Real world partial panel is not like training

[MAKG1]

I flew an IFR cross country up to Napa today for a $300 burger. The aircraft was an Archer equipped with 650 and Sandel HSI. Autopilot is Piper original and inop. Weather was 5000 overcast at Reid and 1500 overcast at Napa, with light rain. Outbound was uneventful except Center dumped me on the ILS too close and too high after telling me to expect the VOR. Needed a lot of descent, but it was fine.

On the return, the wind picked up. I was tracking nearly 20 deg WCAs and had occasional smooth but fairly strong up and down drafts. Once, I needed full power to maintain altitude (but at 100 knots rather than Vy, so I did have some margin). That had me looking for ice, but there was no trace at all and the OAT was 8 C, and carb heat dropped 100 RPM and never increased (so no carb ice either).

All these are great exercise. I've been flying a fair amount of actual lately, and a little light turbulence and some winds are just perfect.

Then, over Marin, in the clouds at 5000, the HSI alerted "Gyro" and started jumping 20 deg intermittently. Vacuum is fine and the AI cross checks just fine. I figure out that the GPS is set for heading up (what kind of MORON came up with that?), but decide I don't want to fixate on GPS settings given workload, and I'll ignore it shifting discontinuously with the heading jumps. The turbulence makes the mag compass very difficult to use, as it won't hold still. The HSI can't be ignored entirely because it has the CDI. It's changing instantaneously 20 deg occasionally, and is no longer flagging the gyro (!). Distracting as hell, and it's a LOT harder to ignore the wrong heading information than it should be. NorCal complains about my heading and I tell them I'm flying the mag compass due to a DG failure. Eventually, I figure out the best way to hold a heading is to hold a track and use the desired and actual tracks off the GPS. Then, I ask NorCal for courses rather than headings. They take a while, but eventually give me direct OSI VOR. Good enough. I notify NorCal at each handoff that the DG has failed.

Now for options. SFO and all the Peninsula airports are reporting 1500 overcast. I don't think shooting an approach like this is a very good idea. I can hold altitude pretty well despite the occasional turbulence, but heading is not good enough. San Jose and Reid both report overcast 5000. Good. So the best option is to exit the clouds from the side into VMC. NorCal gives me a descent to 4000. A few minutes after level off, I break out. Then I get 3000, fly direct Reid, report in sight. Visibility is 30+ miles under the deck, no problem. Cleared for the visual, and landing is uneventful.

Lessons?

1. Check the GPS settings before takeoff.
2. Get some partial panel practice with foggles in turbulence, or at least with an instructor. It is NOT the same -- actually, not even close -- but I should have a strategy for dealing with this failure, rather than working this out on the fly.
3. Don't fly that damn Archer in IMC, or at night. It has done this before.
4. Consider a separate DG and CDI so I can put a sticky on the DG. Maybe use the GPS default nav page for this.

 

[azure]

Then, over Marin, in the clouds at 5000, the HSI alerted "Gyro" and started jumping 20 deg intermittently. Vacuum is fine and the AI cross checks just fine. I figure out that the GPS is set for heading up (what kind of MORON came up with that?), but decide I don't want to fixate on GPS settings given workload, and I'll ignore it shifting discontinuously with the heading jumps. The turbulence makes the mag compass very difficult to use, as it won't hold still. The HSI can't be ignored entirely because it has the CDI. It's changing instantaneously 20 deg occasionally, and is no longer flagging the gyro (!). Distracting as hell, and it's a LOT harder to ignore the wrong heading information than it should be. NorCal complains about my heading and I tell them I'm flying the mag compass due to a DG failure. Eventually, I figure out the best way to hold a heading is to hold a track and use the desired and actual tracks off the GPS. Then, I ask NorCal for courses rather than headings. They take a while, but eventually give me direct OSI VOR. Good enough. I notify NorCal at each handoff that the DG has failed.

Bingo. This is the strategy my CFII and I worked out during training. Not sure what GPS you have, but with the 480 the HSI tool on the NAV page is almost as good as the real thing, as long as you remember that it is track-based instead of heading based. In NAV mode (as opposed to GPS) you can even use it to fly a ground-based approach.

But getting out of the clouds as you did is an even better strategy. Sounds like you handled the emergency well.

 

[Arob16]

Yeah I have slowly transitioned to paying more attention to the GPS track over the years. It's just more reliable, especially if the DG in the airplane stinks or needs re-set constantly. With other tech in the cockpit (iPad, ADS) it's easier to have redundancies to the GPS as well.


Sent from my iPhone using Tapatalk

 

[James331]

Think any gyro talking to a Sandel HSI is going to be electric and not use vac, or at least that's the case with the 3308.

 

[MAKG1]

Think any gyro talking to a Sandel HSI is going to be electric and not use vac, or at least that's the case with the 3308.

Yeah, I would think so as well. Nevertheless, it isn't labeled, and it didn't take much to check vacuum and AI.

I'm not impressed with that HSI's failure modes. Displaying wrong data is not acceptable. It only flagged it occasionally, and for some time, just gave me a heading that was 20+ deg wrong with no flag. I should have been looking at a red X, for heading, but that never showed up at all. If it did that at a lower altitude or on an approach, that could have been very messy.

 

[coloradobluesky]

If you use the IFR GPS in track up mode with highest sensitivity (.3), you can use the track line for right and left much like the needle in a VOR head. That can come in handy. Also, if you lose your airspeed, dont forget GPS speed, its better than nothing especially if you are aware of what it is before you lose the air data one.

 

[James331]

Yeah, I would think so as well. Nevertheless, it isn't labeled, and it didn't take much to check vacuum and AI.

I'm not impressed with that HSI's failure modes. Displaying wrong data is not acceptable. It only flagged it occasionally, and for some time, just gave me a heading that was 20+ deg wrong with no flag. I should have been looking at a red X, for heading, but that never showed up at all. If it did that at a lower altitude or on an approach, that could have been very messy.

Not the HSIs fault, it's only showing what the gyro is telling it.

As far as Sandel gyro combos, I've been very happy with mine, it's just getting a good combo and taking care of it.

 

[MAKG1]

If you use the IFR GPS in track up mode with highest sensitivity (.3), you can use the track line for right and left much like the needle in a VOR head. That can come in handy. Also, if you lose your airspeed, dont forget GPS speed, its better than nothing especially if you are aware of what it is before you lose the air data one.

Yes, absolutely. For some reason, this GTN650 was set heading up. I don't understand why that setting even exists, as I can't find a use for it with the map. Sure, for traffic and weather pages, but for nav, it's stupid. I decided that it wasn't wise to fixate on GPS settings under the given circumstances. Given that the heading input was FUBAR, the map was jiggling.

 

[James331]

Depends, I like heading up for some stuff.

Normally for enroute I'll go heading up on my top box, flight plan or nearest on my bottom box, set my EHSI to display airspace and airports and ring it to my current projected glide range and have my transponders' stop watch started from engine start.

 

[MooneyDriver78]

Yes, absolutely. For some reason, this GTN650 was set heading up. I don't understand why that setting even exists, as I can't find a use for it with the map. Sure, for traffic and weather pages, but for nav, it's stupid. I decided that it wasn't wise to fixate on GPS settings under the given circumstances. Given that the heading input was FUBAR, the map was jiggling.

Been that way since the beginning, still used by ground based GPS units, smart phones, makes more sense on the ground with street navigation. In the air, not so much.

 

[Palmpilot]

One of the troubles with rental planes is that it's often necessary to check that avionics display options are set the way you want.

 

[MAKG1]

One of the troubles with rental planes is that it's often necessary to check that avionics display options are set the way you want.

Yes. That's a significant lesson learned from this event.

 

[denverpilot]

One of the troubles with rental planes is that it's often necessary to check that avionics display options are set the way you want.

I've seen folks flying G1000 stuff who have a massive checklist they wrote themselves to go through tons of pages and set things they way they want them, prior to a solid IMC flight or training. It's wonderful that avionics are "customizable" these days, but in the rental environment, if someone "customizes" the stuff in a weird way, and you don't catch it pre-flight, it can be horribly distracting.

 

[Palmpilot]

I've seen folks flying G1000 stuff who have a massive checklist they wrote themselves to go through tons of pages and set things they way they want them, prior to a solid IMC flight or training. It's wonderful that avionics are "customizable" these days, but in the rental environment, if someone "customizes" the stuff in a weird way, and you don't catch it pre-flight, it can be horribly distracting.

I don't remember: Is there a single menu item to set everything back to the defaults? Depending on one's preferences, that might be a way to significantly shorten such checklists.

 

[denverpilot]

I don't remember: Is there a single menu item to set everything back to the defaults? Depending on one's preferences, that might be a way to significantly shorten such checklists.

I don't remember either. I have a whopping .6 behind a G1000 at this point. Most of it spent taxiing/parked and screwing with trying to get the thing set the way the CFI wanted it.

 

[MAKG1]

I don't remember: Is there a single menu item to set everything back to the defaults? Depending on one's preferences, that might be a way to significantly shorten such checklists.

On the G1000? There are several reset-to-defaults. The CAP quick reference encourages resetting defaults in the MAP page and the AUX page.

We've run into operational complications with excessive configurability at work, as well. In particular, it makes reproducing problems challenging, and my MO is generally to ask the operator to demonstrate a problem so I can see how they have it configured.

Honestly, it's a failure of design to have to build in a lot of preferences. It means the use cases are insufficiently understood.

 

[MauleSkinner]

just out of curiosity, what did your partial panel training look like?

Granted, I'm an old fart, but approaches to near minimums and in actual was par for the partial panel course for me as both a student and instructor.

 

[MAKG1]

just out of curiosity, what did your partial panel training look like?

Granted, I'm an old fart, but approaches to near minimums and in actual was par for the partial panel course for me as both a student and instructor.

Sure, I did a bunch of approaches with DG and AI covered up. And believe me, I put it to work. But it's very different when the failures affect other systems and turbulence shakes the mag compass -- it's hard to figure out what your heading actually is. Did you ever do it in actual? It's not the same.

Real failures also don't look like sticky notes and an instructor saying your whatever failed. You have to figure it out. Part of that is assessing what your capabilities really are, under the conditions you actually have. I wasn't happy with my headings under those conditions, and would only have attempted an approach if there were no alternatives. That I did one in a 182 two weeks ago with a simulated AHRS failure, in much lighter winds and no turbulence, doesn't matter.

I did have a few sim sessions on failures, and the AI failure was quite subtle and very easy to follow into an unusual attitude.

 

[MauleSkinner]

I did most of my instrument training partial panel, a lot of it was in actual, and some in turbulence. You just average the compass, and realize that if you keep the turn needle centered, the airplane isn't turning much. Back in the good old days, we also had to figure out what the compass heading was so we could set the DG...certain approaches tend to come out in wierd places if you fly the wrong headings.

If your AI failure got you into a 20-degree bank or more than 100 feet off altitude, you have some basic scanning/interpretation issues. I've had a couple in real life in IMC, and they haven't been that big a deal.

Edit: advanced avionics do add a level of complexity and potential confusion...you have to figure out whether it's an instrument failure or a systems failure. But that should be part of the training and/or equipment checkout IMO.

Yeah, it's tough to ignore incorrect instruments, but that's why they invented Post It notes.

 

[jsstevens]

One interesting take away from MAKG1's experience: the HSI was partially failed. DG couldn't be trusted but he needed the CDI portion. That would be much harder (IMO) than simply covering the HSI. I never considered that as a failure mode. Thanks for the enlightenment...

John

 

[MAKG1]

One interesting take away from MAKG1's experience: the HSI was partially failed. DG couldn't be trusted but he needed the CDI portion. That would be much harder (IMO) than simply covering the HSI. I never considered that as a failure mode. Thanks for the enlightenment...

John

In retrospect, I should have just covered it up.

The GPS has a "default nav" page that can substitute for the CDI. It's not in a very good place for a scan, but it's there. It can't display GS or GP, but I wasn't doing that.

Of course, the tradeoff is that I can't use the GPS for anything else without losing my CDI. Like monitoring nearest airports or the weather or traffic.

There would be a similar issue in a G1000 or Aspen, were it to get incorrect heading data. All the other data in the PFD is needed -- gyroscopic turn rate, especially, as there is no standby turn coordinator. It's supposed to cover heading with a red X, but so was the Sandel.

 

[MauleSkinner]

It's supposed to cover heading with a red X, but so was the Sandel.

The red X is for an instrument failure, but it sounds like you had a remote gyro problem, and since the heading indicators were still getting heading information (albeit bad information), the heading indicators didn't know there was anything wrong and didn't display the failure.

 

[MAKG1]

The red X is for an instrument failure, but it sounds like you had a remote gyro problem, and since the heading indicators were still getting heading information (albeit bad information), the heading indicators didn't know there was anything wrong and didn't display the failure.

The HSI episodically alerted "GYRO," so it knew there was a fault. At least sometimes.

Heading also can't change instantaneously by 20 deg and back.

 

[MauleSkinner]

The HSI episodically alerted "GYRO," so it knew there was a fault. At least sometimes.

Heading also can't change instantaneously by 20 deg and back.

The indicator doesn't know the heading can't change instantaneously...it only sees specific failure modes.

Sounds like it did lose enough signal intermittently to flag, but if it's getting a signal, it can't differentiate between a "bad" signal and a "good" one.

 

[MAKG1]

The indicator doesn't know the heading can't change instantaneously...it only sees specific failure modes.

Sounds like it did lose enough signal intermittently to flag, but if it's getting a signal, it can't differentiate between a "bad" signal and a "good" one.

I think you need to replace "can't" with "is not designed to." It's not that difficult to validate inputs in a control or monitoring system. In fact, some sort of filtering is almost certainly applied, as gyros (especially rate gyros) are rather noisy.

 

[James331]

just out of curiosity, what did your partial panel training look like?

Granted, I'm an old fart, but approaches to near minimums and in actual was par for the partial panel course for me as both a student and instructor.

Spent a lot of time for my IFR in a frasca with a very high time II, he could fail the instruments realistically, which was really nice. Once was told to just fly the AI straight and level, he failed it and demonstrated how it will put you into a spiral dive. This is one of the reasons logging sim time, even beyond the max required, is a great indicator of the quality of your CFI, know when to use the plane and when to use the sim and when to use the desk.

 

[MauleSkinner]

I think you need to replace "can't" with "is not designed to." It's not that difficult to validate inputs in a control or monitoring system. In fact, some sort of filtering is almost certainly applied, as gyros (especially rate gyros) are rather noisy.

I would suggest that changing heading 20 degrees instantaneously would probably break the airplane, so I'm going to stick with "can't".

But yes,you can program any number of failure recognition modes into the unit...it's just a matter of coming up with all of the possibilities that shouldn't ever happen in every hardware combination that might be used. And once that's done, the FAA can shut down their pilot certification branch.

 

[azure]

In retrospect, I should have just covered it up.

The GPS has a "default nav" page that can substitute for the CDI. It's not in a very good place for a scan, but it's there. It can't display GS or GP, but I wasn't doing that.

Assuming GS or GP == glideslope or glidepath, that's surprising. The pseudo-HSI on the 480 definitely has a glideslope indicator (similar to the one Sandel uses when the course pointer is > something like 45 degrees from vertical). I'm surprised that Garmin didn't design something similar into the 650.

I wouldn't hesitate to use the 480's "HSI" to fly a precision approach in an emergency. Certainly an LPV, and I'm >99% sure the displayed GS on an ILS is driven by the GS receiver in the built-in NAV radio.

There would be a similar issue in a G1000 or Aspen, were it to get incorrect heading data. All the other data in the PFD is needed -- gyroscopic turn rate, especially, as there is no standby turn coordinator. It's supposed to cover heading with a red X, but so was the Sandel.

Out of curiosity, which Sandel do you have?

 

[MAKG1]

Assuming GS or GP == glideslope or glidepath, that's surprising. The pseudo-HSI on the 480 definitely has a glideslope indicator (similar to the one Sandel uses when the course pointer is > something like 45 degrees from vertical). I'm surprised that Garmin didn't design something similar into the 650.

I wouldn't hesitate to use the 480's "HSI" to fly a precision approach in an emergency. Certainly an LPV, and I'm >99% sure the displayed GS on an ILS is driven by the GS receiver in the built-in NAV radio.

Out of curiosity, which Sandel do you have?

It's a rental, and I couldn't tell you which Sandel it is.

Garmin 650 units (and 430 units as well) do not display glideslope on their own screen, but rather rely on an external CDI to display. The HSI was serving this function. They do display a horizontal CDI of sorts on the default nav page.

 

[azure]

It's a rental, and I couldn't tell you which Sandel it is.

If you're going to keep flying that plane, it might be a good idea to find out so as not to be surprised by certain failure modes. Not all Sandels are equal. The 3308's display is illuminated by a halogen projection lamp that can fail unpredictably. I was within minutes of picking up an IFR clearance at a New Hampshire strip last summer, had just finished my runup, when I noticed that my 3308's display was completely dark. Power cycling did not fix it, the projection lamp had gone poof! Of course without a DG my plane was no longer legal for IFR so I limped home VFR, much of that leg at night. Not fun at all.

 

[MAKG1]

If you're going to keep flying that plane, it might be a good idea to find out so as not to be surprised by certain failure modes. Not all Sandels are equal. The 3308's display is illuminated by a halogen projection lamp that can fail unpredictably. I was within minutes of picking up an IFR clearance at a New Hampshire strip last summer, had just finished my runup, when I noticed that my 3308's display was completely dark. Power cycling did not fix it, the projection lamp had gone poof! Of course without a DG my plane was no longer legal for IFR so I limped home VFR, much of that leg at night. Not fun at all.

Yeah, I'm having second thoughts about flying this plane in the clouds. Failures happen, but this one happened before a few weeks ago (VFR), and was "repaired." I did leave a complaint for the maintenance chief, in addition to a squawk, that I had no way of finding that out before flight. That bit of process needs fixing.

The display looks like the 3308 images I can find online, but it's not as bright, so I think it's older. It kinda looks like an LED display. It does have the same buttons and the same symbology. Thanks for the heads up.

 

[azure]

The display looks like the 3308 images I can find online, but it's not as bright, so I think it's older. It kinda looks like an LED display. It does have the same buttons and the same symbology. Thanks for the heads up.

Why not simply check the AFM supplement in the airplane? It should tell you the model of HSI that's installed. If it's not in there, your decision is made for you as the plane is technically unairworthy.

BTW, the projection lamps are supposed to be replaced every year or 225 hours, whichever comes first. The lamps from Sandel are expensive, so there is a "black market", and some bootleg 3308 lamps have diminished brightness that can cause problems with readability. If the display in your plane is uncomfortably dark or dim, I would ask whether the lamp came from Sandel.

 

[Kritchlow]

Not sure if I saw it mentioned as I only skimmed... Any reason for not asking for no gyro vectors instead of headings? You still had a TC I assume...

 

[James331]

Why not simply check the AFM supplement in the airplane? It should tell you the model of HSI that's installed. If it's not in there, your decision is made for you as the plane is technically unairworthy.

BTW, the projection lamps are supposed to be replaced every year or 225 hours, whichever comes first. The lamps from Sandel are expensive, so there is a "black market", and some bootleg 3308 lamps have diminished brightness that can cause problems with readability. If the display in your plane is uncomfortably dark or dim, I would ask whether the lamp came from Sandel.

It's a generic light, no biggie to replace with a non sandel part.

As for the model number, think most EHSIs have them on the top left of the bezel.

 

[azure]

It's a generic light, no biggie to replace with a non sandel part.

That's what I thought too. Before Sandel came out with a threatening SB forbidding the use of knockoff lamps, I bought one of those kits containing 3 lamps. The first worked fine; the second quickly became too dim to use safely. That's when my avionics tech told me about the Sandel SB, which specifically says that using a knockoff renders the unit unairworthy. Whether there's anything behind the SB other than greed, I sure wouldn't want to get dinged for improper maintenance (can't log the install if I do it myself since it doesn't qualify as preventative), so I tossed the remaining one and have used Sandel lamps ever since.

Maybe they really are generic and Sandel is just making a killing. But I haven't seen a Sandel lamp as dim as the one I had to have replaced.

 

[denverpilot]

It's a rental, and I couldn't tell you which Sandel it is.

Might be worth knowing. I'm sitting here thinking an electronic HSI almost surely has a mode to remove the DG portion and provide only the CDI indications.

Certainly any designer of such and the FAA who approved them would only need five minutes trying to fly behind one with an inaccurate or moving DG input or internal failure to know that's a very dangerous failure mode.

You'd think for three to ten times the price of a "non-certified" unit, somebody might have thought of that. We all hear how wonderful the certification process is and how it keeps us all safer by making modern avionics outrageously priced.

But perhaps they did and you didn't read the manual for your gear. Hard to say from here.

Just be thankful it wasn't an old mechanical one and the whole thing wasn't spinning when the DG gyro failed.

I've noticed this tendency toward single boxes doing everything in modern panels and I don't like it. There was a reason the "traditional" six pack had two CDIs, even if one didn't have a glideslope indicator... And a "serious" panel often had two CDIs with glideslopes.

You need to think about all failure modes and systems if you're going to fly in hard IMC, IMHO. Your "outs" need to include the limitations of the avionics with all possible failure modes analyzed.

My airplane with an old six pack and a single glideslope, for example: I'd better always *plan* to be in fuel range of an airport that's forecast to have non-precision approach minimums.

One CDI failure of the top CDI and my ability to fly anything but a ground-radar based precision approach instantly drops to zero. All it takes is one needle not moving.

Thinking about the eventual upgrade to having GPS on board, the glideslope CDI is staying and moving down to Nav 2. The GPS will get a compatible glideslope capable CDI and the King radio with ILS also moves down to Nav/Com 2. The one without a glideslope gets pulled along with its CDI.

Also realized this week that the DME is worth replacing with a used one. Just because the GPS can replace it in a no-failure mode, when the GPS or its display crap out, DME information will still be available, and compared to the price of the "do everything" GPS box, it's cheap insurance.