because within just moments of turning back home, I popped out into VFR with 5000 foot ceilings, and I cancelled IFR. Then I started some tests...
While flying level, I quickly-but-lightly pushed the yoke forward and back, 2 or 3 times, to "simulate" some turbulence, while watching the gauges. In doing this, I made the plane bump a bit, without really changing my attitude at all. SURPRISE! The trouble revealed itself: my AI instantly showed a 45+ degree climb with a 30 degree bank from the light "bumping", for about 3 or 4 seconds, before it "popped" back to indicate the correct level flight. I then wagged the wings, again quickly-but-lightly, two times, and sure enough, the AI went haywire again, for about 3 seconds (this time a 45+ degree dive, with only slight bank), before returning to a correct indication.
Like I said, it sounds silly to me now that I would think that the only way the vacuum gauges fail is from a loss of vacuum, but as this was happening that's exactly what I thought. Oh, and another "fail" on my part: I have an AHRS backup in my Stratus, which I didn't even display until AFTER running my tests and actually finding the AI as the problem (and the AHRS worked fine for the entire VFR flight home, while that AI messed up a couple more times in the pattern). I don't plan to make that mistake again...how easy it would have been to turn that on, at least immediately after I had an inkling that it could be a gauge problem...
 
Last edited:
Mechanical things have a tendency to fail from time to time. I got lucky with my AI, it never righted it's self after startup, one day.
I sent it to AQI, and it is now good as new. ;)
 
I’m going to send it to a local shop: Preferred Avionics in Howell. They’ve done good work for me in the past.
 
I’ve now had a total of 5 AI failures, no vacuum pump failures (yet). Only the most recent was IMC, local practice flight so I continued on and worked on my partial panel skills.

(P.S. - all in different aircraft)
 
Last edited:
I’ve now had a total of 5 AI failures, no vacuum pump failures (yet). Only the most recent was IMC, local practice flight so I continued on and worked on my partial panel skills.

(P.S. - all in different aircraft)
I've never had an AI fail before...I've had two DGs fail, and two TCs fail, but never an AI.
 
AI’s fail just like anything else mechanical. The idea that it wouldn’t fail is odd.
 
I have lost a DG and an AI without losing the vacuum pump. When I pulled the DG out, I found it had a piece of masking tape on the top of it that said $5. Apparently, some flea market special. What was also missing was a filter on the air in side of both devices.
 
I have lost a DG and an AI without losing the vacuum pump. When I pulled the DG out, I found it had a piece of masking tape on the top of it that said $5. Apparently, some flea market special. What was also missing was a filter on the air in side of both devices.
Yeah, I was gonna ask the OP if there was an intake filter.
 
I’ve now had a total of 5 AI failures, no vacuum pump failures (yet). Only the most recent was IMC, local practice flight so I continued on and worked on my partial panel skills.

(P.S. - all in different aircraft)

Exactly the opposite for me - I've lost two vacuum pumps and zero AIs.
 
Nice write up,

FWIW, so far across a few different vac planes, one pump failure, 3.5 AI failures (one worked but was slow and getting ready to check out)
 
Can I hijack a moment?...really just looking for info from instrument gurus...

I flew a 172 a couple months ago, and the turn coordinator was just going nuts. It was bumpy, but man this thing was like it was on too much coffee. The word I'd use would be flopping around.

Also, as I was turning, and trying to keep the ball centered, it was way out of the lines. So much that after putting in more than usual rudder, I quit trusting it. There was 3kts wind only.
Also, regarding the ball....it really doesn't give false readings if it's calibrated does it? It didn't appear crazy when taxiing.

Never seen this behavior before. My buddy who is also a pilot was with me and was actually laughing about how crazy it was acting.
I reported it to the FBO, but haven't spoken to them since so don't know if they addressed it or deemed it ok.
Do the turn coordinators get all wonky like that when/if they are going bad?
 
Can I hijack a moment?...really just looking for info from instrument gurus...

I flew a 172 a couple months ago, and the turn coordinator was just going nuts. It was bumpy, but man this thing was like it was on too much coffee. The word I'd use would be flopping around.

Also, as I was turning, and trying to keep the ball centered, it was way out of the lines. So much that after putting in more than usual rudder, I quit trusting it. There was 3kts wind only.
Also, regarding the ball....it really doesn't give false readings if it's calibrated does it? It didn't appear crazy when taxiing.

Never seen this behavior before. My buddy who is also a pilot was with me and was actually laughing about how crazy it was acting.
I reported it to the FBO, but haven't spoken to them since so don't know if they addressed it or deemed it ok.
Do the turn coordinators get all wonky like that when/if they are going bad?

The ball, presuming it’s not a digital one, is really just a construction type level stuck to the panel, when you turn on the ground it should go to opposite direction as the wings bank on the TC. Also most TCs are electric and they do have issues like all things.
 
The ball, presuming it’s not a digital one, is really just a construction type level stuck to the panel, when you turn on the ground it should go to opposite direction as the wings bank on the TC. Also most TCs are electric and they do have issues like all things.

Yeah, all that is what I thought to be true. That's why it was so confusing to me, and had to ask.

Do the older Cessna's skid more than the newer ones or something? This one was a 1966 model. I usually fly a 1975 or 1977 model.
The ball was entirely out of the lines by a ways, to the right on left base/final. It wasn't a very steep turn either, probably less than standard rate. Not typical, especially adding a lot of rudder. Like I said, really no wind at all.
 
The ball was entirely out of the lines by a ways, to the right on left base/final. It wasn't a very steep turn either, probably less than standard rate. Not typical, especially adding a lot of rudder. Like I said, really no wind at all.

Ball out doesn't necessarily mean more rudder... It can also mean less rudder!
 
AI’s fail just like anything else mechanical. The idea that it wouldn’t fail is odd.
If you had asked me, before this flight, if an AI by itself could fail, I would have replied, "Of course it can." It wasn't so much that I didn't think it could fail, it was more from the conditioning that throughout my instrument training, whenever we would talk about instruments failing, my instructor would either cover just the TC (and say, "You just lost electrical" and/or "Your TC just failed."), or cover BOTH the AI and the DG, and then say, "Your vacuum just failed." I'm not blaming him at all, but we never covered up just the AI or DG and I guess I got conditioned to thinking AI/DG failure=vacuum loss.
 
Last edited:
If you had asked me, before this flight, if an AI by itself could fail, I would have replied, "Of course it can." It wasn't so much that I didn't think it could fail, it was more from the conditioning that throughout my instrument training, whenever we would talk about instruments failing, my instructor would either cover just the TC (and say, "You just lost electrical" and/or "Your TC just failed."), or cover BOTH the AI and the DG, and then say, "Your vacuum just failed." I'm not blaming him at all, but we never covered up just the AI or DG and I guess I got conditioned to thinking AI/DG failure=vacuum loss.

That is a big problem during instrument training. Outside of a sim, it's not possible to "fail" a AI/DG/Vacuum pump on command. You instructor tells you it failed, and/or covers it up. In the real world a failure may be subtle and difficult to diagnose. Even worse is you have instruments disagreeing and you have to figure out which ones are lying, while possibly getting knocked around in turbulence while handflying in IMC. This is why real world instrument failures are so dangerous, such as the Bonanza that crashed outside of NYC after a vacuum failure.
 
That is a big problem during instrument training. Outside of a sim, it's not possible to "fail" a AI/DG/Vacuum pump on command. You instructor tells you it failed, and/or covers it up. In the real world a failure may be subtle and difficult to diagnose. Even worse is you have instruments disagreeing and you have to figure out which ones are lying, while possibly getting knocked around in turbulence while handflying in IMC. This is why real world instrument failures are so dangerous, such as the Bonanza that crashed outside of NYC after a vacuum failure.
And that's why I posted...I think my initial impression, that the problem was all me, and then the impression that ALL the gauges were bad, were compounded by the fact that the problem was intermittent. I knew something was wrong, but couldn't figure out what that "something" was...I suppose it didn't help that I'm a low time instrument pilot as well (40 hours hood, 20 hours actual, 700 hours VFR).
Also, I think my mistake of thinking that AI/DG loss = vacuum loss goes back to the saying, "Under pressure, we don't rise to the level of our expectations, we fall to the level of our training."
 
Last edited:
Good work! And very nice writeup. Yes, I have had an AI fail without loss of vacuum. Fortunately, it was in VMC. Having that happen in IMC would truly be a trial. But you handled it well.
 
The pump failure is pretty insidious. The one I had my wife caught the vacuum gauge going. I continued to watch the AI and DG spin down and concluded that it would be hard to tell before it was too late that the AI had failed. The DG is easier to catch because you expect precession in normal operation and it just starts going faster than usual. This is why I had the vacuum gauge moved right next to my AI and when I get my new MVP50 install done, it will give a better alarm.
 
My Sierra had a vacuum warning light right next to the 6 pack which was helpful, and I did lose the pump once on a VFR flight. I guess G5s would be a step up since I'm thinking most failure modes are generally going to be more obvious with those.
 
The pump failure is pretty insidious. The one I had my wife caught the vacuum gauge going. I continued to watch the AI and DG spin down and concluded that it would be hard to tell before it was too late that the AI had failed. The DG is easier to catch because you expect precession in normal operation and it just starts going faster than usual. This is why I had the vacuum gauge moved right next to my AI and when I get my new MVP50 install done, it will give a better alarm.

Interesting. Both of my vacuum pump failures were pretty noticeable pretty quickly. One happened during a takeoff roll, the other in cruise. Both times, the AI took a dump pretty quickly.

Of course, your gyros may have been in better shape than the ones on those two planes, thus they kept spinning for longer - Or maybe your pump didn't crap out all at once?
 
The AI just gracefully started to lean. It wasn't the case that it just toppled over. The pressure gauge went to zero pretty fast.
 
The pump failure is pretty insidious. The one I had my wife caught the vacuum gauge going. I continued to watch the AI and DG spin down and concluded that it would be hard to tell before it was too late that the AI had failed. The DG is easier to catch because you expect precession in normal operation and it just starts going faster than usual. This is why I had the vacuum gauge moved right next to my AI and when I get my new MVP50 install done, it will give a better alarm.

REALLY GOOD thing to do in a nice sim, during my IFR training my CFII (very high time guy) told me he was going to fail the AI or vac, forget which, and to just keep flying straight and level off the AI, puts you into a textbook grave yard spiral
 
THIS is why the FAA moved paperwork mountains and found a way to (for them) shortcut the approvals on the new digital instruments to replace the vacuum powered instruments. I wasn't counting but you guys have listed dozens of failures from a pretty small sample size!
 
My new (overhauled) AI should be installed by the weekend.

For anyone looking for instrument work, Preferred Avionics of Howell Michigan has done great work for me several times.
 
Of course, when I pulled the dead DG out of my plane I found a piece of masking tape on it that said $5. I wonder what fleamarket the previous owner found that one at.
 
Flew a Cherokee recently on a ferry permit. The AI was sketchy, the TC didn't work at all. Vacuum was fine. VFR so no biggie. Replaced both once we got the a/c home.
 
Of course, when I pulled the dead DG out of my plane I found a piece of masking tape on it that said $5. I wonder what fleamarket the previous owner found that one at.

Fly market, you mean? ;)

Don't knock it... How long was that DG working? Did you get $5 worth out of it?

I once bought an MP gauge for a club plane from Wentworth at Airventure (gauge happened to fail during the show... convenient). It cost $38 and had a 60-day warranty and of course zero lead time. The FBO was asking $1900 for an official Cessna MP gauge, with a 90-day warranty and a few days' lead time in a very busy flying season. We did end up replacing it later, but during the winter, and with an EI electronic one because it was way cheaper than the Cessna one.

I'd say we got our $38 worth.
 
I don't know how long it worked for the previous owner. I got less than six months service out of it.
 
Old Thread: Hello . There have been no replies in this thread for 365 days.
Content in this thread may no longer be relevant.
Perhaps it would be better to start a new thread instead.
Back
Top