What's new
Van's Air Force

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

Viking 110 notice

It seems fairly obvious that the crack initiated at the outer diameter of the boss in the center of the drive flange where the bending load was concentrated. Likely the outer crack originated where the boss is located on the ear to hold the drive pin, there is another sharp change in section there.

Section changes are obviously bad practice. However, consider the torsional load on that free-standing lobe. Think of the pin as the handle of a wrench.

Real analysis would be best, but if it has to be TLAR-based, use a disc with three pins.
 
Whirl Mode

Granted, I didn't consider that. On further "associative memory" reflection, recalling the Lockheed Electra in-flight failures there could be a "whirl mode" issue?

This is a "just thinking-what if" observation. Not saying it's necessary.

Hoping you have an additional cogent insight.

:D

The whirl mode issues on the Electra were actually wing resonance/flutter failures induced by a tendency of the propellor/prop shaft to move in an orbital motion instead of rotating about their true centerline. This was not induced (AFAIK) by gyroscopic forces from the propellor.

Skylor
 
The whirl mode issues on the Electra were actually wing resonance/flutter failures induced by a tendency of the propellor/prop shaft to move in an orbital motion instead of rotating about their true centerline. This was not induced (AFAIK) by gyroscopic forces from the propellor.

Skylor

You defined whirl; Imbalance, worn bearings, resonance, alignment, gyro forces (secondary?), flutter, and probably more are contributors. Sounds as though you're saying this cause/effect excited structure and/or aero-surfaces.

For you rotating equipment types; can you have super-synchronous whirl? I'm aware of the sub-sync variety and can wrap my head around that. Any real knowledge you could share would be appreciated.

To quote the phrase "TLAR" that is used here, its application to PRSUs is down right scary to me. People want to defend their choices, awesome. Don't do it to a fault. Keep both eyes open. No one has built the perfect aircraft.
 
To quote the phrase "TLAR" that is used here, its application to PRSUs is down right scary to me. People want to defend their choices, awesome. Don't do it to a fault. Keep both eyes open. No one has built the perfect aircraft.

This is precisely what scares me about auto conversions. The torsional vibration issues in a gearbox CAN be overcome, they ARE well understood - but you're not going to get there with TLAR.
 
Yup. I can only think of two companies who actually engineered and tested their redrives properly for the Experimental market. All the others were designed with TLAR and most of these have had issues to varying degrees (some catastrophic and deadly).

TLAR can work if you deal with each failed part as they break but it's time consuming and potentially dangerous. Certainly not the way to do it for commercial products.

All redrives really need to be validated regardless of how or who designed them though. Even the experts can overlook something and have failures. Simulation and computer modeling isn't a substitute for actual testing on the engine with the prop in place, preferably in flight.

These small companies rarely have the resources to do this and most won't even bother consulting an engineer, figuring they are quite capable of designing a reliable drive. We see the results here, yet again.
 
Last edited:
I now have the second failed engine, the one with 800 hours and the earlier type gearbox input flange. I dont think it is a coincidence that the older unit lasted longer, it had not been welded on. Whereas the 134 hour failure had the 3 pins welded on with a high heat input given away by the blue discoloration around the pins.

I have not yet had a chance to even look at the Hubelbank engine, I literally got the pallet off the trailer and inside the garage seconds before a thunderstorm took a big dump on the area... So the attempt at getting the first drive flange off the 134 hour gearbox will have to wait till tomorrow.

I did do some measuring on the aluminium backplate that supports the input shaft bearing. The bearing OD is 63mm and the "boss" on the back of the plate is 66mm. So folks, all that is holding that bearing in place is 1/16" wall thickness of aluminum. The lightening measures on this gearbox have clearly been taken way too far. Im pretty sure that section will be crushed from the force needed to press out the input shaft from the drive flange, so I am going to try to reduce the load by cutting most of the way through the circular section of the drive flange in 3 places where there is the best access for a 3" cut off wheel.

I'm posting a picture of the bearing splitter that I got with the hope that I could close it enough to support the inner race of the bearing so that the aluminum back plate that is so thin doesnt have to bear all the press force. I also got a heavy wall tube to support around the gear in case the bearing splitter plan does not work out.
 

Attachments

  • IMG_0387.jpg
    IMG_0387.jpg
    649.8 KB · Views: 80
  • IMG_0381.jpg
    IMG_0381.jpg
    600.3 KB · Views: 96
This is precisely what scares me about auto conversions. The torsional vibration issues in a gearbox CAN be overcome, they ARE well understood - but you're not going to get there with TLAR.

Not just the TVs that keep me shacking my head on this one. While I haven't seen the associated fracture surfaces personally, the pix sure look like they're the result of fatigue. While I'll admit that I haven't seen the entire assembly, that thin, cantilevered "base plate" is a real head shaker. Does anyone with any appreciation/understanding of the associated art think that axial and angular alignments remain constant after loading? While this probably won't affect the related freqs/cycles, the associated amplitudes would increase. Alignment deviation could very well be the primary driver why some failed earlier than others. All speculation from the limited info available here, of course.

The entire approach is a tough one for me to accept. If I'm wrong, educate me. Willing to learn here.
 
The biggest mystery right now is why just 1 drive ear failed. If the remaining 2 were close behind the first, then one could say that misalignment was the primary culprit. Also the design of the drive flange and the type of material used with "accidental" heat treatment through the welding process could be another factor. If a low alloy/CrMo type material had been used and it had been ductile it may have taken much longer to fail.

But to break off a single ear with no obvious cracks in the 2 remaining ears, it suggests that the guibo itself may have inconsistent segment lengths. Such that just a single segment was loaded and the remaining pair just loafing until the tight one failed. Of course the guibo is very strong given the fact that it was designed to go behind the transmission where max torque may be 4x engine torque of 250hp+ engines. Thus it comes down to the weakest link in the chain which was the input flange.

I think that the alternate guibo with a carrier plate with bushings will be a lot more dimensionally consistent and spread the load more evenly. The drive flange needs the sharp changes in section mitigated so that you do not get load concentrations and the welding should be eliminated, in other words go back to bolts on the input side. The flywheel side drive flange seems to have withstood all the abuse without any cracking so that side could potentially stay unchanged for now.

Of course the gearbox mount has to be fixed and that cantilever effect has to be removed. That is going to be the toughest design change since most of the bell housing flange was cut off the block by Jan as part of prepping the block for the conversion. It seems that there are still some bosses left on that side of the block (the underside as installed) that could support a new bracket. Worst case scenario the fix involves building up a new base engine. $600-1300 for a lowish mileage engine and swap all of the parts over. Perhaps freshen it up with new pistons, rings, bearings, a hone and it gives one a chance to see the inside of the engine before you commit to putting it on your airplane.

Obviously given 450 installed engines, if a solution can be found that is compatible with the sawed engine blocks, that would be best for everyone and will be the primary focus. The gearbox will probably need a new back plate with extra material for the new bracket and less material removal in some critical areas. I would like more than 1/16" of material holding the bearing in place, for instance...
 
The biggest mystery right now is why just 1 drive ear failed. If the remaining 2 were close behind the first, then one could say that misalignment was the primary culprit..

But to break off a single ear with no obvious cracks in the 2 remaining ears, it suggests that the guibo itself may have inconsistent segment lengths..

I would be @ 100% understanding and agreement if the fracture surface (pictures) appeared to be an ultimate failure. If that were true, the other flange segments would have followed in very close order; assuming the associated stress/strain was not caused by installation irregularities. It is very conceivable that the asymmetric assembly and operational stresses would create higher amplitude cycles at one location versus the others.

This is not to say that your other observations aren't influencers. Again, I'm talking primary only base on the fracture surface pic alone. Thanks for your insight. Keep it coming.
 
The fatigue cracks had propagated over time and the "ultimate" failure only had a fraction of the parent material left when it let go. Yet magically the 2 other ears did not have the same sort of overload condition. All the parts but the guibo were cut on a CNC which "should" have had good enough accuracy for a consistent pin distance for the guibo segments.

If the shafts were then moved out of alignment I have not yet come up with a method that favors 1 pin, all of the pins should in turn have been overloaded as the pair of drive flanges rotated through the offset distance. One should actually been able to feel this tension load if one removed the spark plugs and rotated the prop by hand. It would go tight, loose, tight loose as each drive pin reached the point of maximum travel away from the flywheel side guibo pin, which cant move.

One would need to obtain several guibos and measure the bushing to bushing center distance while the link was under a specific load to see how much variation there was. What we do know is that the failure mode was identical in 2 cases and both used the same type of Guibo.

Aeromomentum I believe uses the same type of guibo in their gearbox design and uses bolts and not pins to hold them on (just like BMW) and so far they have not yet had any failures that I know of. Bot of course they also have a "proper" mount on their gearbox.
 
Yup. I can only think of two companies who actually engineered and tested their redrives properly for the Experimental market. All the others were designed with TLAR and most of these have had issues to varying degrees (some catastrophic and deadly).

TLAR can work if you deal with each failed part as they break but it's time consuming and potentially dangerous. Certainly not the way to do it for commercial products.

All redrives really need to be validated regardless of how or who designed them though. Even the experts can overlook something and have failures. Simulation and computer modeling isn't a substitute for actual testing on the engine with the prop in place, preferably in flight.

These small companies rarely have the resources to do this and most won't even bother consulting an engineer, figuring they are quite capable of designing a reliable drive. We see the results here, yet again.

The redrives that I am using on my Jodel and now RV3 have been designed by someone who is doing this for a living. I believe he also has the Robinson helicopter and several Formula 1 teams on his resume. They have then been built by people who know their stuff. The company was preparing to race their engines at Reno (in Lancairs). Several millions of dollars of R&D money have been invested. And in the end, even these guys ran into a brick wall with their 6 cylinder turbocharged engines and Hartzell propellers. So much so that they have discontinued their operation. Leaving me with two of their very few redrives in existence.

My 4 cylinder engine and lighter, wood-core MT prop run free of torsionals, as long as I stay away from the 1100-1500 rpm range with driven motor. This means per flight two times - during startup and after landing when leaving the runway for a brief half a second. I can live with this, but it would not look good on a production engine.

What I am saying is - this stuff is hard.

Jan doesn't have millions of dollars to his disposal to perform proper research and product development. He didn't when he was doing the Subarus and most likely he still hasn't.

It's sad to see that some customers have to suffer the consequences, apparently.

(And I'd be interested to learn which are these two companies with proper R&D Ross!)
 
if the 3 pieces of the guibo slip on and off of the pins then i don't see how there can be a strain on one pin. seems like this test would make the focus more on the dimensions of the metal piece
 
Bob, given that the broken drive ear went out the side of the cowling at high speed in both cases, there is now no way to determine the geometry of the overloaded pin position. But I think that everyone agrees that if both drive flanges had been symmetrical and if the guibo links were the same length, then either all pins should have been loaded up or none. In which case all 3 drive ears would either be cracked or not.

To have total failure on 1 drive ear to the point that it breaks off completely and then have the 2 remaining drive ears look nearly perfect suggests that something has to be way off, either in the geometry of the drive flange itself or the guibo. If the spline features in the hub of the drive flange were to be non concentric to the axis of the hub, that could be another potential problem because that could put 1 drive pin out at the longest radius and it would always be taking the load.

Usually it is quite difficult to get good concentricity on broached features. A lot of gear makers would do the broaching operation first and then locate off that feature when cutting the gears so that they run true. Unfortunately I have no idea on what the manufacturing process is for both the shaft and the drive flange itself, although seeing the shaft slice chips off the drive flange as it is pressed on is not very re-assuring about the fit up of the 2 parts.
 
This evening had a little help from my neighbor with his 20 ton arbor press. Pressing out the input shaft from the flange proved to be uneventful. I just used the tube spacer that was a close fit to the gear OD and it came out with no drama or strain.

There appeared to be no visible cracks on the 2 remaining drive ears. Dye penetrant inspection will be tomorrow. Then it goes to the metallurgist for analysis. Based on the totally asymmetric damage, I am concerned that its possible that the splined connection may not be concentric. So if you own one of these, remove the guibo and attach a dial indicator to the gearbox housing with the plunger perpendicular to the input shaft centerline. Then adjust the dial indicator so that when one drive pin on the input shaft is at maximum radial dimension the meter is zeroed. Now check the other 2 drive pins. Are they at the same radial distance ? If not that is a problem. The pin that has the largest radial distance will be the hero pin that takes all the load and ultimately breaks off...
 

Attachments

  • IMG_0388.jpg
    IMG_0388.jpg
    111.8 KB · Views: 74
  • IMG_0389.jpg
    IMG_0389.jpg
    145.2 KB · Views: 70
  • IMG_0390.jpg
    IMG_0390.jpg
    165.5 KB · Views: 77
Did the dye penetrant inspection on the drive flange. Confirmed that the 2 other ears are not cracked. There were cracks on the welds holding the pins in place but it is quite likely that because of the high carbon of the pins themselves that these cracks formed when the weld solidified, in other words on the day the part was made.

So we are still dealing with a scenario where somehow 1 pin was carrying all the dynamic load until it failed.
 

Attachments

  • IMG_0392.jpg
    IMG_0392.jpg
    31.7 KB · Views: 57
  • IMG_0393.jpg
    IMG_0393.jpg
    33.6 KB · Views: 61
  • IMG_0394.jpg
    IMG_0394.jpg
    28.1 KB · Views: 45
  • IMG_0395.jpg
    IMG_0395.jpg
    31.5 KB · Views: 54
  • IMG_0396.jpg
    IMG_0396.jpg
    33.9 KB · Views: 56
  • IMG_0397.jpg
    IMG_0397.jpg
    218.8 KB · Views: 54
There were cracks on the welds holding the pins in place but it is quite likely that because of the high carbon of the pins themselves that these cracks formed when the weld solidified, in other words on the day the part was made.

Perhaps, but maybe not. I've seen homebuilt axle assemblies with the same cracks, a function of poor design.

Easy to check. Grind off the weld and see if the axle will come out of the flange without application of significant force.

See the sketch. The assembly will not survive unless the axle is a shrink fit in the flange. Even if designed for it, production tolerances can make it iffy.
.
 

Attachments

  • Bad Axle.jpg
    Bad Axle.jpg
    12.9 KB · Views: 75
Last edited:
Dan, the pins are pressed in, then welded. There was no incidence of pins falling out prior to welding, but then apparently it was considered "safer" to weld them.

I was technically responsible for urea injector production where a hardened 440C orifice plate was laser welded to the injector body following a press fit. The annual volume was approximately 50k/yr from 2012 to the present so 10 years at least. I can assure you that welding high carbon steel successfully is a very difficult proposition and required a very narrow beam (200um) with very accurate position of the beam relative to the joint and very high weld speed (600rpm on the collet chuck) with only 5um total runout of the part.
 
Dan, the pins are pressed in, then welded. There was no incidence of pins falling out prior to welding, but then apparently it was considered "safer" to weld them.

Works for me. Apparently some were flown without welds?
 
Some are being flown without welds. Jan knows who has which parts in what state. The point its that the SB version of the system failed at 134 hours whereas the original input shaft drive flange with the bolts went 800 hours before it failed.

So I hesitate to accept that the SB version is in fact an improvement on the original version. The main issue with the original design was the aluminum flywheel which would crack and some people are still flying that...
 
Perhaps, but maybe not. I've seen homebuilt axle assemblies with the same cracks, a function of poor design.

Easy to check. Grind off the weld and see if the axle will come out of the flange without application of significant force.

See the sketch. The assembly will not survive unless the axle is a shrink fit in the flange. Even if designed for it, production tolerances can make it iffy.
.

Dan if you look closely around the drive pin you will see the slightest witness mark from the dye. That was just a tiny bit of seepage from that joint, you can see it hardly bled out at all compared to the bleed out from the end of the crack. So that also indicates a tight fit of the pins to the hole in the flange.
 
If the shafts were then moved out of alignment I have not yet come up with a method that favors 1 pin, all of the pins should in turn have been overloaded as the pair of drive flanges rotated through the offset distance.

Just a sketch to illustrate the above, something I was playing with to satisfy my own curiosity. Agree with Keith; non-concentric rotation should load and unload the three ears equally, in turn, at first order. Note the change of length between the driving/driven pin pair given 180 degrees of rotation.
.
 

Attachments

  • Axial Misalignment.jpg
    Axial Misalignment.jpg
    33.2 KB · Views: 85
Dan, so far the only scenario I have come up with is if the entire drive flange was offset to the shaft centerline. That would give you an offset distance that varies with the rotation angle. That could favor a single pin if the offset was towards that pin, so that the pin found itself at the greatest distance from the shaft centerline.

Jans video on fitting the drive flange to the gearbox seems to show a lot of difficulty to get it to go on straight and keep going on straight. The splines on the shaft also slice material out the inside of the drive flange, potentially making any shift permanent. Here is the link to that video:
https://youtu.be/gW0IyuERbcU?t=147

His assembly process seems geared toward creating exactly the type of defect that would cause this problem, be highly variable, and all down to human error because its all "eyeballed", If he actually used a dial indicator to validate that the radial distance of the pins was the same (so it was centered) that would at least be a check and balance. But no evidence of that at all.
 
Last edited:
Well, I have news. I let the drive flange sit for hours since I did the inspection this morning and as it turns out there is indeed the start of a crack at a second drive ear. It took a long time to bleed out to the point that I could recognize it, so its probably very shallow still.

Interesting that it is not at the inboard end of the ear, its at the outboard end and tangential to the boss I mentioned earlier which holds the pin. The broken ear shows 2 different cracks, one outboard and 1 inboard. But the overall assumption is the same, far more damage to the ear that failed and damage now starting on one of the 2 remaining.
 

Attachments

  • IMG_0398.jpg
    IMG_0398.jpg
    866.7 KB · Views: 99
Jans video on fitting the drive flange to the gearbox seems to show a lot of difficulty to get it to go on straight and keep going on straight....His assembly process seems geared toward creating exactly the type of defect that would cause this problem...

I especially like the part where he recommends hitting it with a hammer for alignment (about 3:35). The only place to hit it would be the ends of the lobes. Ack!

You're going to replace this thing with something better anyway. 100 lbs-ft nominal torque suggest a size 12. Replace the driving end with a flat disk (or better, integral, a new flywheel with a thickened center) drilled four places for Type S pins. Used a splined Type 1 slug on the driven side. If the existing splines are standard SAE, use a Type 1 L-LOC, which would allow slipping it onto the splines and clamping rather than using some variation of the press method. See attached.

Note the Centaflex A coupler places the rubber elements in compression when the radial fasteners are tightened. There are no tension fibers buried in the rubber.
.
 

Attachments

  • ScreenHunter_1444 Dec. 01 13.09.jpg
    ScreenHunter_1444 Dec. 01 13.09.jpg
    39.3 KB · Views: 87
  • Dimensions.jpg
    Dimensions.jpg
    85.6 KB · Views: 53
  • Torsional Data.jpg
    Torsional Data.jpg
    122 KB · Views: 73
  • L-LOC.jpg
    L-LOC.jpg
    224.7 KB · Views: 71
  • Type 1S.jpg
    Type 1S.jpg
    41.1 KB · Views: 85
The barrier to experimentation right now is my ability to get some of the machined parts. I would have to reverse engineer the splined joint to be in a position to make my own input flange. From the discussion yesterday with Jan his gearbox supplier is legit with broaching, hobbing, heat treatment and gear shaving all in house. Part of the problem is that the 110 is one of the smallest engines in the lineup and its using a guibo that was designed for much more powerful engines. This makes the guibo "hard" for this application and it would for sure be better to use a product that has examples made for lower nominal torque values.

DanH I appreciate your recommendation and will try to educate myself more on the option that you raised. I do believe that a Civic manual transmission flywheel may fit onto the Fit crankshaft, they are available in CrMo with the starter teeth machined right into the billet flywheel, 1 less thing to go wrong. The flywheel would need to be lightened since its not accepting the thrust of a clutch plate, so the wear surface can be removed.
 
Same thing with Airtrikes out of Canada, bolts instead of pins, and NO CUT UP
BMW/MERCEDES rubber u joint, auto flywheel is lite, but factory made
 
I am guessing (which is bad) that the only reason to cut up a BMW OR MERCEDES “ rubber donut” is if alignment is questionable on mating pieces
 
I was at the Work Boat Show in New Orleans yesterday and talked to several companies that specialize in TV dampers (Rexnord Centaflex-A, Geislinger, Timkin Lovejoy LF, Vulkan Megiflex-B, Renold HTB-MP, Isoflex, VibratechTVD). The guys at Vulkan had one of the best solutions for a 125 hp/6000 rpm engine-gearbox combination. Vulkan recommended using their Megiflex-B (similar to Centaflex-A and Lovejoy LF) with a split splined collar that would attach to the splines in the GB drive pinion shaft. The collar would clamp to the splined shaft with a cross bolt to close the split and prevent the collar from moving. The collar would have three threaded radial holes with 120 deg offsets for attaching the damper with hex socket bolts. The damper would be face bolted to the crankshaft side drive plate with hex bolts. Assembly would be similar to the current setup. The Vulkan folks said they would be happy to assist in the design, TV analysis, and manufacturing as desired (all the others likewise offered design assistance). The trick of course is finding and testing the correct Shore hardness for this application. The photo is the Centaflex-A version of the above.

OXE had a cutaway of their 300 hp Diesel outboard that uses a modified BMW turbo diesel engine for the powerhead. The TV coupler shown is on the business end, some type of molded plastic housing with puck style elastomers for light weight. Nice compact engine package designed to run all day at high rpm.

John Salak
RV-12 N896HS
 

Attachments

  • Rexnord Centaflex-A Coupled example.jpg
    Rexnord Centaflex-A Coupled example.jpg
    420 KB · Views: 89
  • OXE 300 HP Diesel Outboard, BMW Engine TV Coupler.jpg
    OXE 300 HP Diesel Outboard, BMW Engine TV Coupler.jpg
    590 KB · Views: 103
  • OXE 300 HP Diesel Outboard, BMW Engine TV Drive Cutaway.jpg
    OXE 300 HP Diesel Outboard, BMW Engine TV Drive Cutaway.jpg
    800.5 KB · Views: 91
Less than a new IO-540, but not by much. There is a reason the primary user market is government, military, and law enforcement. A 12-meter interceptor with 3 of these engines cost about the same as a house and will only do 54 knots on 900 hp. This is a well-engineered machine that has many millions of development dollars sunk into it.

John Salak
RV-12 N896HS
 
Took a look at the Hubelbank drive flange today. This went 800hrs to failure and is the original type of input flange that takes bolts to hold on the guibo. I compared the material thickness, it is identical to the later type at 1/4" thick for the flange on both. The bolt style flange does not have the bosses to hold the pin, in fact it has a shallow counterbore on the back side that is 2mm deep with a sharp inner corner radius.

The fracture itself is straight until it reaches the point of ultimate failure where it has a slight hook shape.

This flange was not a press fit on the gearbox input shaft, it is a slip on. There is a slight indication of some fretting where the flange touches the step on the input shaft from the slight rotational clearance.
 

Attachments

  • IMG_0400.jpg
    IMG_0400.jpg
    54.6 KB · Views: 66
  • IMG_0401.jpg
    IMG_0401.jpg
    39 KB · Views: 69
  • IMG_0402.jpg
    IMG_0402.jpg
    39.9 KB · Views: 64
  • IMG_0407.jpg
    IMG_0407.jpg
    546.2 KB · Views: 63
Some views on the fracture face.
 

Attachments

  • IMG_0403.jpg
    IMG_0403.jpg
    132.7 KB · Views: 75
  • IMG_0404.jpg
    IMG_0404.jpg
    133.9 KB · Views: 70
It looks like a Centraflex A size 12 coupling (140Nm nominal torque rated) is about Euro 120. The Michigan based distributor is still going to get back with me... This is for the outer element. The inner element is probably going to have to be custom made to suit the spline on the gearbox shaft. Its circular on the outside with 3 tapped radial holes to transfer the torque.
 

Attachments

  • Centraflex type a.JPG
    Centraflex type a.JPG
    22.2 KB · Views: 68
The only news from the dye penetrant inspection is that there is the start of a second crack on a second ear, just as with the first flange.
 

Attachments

  • IMG_0408.jpg
    IMG_0408.jpg
    409.9 KB · Views: 98
Last edited:
Giubo vs Guibo vs Guido

Took a look at the Hubelbank drive flange today. This went 800hrs to failure and is the original type of input flange that takes bolts to hold on the guibo. I compared the material thickness, it is identical to the later type at 1/4" thick for the flange on both. The bolt style flange does not have the bosses to hold the pin, in fact it has a shallow counterbore on the back side that is 2mm deep with a sharp inner corner radius.

The fracture itself is straight until it reaches the point of ultimate failure where it has a slight hook shape.

This flange was not a press fit on the gearbox input shaft, it is a slip on. There is a slight indication of some fretting where the flange touches the step on the input shaft from the slight rotational clearance.

I just want to point out that the correct spelling is "giubo" not "guibo", but the misspelling is common.

Every time I read "guibo" in this thread, it sounds in my head a lot like a male Italian name that gives me flashbacks of the classic early eighties movie "Risky Business".

"I can't believe this. I've got a trig midterm tomorrow and I'm being chased by Guido the killer pimp"

Skylor
RV-8
 
So far, my overall impression is that the web of the drive flange is too thin for the cantilever loads it has to carry. And the material has too little ductility. And the welding of the pins makes everything worse rather than better.

It could be that the giubo (thanks skylor) does not do such a good job of damping on this, what used to be the smallest engine in the lineup before the 3 cylinder Mitsubishi was added.

I would abandon the BMW style giubo at this point and start looking at the Centraflex couplings because there does not seem to be a cost penalty and they are an engineered coupling all the way to the attachment points. The gearbox side drive flange would get replaced by a cylindrical part that is again a slip fit on the splines and attaches to the rubber coupling with 3 bolts. Use some copperslip high pressure grease on the splines to prevent fretting.

To continue further, I am looking at the possibility of a custom CrMo flywheel of reduced diameter, which does not project past the block as it was modified by Jan. If I can find a short enough starter, it could now be attached to the gearbox back plate which is anyway going to be replaced. But of course the starter has to be short enough to fit in the cowling. Or, we would have to add a 3rd gear the same as the input gear running in some bushings that would allow the starter to go behind the gearbox in a suitable spot. Reducing the diameter of the flywheel opens up the possibility to add the 5th support to the gearbox to get rid of the overhang and stiffen everything up.
 
If there is anyone who owns a running Viking 110 engine in the Michigan, Northern IN or OH or NE IL who would allow his engine to be instrumented for a vibration sweep, please contact me. This would just be a ground run. The object would be to determine if there are any obvious resonance modes we can uncover which may be the cause for high loads on the drive pins.
 
The inner element is probably going to have to be custom made to suit the spline on the gearbox shaft.

That input spline is not some SAE standard? Sure would be nice to use a clamped L-LOC.

Its circular on the outside with 3 tapped radial holes to transfer the torque.

Four.
 
DanH I see there is a way to register and get CAD files. PDF Catalog shows 3 holes in the photos and no axial view so no way to determine how many holes there are. But I see the online product catalog (not the PDF) shows individual elements with a picture and basic spec.. Would you suggest trying the lowest shore first ?

 

Attachments

  • CF-A-012-0-50.JPG
    CF-A-012-0-50.JPG
    38.6 KB · Views: 67
DanH I see there is a way to register and get CAD files. PDF Catalog shows 3 holes in the photos and no axial view so no way to determine how many holes there are. But I see the online product catalog (not the PDF) shows individual elements with a picture and basic spec.. Would you suggest trying the lowest shore first ?

Complete dimensions in the catalog, see below. A #12 has 4 lobes at 90 degrees.

Selection is the usual balanced compromise. As you know, it's not really a matter of Shore hardness, but rather, dynamic torsional stiffness. Design goal is to move the fundamental resonant RPM below the normal operating range without allowing the F2 to creep into the upper end of the range. Less stiffness will move it lower, more higher. In the optimum case, you would model the system with the available stiffness values to see where things land. However, in this case I think it is reasonable to experiment.

So take a good look at the torsional data for the #16 as well as the #12. The #16 is a three lobe in a slightly larger overall diameter and thickness, and thus offers even lower dynamic torsional stiffness values...while at the same time increasing the allowable nominal and vibratory torque. The issue might be max RPM, 6000 vs 6500. I don't know the max for the Fit.

Two plots below, live vibratory propshaft telemetry from 1999, 3 cyl Suzuki with a cog belt, driving through a 50 Shore #12. The test goal that day was to compare vibratory behavior with two different props, one maple, one mahogany, thus quite different in stiffness. Note there was very little....but I digress.

Some items of note, clues regarding your app. First, the fundamental resonant frequency of the system was 30 hz. That's the 1200 RPM shown here with the 3-cyl, but remember, it would be 900 RPM with your 4-cyl. (RPM x #cyls) / 120 = hertz, forcing frequency for a four stroke.

Vibratory amplitude was 160~180 lbs-ft measured at steady state throttle. Vibratory torque is almost proportional to manifold pressure, so it is important to minimize throttle opening while passing through the resonant range. That's the most important reason to push the F1 below the operating range, so you pass through it only at start, preferably with throttle at idle.

Note the F2 creeping in at the upper end of the RPM range on the plot for the 68" prop. I wasn't able to get enough static RPM to see it with the 72" prop, but you can bet it was there. Because of the greater output torque oscillation of the 3-cyl, I was using a high inertia flywheel, certainly as compared to the light ring gear disk I see in the Viking videos. In your case I think a low mass flywheel will push F2 up the scale, but I've not done any analysis.

If you have the physical space for it, I think I'd try the #16 in Shore 50, with the backup plan of a Shore 60.
.
 

Attachments

  • Dimensions.jpg
    Dimensions.jpg
    71.7 KB · Views: 68
  • Torsional Data.jpg
    Torsional Data.jpg
    98.4 KB · Views: 46
  • Fig1.jpg
    Fig1.jpg
    48.9 KB · Views: 66
  • Fig2.jpg
    Fig2.jpg
    45.6 KB · Views: 64
Great post Dan. The size 12 is approximately the same size as the current giubo and the 16 is basically 6" OD.

I actually think the #16 will be easier to mount because its bolt circle will be outside the flywheel bolt circle where the 95mm pitch circle of the BMW part puts it very close to the flywheel pitch circle making it harder to accommodate. The pitch circle of the #16 being 125mm should work nicely on a manual transmission flywheel. Unfortunately my R18 engine is in the hangar so I couldnt take a picture of that, since it has a manual transmission flywheel on it, but attached a pic showing the 125mm bolt circle dimension on the existing flex plate.
 

Attachments

  • IMG_0410.jpg
    IMG_0410.jpg
    615.8 KB · Views: 87
Complete dimensions in the catalog, see below. A #12 has 4 lobes at 90 degrees.

Selection is the usual balanced compromise. As you know, it's not really a matter of Shore hardness, but rather, dynamic torsional stiffness. Design goal is to move the fundamental resonant RPM below the normal operating range without allowing the F2 to creep into the upper end of the range. Less stiffness will move it lower, more higher. In the optimum case, you would model the system with the available stiffness values to see where things land. However, in this case I think it is reasonable to experiment.

So take a good look at the torsional data for the #16 as well as the #12. The #16 is a three lobe in a slightly larger overall diameter and thickness, and thus offers even lower dynamic torsional stiffness values...while at the same time increasing the allowable nominal and vibratory torque. The issue might be max RPM, 6000 vs 6500. I don't know the max for the Fit.

Two plots below, live vibratory propshaft telemetry from 1999, 3 cyl Suzuki with a cog belt, driving through a 50 Shore #12. The test goal that day was to compare vibratory behavior with two different props, one maple, one mahogany, thus quite different in stiffness. Note there was very little....but I digress.

Some items of note, clues regarding your app. First, the fundamental resonant frequency of the system was 30 hz. That's the 1200 RPM shown here with the 3-cyl, but remember, it would be 900 RPM with your 4-cyl. (RPM x #cyls) / 120 = hertz, forcing frequency for a four stroke.

Vibratory amplitude was 160~180 lbs-ft measured at steady state throttle. Vibratory torque is almost proportional to manifold pressure, so it is important to minimize throttle opening while passing through the resonant range. That's the most important reason to push the F1 below the operating range, so you pass through it only at start, preferably with throttle at idle.

Note the F2 creeping in at the upper end of the RPM range on the plot for the 68" prop. I wasn't able to get enough static RPM to see it with the 72" prop, but you can bet it was there. Because of the greater output torque oscillation of the 3-cyl, I was using a high inertia flywheel, certainly as compared to the light ring gear disk I see in the Viking videos. In your case I think a low mass flywheel will push F2 up the scale, but I've not done any analysis.

If you have the physical space for it, I think I'd try the #16 in Shore 50, with the backup plan of a Shore 60.
.

Great post Dan!

Interestingly, the first order TV peak manifests itself at exactly the same place where my 4 cylinder EJ-25 seems to peak. And where Ross mentioned his is as well.

And your comment on throttle opening determining the gravity of the torsional vibrations makes sense of why I don't notice any torsional misery when landing (engine idling, speed coming down, I pass through the Forbidden Zone), but DO notice it when I throttle up a bit to get the engine speed back to >1700 for taxiing.
 
Yes, almost all 4 cylinder engines I've seen with PSRUs exhibit a noticeable TV period in this 900 to 1200 rpm range roughly, including the Rotax 912. There may be other periods above this range as well.
 
Yes, almost all 4 cylinder engines I've seen with PSRUs exhibit a noticeable TV period in this 900 to 1200 rpm range roughly, including the Rotax 912. There may be other periods above this range as well.

Roughly similar prop inertias, roughly similar crankshaft inertia and stiffness.

The systems with some sort of soft element between the the two bulk inertias tend to ballpark an F1 of 25~35hz. Older belt drives, like the Blanton, Reductions, and Belted Air Power with the lower sprocket bolted directed to the flywheel are closer to 45 hz, depending on belt choice.

Again, remember, the RPM at which it resonates is dependent on number of cyls. A system with a 45hz F1 on a six and a 30 hz F1 on a four both resonate at 900 RPM.

The "highly tested" Reductions Inc drive I purchased for the 3 cyl, circa 1995, hammered like a 50 caliber at about 1800. I'm like "What the ????..". Young and dumb.

BTW, if I may make an observation, something I'm sure Hans and Ross already understand. Total operating hours means very little, because component stress is relatively low at any non-resonant RPM. What counts is the (1) number of times it has been through the resonant range, and (2) with how much throttle. It's one of the reasons why two otherwise identical installations may exhibit very different time between failures.
 
Last edited:
The "how much throttle" remark really makes a good point. My standard operating procedure upon landing is to "push through" the rough range, back to a safe 1700 rpm. This by giving it a fistful of throttle. Maybe better to take it slow there, and ease it through instead. Next time I go flying, I'll have a look how that goes.

(And 99% of my flights are short 20-minute hops, so the number of cycles do add up)
 
I've been involved with several belt PSRUs also on Subarus, very similar lower resonance RPM range. It's always there, regardless of flywheel mass and fairly different prop MMOIs too, just varies in severity.

I push quickly through the range as Hans does. My amplitude was greatly reduced and the range narrowed a bit by doubling flywheel inertia but it's still there. I do a lot of short flights for testing purposes mainly now so it's seen many hundreds of passages through this TV zone.

As I decelerate through the zone, TV is minimal as Hans also observed.

You should see what a Rotax 912 does at 700-900 rpm.. Wish I still had the video of that on my test stand...
 
Last edited:
vib limit

Interesting thread.

The data that Dan posted in #74 for the size 16 damper shows a continuous vibratory limit of .08 kNm ~ 60 foot lbs.

Is that OK? It looks like there is plenty of headroom for the DC torque, though. (EE) That vibratory limit may be set by endurance or heat, dunno.

Some TV thoughts:
Since resonance will generally be way below the normal op point, therefor in flight the engine/flywheel will be doing its own thing angular velocity wise and separate from the prop which will have a relatively constant velocity. So the angular differences of the two masses times the coupling stiffness will be the vibratory torque.
Then the game seems to be flywheel size and coupling softness to reduce TV. That means energy storage. Big flywheel and lotsa rubber. Or at lease enough to keep things from breaking.

I agree that the drive pins are inducing twist and bending on the arms. But I also think there may be way more than 100 foot lbs peak alternating loads.

Gots to do some modeling and finite analysis.

Ron
 
You should see what a Rotax 912 does at 700-900 rpm.. Wish I still had the video of that on my test stand...

In our high-fuel-prices-Europe, the Rotax 912 is everywhere. And indeed, the TVs are brutal on that one. Even when simply shutting off the engine (so no throttle). I cringe every time I hear one doing that.

In fact, when explaining torsional vibration to fellow pilots in layman's terms, I refer to it as "the Rotax rattle". Gets instant recognition...
 
I dunno gents...does anyone in the alt engine world really have standing to knock a 912? It has been a rather successful design.

Propeller inertia is the relatively immovable object against which the rest of the system oscillates. Given the same shaft load, low engine mass means the block oscillates more than a Subaru with a 50 lb gearbox.

The metal-to-metal rattle in the resonant range is the ramp-type dog clutch doing exactly what it was designed to do. There are several styles, and there is also an available friction clutch to further limit maximum shaft torque in some versions.

Photo here, two different dog clutch hubs. Driven side stacked on driving side, which is integral with the spur gear. Left is no-clutch, right is driven hub with external spline for overload clutch plates:

http://contrails.free.fr/images/rotax/DSCN5156.sized.jpg
 
Last edited:
I dunno gents...does anyone in the alt engine world really have standing to knock a 912? It has been a rather successful design.
]

I bring it up only to illustrate that even with all of their engineering and funds available, Rotax didn't solve the TV issue down low, they do what all of us are doing- avoid idling in the resonant range with the recommendation to be over 1400 rpm. The 912 I ran on my test stand was much worse than my Subaru ever was below this rpm.
 

Attachments

  • rotax1.jpg
    rotax1.jpg
    152.5 KB · Views: 82
Back
Top