Testing the new collet

With the new collet installed I then moved onto testing the engine. Since I was on my own you'll see that I rigged up some eyes in the walls of the garage and secured the hovercraft from moving significantly laterally.

Here's my first flight.


Everything went relatively well but there's some ominous rattling near the end. It's easier to notice now that I listen to the recordings. But what do you do after a successful test? Try again...


Yeah, not so good. It was clearly not doing things properly since it didn't even start hovering. And by then it had been loosened so of course it fell off again. But it's nicely contained in the housing.

So I took it all apart again. Measured that it hadn't worn more than a few thousands off the bushing, and put it back together again. This time I put more than the recommended 95 inch pounds, but not excessively. But then I ran a very short test:


It hovered well, but I took it apart anyway and there was an extra quarter turn I was able to get out of the tightening bolts. So I tested again.



This time I throttled up a bit and everything seemed ok. But I took it all apart and was able to tighten each bolt a quarter turn again. So one more check.


Things sounded better, and the bolts did not have freedom when I took it apart. So I decided to try again.



Everything sounded good and it was hovering very high. I've had some balance issues so I tried adding some weight at the front to feel the balance. It was riding super high and super stable when I throttled up. However as I throttled down to feel it settle, the propeller started hitting the underside of the engine as it had slide up the shaft during the testing. That's what made the horrible noise at the end.

So I'm a little frustrated now...

I think that there's some work to be done. Leveraging the keyway seems like a good idea. And also using the end of the shaft, and the end bolt would make sense. This is the mounting technique for most lawn mower blades.

New collet installed

So after waiting several weeks for Univeral Hovercraft to send me a replacement collet I was excited to get it mounted up again and get hovering again.

Some new parts

When dissassembling I found a lot of metal filings.

The removed propeller with worn collet and filing visible in the old housing.. Also note the shorn off bolt. 

And based on those metal filings I thought I'd check the size of the collet new vs old.

New collet lower edge measurement

Old collet lower edge measurement

There's an exciting ~15 thousands of an inch extra clearance which

New collet top side measurement

Old collet top side measurement

And the wear was almost as much on the other side of the collet too.

For reference I also measured the shaft. And it's pretty close to the expected size still.

Shaft size near the end

Shaft size where there appear to be the most wear

So I moved onto reassembling the lift fan. Taking note of the bushing instructions.

Collet bushing instructions

I did notice some differences in dimentions of the bushing mounting blocks, but the critical dimension where the bushing rides appeared to be the same.

New and old bushing housings side by side

New and old bushing housings side by side.

I put everything back together. And I installed the collet to the shaft before installing the propeller so I could see and access it better. Everything looks good.

Reassembled collet end view

Reassembled collet side view

The bolts were all secured with lock-tite.

A close up trying to show the lock tite

I added the propeller on last before reinstalling.

And fully assembled

First parking lot test

On this sunny Easter day I got a chance to head out and test the hovercraft in a parking lot.

Thanks to all the help from Jenny, Wim, Amy, Adrian, Ayse, and Lars!

Everything prepared to go in the morning

Adrian came over early and we loaded the hovercraft into the trailer. The upside down hand cart worked well.  The two of us were able to load it without a significant challenge rolling it up the ramp.

A freshly loaded hovercraft

After loading up hovercraft we headed out for the Redwood City Port Boat Launch

The trailer about to leave the driveway carrying the hovercraft. 

Arriving at the Redwood City Port Boat Ramp

The launch permit for the day. 

After a break for brunch we returned and rolled the hovercraft out of the trailer. And tried it out.

We started testing travelling using the docking lines. And it hovered for over a minute. And we crossed the puddle uneventfully! Here's the view from the trailer.

Unfortunately if you listen closely at the end the lift propeller fell off again.

So we found the right wrenches again and disassembled the lift engine again. I had tightened the bolts pretty hard but I guess it could have had more.

Unmounting the engine in the field

The propeller collet after falling off the engine shaft.

There was only a little scuffing on the bottom of the blade this time. It didn't fall all the way off the shaft as it hit the deflector baffle for the skirt.

We got started reassembling the engine and decided to use a larger wrench to make sure the bolts were tighter. However the 15" wrench handle prooved too strong for the collette bolt. See below...

The broken bolt

Unfortunately this brought the test to an end since the other half of the bolt remains in the collet and there's no way to secure the collett securely anymore.

I think I need to find a new mechanism for securing the propeller. 

Shrink Wrapping the Rudders

I've successfully covered the rudders with shrink wrap. They look pretty good.

Two shrink wrapped rudders.

I've still got quite a bit to learn about how to effectively use the shrink wrap gun. But I only had to redo one of the rudders. 

The first one I took longer to do. 

You'll note that I spent more time at a distance and definitely burned some holes. I also tried to adhere the edge using heat bonding. However it was problematic as I was pushing against the open part of the frame, thus couldn't get much pressure between the pieces I was squeezing. 

Results of first attempt to shrink wrap a rudder

A close up of the burn hole, and some of the edge attachment

Where the bolt ends pierced the shrink wrap

At the other end, 2/3 bolts burned through. 

I tried again on the 2nd rudder. This time I clipped some holes for the bolts first and ran tape over the ends. I also used the shrink wrap tape down the trailing edge and over the ends with the hopes of getting a smoother surface once shrunk. 

Taped edges visible with bolts cut through early.

 Here's how it went:

The results were notably better. With a few caveats. Clearly the tape worked better, but there's still some curling up at the edges of the tape.

Overheating at the ends, but bolts are much better. 

 And I clearly discovered that applying the tape down with full pressure is important. I missed a few inches that were not pressed down well, and consequently the stretching pulled the adhesive apart before it could fuse. (I probably also heated the front edge first too.

A tear where the tape was not adhered well enough.

But I was able to patch it over relatively easily with another strip of tape, and a quick heating by the heat gun.

I patched the tear with another layer of tape and tensioned it again to look like this afterwards.

 I cut the shrink wrap off of the first rudder and tried again. The biggest difference I note is that I'm more confident. I move more quickly at closer distances. And the other thing is that the wrap shrinks when cooling, not when hot. When it's hot it's actually softer. So when applying heat, you don't apply heat until it's taught. You apply heat quickly and then move on. After it cools come back and check on it, and if it's not tight enough re apply heat. Continuing to apply heat to an area basicaly melts it and or burns it.

Here's my 3rd attempt.

The results worked out well. I used the above quick heating and trust that it will shrink later. As well as I trimmed more of the excess film away. And I taped at the trailing edge, and made sure to apply the tape firmly and press the edges down hard against the frame underneath instead of taping where there was an air gap.

3rd attempt size A

3rd attempt side B

Rudder improvements and improving ground handling

Much of today's effort was spent wandering the isles of Home Depot, but Michael and I made some good progress. Unfortunately we forgot the gopro so no timelapse for today.

Our first challenge was to decrease the weight of the rudders. As such we started by removing most of the wood from the poplar boards. This cut the weight down significantly. My searches for local suppliers of shrink wrap came up empty so I'm going to have to order it online. So the rudders will need to be covered later.

Here are the rudders propped in place awaiting the shrink wrap

The weight of the rudders is probably 1/5th to 1/10th of the previous weight and I think they're still quite a bit stronger than needed.


From there we moved onto working on eliminating the rudder backlash. The pin at the rear is a 1/4" bolt, but the socket for mounting the steering cable is closer to 5/16". This provided quite a bit of slop. I found some nice spacer flanges that could expand the bolt from 1/4" to 3/8" but that was way too big. And drilling out the socket on the cable looked like it would significantly decrease it's strength. So we kept looking.

We found some plastic tubing with 1/4" ID and 3/8" OD in the plumbing isle. And after some experimentation with various ways to slim down the plastic tubing. We put a longer segment (1.25" ) of the tubing over a 1/4" bolt, then chucked it in the drill. And then ran it over a piece of sandpaper. At first the sandpaper was handheld, but for more effective pressure we put the sandpaper on a fixed block of wood. (The 1/4" bolt allowed us to apply a lot more lateral pressure as well to speed up the process)

The slimmed plastic tubing.

 After sanding it down we cut off the last 3/8" which we needed for the fitting. And the backlash in the steering was cut in half. Now I think we're approaching the limits of the steering cable itself.


The other problem that we tackled was starting to think about how to load the hovercraft into and out of the trailer. We spend a lot of time looking at wheels and how to mount them to the vehicle etc. However, after quite a few trips through the isles we realized that we were thinking too hard about this and that maybe there was a dolly we could just use. And after that I noticed the hand trucks near the back, and we realized it would be much easier to use them with the axles and wheels already assembled instead of engineering it all ourselves.

We picked up the sturdiest looking of the simple hand trucks, as well as some straps to secure it to the hovercraft and headed home.

The first challenge was to get it into my car.

The hand truck barely fitting in the back of the car

To get it in required a little bit of squeeze and all the seats and cargo netting/protection to be removed. 

Once home the first order of business was to remove the handle. The grinder with a cutoff blade made short work of it. Something like 20 seconds of grinding for each end. 

The handle removed. 

The handle removed view 2

Of course we spent a little bit longer grinding down the exposed results so there's no sharp edges.

Michael having fun with the grinder.

 With the handle removed we could simply slide the hand truck under the hovercraft, "the wrong way up".

The hand cart inserted under the hovercraft.

And we used some straps to hold the "top" of the hand truck centered and off the ground.

The strap holding the hand truck to the underside. 

In our testing we were able to move it around relatively easily with just the two of us lifting the front. In a pinch I think I could do it myself on flat ground. But at least two people will be necessary for getting it up the ramp.

Though I think that the cleats will be of great use for us when we get to that. With the fiberglass around the perimeter the cleats are much stronger.

There are a few improvements I'd like to make to the hand cart. The straps should probably pull forward a little. And I'd like to make them a fixed length and go to special screw eyes so that it remains centered and doesn't take much thought to attach.

I'd like to add a little bit of fixturing at the rear to both keep it centered and from sliding backwards. With the handle removed the truck slides under the craft nicely. However it also slides out from underneath well too. And I think that a little bit of a block near the base could also keep the "top" of the truck from pointing down as it rides on the skirt attachment more than the hull surface.

Fiberglass reinforcements

Michael and I got an early start and tested the rewired kill switch for the thrust engine. And we noticed that I'd not quite lined up the lift fan's duct with the baffle underneath. So we realligned that and verified that it filled the skirt better.

A little later Adrian and I first looked into methods for reducing the rudder's weight, mostly by hawging out the rudder wood. The 1" poplar is way stronger than we need. I'm hoping to keep the frame but cut out all the interior space and stretch a fabric/plastic across it to block the wind but be much lighter than the whole volume of wood.

But we decided to first take on reinforcing the perimeter of the body with fiberglass. We started by sanding down the surfaces and rounding things off slightly near the cleats.

Below you can see the timelapse of us adding the fiberglass.

With the reinforcements around the perimeter I expect the craft to be much stronger. Previously it was relying on the foam to transfer loads between the sheets of plywood. Now the foam can just be a spacer. The full box beam has not been completed as the bottom corners are not completed, but it's way stronger.

The bottom corners are going to be much harder to do as they're both inside the skirt and include the bottom edge. The epoxy will hold somewhat on the bottom, but it won't be as clean as if I flip the whole vehicle over to fix it. I'll probably defer it to later when I am ready to redo the skirt with the final material.

With Adrian applying the epoxy and me flushing it down with my fingers we went through a lot of gloves. It wouldn't have been too many, but we needed to break out the scissors for the handles, and cleats, as well as the corners. And I needed to break out and cut more strips at some point.

At the rear we removed the rudder hardware and applied the fiberglass to the surface. Once the epoxy has hardened we can remount the hardware. (It's the oarlock mounts.

The hovercraft with the epoxy setting around the perimeter.

And as a bonus. Adrian brought his new FLIR camera which plugs into his phone and took these pictures of the heater running. It's way hotter than 216 degrees but you can see that it's overloading the sensor. The blackspots on the top of the heater register as 4 degrees F according to the camera.

An IR view of the heater running. 

The temperature on the ceiling above the heater, even with the fan running. 

Duct, rudder, and kill switch tuning

Today I spend the day filling in various small projects.

I've been hoping to reinforce the corners with fiberglass. On Saturday I attemped to warm up the garage above the minimum 55 degrees necessary for the epoxy to cure. It was just below 55 degrees so I opened the garage door hoping the afternoon sun would warm things up. At which point it clouded over and that plan didn't go anywhere.

As such I did some research and looked around for some other techniques for heating the garage. There are quite a few options that Home Depot advertises however most of them are not available anywhere nearby. And even better is when you get to Home Depot, no one knows where they would be. They don't seem to carry the more classic horizontal tube like heaters anymore. And I found one high capacity one finally, in the "light cloud". The "light cloud" is the section of the store where they have all the sample lights??

After way too long at the store I got it home and fired it up. It's rated for up to 1350 square feet, so it warmed up the garage quite quickly.

The heater warming up the garage with the fan next to it to circulate the heat better. 

I took apart the thrust engine kill switch, but realized I needed some more connectors to connect the kill switch properly, so paused that.

The lift engine cowling had been rotating while I was testing so I pinned it from each side using just some wood screws. I would like to beef it up with 1/4-20 bolts in the future, but didn't want to disassemble the lift engine just to secure it. I'll try to remember to do that next time it's open.

The rudders are very heavy so I'm planning to take the wood and remove most of the interior material and then just cover the outside with cloth. I haven't decided if it should be plastic, skirt material or fiberglass. I'll leave the perimeter from the existing board to be the main structure, as well as a lattice of the original boards to take the main load.

One other bit of tuning for the rudders is to find a spacer washer to give it less backlash in the controls. They are passing over a 1/4-20 bolt as a pin, but the hole is at least 1/4" clearance giving I estimate a 16th of an inch of backlash, which is a lot for a 2" throw overall.

Here's my first timelapse.

I had to take a break and get some more supplies.

Coming back several hours later, everything was cold again. In the 15-30 minutes I ran the heater I got the ceiling up to 70 degrees, while the hovercraft body got to close to 60 degrees. The floor remained approximately 53 degrees. I was hoping that the heat would remain longer and I would not need to keep heating while the epoxy cures.

While at the store I also found some end caps for the throttle leavers to make them a little bit safer.

End caps for the throttles

And I finished up by getting the thrust kill switch wired up. Though I didn't test it due to it being late at night.

Here's the timelapse of part two.

Improved Lift Flow Splitter

Today Adrian and I worked on increasing the effectiveness of the lift fan splitter.

From reading through the design manuals they talk about having the splitter as close to the fan as possible for greatest effectiveness. Some of the lift fan's output is directed into the lift bags, while the rest simply goes into the cavity under the vehicle. The bag has lower flow but higher pressure to maintain the perimeter and keep the air cushion underneath. My splitter was mounted too far down, almost 12 inches below the fan. And with that distance, the bag pressure was allmost equivalent to the rest of the air cushion giving very little force to keep the air underneath.


To make working on this easier we started out by rigging a hoist mechanism to make it easier to lift the hovercraft when in the garage. A few pulleys in the rafters and it worked pretty well. I'd really like to get some better hard points, but until I reinforce the perimeter I"m afraid to use the cleats or the handles too hard. So we attached to the superstructure.

A few notes on the hoisting. I tested the pull using a luggage scale and the front was ~ 55lbs force to hold the front in the air. Two more pulleys at the bottom would be a great help, there was a lot of friction in the system. Since I can relatively easily squat lift the front end. With the pulley system rigged it should be ~4x the force.

To fix the splitter we first installed a wood sheet baffle to traverse the thickness of the hull and keep the bag flow separated from the main lift flow. Then we attached a piece of sheet metal across an equivalent section of the cowling between the lower grill and the fan.  There's over 5" inside the cowling behind the back of the fan.

While we had the cowling out we cleared out half the wires from the lower protective grill. Having verified that there's more than 5" of clearance between the grill and the fan, and it's the output side of the fan, on the undersaide of the vehicle, recessed by 6" from the bottom of the hull, it felt like overkill to have < 1/2" gaps in the grill.  Holding my hand over the grill when testing it did feel like we'd made a noticiable improvement in the airflow.

Also, when we were cleaning up the cut out parts we noticed it was a non-trivial amount of weight too.



Here's a timelapse of the first half of our work. Hoisting and adding the lower baffle.


We took a break to buy the needed sheetmetal. And then returned to work on the cowling.


Here's the video of our final test at the end of the day.




The hovercraft did seem more stable side to side with the improved splitter. And I've learned where to position myself to be in the center, but I think it did better at not dragging the stern when at low speed.

You can't see it in the video but the cowling did rotation slightly when we were testing. Below you can see the picture. Through the whole fan you can see the wooden baffle parallel to the main structure, and the sheet metal splitter just below the fan rotated by ~ 15 degress. The whole cowling actually rotated by that amount. We just need to pin it on the side. It will add one more step for getting it in and out but it will help to keep the air pressure up.

The splitter inside the cowling showing rotation after testing. 

Steering linkage tuning and lift throttle assembly

I've made some more progress this weekend with the help of Adrian.

We had two main tasks for the day. 

First off we had to switch the direction of the steering linkage, and while we were at it we needed to recenter it as well. 

And we replicated Jeff's highly professional linkage on the other side. I'll get some closeups later when there's more light out. I forgot to get them at the time. 

Here's the full timelapse:

For good measure we also tested the linkages.

We started with testing the rudder linkage. The ambient temperature was cooler than I've run the engines before and I was surprised by how easily I flooded the engine before it warmed up. 

You can see that it is approximately centered now. There's a little bit of a question whether it's a wide enough range of motion. If we need more range of motion I can easily add a new pin to the bottom of the rudder with a shorter lever arm. 

The lift engine also flooded when throttled up, but I was able to react and it kept going due to momentum enough for me to lower the throttle. 

Redone handles, new cleats, kill switches, and throttle control.

Thanks to all the help today we knocked a lot of items off the todo list.

It was a great afternoon with Jeff, Derek, Michael, Angel, and Niharika.

I started out the day working on getting the kill switches mounted. I had two proper power boat deadman switches that clip to you to shutdown the engines in case of you falling overboard. However they needed a good mounting housing.

Jeff started out working on attaching the cleats at the corners. We used the 1/4" poplar boards as a backing and the toggle bolts for securing into the plywood. The toggle bolts would go all the way up through the cleat and the board would also be stapled down to distribute the load to the plywood.




As I believe I've mentioned the lift handles were wiggling loose. So Jeff and Derek unscrewed them and pried them from their epoxy. We screwed the handles back in with lock-tite this time. And scuffed them up to try to get the epoxy to secure better.

Meanwhile Michael figured out how to wire in the kill switch for the thrust engine and ran the cable forward to the box I mounted earlier.

After preparing both the new handles and cleats we broke out the epoxy. There were a few complications. Unfortunately the stackup of the spacer board + cleat was too think such that the bolt couldn't reach the toggle bolt. We had to drill out the cleats to get the head of the screws slightly lower.

To test whether the cleat was secured well I pulled upward. Unfortunately this revealed that the port bow plywood was not secure down well at all. And the cleat had been secured down well, but the plywood gave, bending/partially breaking. So we had to mix up an additional batch of epoxy, and both secure the cleat, as well as glue the plywood down. This reinforces the urgency of running reinforcing fiberglass around the perimeter of the body to transfer the loads between the plywood on the surfaces and sides.

Unfortunately that's about when the batteries ran out in the GoPro. After securing everything we added my standard weights of barrels of water. In the beginning of the next clip you can see the buckets have appeared suddenly. The 3 parts of the video are from the 3 batteries used for filming. I'd love to have longer battery life. But at least in the Hero 4 the batteries can be changed quickly, and I have several to swap out.




In the final session we moved onto focusing on the throttle mechanism. Jeff and Michael discovered that they could loosten a nut and make the throttle move much more freely. And Jeff created a throttle mechanism out of an angle brace, 18" of aluminum tube. And most cleverly discovered that 1/4" post electric crimpable connectors can work well to crimp onto throttle cable housing. Experimenting with a few different ratios we successfully made a handle which can easily be moved, gives full range of throttle.

The one problem is that it is mounted at the aft end of the cockpit. We knew this would be a problem before we mounted it, but we wanted to test the mechanism anyway. I just need to find a longer throttle cable with housing. The current one is only 4-5 feet.



The end of the day concluded with testing the throttle and kill switch for the thrust engine. We held off on the lift engine due to all the epoxy around the perimeter. And the buckets balanced on the corner...

We started with testing the kill switches and when we tried the first kill switch it didn't do anything. So I turned it off at the engine and that killed it. After a little puzzling we remembered that the engines required an active connection to kill it instead of interrupting the signal. So it needs both the engine switch and the kill switch in the off position to successfully disable it since they are wired in series. We'll change it in the future to be in parallel such that if either are in off, it will be disabled, and it will require both to be in the on position to run.

The throttle mechanism testing was great. It worked just liked we planned. Which was a surprise due to it being made of ~ $5 in parts, an angle bracket, a bolt, several washers, and a hammered aluminum tube.