Mini-Mill Improvements, A Companion WIP

[epicfail48]

Behemoth of a dovetail cutter finally showed up, so im back to work on the saddle. Quick refresher for anybody who doesnt feel like reading through, im presently finish machining a replacement saddle assemble to correct the X and Y axes not moving square to each other. Ive already got the outside bits surface ground flat to use as reference surfaces, now im machining the 55 degree dovetails.

Now, in this next part im going to be explaining how and why im cutting these dovetails and what im focusing on. I realize that a lot of people already know how to cut a dovetail, if you do feel free to skip this part. This is more for Joe Newbie who stumbles across this thread. Now, some visual aids:



Some color-coded photos of the saddle, both as its own thing and installed on a mill. Itll make explaining easier for me. Now, first face we want to look at is the black face, what im going to call one of the working faces. This face is going to be what makes direct contact with the dovetails on the base of the mill and underside of the table. These faces, on top and bottom, need to be parallel to each other, otherwise the face of the table wont be parallel to the travel of said table. Flatness and parallelism of these faces has already been taken care of at the surface grinder, and theyve been ground as flat and parallel as i can measure in my shop.

The red and green faces make up the portion of the dovetail that ill be milling in this setup, and can be referred to as the actual dovetail. In this instance, the red face is a working face, it makes direct contact with a mating face on the table or base, depending on which side of the saddle youre looking on. This red face needs to be parallel with the other half of the dovetail making up one axis, and perpendicular to the dovetails on the opposite side of the saddle. The green face is clearance in this application, it has no real bearing on anything.

Now, to re-cut the dovetails ive got the saddle clamped in my milling vise, with the ground faces of the dovetail trammed in to be perfectly parallel to the table, and the vise trammed in the X direction. In order to get the new saddle square on a mill that doesnt presently move square, im not going to make any cuts along the Y axis, instead im going to rotate the vise and attempt to transfer the accuracy of the vise onto the part. Now, the first setup:


Ground face up against the fixed jaw, long edge of the vise trammed along the X axis. The cutter ill be using:


Thats a 50mm 55 degree dovetail cutter. Yes, the angle is bolded for a reason. The dovetails on these machines are 55 degrees and require a non-standard cutter, as i found out to my detriment. Bit of a pain to find, only ones i could find were cheap imports, and yes, thats the bad cheap. This one is HSS, ran me about $30, and has an annoying amount of runout. Still, the angles proper and it cut well enough. Its being held in a 5/8" er32 collet, which is interesting for me because my mill doesnt like the stickout that collet game me. At least, my mill didnt used to like it, after upgrading the column, 0 problems. Money well spent im feeling. Anyways, lets move on to the part:


Thats a closeup of the dovetail as i got it. It doesnt show up quite as bad in a photo so youll have to take my word for it, the surface was pretty rough and franky didnt feel flat at all. Color the surfaces of the dovetail with a sharpie and take a quick pass to see what gets cut:


Surprising nobody, its not flat. I wont bore you with the step by step on how to cut a dovetail, ill just skip to the end where i got the entire surface remachined and looking good:


Actually got a bit of reflection going on the machined surfaces, i was pretty happy with that. Thats one half of the dovetail done, now im going to step over and cut the other side to be parallel to the first, and just as smooth, before breaking down the setup to work on the other side

[epicfail48]

Now, heres the setup after being rotated:


I have the same side of the saddle referenced against the fixed jaw, ive flipped the saddle vertically, and the saddle itself is referencing off a set of match-ground 123 blocks from the mills table, the vise is just holding it square and in place. Here again i trammed the part in along the X axis as fine at my tools allowed, and again, if my theory works then the X and Y axes on the new saddle should be as square to each other as the corners of my vise, so pretty square. The same sequence of cuts is made on these dovetails and thats it, were finishedL



Got a set of nice, freshly cut, smooth, flat, parallel dovetails. A word of note, im saying flat but honestly i dont care enough to blue up a straight edge to see how flat. In this case, flat enough is flat enough for me. The scrape marks on the ground surfaces are intentional, its redneck flaking. Frankly i didnt care enough to fully scrape in those surfaces but still wanted to try to prevent any stiction. Works on the other saddle and the ground surfaces i did on it, so i figure it cant hurt here.

[epicfail48]

So, conclusion to this part of the project. I got all the bits and pieces from the old saddle moved over to the new one, no pictures of that process because franky, its boring. Theres like, 10 screws and you can find guides on how to disassemble a mini-mill anywhere, you dont need my hackjob version of it here. So, howd i do?

Well, first impression was that the sliding surfaces felt a lot better. Honestly, thats probably confirmation bias, but it really does feel like everything meshes together just a tiny bit better and makes for smoother movement. As far as accuracy goes, huge success! Now, before i start going into my numbers i want to be very clear on a few things; I am measuring with the standards available in my shop, and i do not have fancy standards. My reference surfaces are 123 blocks and my screwless vise, i dont own a granite square or any other fancy gizmos, no super precise certification sheets are hanging on my walls, my tests are not NIST traceable. All these numbers are just measured with my basic equipment in my basic shop on my basic tools by my basic self.

So, disclaimer out of the way, i started off by measuring on the machine. I got the fixed jaw of the vise trammed along the X axis, then swept a dial indicator along the side of the vise using the Y axis. With the original saddle, this test got me about .006" difference over about 4", so an error of .018" per foot. Not really an acceptable amount of error. Same test with the new saddle gave me about .0005" over the same 4" distance, so about a .002" per foot deviation. Hardly toolmakers accuracy, but frankly good enough for me, doing the math that puts the angle at 89.993 degrees, or 26 arc seconds off square. Could it be better? Yeah, probably. Do i need it to be better? Not for the work i do.

Next thing i tried was squaring off the end of a bar of steel, then checking it over at the surface plate with a comparator. The steel was a piece of 1.5" wide 5/8" thick 1018, held in the vise, the end squared off by moving the Y axis. Took it over to the plate stood it up on the freshly cut end, used a 123 block as a self-proving reference square to calibrate the comparator, then checked squareness on the part about 3" up from the surface plate (making sure to measure the face of the bar that was up against the fixed jaw of the vise). End result? Bout the same as what i measured on the mill, about .0004" off from square.

Too much detail that makes very little sense and doesnt prove anything to anybody but me? Probably. Bottom line is that my mill now cuts square to a decent enough tolerances, plus the overall smoothness and rigidity have had a subjective improvement from this. Time well spent to me. Now i just need to get the DRO scales reinstalled and im off to the races, at least until the next project to improve this thing. Already got some plans brewing. Unfortunately i dont have access to a larger machine so i cant get to work truing up all the other dovetails, id love to recut the base and table dovetails, so the next project will probably be something a little smaller. Ive got some ideas for replacing the leadscrews with ballscrews to start making steps to a CNC conversion in the distant future brewing

[Ringo]

there is way to confirm and/or measure a accurate squareness without any standard square.
it's an old wood tablesaw trick.
take a piece you 'think' is square, on table saw lets say 6" sq,
using your fence or mitre make a skim cut, rotate the part 90 degrees that the fresh cut is on the fence, take another skim cut.
repeat this till you go around the part 5 times.
the last skim cut is in fact a parellel if your fence/mitre is in fact square.
Should be the same with your mill, the fixed jaw is your fence, and if you go around 5 times and the last cut is a parrellel, then you are square.
What this does is magnify the error every time you place the fresh cut on the fence, turn it 5 times you magnify error 5x.
Go around 10 times and you get 10x the error, this makes a small error measurable.
If the last cut is a taper, you are not square