Sunday, April 21, 2013

Pot Metal Prosthetic

One of my neighbors dropped off a small part with my wife to see if it could be welded. Scott is a metalworker artist that lives down the street. He really doesn't have much of a shop and ends up building most of his sculptures out in the alley with his MIG welding machine. He builds metal animals and furniture with found objects, typically metal and aged wood. For this reason he is always on the lookout for metal objects for his work. I think the word is out to all his friends too because his pile seems to grow when any of them stops by or has some metal bit they don't want any more.

One of his recent acquisitions was a wood cutting vertical band saw. I didn't hear the story about where it came from but somehow it was damaged. If you have been following the blog then you know I have a soft spot for machinery repair. Add this onto the fact that Scott is a cool cat and fellow metalworker and you have the makings of a good project.

Looking at the part it appears the saw was tipped over and fell on the band wheel side of the machine. The part that he brought over is the bracket that does the upper wheel tensioning and blade tracking. Obviously this is a pretty important part. Scott was thinking it would be an easy welding job for TIG so he gave it to my wife to see if it could be welded. My wife is an extremely talented welder and immediately noticed it wasn't aluminum but non other than the dreaded pot metal.

Most welders have a couple of stories in their inventory about trying to weld pot metal. Nobody can really tell you what it is other than pot metal 101. The alloy contains all the most difficult to join materials all rolled into one alloy. Zincalumagleadalloy would be as accurate a description as pot metal as any. It is the scourge of the welding community. To a person it is hated everywhere.
The density is always the give away for me. Many of these materials that are collectively called pot metal contain high levels of zinc in them which is fairly dense. This is a common permanent mold casting material and was never intended to be repaired by any welding process or mere mortal. You can see in the picture the large chewy coarse texture of the break.  My wife smartly tossed the job over the fence to me to see if I had any special tricks up my sleeve.
I took a look at the part and came to the same conclusion as my wife. Welding was not the way to repair this part. My second instinct was to see if you could easily buy one. Zinc is used for a reason, its very inexpensive to produce. If spare parts are available for this saw then I would guess this is a sub $10 piece. But where's the fun in that? I do this stuff because I like to. Its pretty cool when you hand back a job and the customer is totally blown away by something you had fun doing. For this reason I decided to make a completely new part from scratch. I shot a bunch of video so check out the oxtoolco YouTube channel in a couple of days to see live action.

The first step was to take some measurements of the original part and make a shop sketch. I find this light marinating helps me form the plan of attack for the work sequence as I measure and examine a piece like this.
 The first step was to rough out a rectangle the same size as the exterior of the sample part. I started with 1-1/4 thick 6061 plate. Band sawed then milled on all six sides. In general with a part like this with some curves and steps you try to keep it square and blocky as long as you can so its easy to pick features up and hold on to the part. I poked all the holes in it when I had it in the mill. The original part had a shaft pinned in place in the .590 hole. I bored this hole so the pin is a press fit. The reason for this is the roll pin was put in willy nilly and would be a pain to locate and re-use. With a press fit and maybe a setscrew as a chicken bolt the shaft is not going anywhere.

I decided to do the .940 bosses first. I made a quick holding spud to fit the large hole in the part accurately. This spud will be used to turn the bosses on either side of the part.
 The spud is tapped to take a 1/2-13 bolt. The length is a whisker shorter than the length through the part so when the bolt and washer come up against the part it is clamped securely. The boss will be an interrupted cut so I used a large fastener to hold it to the fixture.
In this picture you can see the washer and bolt holding the part to the spud. I just tapped the washer into alignment so it didn't run out so much and get cut by the tool as I turned the boss to the correct diameter.

The next operation was to cut the small round bosses on the end of the part. I decided to do this in the four jaw chuck. I suppose I could have done it in the mill but I was in the mood for lathe work.
I had already put the small holes in so I had something to indicate on. With two chuck keys and a good indicator its easy to align something in the four jaw. I used to hesitate to use the four jaw chuck, but since getting my chuck changing setup down pat and having a second jaw key its really not a serious roadblock anymore.
 This is after the four jaw work. The part is starting to look like something. Next step is cutting the radius on the outside of the part. For this I'm going to use an old time toolmaker technique to step the radius with a ball nose end mill. I have heard several names for this technique but the one I like the best is "Kellering".Keller machines generally followed templates or models but they made their cuts as a series of stepovers that look very much like this method.
Here is a screen shot of the layout for 5 degree steps with a 1/8 ball end mill. I plotted the coordinates in AutoCAD for the Y and Z positions on the left and the X positions on the right. Because of the boss I need to stop the tool short as I come into contact with the curve of the boss. All the coordinates are off the centerline so I can use the same numbers for the opposite mirrored half. Its a lot a numbers to keep track of but you can do some cool stuff with manual machines using this technique.
Here we are partway through the process. Its the same technique that modern day CAM software calculates the toolpaths for three dimensional milling.
Here is my cheat sheet. You see I marked off each set of coordinates as I used them. If I didn't do that then I would be lost after a few passes.
Here is the part right off the machine.
The X positions bumped into the boss by a small amount. At first I was puzzled why that happened. It turns out it was the small difference in the radius in the sample part. My Z touch off was on the smaller diameter when it should have been on the slightly larger size. For this part its not a big deal but it easily could have mattered. I like to go back and understand what went wrong so hopefully I don't do it next time.
Here is the completed part. I re-used the shaft out of the original part and pressed it into the new bracket with a couple of grand of interference. I heated the part with a propane torch to expand the diameter and make the assembly easier.
The curve was cut on the band saw and hand draw filed until it matched the original. I added the chicken set screw just in case.  Here is one of the video's I shot that shows the fitting of the shaft into the new part.
Be sure to check out all the video's in this series. I had a some fun making the part and Scott is jazzed to get his saw going again. Its a win win dealeo.

Thanks for looking.

Thursday, April 11, 2013

Manual Helical Mill Turn

We had an interesting job come through the shop the other day. We have been working with a physicist who is taking magnetic measurements of the magnetic field quality of one of our large superconducting dipole magnets. He measures and maps the field generated by the magnet when the magnet is operational.

This entails that his magnetic measuring probe must move up and down through the bore of the magnet that is cooled to 4.2 Kelvin. He uses an interesting anti cryostat device inserted down the bore of the large magnet. And inside this his probe rotates, and is moved up and down axially through the bore of the magnet to map the magnetic field. The purpose of the anti cryostat is to keep the measuring probe at room temperature when it is inserted in the super cold bore of the magnet.

The problem we were asked to solve was a failure of a plastic part at the end of the magnetic measuring probe that couples to the rotational motor. During the test run this part somehow got warm, and the small set screws holding the coupling to the probe collapsed the probe end. There are very small instrumentation wires that pass through the center of the coupling and the probe end that really don 't like to get twisted around.

The part that had failed had some semi-complex geometry so the way they were manufactured was by the 3D printer or FDM process. This was fine for previous tests and the parts performed well. On this particular test there was a heating problem that caused the FDM parts to soften and fail. The first step was to make the probe end out of something a little more durable. We had a good sample and a drawing so I gave the job to one of the new technicians to fabricate. We chose PEEK for the probe end material. This is a tough strong high performance plastic that has a wide operational temperature and excellent machinability.

While the probe end was being fabricated I had another discussion with the scientist and we decided that the failure was most likely from eddy current heating of the metal bearing spacer that happens to ride on the probe end. With this realization we decided to try to eliminate as much metal from this particular area as possible.

One of the parts that needed to be made of some non metallic material was a small helical shaft coupling. This type of coupling is used because of its zero backlash ability. The magnetic measurements are correlated to the probes position inside the magnet with rotary and linear encoders so a map of the field can be determined for that particular magnet configuration.
Our task was to duplicate this coupling in PEEK as quickly as we could, with stuff we had around the shop. There is limited measuring time and the cost per hour of the large magnet test is quite high so speed was important. I scrounged some tooling together and formed a plan of attack.

This is just the kind of job many machinists love to do. A tricky job with a minimum of detailed constraints, and the freedom to do the job anyway you can. The only thing that machinists generally don't like is the time constraint. When I was learning this stuff I would have liked to sink my teeth into a job like this every day of the week. In fact, I have never done this type of operation before. Sure, I've cut lots of helices and all manner of threads, but I never had to make a coupling like this. Honestly I really wanted to do this one myself but now its more important for me to allow other folks to have the experience and successes in the trade.

Part of the tooling we needed was a special holder to fit in the toolpost of the lathe. I made a quick hand sketch and gave the job to the other new technician I'm looking after. If this was going to work we would do it as a team. The tool block was needed to hold the hand piece of a Foredom tool

The general plan of attack was to use a thin saw blade mounted on the tool post to cut the helical groove in the coupling. I had a couple of small saws with arbors that would fit in the Foredom tool.
The blade width with was pretty close to the cuts on the sample. Definitely close enough for this operation. So both technicians, Matt and Nick were busy prepping the tooling and coupling blanks at this point.
There was a little more to the setup than mounting the slitting saw in the lathe. We wanted to match the pitch and number of flexures on the sample coupling as closely as we could with the tooling we put together. A quick calculation gave the helix angle of the "thread" we were going to cut. The pitch on the coupling was 12.5 TPI which is a bit of an oddball. We went with 12 TPI because this would thicken the flexure membrane a little and the Monarch could do this pitch.
 From this calculation we get the helix angle of the thread which we use to set the compound rest and the tool holder.
We also had to pitch the toolholder at the same angle since the tool we were using extended below the machine center line. If you look at the picture above showing the hand piece clamp you will see a set screw in the tongue of the holder. This was used to tip the tool in the Aloris holder to the helix angle. 
The idea was to use the lathe as a synchronization and holding tool as opposed to a machine tool. This little Monarch 10EE has a nice DRO so we could control the Z axis position well. We set the threading levers to our pitch (12 TPI) and engaged the half nut. We then ran the saw up to the face of the part by hand with the threading lever engaged and picked off the Z position.
The saw was set on centerline because I couldn't think of a reason why is should be anywhere else.
The first pass was to see how things went. PEEK is easy to get hot so we had the koolmist setup going. The saw blade would be passing through quite a bit of material so I was worried about overheating the material.
Here Nick has done a few passes along the helix. We used the DRO to reference the start and stop positions of the groove. When the end was reached he backed out the compound that was set at the helix angle and then just hand reversed the spindle by hand to get into position for the next pass. There was a fair amount of backlash in the system but with the DRO it was easy to get back to the start.  Nick and Matt cut the helix on their own. I had to go to a meeting right around the time they were cutting the helix. There was enough time for me to try my hand at one pass. So I did get to actually try it. Thanks Nick and Matt for giving me a go at it.
The results speak for themselves. Not bad for throwing some stuff together in a few hours. So with some Yankee ingenuity and some curious and willing teammates you can get a tricky job done and have some fun too. I know I still feel satisfied and proud even when I didn't turn the cranks myself. It would have taken me three times as long if I had to do all of the work myself. This is an excellent example of how team work can pay off.

We shot some video of the operation which I will put up on my YouTube channel in the next couple of days.

Thanks for looking.

Wednesday, April 3, 2013

Squareness Comparator Gage Finale, finally

I finally felt bad enough to go ahead and really finish the squareness comparator I started a few months ago. Part of the reason is I'm close to heat treating a couple of 8620 cylindrical squares that will be used with the comparator to check squareness in the shop to very close limits. There will be a separate article on the cylindrical square fabrication. It all needs to be done to make a nice article about it.
The first step toward completion was to break down and buy a few carbide balls for the wear surfaces on the bottom of the gage. These are standard McMaster Carr items. In fact everything on the gage was sourced from our friends in yellow and green. These ball were then pressed into holes in the bottom plate and then ground flat into a good size bearing spot.
The carbide balls are .375 diameter and the press fit was about .002 on the diameter. With a ball you have very little diameter in contact with the hole so they can use more interference and still be pressed together easily without having to bore the holes to a .0002 tolerance. If the rest pads were cylindrical I would have only used .0002 interference for the amount of axial engagement.
The Bridgeport quill and a good chuck make a great precision arbor press. If I need to remove them for some reason I can drill from the other side and punch them out. There is plenty of meat left on the balls for several lifetimes of re-surfacing for even the fussiest toolmaker.
I started to grind the balls flat on the surface grinder but it was taking too long with all the traversing across the part. I took the base out and roughed them on the carbide tool grinder until I was happy with the spot size then came back and finish ground the flats all smooth and coplanar. They don't need to be ground for function specifically but its nicer on the surface plate with larger flats.
The finished base plate of the comparator gage.
Close up of the wear feet and spot size.
Finally finished. It took a couple of months of walking by the thing every day to make me finish it.
I did make one modification that is not on the drawings I posted on the blog. I was using it one day for a squaring job and I realized that the indicator wouldn't go low enough for the part I was checking. I looked at the indicator bracket and saw a simple solution to  allow the indicator mount to drop another inch lower.
By flipping the indicator mount over the offset goes down lowering the indicator stylus. The only thing I needed to do was cut a relief into the mount for the back cap of the indicator. It was literally a five minute job. Depending on the indicator you use it may not need this relief cut.
So that concludes the squareness comparator gage. Sorry it took so long to get to the end.

Thanks for looking.