Post By toolmaker35
Squaring parts without fixturing
There was a recent thread where the OP wanted to know how to square parts on a surface grinder. I mentioned how to do so without any fixturing (other than a magnetic chuck) and thought that there was an older thread illustrating this, but couldn't find it to save my life. The actual practice of this is to find out how much error exists on a perpendicular side (how out-of-square), then grind that amount of error into the side facing the wheel to counteract the initial error and produce a true 90* angle between those sides. I've been asked about this before in the shop, so I figured that I would draw some illustrations of this for reference to anyone running into this problem in the future.
Fig 1 shows a true square, where all 4 sides form 90* angles. This is what we are shooting for. Length of sides in this example are irrelevant, although the length of the sides will need to be accounted for on an actual part to proportion the amount of stepping needed.
Fig 2 shows what typically happens when we grind 4 sides of a square, 1 of each parallel to it's opposite. This is just a matter of grinding one side, then flipping the part 180* and grinding the opposite to it's mate. Parallel, but certainly not perpendicular to adjacent sides.
In Fig 3, you can see that after a measurement is made using a cylindrical square (or equivalent) and indicator, that there is an actual number to work to. This can be used to gage the amount of total "step" that needs to be ground throughout the surface of the side facing the grind wheel. This will give a total amount that needs to be ground away while stepping the wheel down. The white line indicates the amount that the block was originally ground to (84.289*), while the yellow line shows where it needs to be ground to to produce a 90* angle.
In Fig 4, the yellow line is still present for reference as in Fig 3, but now the white line is shown in a series of steps, each of which produces peaks along the length of the yellow line thus producing a 90* angle to the side. These peaks will be the only points that are perpendicular to the side, but they will create a datum point to use afterwards. After this series of steps are ground, then the part can be flipped 180* placing the steps down toward the mag chuck so that the side facing the wheel can now be ground parallel to them. The part can then be flipped again to grind the steps away, producing two parallel sides with a third side perpendicular to both of them.
I've used this method before on very picky parts where my fixturing didn't have the tolerance that was required on the parts that I was grinding. This is kinda long-winded and probably elementary to some, but I thought that it may be of some use to someone.
This is a very good method, and one I've used as well, but I grind only one step rather than a whole series of steps.
When I've finished with this correction grind, I end up with a thin ridge on one edge of my part that is tall enough so the part rests on it and the opposite corner, such that the part sides are vertical when the ridge and corner are touching the surface plate just as you do, but I've only got two corners touching so mine's more fragile but faster to tune.
I check the verticality by using a height gauge and an indicator with either a mag cylinder square and tenths clock, or by touching each vertical side in turn with an indicator in a height gauge at the top of the vertical side while the base of the height gauge touches the bottom of the vertical side.
Comparing the readings of course, gives double the error of each side.
I can fiddle the step I grind a tenth at a time if I want, until the verticality on the surface plate is as perfect as I've got the patience for.
I can also stone the edges of the step to tweak it the last tenths.
If I'm impatient, I'll sometimes just burnish the edge I want to drop a tenth by putting heavier pressure on the high side and bearing down on the workpiece while rubbing it on the surface of a crappier surface plate (but we're NOT going to endorse abusing our fine reference surface plate like that, are we!!)
Once I'm dead nuts, I can set the stepped side down on the mag chuck and kiss the other side, then flip again and grind away my step.
If all goes well, I can get awfully close that way, and the flatness and parallelism of my ground surfaces start to become a bigger source of error, so time to man the scraper or the lapping plate.
Once I've got a master cube that's as square as I can make it and measure it, other shop tools like angle plates can easily be ground very close to square just by clamping to the cube.
It's a great old-school method, not all that well known: thanks for posting it!
Implant Mechanix – Design & Innovation - home
Vancouver Wire EDM -- Wire EDM Machining
Forgive my ignorance - I've never actually ground anything (though I did buy a surface grinder which I am storing...)
In both approaches above, the workpiece surface to be ground is always facing "up" (180* away from chuck) - right? And we're depending the magnetic chuck (or equivalent) to give us a flat reference plane?
Yes, I do depend on the chuck, but not in the same way as I would for a surface plate. You brought up a very good point in this, actually. The chuck is typically ground in after mounting to the grinder so that it's surface coincides with the movement (ways) of the grinder; i.e. the top of the chuck is mirroring the grinder's ways. Usually, if the grinder is in good shape, it's ways have been ground or scraped to such a degree that their error in an application such as this is negligible, but not always. This is just a case-by-case situation where the person grinding has to know what kind of error their machine produces over a given distance. If the machine in question does have some measurable error present that cannot be tolerated, it can be figured into the sum of the stepping process that I described above.
Originally Posted by bryan_machine
Let's say, for example, that the cross-slide on a given machine "drops" .001" across a given part, that is to say that after grinding two sides 180* apart (top and bottom), they are not parallel by an error .001". The side facing the operator is actually +.001" taller than the opposite side. This can be seen easily on a surface plate with an indicator, but with an indicator mounted on the wheel head on the grinder, it would still register zero across the whole face of the part since the movement of the table is "sloping" downward by .001". By knowing (by indicating from a surface plate) that there is .001" error present and in which direction (+.001" toward the operator side in this example), that same .001" of error can still be added (or subtracted, depending on how the part is oriented) to the amount that the wheel has to be plunged into the part to correct the perpendicularity that's being chased down to begin with.
In the above example, one way to get around this out-of-parallel problem (if tolerances allow it) is to simply rotate the part 180* on the same face that it was just ground on while keeping the grind wheel at the same location. You're still keeping the same face toward the mag chuck, just re-orienting the part front-to-back. Another pass across the part will split the difference of any errors that were present to begin with. It will appear to grind on roughly half the width of the part, then spark out. This isn't necessarily fixing the problem, but moreso just averaging out the errors. I've chased my tail more than once with this exact scenario. It's just a matter of keeping all of the known variables in check while grinding, or even milling.
Hello Marcus. Yes, just grinding one step is how I was shown to do it several years ago, but I noticed more than once after placing the stepped side down toward the chuck & fencing it in, that I was getting some chatter across the face while correcting the opposite side. I was (and still am) running a variable field magnet turned down (as little distortion as possible), so I attributed it to not having enough surface area in contact with the chuck to stabilize the part. I tried it with several smaller steps after that and it seemed to provide more "meat" to hold the part steady while I ground the back, but that's dependent upon the size & mass of the part. You are correct though about it being more time consuming and fussier to do. It's also an open opportunity to screw up & get off-track just that much more. I don't do much gage work anymore, so I rarely use this method for squaring nowadays, but it is nice to know about. It let's you get that last couple of tenth's out of a part that would normally be a major pita. You're also spot-on about having to deal with flatness issues at this level. When tolerances get down to this, everything comes into play. Having the patience to walk away from a job while the part's temperature settles out can be mind-numbing sometimes.
Originally Posted by implmex
I guess you could do parts this way.
But at the end of the day for production it makes life much easier to have square blocks to work off of.
Horses for Courses
Very true Cash, but this is for when you're too cheap to buy the goodies, or you really really really want to roll your own.
Don't forget we toolbreakers love to fiddle fart around building things that we can admire and go " Man I'm good!!"
All joking aside, it has been a most useful technique for me when I had to get something as accurate as I possibly could, and I do have all the goodies including granite squares, grinding vises and angle plates, so I'm not scraping bottom with either my setup or my technique.
When I need to do it, I charge appropriately for it, and no one seems to complain.
Normally of course, I do it like everyone else does too; but for that special project or that plate that's just too big for my biggest angle plate, I like to have it in my back pocket.
Implant Mechanix – Design & Innovation - home
Vancouver Wire EDM -- Wire EDM Machining
Last edited by implmex; 05-28-2012 at 10:19 PM.
Reason: Frickin stupid program re-formatted my post and won't let me change it.
yeah, clamping to an angle plate seems to be the easiest for us. I think the biggest concern is to ensure you are doing a good job of grinding, and to make sure your grinder's magnet is flat. I've had a part clamped up to a square before, and removed it to find it was only square within 0.001. ground it again, same problem. Next time, didn't unclamp the part, and ran an indicator over the ground surface, which showed 0.0007 or so taper. This was across a 6" D2 piece btw.
Clamping to a fixture is what I do as well 99% of the time. Sometimes, though, the fixture doesn't give you the accuracy that you're after. That's where this method shines. It allows you to squeeze that last couple of tenths out of a part that can be such a pain to get. When I was shown how to do this the first time, the toolmaker that was training me told me to grind the part as I illustrated, with no fixturing at all, to prove how it works. Obviously, it was out by a mile from just grinding the sides parallel with no reference to go off of, which meant that I had to step the top out quite a bit to square it up. The ideal situation is to rough grind from a vise, knee, ect., then step the sides out to get where you need to be. If the fixtures used are in decent shape, that should leave very little to step out for the final finishing. The last time that I did this, I made a rectangular surface plate square similar to the one Hermann Schmidt used to make. Lapped all 4 sides in for flatness, surface ground the bottom pads from a knee, then step-ground the top to square the pads from and lapped them in. I sent it to our cal lab afterward to be checked, & the inspector said it was closer than he could reliably measure (within .0001") over the length, about 6 1/2". Not too bad for a knee, old Parker-Majestic grinder and a lapping plate. I wish that I could've kept that one for myself.