No announcement yet.

Repeatable Micro Accuracy

  • Filter
  • Time
  • Show
Clear All
new posts

  • Repeatable Micro Accuracy

    I have had and used the chinese made Grizzly G9729,

    for a couple of years now.

    For those of you with the (HF) mill/lathe, this is the exact same machine with the quill shaft changed from R8 to MT3 and the tail stock shaft changed from MT2 to MT3. Oh yea, and its green.

    I've managed to get above average results from mine with some effort.

    Notice the 3/8"X3"X4" table frame, the 3/4" table top, and the extra foot under the middle of the machine. Not visible is the dedicated 12"X12" extra reinforced concrete footings, the 8 1" and 8 1/2" adjustable anchor bolts, and of course the leveling of all the structure and machine and the tightening of the machine. After much attention to detail, I'm able to get some pretty good work out of it...

    ...for what it is.

    After reading Jerry Kieffer's article on "Making Miniature Taps and Dies", (HSM - Jan/Feb '08), I'm left salivating for micro accuracy. Running my machine as tight as I can run it without causing failure, I can get pretty dang accurate for a couple of runs but, I'll have to retighten a gib here or reset a half nut there, and that machine just flat out ain't designed for repeatable micro accuracy.

    If money wasn't an issue...

    ...if money wasn't an issue, I'd get me a nice toolroom lathe and set it up in my climate controlled shop.

    Did you notice the tarp that serves as the wall in my shop? (Second picture)

    So, realistically, I'm looking at these Sherline machines and Taig machines and thinking that I might get some good repeatable accuracy out of these. Mr. Kieffer makes mention of Sherline and that "imports" should be avoided.

    So are there any Micro lathe or mill owners here with solid opinions on these little machines? Got any headaches to report with them? Would you purchase differently if you had it to do over? Any horror stories?

  • #2
    Can you define just what you mean by "micro accuracy" (in fractions of an inch) Its easier for folks to tell you what will and won't meet your expectations if we know what they are.

    If you are talking about machining to tenths, that's not easy on any machine and takes lots of attention to detail. There are techniques like using the compound rest on a lathe set at a short angle to allow you to more easily dial in a few tenths. Even then, temperature changes in the machine and material can create shifts. Using a cheap live center could be as big an issue as gib fits, for example. In short, it may be that you can get what you need with what you have with some careful analysis of what's wrong.

    When you talk about "tightening gibs" are you talking about carefully adjusting them or using a locking screw in the one and only location it contacts for a quick and dirty snug-up? I would think that carefully adjusted gibs, checked for proper adjustment along the full length of travel, with adjustment screws then locked in place, should not change after one or two movements. If your gibs are adjusted with set screws in multiple locations as opposed to being a tapered gib with one adjustment, you may find that nylon locking insert type set screws are preferrable to a standard set screw and lock nut. I have that on my list of things to do on my mini-lathe. The latter tend to make for movements after correct adjustment.

    One other thought-- did you separately level the lathe on top of your table? This is the only way to do it and requires a precision level that will show variation of about .0005 in 12" or less. Twists in the bed are a good source of variation along a length of work.

    Last edited by pcarpenter; 02-13-2008, 02:06 PM.
    Paul Carpenter
    Mapleton, IL


    • #3
      Can you define just what you mean by "micro accuracy" (in fractions of an inch) Its easier for folks to tell you what will and won't meet your expectations if we know what they are.I would really like to have the ability to intentionally machine within a tenth or three, yes.

      When you talk about "tightening gibs" are you talking about carefully adjusting them or using a locking screw in the one and only location it contacts for a quick and dirty snug-up? Carefully adjusting them for what's possible on this machine. The gib adjusting screws on the main ways are locked in place by placing another short screw on top of them after adjusting them. Has to be the poorest possible design.

      I would think that carefully adjusted gibs, checked for proper adjustment along the full length of travel, with adjustment screws then locked in place, should not change after one or two movements.I do dream of such dependability.

      If your gibs are adjusted with set screws in multiple locations as opposed to being a tapered gib with one adjustment Main ways has two points approx 5" apart. Cross slide has a tapered gib, remaining cross slide adjustment has to be done with lock down screws.

      One other thought-- did you separately level the lathe on top of your table? This is the only way to do it and requires a precision level that will show variation of about .0005 in 12" or less.A 12" No.98 was used from the ground to the headstock.

      Twists in the bed are a good source of variation along a length of work.That was the purpose for the added foot under the middle of the bed.

      I do appreciate your advice. Please tell me more about your "To-Do" list for your machine. Also, what machine do you have?
      Last edited by brokenarrow; 02-13-2008, 02:46 PM.


      • #4
        Micro accuracy... whatever that means... microinches?

        Anyway... That kind of accuracy will not be found in a taig or sherline. Micro-accuracy will require Macro-rigidity. Think Monarch 10EE or Hardinge HLV on the medium size and some of the smaller heavy cast iron lathes like the Rivett 608 and Hardinge Cataracts. Heavier is better. The Hardinge and Monarchs came with 30 to 50 millionths runout in the spindle. Just how accurate do you want to get?


        • #5
          The best advice I can give is that turning to .0001 is very difficult under ideal temperature controlled circumstances. I am sure that some of the professional machinists will attest. More importantly, the necessity of that is rare. If you really need it, you might start shopping for a Monarch EE in really good shape...and then make sure your shop really is climate controlled.

          I actually have a 13x40 import lathe that gets used most of the time now days, but kept the mini-lathe that I bought maybe 8+ years ago. I just haven't made that many more improvements to the mini-lathe after the larger one came along....but the gib screw change was on the list. The mini-lathe is less subject to twisting in spite of the fact that its just plopped on a table. Its stregth to length ratio makes it rather rigid in spite of its diminutive size. I also never worried about turning to .000x precision on that lathe.

          I would do a tune up on what you have. It may be that the center foot you added will make things harder to level correctly. My 13x40 has 6 feet under its base which made levelling a real pain. I finally realized this, levelled using the outer 4 points, and then brought the inboard pair of feet down to touch and re-checked level. A Starrett 199 series master precision level is graduated 5 *tenths* per graduation and not 5 thousandths like the #98. If you were off by 5 thousandths and set the master precision level in place, I think you would find the bubble completely off scale.

          However, I was talking to a local tool and die maker a week or so ago as I was looking at a used lathe he had. I found .012" dip in the ways and we started talking about the importance of levelling. He went on to talk about his rather large lathes and he said that they are quite level, but still not perfect. When he has to turn to a few tenths across some length (rare) they still have to know where to "bump" the compound to take out the variation that exists in the ways--some due to wear some due to imperfect conditions. He also said that temperature changes play hell with that sort of precision....but its just not often called for. Its easy to get wrapped up in that sort of precision for precision's sake...and forget that its not often needed. As he pointed out, most things that need that sort of precision are centerless ground.

          If you want to see some runout, check your three jaw by clamping on a ground dowel pin or round ground tool blank. There's just all sorts of locations for runout and not all will affect the precision you seek, but many will...and its not just *your* machine. Few lathes will easily give you the precision you seek.

          You can't use the cross slide reasonably to adjust depth of cut in tenths. The better bet is to use the compound set to a known shallow angle and then calculate the amount of cross slide advancement that is then needed to equal .0001 inch.

          The tune-up you may want to do on the flat gibs is to pull them out and make sure you have "dimples" correctly ground in (using a carbide burr or a drill point) where the set screw points will hit. Be gentle as this is the weak point in a gib. Toss the set screws you have and use either nylon insert locking types, make some of the same by cross drilling longer screws of the same type and inserting a dot of string trimmer line, or just hold a piece of monofilament fish line in the hole when you thread in the screws (a tip from a guy on this BBS). These methods will give you a friction locking set screw without the need for a lock screw on top. I would also make sure that the points on the screw heads are turned nicely and then the point blunted with a file. Having an out of round point will make for difficult adjustment on the gibs.

          If there is room, you can drill and tap, adding more gib screw locations for more even pressure on the gib. Be sure to add dimples to the gib with a drill bit or carefully with a burr.
          Paul Carpenter
          Mapleton, IL


          • #6
            Accuracy is relavant to the cost and rigidity of the machine.

            It would have to be a very expensive lathe or mill to take .0001" off with each succeding cut and have an extremely fine surface finish with little or no tool marks. Any tool marks at all will give an in accurate size.

            Even with a round nose tool and the compound set to 6 deg off the axis and each .001" of the compound I don't think many lathes would really take off just .0001" per cut.

            I would like to watch a manual lathe that can take .0001" per pass and measure it after each pass. I guess for enough money you can buy such a lathe. Maybe $50,000 or less perhaps.

            Paul, that is a good post and right to the point. Making .0001" cuts per pass is best left to NASA shops and climate controled die shops. The rest of us should be happy to take .0005" per pass if posible and most lathes are happy to do .001" per pass.

            A lot of people on these sites are real anal about getting .0005" or .0001" per pass and in the real world it won't happen. That's why real machinists use files and emery cloth or grind a finish on a part.
            Last edited by Carld; 02-13-2008, 04:56 PM.
            It's only ink and paper


            • #7
              A couple other thoughts popped into my head. I hope Evan will pipe up as he has mentioned a project in an old thread where he needed that sort of precision...but its still rare. What made me think of that was my reaction to his work in that case....just done on "an old 9" South Bend". But then that brought up another point:

              Some of the most precise lathes have plain bearings (run on an oil film) and not ball bearings. Little esoteric stuff like this all goes into gaining on great precision.

              Other little tunable things in the persuit of precision:

              A 4 jaw carefully set up is likely more precise than a three jaw. I found that the three jaw that came with my lathe was good for .0002 runout if I carefully worked around all the screws rather than just one...where I got .005 or so as I recall by tightening just one. Thats a big improvement, but many 3-jaw scroll chucks will not do this. Likely mine will not either with some wear on it.

              Careful tailstock alignment means a lot, and few do it. A fixture that goes in the chuck and revolves an indicator around a center in the tailstock is perhaps the most precise way to check this.

              Maybe we can turn this thread into a listing of things to do when seeking low tolerances, even if .0001 is not very realistic in most cases.

              Paul Carpenter
              Mapleton, IL


              • #8
                Many interesting questions and much to be learned in the pursuit of accuracy.

                Some observations and thoughts:

                - It seems easier to achieve accuracy in a lathe than a mill. That's an over generalization, but it has been my experience.

                - Technique is hugely critical. There are so many ways to do everything. HSM's don't necessarily have anyone to teach them good techniques either. Do we turn between centers or use a steady often enough? Or do we just slap it in the 3-jaw and git 'er done? There are lots of rigidity tricks to learn, whether that be to increase workholding rigidity or tool rigidity. For example, take you boring bar in at as much of an angle as the bore allows. That's a tip from Machine Shop Trade Secrets.

                - RE taking 0.0001" cuts, that's an interesting problem. My accuracy improved when I quit taking really small cuts. I use CCMT carbide inserts, and they don't like too fine a cut. Accordingly, I shoot for 5-10 thousandths to be my final finishing depth of cut. That worked out better than trying to dial things in with the compound at an angle.

                - Tooling matters. I've found the most difference in boring bars. The imports I've used stink compared to quality solid carbide bars. There is just no contest. Seems like that particular op is extra touchy about rigidity, or maybe its just me. I've ditched my OEM 3-jaw in favor of a Buck Adjust-Tru 6-jaw for most jobs and that made a big difference.

                - If you can't measure to much finer accuracy than you're cutting to, you are doomed. If you want to cut to 0.0001", man you better have some nice measuring tools in the old toolbox. Convenience of measurement also matters. DRO's are a good thing as are attachments to put indicators on tailstocks or other parts of your lathe.

                - Machine tune up. I've noticed a difference in gibs for sure, but also in terms of the lube status of the ways. My machines run more accurately when the ways are freshly lubed. I manage to do it at the beginning of every project for sure, but a one shot oiler is on my project list. On the gibs, I notice you really have to experiment to find the "happy medium". I'm sure there is a lot of tuning that can be done on all these machines. Whether that's an exacting tailstock alignment, headstock alignment, column tramming in 2 dimensions, and all the rest is up to your knowledge and skills, but it will matter.

                - I totally agree with Macono about the Macro-Rigidity issue. When comparing machines, weight is an important consideration. More is better. Lately I am very interested in the idea of filling hollow castings with an epoxy granite mixture to increase rigidity and damp vibration. It's not hard to do and I have talked to some folks who've had success with this. I intend to try it out on my mill at some point.

                - Experimentation matters. Try to give yourself time to experiment with accuracy alone. Make a perfect cube, accurate to whatever degree you're able. Try making yourself a cylindrical square in the lathe. Experiment and keep a notebook on what you've done.



                Try G-Wizard Machinist's Calculator for free:


                • #9
                  There are a couple of different dynamics involved here. Jerry Kieffer and other very talented machinists work to very small dimensions. Some of them are featured in the Sherline Craftsmanship Museum;


                  While several of them do use Sherline equipment, it is not a requisite for inclusion.

                  The Sherline machines are capable of this degree of accuracy as are almost any watch or clock making machines. However, it is the machinist, not the machine that will ultimately determine the results.
                  Jim H.


                  • #10
                    Brokenarrow, Have you seen or performed any of the "tuning" that's suggested in various 9x20 and 7x12, etc. articles?

                    They include things like checking gibs for flatness, flattening if necessary with abrasive sheet on a very flat surface, careful adjustment of gibs, backlash on feed screws (leadscrew and cross slide) if available or mods to improve the backlash if not provided.

                    Have you "run in" the headstock and changed the oil (perhaps a few times) until it stays clean?

                    Another thing that may help is to use way oil on all sliding surfaces. It helps to reduce slip-stick.

                    Proper gib adjustments on the cross slide and carriage, adjustment of hold downs for the carriage and removal of backlash should get you close to where you'd like to be.

                    One last thing that comes to mind, as long as the headstock is relatively vibration free, is to vibration mount the motor and use good quality belts. A VFD and three phase motor would be icing on the cake but isolating other vibration sources will improve the surface finish so that it has less impact on diametrical tolerances.

                    Last edited by nheng; 02-13-2008, 08:50 PM.


                    • #11
                      All good posts, only thing I have to add about things jumping around on you some --- when you get down to tenths its important to have your tools sharpend accordingly, very very sharp with proper angles -- without this factor you can actually have a great machine but have mixed results due to varied pressures to make the tool work properly -- a tool bit that takes too much pressure to make it work properly and then if you push it just a slight bit more it gobbles more than the last added or a tool that is a little too hungry upon contact and removes a slight extra to begin with -- this can not only be tricky to dial in, it can make a good machine look bad as it give varied results.


                      • #12
                        Minimum depth of cut, as alluded to by several members, is controlled by the edge sharpness of a tool. Only a properly lapped tool is going to take .0001" off and actually do any cutting. Most tools will just rub and kick up some dust. I've put chips from .000100" under a microscope versus dust from .000100", and there is a tremendous difference.

                        Misalignment of the tool, a dynamic structural loop, and temperature are the three biggest items that will cause error in any normal shop. All three are not something adjustable in a home shop environment unless reconditioning machine tools for high precision work is what you consider your hobby.

                        Saying .000100" accuracy by itself is meaningless. You can create a cam that has a constant .000100" diameter, but does not resemble a cylinder at all. So do you want that .000100" diameter, or .000100" cylindricity? Maybe you just care about .000100" taper in 6" inches. Or you don't care about taper at all and just want the part to be circular within .000100". You need to start getting real specific about what you want down near .000100", because it's easy to provide SOMETHING that meets a .000100" requirement, but isn't what you intended.

                        Temperature is something worth mentioning by itself and somewhat in depth. Temperature is something that virtually nobody understands in a shop. I've seen people turn the AC to 68 degrees in inspection rooms and suddenly feel they're a world-class inspection facility, when that doesn't control anything. Much like the issue of "what do you mean by .000100" is the issue of "what do you mean by 68 degrees?" How much does it vary, and over what period? There is going to be a long period sweep as the day progresses, and short period waves as the AC kicks on and off. What about stratification - the temperature difference between the top and bottom of a room is usually several degrees. Are you handling the part? Is the machine generating heat unevenly, i.e. from motors? Incandescent lightbulbs put off a lot of radiant heat, which does change the temperature of the part and doesn't show up on the thermostat.

                        With AC, your shop might wander by five degrees or so, depending on where in the room you stand. That's .000060" of your .000100" right there, in pure, unmeasured drift. In steel. For aluminum, you are screwed. You have .000110" of dimensional error drift in that environment. Then figure you've got .000050" in a 72 degree shop, away from the 68 international number. So you go look up the coefficient of thermal expansion for A36 steel or whatever, and see it's 6.5x10^6 per deg F. Except that's just a ballpark, and you have a +/- 20% tolerance on that value. So you can't "do the math." You have to set up a controlled drift check with a ground specimen to figure out how much it expands and contracts. The thermal conductivity is also important.

                        The most accurate machines in the world are very lightweight, and ride on active damping equipment. They are lathes, either using single point tools or running a grinding head. They usually run on aero-or-hydrodynamic bearings. Not plain journal bearings. Plain journal bearings and rolling element bearings are poor choices for high precision work.

                        In conclusion, get what you can out of your machine but don't ever fool yourself into thinking you can work to .000100, even if your micrometer says you nailed it. Your micrometer is just reading .000100, it doesn't mean you actually have it down.


                        • #13
                          Odd that

                          All or most of the comment thus far seems to be correcting things in a static state without due regard for the dynamic state in which the machine operates.

                          Things don't have to out of dynamic balance by much for it to be a real problem in a dynamic state.

                          Have you ever looked at how very small weights on car/truck wheel rims make a huge difference in the dynamic state when the tyre fitter runs the wheel up to speed? Did you notice how and why he puts the safety guard over the wheel and tyre? Ever notice how little is needed - on either or both sides of the wheel/rimto correct that "out of balance" tyre/s on your car/truck? Not much is it to cause such a big problem?

                          Notice how both sides of the wheel are balanced?

                          Notice how the small weights come nowhere near making the wheel rotate until the small weight is at the bottom of rotation when in the static state?

                          You know all this - right?

                          You carefully balance your surface and cylindrical grinding wheels too - don't you? Even 3", 5" 6" or even 7" running at say 2,000RPM in a HSM shop let alone a real big industrial wheels running flat out.

                          And you are super-careful about rigidity and low speeds when using a boring head in a mill aren't you?

                          Then why only deal in statics for your lathe that might be running at 2 - 3,000 rpm?

                          I won't get into harmonics here - that is a lot of work - and has been dealt with very well previously in the last 12 months. Suffice to say that it can have enormous effects and complications.

                          A 3 or 4 jaw chuck would have to be the most out-of-dynamic balance bit of equipment you have got. The "gunk" trapped inside the chuck has a lot to do with it as does uneven and unbalanced mounting screws and bolts on the chuck and backing/flange-mounting plates.

                          Centrifugal and centripedal force/s are dealt with well enough in:

                          A better approach for HSM work is in my Edition 27 of "Machinery's Hand-book" - commencing at page 179.

                          The chuck can be considered to be a cylinder.

                          The centipedal/centrifugal force in pound = 0.000341WRn>2

                          This might not seem much but the force varies as:
                          - directly with increase in weight;
                          - directly with increase in eccentricity;
                          - as the square of the change in speed (RPM)

                          Self-evidently, changes to weight and eccentricity might not be much and neither might low speed (RPM).

                          If the speed were changed from say 100RPM to say 1,000RPM for a smallish job the n>2 at 100 becomes 100>2 = 10,000 and at 1,000 n>2 becomes 1,000,000.

                          Now, any change in any of the quantities are becoming very significant.

                          Now if the chuck, and work are perfectly dynamically balanced then all will be OK.

                          Each component of the job from the movable chuck jaws and the mounting bolts and swarf etc. will or can be be off-centre from the body of the chuck and their separate effects are additive. This is more to do with the centre of gyration of the separate parts.

                          If the chuck and job have almost any centrifugal force the chuck, even if set "dead nuts" in the static state will gyrate in a new circle at ever increasing error as speed is increased. Just like the weights on car tyre.

                          So what to do?

                          Remove the chuck and use collets - preferably R8 which have minimum mass and are small and fit neatly into the lathe spindle.

                          That is why smoother running is achieved with collets.

                          Further, if there is sufficient centrifugal force the whole damn lathe will start to move in harmony with the reciprocating force. The cross slide and top/compound slide will certainly move as well. And if the force is large enough the whole carriage will be injected with a "dither" which may overcome static friction (aka "stiction") and cause the cross-slide lead-screw to try to oscillate within the back-lash of the cross-slide lead-screw.

                          Keeping to a "tenth" (of a "thou") on a lathe means accurate movement of the cross-slide of half that (ie "half a tenth") to get the diameter within a "tenth".

                          While I agree with the "set the compound rest to get 1 tenth feed in for one thou feed along the top slide" I don't agree with it in practice at that level of accuracy (ie diameter to a "tenth").

                          (The angle set for 1 in 10 = Asin 0.1 = 5.74 degree).

                          Setting the infeed this way is only a case of setting it and thinking/hoping it will "work" with the cross-slide moving "in" (or out for boring) by 1/10 of the top slide lead-screw setting. I reckon it is never checked to see that that is what the tool is advanced.

                          If it were me I would either set a very accurate Test Dial Indicator (TDI) on the tool post near the tool or use a very accurate Digital Read Out (DRO) to confirm the tool movement.

                          I would tighten every gib screw on the top-slide as well, or as I think John Stevenson does, remove the compound slide altogether and fit a rigid spacer block in lieu of it for those reasons.

                          The "last bit" can be removed the very fine emery cloth which is "blunt/ed" - or even blunt "wet and dry" or better still, some of the pot-scourer pads from the Kitchen.

                          So, for high speed high accuracy work - use collets.

                          For medium speed - just be careful in your choice.

                          For low speed work - use pretty much what ever suits.

                          And don't EVER, EVER apologise for having or using a mini/micro mill/lathe, or "el-cheapo" 3-in-1 as there are some superb works done on these small machines by model-makers, watch-makers, Instrument, Tool and Die Makers.

                          For small - or any - work that requires high accuracy of size and finish, it is very hard to beat a reasonable tool-post grinder.


                          • #14
                            A damn fine post!!

                            Originally posted by toastydeath
                            Minimum depth of cut, as alluded to by several members, is controlled by the edge sharpness of a tool. Only a properly lapped tool is going to take .0001" off and actually do any cutting. Most tools will just rub and kick up some dust.


                            In conclusion, get what you can out of your machine but don't ever fool yourself into thinking you can work to .000100, even if your micrometer says you nailed it. Your micrometer is just reading .000100, it doesn't mean you actually have it down.

                            Thanks again TD for a very fine precision, focussed and "on the money" post - I really did enjoy it.

                            It seems that we were typing at the same time. I had several goes at mine between doing other things (Nope - nutthin' to do with "Getting lucky" on Valentine's Day either - at all!!).


                            • #15
                              I agree with what has been stated here and especially what Bob pointed out....which may seem contrary to what I was trying to say. I don't know that I would every advocate a .0001 cutting pass. Likey, any given pass would be larger than that. Rather, the idea of using the compound set at angle is to enable you to more carefully dial in any given depth of cut +/- .0001. Even then, as others have stated, this is not the realm of operation of most equipment...including some pretty good stuff on a good day when the mooon phase is right and you're holding your tongue just right dial in .0001 on a micrometer and look at it....a piece of notebook paper is .003 so this is a tenth of a third of that....Now why do you need that from a lathe?

                              This might be a worthwhile discussion from the standpoint of knowing how to wring out all you can when you need to, but the initial presumptions of working to .0001 precision day in and out is not very realistic with very many lathes. I think having to fret that last tenth all the time would get very old very quickly. Its not a centerless grinder after all.

                              Paul Carpenter
                              Mapleton, IL