What's new
Van's Air Force

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

GSU 25 remote mount vibration issue

Draker

Well Known Member
I recently "derailed" this thread with my report of extreme vibration issues with the Garmin GSU 25 mounted remotely, aft of the baggage compartment. My device was attached via four bolts/nutplates to Van's AHRS mount, which consists of a pretty thin sheet of aluminum riveted to the exterior skin and the center brace between the bulkheads.

The symptom: AHRS display malfunction (incorrect ~20 degree roll depicted) at high RPM, with CAS messages indicating ATT MISCOMP and AHRS REVERT. Upon analysis, AHRS Dev (%) as logged by the G3X was spiking up to 700% during these episodes. Analysis across multiple flights indicated AHRS Dev (%) correlated most strongly with Eng. Pwr and RPM. Pretty likely vibration related.

It looks like the VAF consensus is that installing these ADAHRS units on the back of a GDU display is acceptable and indeed produces the best results for most RVs due to the strength and thickness of our panels. Well, I'm just too lazy to re-route my tubes and wires forward through the wing spar, so here we go.

After looking at a few examples sent to me, I set out to build a chonkier mount for this thing, while allowing me to keep it in its current location. After a trip to Metal Mart and a couple of designs I settled on a thick 1/4" 6061-T6 aluminum angle with another 1/8" angle attached. This guy is thick and beefy:



It fits right under the GSU unit, sandwiched between it and the original Vans mount. The 1/8" angle gets bolted to the long angle that goes down the center between the bulkheads:



So it adds some mass and is cross-braced to a solid, structural angle in the plane. I'm not a mechanical engineer, but I convinced myself it at least wouldn't hurt to try.

Well, three flights later results are so far so good! After 6 full-power takeoffs and 7 full-power climbs from 2500MSL to 5500 (tested every 5kt between between 65 and 95), I did not record any AHRS failures or CAS messages. Great. So let's analyze the logs.

Here, you can see plots of RPM (blue) vs AHRS Dev (red) before the new mount:





Observe the large spikes > 400 whenever full power was applied. These corresponded to the AHRS failures and CAS messages.

After installing the GSU on the thicker, braced mount:





As you can see, it's not perfect, but extreme spikes during full power ops are gone, and visually, the attitude display has been corrected. I'm going to spend another 10 or so hours testing this mount, and if I don't see any failures in that time, I'm going to consider this problem corrected!
 
Great Idea to start fresh with this.
I performed exactly the same analysis (for a few others too).

PastedGraphic-1.png

https://vansairforce.net/community/showpost.php?p=1683654&postcount=27

What id like to add:

Performing the ground vibration test guarantees nothing.
In fact, I never suspected vibration as mine had passed all of the ground testing, and I initially had it mounted to the spar carry through. The stiffest structure I could think of. This wasted a LOT of time. rewired canbus, isolating various LRUs in turn. you name it, I tried it. Even once I identified vibration as the culprit, I moved them two further times.

It only manifested itself in flight transitioning through certain combinations of airspeed, RPM and MP.

So in summary there's probably more to it than just the localised stiffness of the mount. But the ultimate solution is to keep moving it until it works.
BTW, my G5 has never missed a beat.

FWIW, I have an IO540 D4A5 9:1, with one Mag and one Surefly EI. Hartzell 3 bladed composite.
 
Last edited:
Good work, but I'm going to guess that that belly stiffener on the right is not nearly as stiff as it needs to be. If you can "oil can" that part of the skin, then it is insufficiently rigid.

Try clamping a diagonal stiffener to that side of the mount from the longeron.

You need to be well under 100% average. I got mine down to about 20%, from 400+
 
Good work, but I'm going to guess that that belly stiffener on the right is not nearly as stiff as it needs to be. If you can "oil can" that part of the skin, then it is insufficiently rigid.

Try clamping a diagonal stiffener to that side of the mount from the longeron.

You need to be well under 100% average. I got mine down to about 20%, from 400+

+1

The new stiff mounting plate is riveted to a very long and thin stringer, riveted to a large and thin skin. To me, that stringer is going to have A LOT of up/down movement and therefore anything attached to it will have the same. IMHO, you want to be somewhere stiffer. You want to be on more of a boxed type structure that has multiple attachment points for rigidity. Also, length is you enemy here, especially with thin metal. Must of us are familiar with the exhaust drumming on the lower skins. That confirms that those large unsupported areas of skins are not stiff. Point your exhaust at a 1/16 steel plate you hear nothing. Point it at an .025 sheet of alum and it sounds like a rock concert. That is movement.
 
Last edited:
Ryan, in the first plot above, note significant variation at what was probably a 2700 RPM climb at high AOA, as compared to five later periods of 2700 with around 1/3 as much deviation. If I'm guessing correctly, the five instances were 2700 in level flight.

If so, the major source is likely propeller beats, strong at high AOA and low airspeed, less powerful at low AOA and higher velocity...a drumming skin issue, as suggested.

As Mike said, try a brace, or simply try an addition of mass. I'd be curious about adding a small rectangle of steel or lead under the LRU footprint.
 
Ryan, in the first plot above, note significant variation at what was probably a 2700 RPM climb at high AOA, as compared to five later periods of 2700 with around 1/3 as much deviation. If I'm guessing correctly, the five instances were 2700 in level flight.

Without commenting on the structural aspects... Dan is on to an important issue here. The AHRS deviation values recorded in the G3X data log are only useful when the aircraft is stationary on the ground, or in totally unaccelerated flight in calm air. Any kind of maneuvering, power change, or even turbulence will pollute the data, since the numbers don't distinguish between movement due to vibration versus actual movement of the aircraft. So although different power settings can certainly have an effect on the AHRS, I would expect to see large values in the AHRS Deviation column any time the aircraft experiences any acceleration or rotation. It's only in truly steady-state flight that you can really make sense of this data.
 
Last edited:
To Matt's point, here is my before and after performance.
Replaced a stiff "engineered" x-y-z mount that was not rigid torsionally with a 3/16" 6061-T6 plate.
Tried all sorts of dampening, dressing tubes and wires, etc. before adding a temporary diagonal brace to stiffen the old mount torsionally, because I found that I could slightly twist the mount with my hand, and I was out of ideas!
That resulted in a marked improvement, so I went with "beef". The resulting improvement in AHRS Dev (%) was dramatic. No more issues. The spikes in the "after" are when I was trying to trip it up by pulling g etc.
 

Attachments

  • Clipboard02ab.jpg
    Clipboard02ab.jpg
    354.3 KB · Views: 43
  • Clipboard01ab.jpg
    Clipboard01ab.jpg
    303.3 KB · Views: 42
Last edited:
Thanks for the additional insight!

Ryan, in the first plot above, note significant variation at what was probably a 2700 RPM climb at high AOA, as compared to five later periods of 2700 with around 1/3 as much deviation. If I'm guessing correctly, the five instances were 2700 in level flight.

The in-flight high-RPM tests on "After" plots 1 and 2 were all climbs, at 95kt, 90kt, 85kt, and so on. (Im in the middle of my Vx, Vy tests) So the AOAs should be comparable. I wanted to simulate as much as possible takeoff conditions, without doing a bunch of touch and goes at my home airport where noise abatement requests discourage. The tests on plot 3 were all actual takeoffs from runways.

Good work, but I'm going to guess that that belly stiffener on the right is not nearly as stiff as it needs to be. If you can "oil can" that part of the skin, then it is insufficiently rigid.

As for the rigidity of the new mount, I agree, it's not all I could have done, and I'm also suspicious about the side of current mount that is only supported by the bottom skin. Before this first attempt, I could reach in and wiggle the mounted AHRS unit around by hand. Obviously terrible. Now, I still can move it but only with a lot of strength and only on that side. My condition inspection is coming up in a week, so now is the perfect time to get in there and do more.

The AHRS deviation values recorded in the G3X data log are only useful when the aircraft is stationary on the ground, or in totally unaccelerated flight in calm air. Any kind of maneuvering, power change, or even turbulence will pollute the data, since the numbers don't distinguish between movement due to vibration versus actual movement of the aircraft. So although different power settings can certainly have an effect on the AHRS, I would expect to see large values in the AHRS Deviation column any time the aircraft experiences any acceleration or rotation. It's only in truly steady-state flight that you can really make sense of this data.

For the usefullness of the AHRS deviation values: Matt thank you for the additional insights. Understood that in flight, this value is affected by all forces, not just vibration. I was looking for the closest proxy for "things that might cause AHRS problems" and that column looked like the measurement to try to minimize. Absent a debug build from Garmin that might provide more data, that column is the best I have!

On that note, I wonder if the higher deviation values on actual takeoffs are due to the surface of the runway? My home base runway's surface is asphalt, but it's not a slick, meticulously maintained international airport. Plus, I'm still an RV newbie, and my takeoffs are not exactly straight as an arrow yet.

Finally, does it even matter? As long as I'm not getting AHRS REVERTs anymore? Correcting that was all I set out to achieve here. I now pass the on-ground runup test and the AHRS doesn't go haywire on takeoff so is there any more benefit to further minimizing this "not very useful in flight" data log number?
 
Last edited:
Finally, does it even matter? As long as I'm not getting AHRS REVERTs anymore? Correcting that was all I set out to achieve here. I now pass the on-ground runup test and the AHRS doesn't go haywire on takeoff so is there any more benefit to further minimizing this "not very useful in flight" data log number?

If the ground vibration test is passed and the behavior is correct, then technically you have done everything specified in the installation manual. If you still have any doubt, the log file does contain some other encoded data that can be further examined if you send it to the g3xpert email address. You'll make their lives easier if you identify timestamps where the log file contains straight-and-level flight.
 
AHRS Data

Thanks for the additional insight!



The in-flight high-RPM tests on "After" plots 1 and 2 were all climbs, at 95kt, 90kt, 85kt, and so on. (Im in the middle of my Vx, Vy tests) So the AOAs should be comparable. I wanted to simulate as much as possible takeoff conditions, without doing a bunch of touch and goes at my home airport where noise abatement requests discourage. The tests on plot 3 were all actual takeoffs from runways.



As for the rigidity of the new mount, I agree, it's not all I could have done, and I'm also suspicious about the side of current mount that is only supported by the bottom skin. Before this first attempt, I could reach in and wiggle the mounted AHRS unit around by hand. Obviously terrible. Now, I still can move it but only with a lot of strength and only on that side. My condition inspection is coming up in a week, so now is the perfect time to get in there and do more.



For the usefullness of the AHRS deviation values: Matt thank you for the additional insights. Understood that in flight, this value is affected by all forces, not just vibration. I was looking for the closest proxy for "things that might cause AHRS problems" and that column looked like the measurement to try to minimize. Absent a debug build from Garmin that might provide more data, that column is the best I have!

On that note, I wonder if the higher deviation values on actual takeoffs are due to the surface of the runway? My home base runway's surface is asphalt, but it's not a slick, meticulously maintained international airport. Plus, I'm still an RV newbie, and my takeoffs are not exactly straight as an arrow yet.

Finally, does it even matter? As long as I'm not getting AHRS REVERTs anymore? Correcting that was all I set out to achieve here. I now pass the on-ground runup test and the AHRS doesn't go haywire on takeoff so is there any more benefit to further minimizing this "not very useful in flight" data log number?

Hey Ryan,

Nice work!

The AHRS Dev column can be useful in diagnosing AHRS problems, but there are pitfalls in analyzing the data in that column absent of other supporting data included in the diagnostic data we can pull from the equipment.

In this specific case, the 2nd set of ARHS data does appear to be objectively better. We can run through the log(s) for you if you would like, we can get a better idea of the overall health of the AHRS.

Like Matt mentioned, the ability to compare 2 plots depends very much on the nature of the two flights. Ideally, the two flight would be performed back to back, and be identical in maneuvering (or as close to identical as you can get).

Thanks,

Justin
 
I'm curious if this is a common problem or not. I too have used the Van's provided mount and wonder if Van's will investigate / improve the mount. In my application I have my GSU units turned 90 degrees from Ryan, but at this point I hope my results will be different.
 
Back
Top