Saturday, July 15, 2017

Lathes + Flywheel Troubles

Having a metal lathe, even a small one, is a fantastic way to avoid paying a large amount of money for a small amount of work. I started by chucking up the crankshaft and adjusting until I had less than a thou of runout. That's still too much for bearing work, but I'm not doing bearing work. I put a 45º chamfer on the flywheel mating face, and knocked the burrs off of the pulley hub mating portion of the shaft. A little polishing to make sure the pulley hub slips on, and we're off to the races. The pulley hub was heated to ~400º F, the crank cooled to ~10º F. I still had to use some persuasion with a hammer to seat it fully, but she's on there now.


The two aluminum rings are the replacement seal housing, complete with seal installed. I must've walked through the assembly process a dozen times to make sure I put the seal housing on the correct way, and I still checked it half a dozen times afterwards. The red material is a cut up silicon baking sheet to help keep the rubber on the seal from melting/losing its elasticity.


After I turned the chamfer on the flywheel end, I re-installed the 4 dowel pins. The bottom pin in the picture is larger than the rest, allowing me to index the flywheel. My plan is to mark TDC on the flywheel, since it can only line up one way, and with very tight tolerances.


When I attach the flywheel and snug up the nut, I still end up with about .008" of runout at the edge of the flywheel. Not a lot, but more than I'm willing to put this together with. I used some more PlastiGage between the flywheel and the crank surface to make sure I was fully seating the flywheel... sure enough, I wasn't, and the place where I wasn't touching the crankshaft is the same angle where the flywheel is furthest out. There are four factors at play here; the crankshaft face, the dowel pins, the flywheel face, and the flywheel holes for the dowels. The crankshaft face is most likely true, since that's how it was run successfully before, and my PlastiGage test seems to confirm it. I proved that the flywheel face was true by using a rotary table and a dial indicator and getting less than a thousandth of runout. I think that leaves the dowels. How did that happen?

It happened pretty simply - I used the flywheel to drill the dowel holes. Normally, this would've been the best way to go about it. However, because of that radius on the flywheel that I chamfered the crank to match, the flywheel wasn't sitting quite right on the crankshaft, which means the dowels are *slightly* cocked to one side, which means they *want* to pull the flywheel that way.

So my next step is to torque the flywheel to spec, which is some ridiculous number above 200 ft/lbs, and see if that fixes the problem. If not, I may have to get creative. On a positive note, the only bit of clearancing that I still need to do involves grinding a little material off one of the connecting rods. Once that's complete, I can start setting things out for assembly. I do, however, need to get this flywheel thing straightened out before I can move forward.

Monday, July 3, 2017

Dowels are hard, PlastiGauge is soft.

It took a lot of work, but I got all four dowel pins out without damaging the crank. The two I put in came out easily enough with some vise grips and a bit of twisting and pulling... the two that were in from before were another story entirely. Nothing would touch them, including the hardened jaws of a *nice* set of vise grips. (Those are no longer a nice set of vise grips, by the way.) In the end, I discovered why machinists have drill bit sets up to half an inch with just a couple thou of difference all the way up...

...It's because hardened steel pins can't be drilled out with big steps between bit sizes! It took me a solid hour and a half to drill each pin out. However, both of them twisted out right before I broke through the wall of the pin (after a couple good taps with a hammer on the side of the pin to break it up a bit.)

I checked the face on a dial indicator, and at most it's out a single thousandth at the edge. That's about an inch from center, which means that might account for about 5 thousandths at the edge of the flywheel.... which means something else is amiss. I checked the flywheel by placing it on an indexer, supported by placing a tight tolerance block between the indexer and the mating face on the flywheel. Spin the indexer, check for runout.... I forget what I got, but it was insignificant (which is why I forgot it. :D) I was pretty puzzled at that point. However, since the pins were out of the crank, I was able to rotate the flywheel on the crank. How does that help us? Well, before I pulled the dowels, I'd marked on the flywheel exactly *where* the runout was, as well as *how much.* When I put the flywheel back on after the pins were out, I lined it up pretty close to where it was originally... and got pretty much the same results. (I didn't use an indicator, as I was less interested in how much it was out as *where* it was out.... and it was wobbling pretty obviously.) Ok, repeatability, that's good. So I rotated the flywheel 180 degrees and tried again. If it was the flywheel, that should've made no difference; the marks on the flywheel still should have been correct. If it was the crank face that was out, then the marks on the flywheel should have been exactly (or close to) opposite what the flywheel was showing the second time around.

So what did I find? Well.... actually, it was about 70 degrees off from where it was originally. I didn't know what to make of that, until I started taking careful measurements of everything. What I found was that the flywheel had a bit of a radius where it mated with the crankshaft, and the crankshaft did not have that radius. I think what's happening is that there's a tiny bit of play between the *side walls* of the flywheel and the crankshaft, and the radius causes the face of the flywheel to be tighter than the crankshaft can fit into. Add the radius and a little bit of slop and, unless the radius on the flywheel and the edge on the crankshaft are perfectly uniform and smooth, odds are the flywheel just isn't seating fully and is staying at a bit of an angle. The fix is simple; just cut a matching radius (or, much easier, a slightly relieved chamfer) on the crankshaft. The side walls and dowel pins will still locate the flywheel, and the face where the two meet won't lose much area. I like that idea better than cutting a sharp corner into the flywheel. 

I should mention, the reason that this is an easy fix is because I recently picked up a Jet 9x20 metal lathe. Just as soon as I finish getting that set up in the basement, this will be a pretty straightforward job. I'll be spending my evenings this week cleaning it up and, as soon as my insert tooling shows up next week, I'll be making chips and breaking tools. Also touching up airplane engine parts. 

I also got my oil galley plugs straked in place so they can't rotate out. I found one of my case savers was loose, so I used some red LocTite, filed a notch in it, and straked that in place as well. 

The last thing I wanted to do today was check my connecting rod bearing clearance. If you've never heard of PlastiGauge, you're in good company; none of the auto parts stores I went to had ever heard of it either. I ended up at a local NAPA, a company I spent 6 years working as a counterman for, and gave the kid at the counter the part number I remembered as being correct. "Huh, it says we've got 12 in stock. What is it again?" It comes from the warehouse packaged as a dozen, so I guess it shouldn't surprise me he'd never heard of it; he'd never sold any of it. Anyways, PlastiGauge is a kind of plastic-y wax string that's very precisely formed. A strip of it is placed between two surfaces you want to check clearance on, and those two surfaces are brought together however they will be brought together during final assembly. In this case, those surfaces are the connecting rods with bearings installed, and the crankshaft. Blurry picture, but the green little stripe in the middle of the white grease is the PlastiGauge before the con rod is tightened....



...And after:


The lighting was pretty awful. However, I can tell you that the strip of green that's *not* a weird reflection is the one above the .0015" mark. Basically, that wax gets squished to different widths based on how much clearance is available, and the package the wax comes in has a gauge on the side to check it with. All four journals had *exactly* the same width as the .0015" mark, and they were all very even stripes from one side of the journal to the other. (If the journals or the con rods had any taper, that would've shown up as a tapered stripe.) The specs for this engine are 0.0008-0.0025, so I'm smack in the middle of where I want to be. Any wear is going to cause that clearance to open up, and I've got a solid thousandth of an inch for it to open up before I make it out of the range I want to be in.

I'll putz around with little things here and there, but mainly I'll be setting up that lathe and taking some test cuts to get to know it. I've got to do my touch up on the crankshaft, and a couple other things, and then I can start test assembling everything. 

Sunday, June 18, 2017

MORE Case and Crank prep...

It's been slow going lately, but progress seems to be taking place. I got lost in a lot of thought experiments and what-ifs, and (as is usually the case,) ended up right back where I started. So I cut the top off of the flywheel flange on the case, did a little bit of clearing on the inside of the case (some spots didn't seem to have more than a few thou of room for rotating parts), and used a Dremel to carve away a couple of channels for the engine mount in the back.



I also pulled all the aluminum plugs to get an idea of how much stuff was trapped in the case...



...And there was a lot! Much of that was from walnut blasting the case, but I found a good number of metal shavings as well. I'm glad I pulled the plugs... Next I tapped the holes for various pipe thread plugs, ranging from 1/16" NPT up to 3/8" NPT. I used a 1/2" NPT to tap a larger hole in the front for an oil temp sensor, because I've read bad things about the current temp sensor location (near the flywheel end of the case) actually reading quite a bit lower as a result of not being in a high-flow area. The new place is directly in the oil flow as it enters the oil pump, so it'd be hard to get a better reading than that!

Drilling and tapping the magnesium was quite easy, actually... in that the magnesium tapped without much force. There were a few places that I had to tap a few threads, then grind the tap down, then tap a few more, then grind, back and forth a few more times until the plug fit the way I wanted it to. There's one hole that I may have tapped just a little too hard and began to mess up the threads, but the plug threads in just fine and seems to have full contact along the entire length and circumference of the plug, so I'm going to leave it.

Now that all of the holes are prepped, I'll pull all the plugs out for another cleaning session. The last step will be to file a notch in each of the plugs so I can swage the case metal into those notches as a final insurance against the plugs working their way back out.





I forgot to take pictures, but I drilled the crankshaft for a couple more dowel pins. When I received it, it only had two, which was probably fine running the prop off the other end, but I'm not comfortable only using two running it with the prop hanging off the flywheel. So I drilled two more holes, reamed them, and tried to put the dowels in.... oops! One of the holes is *way* too loose! So I ordered a 11/32" drill bit blank, cut it down to the right length to be a dowel pin, and re-reamed the hole to be a press fit. This really isn't a bad plan anyway, as it keeps my flywheel aligned the same way every time it's reassembled, and the only extra work I'd have to do on a new flywheel is drill one hole slightly larger.

Upon reassembly of the flywheel and crank, I realized the flywheel is running about 60 thousandths out of true at the outermost edge. Not cool at all. So my next step is to remove the dowels and see if the crank face is out, but my bet is that somewhere along the way some burrs crept up and are keeping the crank from seating correctly. At least, I hope that's what happened... otherwise it's back to the machine shop.

Monday, April 10, 2017

Starting to prep for assembly.... sort of.

With my crankshaft back and polished, I was able to begin reassembling the gears onto the nose.  First the key:




Heat up the cam drive gear, and slip it on:




Next the spacer and the new brass distributor gear.





I'm working out the best way to attach the prop hub - I know what orientation, but if I want it repeatable to exactly the same spot, I'll need to build some kind of jig that locates off of the cam drive gear. I'm not sure how much that will matter at this point.

In the meantime, I finished walnut blasting the case, and decided to try it on the heads as well. They were somewhat improved, but I'll still try soaking them in carb cleaner to try and get rid of more of the carbon that's built up. Ignore the ape blocking part of the picture...




And below, a chamber in progress to show my future self that it was, in fact, having an effect:




Once everything was sufficiently blasted, I used Simple Green to clean the case as best I could. I'll admit, I focused a little more on the outside than the inside, but I think it's pretty clean all the way through.




Once clean and dry, I cleaned the outside with solvent and got ready to paint. Bob Hoover says three things about painting the case: Do it, do it black, and do it thin. When I got the case, it was painted a dark gray... over top of a dark blue. Two layers of paint, the outer one fairly thick, were definitely not going to help much with thermal transfer, and yet it still flew that way. In light of that fact, I took Bob Hoover's advice on two of the points: I painted it, and I painted it thin.... and I painted it red.




I figure the color is less important than the paint thickness, so I cut the paint with Naptha and brushed it on, making sure to run the brush over any areas that looked like they had extra paint. The end result is definitely brush-textured, and not exactly what I had envisioned, but it looks pretty nifty all the same... and more importantly, it's protected from corrosion and it's not likely to overheat from the paint. I'll put this in the oven like I did the cylinders to get the paint nice and hard... it makes a big difference!

Friday, March 31, 2017

Good news! (And also, electronics are finicky!)

Just a quick pseudo-celebratory post - all of the work I've been doing on this engine has been under the shadow of uncertainty. The engine was involved in an off-airport landing, and I've been wondering if my crankshaft had sustained any damage. If it had, this engine would probably be scrap, based on how it had been modified - the major modified parts just don't seem to be readily available, and other manufacturer's modified parts wouldn't fit, and that would mean I'd need a new crankshaft, prop hub, and engine case, and they'd all have to be machined... Got it back from the machine shop today with a clean bill of health. Now I just have to finish cleaning and painting a few things, and the reassembly can begin.

In other news, apparently circuit design is a little tougher than I thought... received the PCB I had printed for my landing lights, (they look FANTASTIC, by the way - really like the purple solder mask with the gold solder pads) soldered everything in place, and... nothing. Back to the drawing board...


Tuesday, March 28, 2017

Rust removal sucks, Cleaning sucks, Painting sucks...

....but they're all pretty necessary if this engine is going to last. So I've been doing due diligence to remove corrosion, clean up, prep, paint, and heat treat all my engine components according to best practices as instructed by Bob Hoover and Tom Wilson. There's a fantastic resource online at http://www.eaa691.net/images/pdf/VW%20aircraft%20engine%20building.pdf , which is a hundred some pages on what to do, why to do it, and what happens when you don't. I've been just collecting parts and trying to get everything ready for the first trial assembly. Tomorrow I'm taking my crankshaft in to have it magnafluxed, and if it comes back ok I'll start putting stuff back on it. First the woodruff key and the gears, then the prop hub. The next step will be installing the bearings in the crank case, and dropping crank and camshaft, along with some associated hardware, inside and making sure it all turns the way I want it to. 

I got new cylinders and pistons, measured them, weighed them, cleaned them *very* well, and painted them black to prevent corrosion and encourage heat transfer. 

The cylinder on the left has wet paint on it, the cylinder on the right has been cleaned but not painted yet. I made sure to keep track of the cylinders, and marked them after they were painted so I could keep track of them.


As it turns out, the cylinders I designated as "A" and "D" were both 4.490" between the two sealing surfaces, whereas cylinders "B" and "C" were slightly shorter at 4.487". That's right on the edge of too much difference, but because I've got less than a half a thousandth of difference between matching cylinders, I'm not going to send them out to have them machined. Instead, I'll put both of the "long" cylinders on one side, so the heads will be able to seal correctly, and either make up the difference between the two sides with a .003 shim, or just not worry about it.... for all I know, once it's assembled, other tolerances may stack up and either make it a larger difference or even it out. That's why I'm calling this a "trial" assembly, because there may be adjustments to make after it's all together. 

I'm working on removing some surface rust on my prop hub with vinegar. Most of it came off after soaking for about two hours, but I'll give it another shot tomorrow after work. I still have a fair amount of cleaning to do on the engine case, which is about the last thing I want to do... but it's going to be the thing that's holding me up in just a few days, I think. By next week, I hope to be posting pictures of a freshly painted, partially assembled engine.

Wednesday, March 15, 2017

Lights! Cameras! Snow drifts! (aka, Lighting Part 1)

LED lighting. Pretty expensive for normal people, not so shocking to airplane people. But you can pay upwards of a grand for the basic strobes and a landing light. Once again, something in me says, "I can do it cheaper!" And in this case, that something in me is right.

In the magazine "Kitplanes," a guy named Jim Weir writes a column on all sorts of electrical topics, from basic concepts to DIY projects. One of the tasks he tackles is building your own LED lighting kit for significantly less than you can buy one. Naturally, I was impressed, and after purchasing the components listed, I built the circuits in their entirety on a breadboard and plugged them in. Unfortunately, nothing happened. I started taking the circuit apart and focusing on individual components, making sure they worked as I thought they should. The only part that wasn't working was the part of the circuit that caused the strobes to.... well.... strobe. I sent email after email to several of Jim's email addresses, and email after email to the contact address for the magazine. After a couple of months of no response from anyone, I decided to give up on the strobe circuit as published.

My solution was to digitize. I've been playing around with Arduino at work, and decided to see what I could accomplish here. 

The LEDs I'm using are made by LEDEngin. The strobe part numbers are LZ1-00R102 (red), LZ1-00G102 (green), and LZ1-00CW102 (white). I'm running these at 800mA, which isn't quite as bright as they can be, but should keep them from burning out for at least a decade. The landing light part number is LZ9-00CW00. 

So, the difficulty is in powering these guys. That part of the circuit I went ahead and borrowed from Jim. Basically, you use a transistor and a resistor wired up as a current limiting circuit. To tune the current, you adjust the resistor. 

Now, for the landing light, that's it. Case closed. Toss a switch in there and all the heat shrink and wire loom you can find, and then cover the whole thing over with electrical tape and dunk it in that Plasti-Dip stuff for good measure.  (Pretty sure that's not how the Designated Airworthiness Representative wants to see your electrical work done.) The strobes still need some work, though, and for that we turn to a tiny electronic arachnid called an ATTINY85.

The AT-Tiny-85 is programmed with the Arduino IDE. Arduino borrows heavily from C++, but it's so well documented that even a complete ape like me can copypasta (that's internet lingo for copy and paste) and make things work. The ATTINY85 has 8 pins, but we only need 3 - one for each strobe. The ATTINY85 puts out 5v, which is plenty for our LEDs, but it only sends a measly 40mA per pin. That's not enough to light our LED, but it *is* enough to trigger a Darlington transistor... which is exactly what we'll do.

A transistor can act like an electronic switch - voltage won't flow across two pins, unless voltage is applied to the third pin. A Darlington transistor is akin to two transistors paired in order to handle large amounts of power, while being switched by just a little bit. In our case, we're controlling 800mA at 12v by applying >40mA at 5v. This means that not only can we turn our LEDs on and off electronically, but we can write code for any strobe pattern or brightness (by using Pulse Width Modulation) and upload it to the chip, rather than swapping analog electronic components for different timing or brightness. What's more, we can control each LED on its own pin, so they can be independently programmed, but they're all running from the same chip, so they can be synced perfectly if we so desire. And we do so desire.

So, in the end, I spent >$150 for everything, including lenses and having the PCB printed and shipped to me. That's pretty stinking good, considering these lights are significantly brighter than what's on the old Cessna 172 I learned to fly in. Total power consumption is >5 amps, which is significantly less than what the old Cessna draws for its dim lights. This may not be the best solution, but it is *a* solution, and I'm pretty happy with it. 

In the spirit of sharing, I've added the board and parts list below, and the board can be ordered at https://oshpark.com/shared_projects/6PehzB0Z . 

*edited March 29,2017: Turns out the board isn't quite right... Upon population, the landing lights worked just fine, but the strobes weren't working at all. I'll link to the new board once I figure out what's wrong and fix it...

Parts list:
1x LEDEngin LZ1-00R102
1x LEDEngin LZ1-00G102
1x LEDEngin LZ1-00CW102
1x LEDEngin LZ9-00CW00
1x ATTINY85
3x 2 watt 1.5 ohm Resistors
3x 2 watt 1.8 ohm Resistors
3x 1/4 watt 1k ohm Resistors
6x LM317 Transistors
3x BDX33C Transistors (TIP-120s will work just as well)
You'll also need a way to program the ATTINY85.  I use the AVR USB programmer from SparkFun. The code in its entirety can be copypasta'd from below:
_________________________________________________________________________________
/*
LED_Flashers_Rev_3

  Designed for a homebuilt airplane's LED strobe lights using an
  ATTINY85 chip. This code will flash the LEDs 10 times in 800ms,
  then leave them off for 800ms.

  modified March 15, 2017
  by Matt Quimby
 */


// the setup function runs once when you press reset or power the board
void setup() {
  // initialize digital pin 13 as an output.
  pinMode(0, OUTPUT);
  pinMode(1, OUTPUT);
  pinMode(2, OUTPUT);
}

// the loop function runs over and over again forever
void loop() {
  for(int i = 0; i < 10; i++) // using a "For Loop," int = integer, used as a counter to count "i." Starts by assigning a value of 0 to "i," and states that the bracketed code should loop if it reads "i" to be less than 10. "i++" tells the counter to add 1 to "i" each time the sequence is run.
  {
  digitalWrite(0, HIGH);   // turn the LED on (HIGH is the voltage level)
  digitalWrite(1, HIGH);
  digitalWrite(2, HIGH);
  delay(40);              // wait for 40 milliseconds
  digitalWrite(0, LOW);   // turn the LED off
  digitalWrite(1, LOW);
  digitalWrite(2, LOW);
  delay(40);               // wait for 40 milliseconds
  }                        // completes the inner loop
  delay(800);              // wait for 800 milliseconds
}                          // completes the outer loop and starts from the "void loop" again
_________________________________________________________________________________

And the board: (2.51" x 2.61")



And finally, the followiing is how I spent my afternoon and evening.