Mini-Mill Improvements, A Companion WIP

[epicfail48]

As some people are doubtless aware, ive made a long-term project out of making improvements to my Grizzly G8688 mini-lathe:

http://bbs.homeshopmachinist.net/threads/75291-Mini-Lathe-Improvements-A-WIP-Thread

As a companion to that tool, ive also got a Grizzly G8689 mini-mill, and i figured "hey, if im spending this time working on the lathe, why not put some effort to the mill as well?". Much like the lathe, the mill is a tool that worked plenty well straight from the tent sale i bought it at. It spun bits, it drilled holes, it made parts, it worked about as youd expect; passably, and thats all you can expect for something you grabbed off a pallet for $100. Not great, but workable.

Being honest, the only reason im touching either of these machines is A) im a firm believer that your work will never be more accurate than the tools you use to produce said work, so it pays to make sure the tools are as accurate as you need them to be, and B) im bored. I have also found some problems with the machine that are affecting the desired accuracy of my work, and those are what ill eventually be addressing.

As it is now, the tool has already had a few mods made to it. The head has been fitted with a gas spring and a belt drive conversion kit, and it has been fitted with a DRO courtesy of a few igauging absolute digital scales and a TouchDRO controller. Planned improvements to the mill are as follows:

-Replace the head bearings with a set of angular contact or tapered roller bearings. As with the mini lathe, i feel that this will improve the machine where the rubber meets the road, at the spindle
- Replace the base and column with a fixed-angle version. My mill is one of the tilting column types, and theres an unfortunate lack of rigidity to it. The solid column conversion kit that LMS offers is a very desirable upgrade to me
-Check and overhaul general alignment. One of the biggest problems ive noticed with my mill is that the table does not seem to be parallel to the direction of movement, which makes for wedge-shaped parts and problems tramming the mill
-Improve fitment of sliding surfaces. Id like to find a way to make the bearing surfaces on the ways just a touch better, be it through scraping or surface grinding.

Thats all that springs to mind at the moment, so off to the races!

[epicfail48]

The first thing ill be addressing is the table of my mill. Ive noticed for a while that parts i face off in the jaws always seem to have a wedge shape to them, no matter how firmly theyre seated on the parallels. Ive also noticed that no matter how in-tram my indicators say the head is, the cuts always heavier on the left side. Before anybody asks, yes, i did test the vise on a surface plate, and parallelism was spot on, so starting this i was 80% that the problem was in the table itself. First step, as it always is, was measuring to see exactly how bad the problem was. An indicator was set up in the spindle, this time my new Mitutoyo digital dial indicator, and zeroed at the back left corner:


Shifting over to the front left corner, still on 0:


So, the top of the table is parallel to the Y axis of movement. Now to check the X axis. Sweeping to the front right corner:


Well that aint good. Halfway across the table and already a .004" drop, which is the same at the rear of the table:


So, this got me 100% certain that the problem was with the machine itself, not the vise. The next step was to disassemble the mill and try to isolate exactly where the problem lies, which i did by starting with the table off at the surface plate.

[epicfail48]

Over at my surface plate, i started by setting the table down on its dovetails and doing the tap test, seeing if any corner was obviously off by listening for a knock, and right here i screwed up my methodology. I didnt realize until after the job done, but i was doing all these measurements by referencing off the TOP of the dovetail, and not the bearing surface like i shouldve been. Now, theoretically the top of the dovetail and the bearing surface are coplaner anyways so it doesnt matter anyways, but i did not verify that, referenced off the wrong surface, and got stupidly lucky. Im presenting this at this point as an example of what not to do. Again, my reference surface shouldve been the bearing surface, not where i measured from.

Now that that disclaimers out of the way, over at the surface plate and verifying that the (wrong) reference surface is flat enough to take some measurements, i swept the surface with a DTI on a surface gauge, just to test and see if one side of the table was actually higher, zeroing at the front left corner and moving from there. The results:


As you can see, the table is a little wonky, with the right side being roughly .006" lower on the left side. At this point, i got a little stuck. I knew i needed to take some material off the top, but at 15" the table is too large to blue up on my surface plates reliably for scraping, and my surface grinder is only a 6x18" so i cant- wait, i know exactly how im going to do this:


There it is, all set up on the grinder ready to true up the top of the table, and take out some dings as an added bonus. Here again, warning, i did not do this correctly. Again, im referencing off the top of the dovetail, not the bearing surface that the table references off of on the machine. What i shouldve done was found a way to fixture the table so that it way sitting on those bearing surfaces, to make sure that the table top and bearing surfaces were coplaner. Anyways, before grinding the top:


And after:


Finish isnt perfect by any means, but its a lot better than it was, and its flat and parallel to the (wrong reference surface!) bottom of the table. Just for giggles, while i was at the grinder i flipped the table over and just barely dusted off the bottom. Serves no real purpose, but was nice to do. Now, another trip to the surface plate!

[epicfail48]

Time to measure the top of the table again, and here its more of the same as the last batch of measuring: reference off the wrong surface like an idiot, check the top to see how consitent the thickness is in a dimension that doesnt matter at all:





DTI reads 0 at all 4 corners, and stays at 0 no matter where you sweep it, so the table top is as flat as i care to measure, and of consistent thickness in a way that has no real use! In case you havent noticed yet, im really cant believe i made such a boneheaded mistake in my process here, because it was sheer stupid luck that prevented me from making things worse. Anyways, now that the measuring is done on the plate, time to reassemble everything and redo the check on the machine, off to the next post

[epicfail48]

Omitting the assembly steps again, time to repeat the same measurement process from the start of all this. Indicator is zereod at the back left:


And then traversed to the back right:


Flat 0 across the full length of the travel. Against all odds, my idiot error actually ended up making things better, since the right side used to be >.004" lower than the left, which was what caused my thickness issues. Somehow i fixed the problem i set out to fix. With the indicator still zeroed, the front corners also need to be checked:



As you can see, the front left of the table is high by ~.0025", and the right is high by ~.0035". This is where my referencing off the wrong face came back to bite me in the ass, once i realized my mistake in reference surface i went back and measure the thickness from the top of the table to the bearing surface of the dovetail. Sure enough, the table is thicker at the front, which confirmed 2 things; the top of the dovetail and the bearing surface were no coplaner, and i was an idiot for referencing everything off the wrong surface. On to the conclusion section

[epicfail48]

There she is, all reassembled and shiney, complete with a few marks already in that nice ground finish. Errors in referencing aside, the end result turned out okay, ive not got a flat table thats parallel to within a few thou of the direction of movement, and even the error is within the acceptable level of tolerance for my work. Once everything was put back together i checked for tram, which i didnt photograph because everybody photographs tramming a mill. Left-to-right rotation was easily dialed in to under a thou off on a 9 inch circle, but theres still about a .004" nod to the head. Adjusting that would require shimming the column, and being that i plan on replacing the column i didnt see it worth fussing with at this juncture.

So, all the effort, was it worth it? Well, an indicator run across the base of the vise shows 0 misalignment, so it looks likely. Proofs in the pudding though, so time for a test:


Bar of 1"x2" aluminium, tossed in the chuck tight against the parallels, then skimmed on both sides with a face mill. First thing i noticed was the double-cut pattern to the surface, where the cutter actually cut on both sides. First time ive seen that, and a nice confirmation that my mill is now actually in tram. The nod in the column didnt present as much issue as youd think either. Beyond subjective measurements, a micrometer read on on both ends of the bar showed a difference in thickness of .0012" across the 9 inch span of the part. Given that im used to seeing about 4 times as much change on a part 1/4 that length, im going to say yeah, it was worth the effort!

[epicfail48]

Been a while, time for an improvement by necessity! In the way that things go, i manage to finish up part of a project, and the second i did, something broke. In this particular case, it was the Y axis, the leadscrew completely seized in the bearing block. What i think happened is that, somehow, a tiny chip managed to migrate into the block, wedge itself between the leadscrew bearing surface and the block, ball up and gall the hell out of everything. Had to press the pieces apart, the results were... not pretty:

The leadscrew, and this is after a cleanup cut on the lathe to take off the worst

The bearing block, complete with crater and massive score mark

The rear of the bearing block has a collar on the leadscrew running against it as a bearing surface. Honestly, poor design

Clearly, i need to do something to fix this. I couldve just cleaned up the leadscrew and bearing block, gone in there with a flex hone or something and taken off the burr, thrown it back in service and called it good, but that seemed kinda hacky to me. I wouldve liked to make an entirely new bearing block, but that would require a working mill for parts, and clearly i do not have a working mill. I ended up deciding to just clean up the leadscrew, which wasnt damaged too badly, and then bore and bush the bearing block.

Some quick planning and i got off to the races. First was material, i needed something to make the bushing out of. Wouldve like to have used some bearing bronze, but i dont have any of that. What i do have is a big hunk of cast iron, and i figure for this application cast iron will work more than well enough. The materials assembled:

Leadscrew, bearing block and a roughly 15/16" slug of cast iron. Dont ask how i made that, hole saws probably arent meant to be used like that...

A quick, garbage sketch of what im planning. The pencil is a non-scale drawing of the existing block, the black line is what the bushing will take up. The flange on the bottom is what the collar on the leadscrew i mentioned earlier will now run against, figured thatd be better than the chinesium of the block. I need to get better at drawing up plans, they make sense to me but theyre really a mess... Basics are the OD of the busing is .625", with a 12mm running fit bore put in it for the leadscrew to run in. Bushing is the full length of the block, with a .125" thick .7" diameter flange on the bottom.

Half hour of turning later and here we are:


Cant see it in that photo, but the bearing block has been bored out to ~.625" (measured with a field-expedient go-nogo gauge as im lacking anything for internal hole metrology), and the bushing being turned to .6265" for a moderate press fit. Little much of a press fit actually, as it took a bit more force that any of my tools could put out to seat the damn thing... Hammers fix everything though. Got it all pressed together, and you can see that i screwed up the counterbore for the flange in this picture:


And thats where it sits for now. Still need to bore out the bushing for the leadscrew, then true up the faces, but thatll be another days work. Managed to melt my lathe belt, have to fix that before anything now. Sorry for no photos of everything in action, my shops a mess and i probably wasnt doing things properly anyways. Picture a fixturing job done by a monkey and youll get an idea of how i turned all this

[Cenedd]

Not sure any of this is of use to you but I suspect I have a similar mill to you and have done some of this myself. I've got an SX2 with the gas spring (from factory/reseller) and I added the fixed column a while back. Two things that may be of interest. Firstly the fixed column kit comes with a new longer leadscrew for the Y axis and secondly you may find that once you put it together you have to start again with truing up the table and the column.

The pads on the base for the column to mount and/or the bottom of the column are not perpendicular to the Z axis so shimming is going to be necessary unless you're damn lucky or can fix that on your surface grinder. I contemplated mounting the column it on the lathe to true up the bottom of that but chickened out when I realised it would have to be held off the dovetails and not just a flat side.

I don't think I'd measured it before I made the change but definitely afterwards I had the same issues with the table. The way I went about it was to tram the head as close as possible, mount three bars of 20x10mm steel square to the T-slots (or running in the Y axis) with counterbored screws and then mill the surface of each bar flat using the Y axis - if I remember correctly, the X was more off for mine and the Y not so bad but also shorter. I then sent the table out for surface grinding and had them chuck it upside down (milled bars against the mag chuck) and skim off the bottom of the dovetails. This then made that true to the plane of the mill and it was then just a case of flip it the right way up, remove the bars and skim the top. If I remember correctly it came out with less than 20 microns (take with consideration for the precision of the cheap import DTI used!) variation.

Pretty sure we also cleaned up the base casting by assembling the table to it (one trued), mounting the whole lot to the mag chuck upside down, skimming the bottom of the casting and then flip, dissemble and skim the pads for the column that were very roughly milled and quite far from true.

All the ways should probably have been scraped true (there's definitely a bow in the Y axis ways) but I didn't have that much spare sanity!

[RB211]

Do you trust these Chinese factories are paying a professional tool scraper to make sure these hobby grade machines are scraped in perfectly at the price points the importers demand? If anything, they feel more pressure today to produce at a lower price point.

[epicfail48]

Alright, been a while but i did a bit more work on the mill. Most noticable, i finally scraped together enough to get the solid column conversion kit from Little Machine Shop:


As with all things from LMS, its ridiculously overprices for what you get but still absolutely invaluable for these machines. Before i get too deep into that, lets talk about the why, because i know someone is going to ask; I made the switch to eliminate the tilting mechanism as a source of deflection in the machine. Im well aware of the fact that i couldve made a bracket to stiffen the existing column, rather than buying the new column. I discarded that idea for several reasons. First off, i dont have a supply of scrap steel i couldve made the bracket out of, so i wouldve had to buy it new, and between the cost of the steel plus the cost of the time, i wouldntve saved any money. Its also worth mentioning that the solid column is custom made to higher standards for LMS, so in addition to it lacking the pivot, the walls of the casting are also twice as thick. Cant do that with an extra plate.

Anyways, moving on to quality, its... Well, its better than stock. Still needed some shimming to get things into alignment, had to drill a hole in the column to mount the gas spring. Absolutely baffled me that i needed to drill into the LMS column to install the LMS gas spring kit thats a default option on the LMS version of this mill that takes all the same parts. Not difficult to do, but odd. The most irritating thing about installing was the paint job, for some reason they painted the mating surfaces too:


Easy enough to correct, but again, irritating. For the price i shouldnt be scraping paint off. Anyways, got everything installed, got the appropriate shims to get everything within tram to about .001" on a roughly 8 inch circle. Not perfect, but anything finer than that needs sub-thou shim stock and i only have .001" stuff. So, hows it perform?

Night and day difference. With the tilting column i could get things to work, but not well. I was limited to a D.O.C of about 1/4", and thats on a good day, in steel no matter the radial engagement, otherwise chatter would leave an flat out unacceptable surface finish. Aluminium and softer metals werent much better. Face milling was about the same story, my 2 inch face mill was limited to about .010" D.O.C in steel, and that was assuming i could live with a bit of chatter and the machine sounding like it was about to fly to pieces. With the solid column, no problems at all on either from, Im pretty confident id stall the motor now before i outdid the rigidity (Motors tiny, keep in mind). I was able to take a .060" D.O.C slotting pass in mild steel with an acceptable surface finish on the walls and floor, a facing pass with the 2 inch face mill at roughly 70% engagement at a .025" D.O.C, and side mill off about .020" from a bar of 5/8" cold roll. Little bit of noise from all those cuts, but again, passable surface finish on everything, no chatter, and the machine didnt sounds like it was going to explode. I was even able to get some really nice chips off the face mill:


Aluminium inserts in an SEHT mill, so the surface isnt super shiney, but seriously, those chips are nice, especially since i used to get dust and flakes instead of nice little lines. Overall, pretty stoked about this. Still need to get the DRO scales resinstalled, and finish washing the taste of dealing with LMS out of my mouth, but overall im pretty happy with this. Tool was already useable, now its just nicer to use.

And before anybody says it, even factoring in all the upgrades ive done, ive still got barely $500 in this thing, counting the cost of the machine, and ive been able to make parts from day 1. Good luck managing that with a craigslist special, especially in my area

[epicfail48]

New column in place, time to address the next big issue with this mill; squareness, or the lack thereof. Unfortunately, the X and Y axis arent square to each other, thanks to a defective saddle. I kicked around with a few ways to fix this for quite some time, before finally deciding on what i think to be the easiest way to get the best results. To that end, i went down to Grizzly, bought a new saddle casting, and im going to completely remachine it to hopefully better standards. Making this just slightly more challenging, ill be doing the machining on this mill as it currently is. I know that my mill can make something flat and straight enough as it currently is, it just wont cut square, so it just became an issue of figuring out how to do all the work with only 2 axes being used, X and Z. Ill try my best to explain along the way what im doing, to start with lets take a look at the new saddle:




As you can doubtless see, its pretty rough for a 'finished' part. Theres the baked on layer of cosmoline, to say nothing of the sliding surface that looks like it was milled with a chainsaw. Ive got it sitting on my surface plate, while i was taking these pictures i was also doing some preliminary checks on the geometry, seeing if the sliding surfaces were parallel and flat, if things were square, etc. Sparing you the gritty details, they werent. Neither side was flat, nothing was square, but thats okay, this is a rough casting for my purposes. A few quick sharpie marks for explanation purposes:


Im going to start off at the surface grinder to get myself some flat surfaces to use as a datum for the machining. Starting at the sides with the red Xs, im going to get those flat and parallel. Once i have both of those sides ground, im going to grab the whole thing in a milling vice with the freshly ground flats up against the jaws and get the flats of the dovetails (black X marks) mostly square to those sides. I say mostly because theres no need to have those 2 surfaces perfectly square and i suck at grinding, so the extra effort id need to put in would be completely pointless. I ended up getting the 2 surfaces square to about .003" worth of deviation measured with a squareness comparator at about 3 inches up the part. Again, not a critical thing, so thats close enough for me. The more important thing was to get the flats of the dovetails flat and parallel to each other, which they now are, pics to come on that

[epicfail48]

And here we are after grinding:




Forgot to get a picture of both sides. Ah well, take my word for it, both sides look passable, and are miles better than what i had at the start. Anyways, now i have 2 flat, parallel, non-working surfaces on the side to grip in the vise on the mill, and 2 flat, parallel surfaces that the table references off of when its installed. Speaking from experience, if i were to install this saddle on the mill as-is the table would be perfectly parallel to the movement of the X and Y axis, and smoother in movement. Ive actually done this same grinding on the saddle installed in the mill, i just didnt machine the dovetails in the original saddle. Speaking of the machining on this saddle, heres how i plan to do that:


The vise is rotated 90 degrees from normal operation and trammed in along the X axis to as perfect as i can get it. I think i measured in the neighborhood of 3 tenths across the length. The saddle is clamped in the vise with what will be the Y axis dovetails facing up. My theory for this is i grab the part like this, machine the Y axis dovetails, then flip the part over so the X axis dovetails are facing up, rotate the vise 90 degrees, then machine in those dovetails. This entire time im only moving the X axis, never the Y axis, with the reference for squareness being however square the fixed jaw of my vise is to the body. Im taking a risk here and assuming my vise is square, but my vise is the only thing i have in my shop that actually has certification paperwork. Its a risk im willing to take. Again, i know that my mill will cut acceptable flat and straight along the X axis, so ill be able to re-machine the dovetails without issue, and im just transfering whatever the squareness of my vice is onto the new saddle. This was the best method i could come up with with the materials i have available. Hopefully itll work out, just have to buckle up and see.

Anyways, onto machining. Now, im aware that the overhang on the part is a bit much and frankly asking for problems, but in this case i feel its a necessary evil. Im limited in space for how i can set up this cut, and im honestly not removing much material, just barely skimming the surfaces, so i dont imagine ill hit any snags. First thing i did, since i already had the face mill installed from the last project, was to skim the clearance surface in the center of the saddle. If you skip back a few pictures you can see what that surface originally looked like, a chainsaw wouldve left a better surface. This is just clearance so the surface finish doesnt matter in the least, but id still like it to look nice and again, face mill was already mounted up and ready to go. Added bonus, this gives me a chance to see how this piece will react to machining and gives me another surface to reference off of if i need it. Final thing to note before the cut, i swept a test indicator over the freshly ground surfaces of the dovetail, just to make sure everything was trammed in and i wouldnt create any weird dovetail geometry. Quick pass with the face mill:


Much, much nicer finish. Not perfect mind you, but a far sight better than what it was. Plus, the 45 lead in of the face mill leaves a nice chamfer on the sides of that relief valley, just looks all around better to me. Now, from here i can move on to actually recutting the dovetails, which i would love to show you if it were for the fact that for some bloody reason Seig ignored convention and made the dovetails 55 degrees instead of the 60 degree standard. I dont have a 55 degree dovetail cutter, not yet at least, so here the project sits until i get that in...

[BCRider]

A nice job on analyzing and correcting items as you've found them. And nicely written up too.

An interesting option for some future time with the solid column would be the addition of an external brace to give an increase in stiffness if you think the scale of the machine would warrant the effort. Something to consider at some point?

[RB211]

Nicely done. The one advantage I can think about the mini-mill is that most parts of it would fit on my surface grinder, if I ever had a need to get one.

[epicfail48]

Nicely done. The one advantage I can think about the mini-mill is that most parts of it would fit on my surface grinder, if I ever had a need to get one.

one of the reasons I'm very happy I stumbled across a 6x18 grinder instead of the more common 6x12 actually, it's just big enough for everything on the mini mill to fit, I think. I haven't tried the column though I'm fairly certain it would. Someone skilled enough could certainly regrind everything, dovetails and all, without horribly much issue.

A nice job on analyzing and correcting items as you've found them. And nicely written up too.

An interesting option for some future time with the solid column would be the addition of an external brace to give an increase in stiffness if you think the scale of the machine would warrant the effort. Something to consider at some point?

I appreciate the compliments! My hope is that if someone happens to stumble across this thread they'll be able to use my misadventures as at least a source of ideas.

With the solid column I doubt that any sort of external brace would add enough stiffness to justify the effort. With the tilting column the brace makes sense as the tilting mechanism sacrifices a massive amount of rigidity, and an external brace serves to fix that. With the solid column the column and based are fixed solidly together and any rigidity issues are with the casting itself, not the fastening method, if that makes sense. Adding to the rigidity of the column would require either filling the column with something like epoxy granite or just bolting massive plates of steel to the sides to try to build up the thickness. Either way I feel you'd hit really diminishing returns. As it is now, the weakest link in the system, rigidity wise, is probably the spindle assembly. Still has the stock radial bearings, and the head castings are a bit on the thin side. New bearings are probably going to be the next thing on the list, still get a bit more tool deflection under heavy load than I'd like and bearings seem like the next logical step. Still though, the column assembly was the biggest weak link in the system, with that replaced it's honestly a different tool