CNC mill conversion - part 14

Working the bugs out of the Z axis

For the life of me, I could not understand what in the world was going on with my mill. I mean, this project wasn’t supposed to be easy, but it should have been achievable. Meanwhile all I’ve had to show for, in exchange for my best efforts, has been a long list of problems to solve, more or less successfully. 

The latest mocking came in the form of an alligator-style death-roll of the Z axis ball-screw, rather than the Swiss-like precision movement I was aiming for.  Looking for an easy scapegoat I decided to blame myself, due to the way I had tapped the big pulley, freehand. 

"Tapzilla" in action

I had not intended to tap it this way, but I just couldn’t fit it all under the head of the mill (to help align the operation), so I took a chance, and lost.

The result was so embarrassing that I couldn't even film it, and once again I strongly considered repurposing the mill as a boat anchor. For a moment that is, since I had way too much time, money, and pride invested in this project. 

What I decided to do in order to rectify the situation, was to make a steel adapter to go between the pulley and the ball-nut, and to single-point thread everything, thus eliminating the “Tapzilla” alignment issue altogether.

First on the list was the removal of the crooked threads from the pulley. I did this by boring the threaded section out on the lathe to an arbitrary diameter of 1.1” (2.8 cm).

Boring bar ready to "go to town" on the old threads

Threads removed

Then, I setup the lathe for internal threading using the 16 threads per inch gear train (6.3 threads per cm, or 1.59 mm per thread)  I already had in place...

Internal threading tool ready to go

... and started threading.

internal threading

I finally had mechanical certainty of parallelism between the thread axis, and the ball-screw.

Looks the same, but isn't.

Next, I identified a suitable steel piece from my scrap bucket, chucked it up...

There is really no such thing as "scrap" in a machine shop!

... and threaded it.

I'm getting fairly proficient at this by now

With the outer thread completed, I mated the adapter to the pulley, and upper flange, and drilled them together to ensure concentricity and alignment of the whole assembly. 

Match drilling the assembly

Using a boring bar, I precisely enlarged the hole for the internal threading I'd be doing next.

Precision boring in steel

Unfortunately, my mini-lathe turned out to be insufficiently rigid for internal threading on steel, and as much as I tried, I couldn’t avoid getting tool and tool-post deflected away from the piece, and unable to do any real thread cutting.

Against my original decision, I opted to bring back “Tapzilla”, but this time it would be centered in the tailstock, once again ensuring mechanical alignment.

"Oh no! Tapzilla is back!"

Freshly threaded steel

Concentricity is paramount

To ensure no more surprises, I drilled and tapped the adapter for two setscrews.

Drilling set-screw holes in the steel adapter

Assembly ready to be mounted

Finally, I put it all together again anticipating a successful test.

Bench testing

But that’s not what I got!

steel adapter thrashing video

Although much better that the first test (the one I didn't tape), this was still an awful performance, and completely inadequate.

Thus, I returned back to square one, once again! 

Things just didn’t add up, I knew I was using standard industry practices, and I had been paying very close attention to the smallest of details, yet it had been a “hit or miss” proposition all along, with a lot more initial misses than anything else.

Statistically speaking, it did not make much sense either, 3 out of 3 axes gave me trouble in the ball-screw alignment department, and I couldn’t even fault the design itself, since other people had already successfully used it before me. 

No, something else was amiss, and that something had touched all 3 axes in one way or another.

Of course, I was still number one on the list of probable causes, but I was no longer willing to accept the blame for it. This meant I had to find another “person of interest”.

Removing myself from the list of suspects, allowed me to consider the unthinkable... what if the tap was bent?

It sure didn’t look like it to me, but then again... I did buy it used on eBay! Furthermore, this would also help explain every other issue I’ve had thus far, like the poor finish on the XY plane, due to the alternating up and down movement of the ball-screw (see CNC conversion - part 12), and if all my 15/16-16 threads were crooked, that could have easily introduced binding in all 3 axes.

Brilliant! 

Regrettably for me, this new operator-absolving theory needed to be proven correct, and that meant machining a whole new adapter, and single-point internally threading it (in case you lost track, this is the 3rd time I remake this part). 

Since I had just established that my lathe couldn’t internally thread steel, I decided to use 2024-T3 aluminum.

I’ll spare all the machining details here, because they were almost identical to manufacturing the previous adapter, but here goes the new test...

Z ball-nut with 0.125 ball bearings

Much better, right? Yet, you might still have noticed the slight wobble. I think this is due to the small 0.125” ball bearings making for a very loose fit. 

Here’s what it looks like after repacking the ball-nut with 0.127” ball bearings.

Z ball-nut with 0.127 ball bearings

Not bad considering it's a 3 foot long ballscrew! 

I now felt good enough to throw the whole assembly up on the mill, and take it for a spin (pun intended). Note that the assembly was held together with just a strap, and the head of the mill was attached to the ball-screw with two clamps. This is far from a rigid setup, and was probably a bit dangerous as well, but I put together as a "proof of concept".

Initial testing of the Z axis (note clamps and strap)

The time had finally come to bolt the complete assembly to the top of the mill’s column.

Z axis inner block positioned inside the mill's column

Same thing, opposite view.

Side brackets bolted to the inner block through the column

Ball-screw mount locked down

Right view of the ball-screw mount

Another Z axis test needed to be run (with the clamps still holding the mill’s head).

Blue clamps holding the mill's head to the ball-screw bottom plate

Secondary testing of the Z axis at full speed (clamps still being used)

I am pretty happy with the results now, and I will go back and remake every part that was done with the 15/16-16 tap (aka "Tapzilla"). While this tap is probably good enough for things that do not move, like threading a bolt hole, it is definitely not adequate for a fast moving ball-nut adapter.

Also on the list of things to modify, is the aluminum plate connecting the mill's head to the ball-screw. I want this part to be adjustable in X and Y to compensate for small misalignments, but I will discuss this new design further next time.

Nose and nose gear - part 2

Retraction system & NG-30s prep (8.3 hrs)

The original nose wheel retraction system consisted of a hand-crank mechanism that was light, and worked quite well. Notwithstanding these advantages, it created the need to turn a handle multiple times right after takeoff, and just before landing, something I have never been looking forward to.

Hand-crank to raise/lower the nose gear

One alternative I had set my sights on was one of the electrical retraction system manufactured by Jack Wilhelmson of EZ-Noselift, I chose the NL1-12E.

NL-12E system components plus NG15A and NG6A (aluminum pieces on left)

You can read all about it on Jack’s website, but I’d like to highlight the fact that it has a back-up battery pack able to power the gear down in case of a complete electrical failure, and the shock is contained within the actuator. Pretty cool!

This system is marketed as a retrofit for an existing short nose Long EZ with a manual retraction system, and the instructions reflect this fact, so they are not as useful as they could have been in my situation (new build), and I will have to modify the way I approach this installation.

So, with plans for a short nose I will not build spread out on one side, retrofit instruction for something that is not even here (and never will be) on the other, all the while keeping the long nose in mind, I started to untangle this mess.

Because of the “retrofit thing”, the installation will use the same bolt holes as on the original short nose with the manual retraction system, except for the lower one (let's call it A hole!).

On the short nose drawings, I first identified the 4 holes (the lower of which I will not use).

Original bolt hole locations

Then, I devised a coordinate system allowing me to locate the holes on the plans...

Locating the holes using a reference system

... and transfer them to NG-30.

Reproducing the hole pattern on NG-30

At this point, I tried putting all the pieces together as best as I could, to see how wide the whole thing would be, but I kept coming up with bigger dimensions that I expected. This was a problem, as the nose structure width must fit the center post of F-22.

Unsure if the problem was an actual one, or just the result of hastily holding numerous pieces up in the air while taking measurements, I decided to take the time to model all the nose structure components as precisely as I could, down to the thickness of the skin, in a CAD program.

Nose structure

Peeling the onion

This took a long time, but it eventually confirmed my suspicions... the nose structure came up 0.308” (8 mm) wider than the part it attaches to.

This looks like a problem

"It's always something, isn't it?"

"If it doesn't fit, you must acquit!"

While a little unsettling at first, I discussed my finding with other builders/fliers, one of which a mechanical engineer, and the general consensus is that making a flox corner between F-22 and NG-30 will restore, and actually exceed the original design strength of the joint.

Hand-drawn sketch of the proposed fix on my iPad

Although I was unable to draw it appropriately in the sketch above, foam will be removed from the core of the two parts, and the inner micro layer sanded off.

If I were to make F-22 again, I’d probably leave it 1/4” (13 mm) oversize at both ends of the center post, and plan on trimming it after NG30 installation.

In light of the recent fitting issues I uncovered on my nose, I decided to also model the actuator jack in CAD as faithfully as possible to ensure everything would fit all-right, and that there would be no more surprises.

Virtual actuator jack

Here’s the way the nose retraction system fits virtually...

Virtual fitting session

... and in reality.

Actual fitting session

Good thing I checked, because the virtual fitting session highlighted one more area of interference where the head of the actuator bumps into the top of NG-30.

"What now?!"

Interference volume highlighted in red

This was never an issue with the short nose NG-30s, since they were not as tall, but it is now, and my solution to this problem is to carve a small triangular niche in the foam 0.150” (3.8 mm) deep.

Modified NG-30 foam depth

Interference removal process

One other issue I identified through CAD, was that of the proper edge distance of the actuator bracket's holes when using my coordinate system.

Obviously something is slightly askew somewhere, perhaps my hole measurements wasn’t precise enough, or my hole pattern reproduction might have been somewhat inaccurate, or my modeling of the brackets... the point is, I made a mental note to not go by those C and D hole locations.

A week later, eager to get back to work, I forgot all about my previous assessment, and drilled those two holes per plans. 

"Measure once, drill twice. Doh!"

It’s really a smaller issue that it might seem at first, but it still pissed me off, and of course I figured it out just as I drilled the last hole.

Since the brackets already had the bottom hole predrilled with the correct edge clearance, I used them as a template to locate the correct position for the C and D holes on the brackets themselves. I will eventually use the bracket holes as templates to drill the correct holes in the NG-30s.

Ensuring proper edge clearance for the new holes

Using the drill press, I finalized their position once and for all by match drilling them.

Match drilling the bracket holes

The former misalignment is very apparent now.

Finding the correct spot

Moving on, I used the router to remove foam in the places where the hard-points will be created...

Free handing it

Nose gear hinge hard point (to be)

Done with the router

... and my “older than dirt” Dremel tool to make craters out of them.

Dremel action shot

Moonscape

Next time I'll fill the craters with "fiberglass disks", and cover the outside of both NG-30s.