Accuracy in my shop is a variable thing due to what I am trying to accomplish. I don't have time to be **** when doing some things but if we are talking bolted stuff, the end result will be a product of machining accuracy. For weldments, a VERY different story (refer to BukitCase for some).
#1 thing I find what wanders is saw accuracy. Spend a LOT of time to get the blade guides and vise to allow a square cut. Do this by making reference cut in square tube (marked for what side is up and 90 degree side with paint marker) of large wall thickness, then roll it over 180 degrees and cut a coupon off, then measure the taper - you will now see why you need 2 x markings to remember what side is what. I cut a fair bit of large steel in my 7 x 12 so 2 more things to consider: the feed rolls need to put the part onto the vise are dead level (i.e. square to the blade vertical cut). I keep a big hex key in the pivot point so I can tighten before clamping. I scribe 0 and 45 reference lines on my table for fast setup.
I die and scribe when machining - but a fine point sharpie will do well for fab/welded joints.
What was mentioned about cutting short and spacing a weld gap is NOT an option, it is a 100% requirement. When you **** up a joint and weld it, the distortion from the weld bead and HAZ (it will be smaller in ALL dimensions when it cools) WILL give you an off-measure angle to the joint and it WILL build in excess stress that will weaken the joint unless fully stress relieved (that not many people are set up to do in any way). If you want things accurate and strong, you MUST fit up the weld joints with a gap - and again as BukitCase pointed out, tack/measure/tack/measure/etc. before making the structural pass(es).
I also do a lot of bolt circles and have on drill press table set up with one row of 24 and one of 10 holes. Once a BC OD is scribed in two short arcs 180 apart with dead center in drill chuck then measured to verify, I will use dial indicator to shuffle table to get BD dia. on spec. Then, an automatic center punch for the first layout hole, verify with calipers and then LIGHTLY with larger CP in line with arm holding table (so it can't knock if off level). Then all holes are piloted with center drill using the table index holes as needed. Final hole must use a bit with flat web smaller than center drill hole OD to keep it centered. I do a lot of BCs that I drill size on size and seldom have any issues, so my tolerance is kept to nominal bolt size hole vs. actual bolt diameter with ease. As others have mentioned, if you need it accurate you need to think things through and take your time. Yes, I could do a bolt circle with dividing head in mill every time (or I could just go out and buy a digital mill) but I can turn out sufficient accuracy with great repeatability and efficient time in a dedicated press.
I also make my own overhead cranes (just bought my first two factory-built ones last year, will go in new shop) but there are things there where accuracy is not just nice to have, it is deadly critical to safe use. To get final numbers that fit, I design most joints with shims to allow for adjustment. It is so difficult to get large weldments that accurate, nor are 20' measurements that easy to land on the button it solves the problem of getting fits and alignments on the button. I thought I was cheating until I met a company that makes a very large quantity of parts I will not identify. Their competitors bolt final pieces together on the as welded pads from their production tooling (talking about the LARGEST manufacturer of same in the world). These guys put EVERY final part (stressed and dynamic) on with a shim stack - and find that they really need to do a high-tech digital laser alignment on EVERY part to get the numbers they need. I felt vindicated in my "cheating".
#1 thing I find what wanders is saw accuracy. Spend a LOT of time to get the blade guides and vise to allow a square cut. Do this by making reference cut in square tube (marked for what side is up and 90 degree side with paint marker) of large wall thickness, then roll it over 180 degrees and cut a coupon off, then measure the taper - you will now see why you need 2 x markings to remember what side is what. I cut a fair bit of large steel in my 7 x 12 so 2 more things to consider: the feed rolls need to put the part onto the vise are dead level (i.e. square to the blade vertical cut). I keep a big hex key in the pivot point so I can tighten before clamping. I scribe 0 and 45 reference lines on my table for fast setup.
I die and scribe when machining - but a fine point sharpie will do well for fab/welded joints.
What was mentioned about cutting short and spacing a weld gap is NOT an option, it is a 100% requirement. When you **** up a joint and weld it, the distortion from the weld bead and HAZ (it will be smaller in ALL dimensions when it cools) WILL give you an off-measure angle to the joint and it WILL build in excess stress that will weaken the joint unless fully stress relieved (that not many people are set up to do in any way). If you want things accurate and strong, you MUST fit up the weld joints with a gap - and again as BukitCase pointed out, tack/measure/tack/measure/etc. before making the structural pass(es).
I also do a lot of bolt circles and have on drill press table set up with one row of 24 and one of 10 holes. Once a BC OD is scribed in two short arcs 180 apart with dead center in drill chuck then measured to verify, I will use dial indicator to shuffle table to get BD dia. on spec. Then, an automatic center punch for the first layout hole, verify with calipers and then LIGHTLY with larger CP in line with arm holding table (so it can't knock if off level). Then all holes are piloted with center drill using the table index holes as needed. Final hole must use a bit with flat web smaller than center drill hole OD to keep it centered. I do a lot of BCs that I drill size on size and seldom have any issues, so my tolerance is kept to nominal bolt size hole vs. actual bolt diameter with ease. As others have mentioned, if you need it accurate you need to think things through and take your time. Yes, I could do a bolt circle with dividing head in mill every time (or I could just go out and buy a digital mill) but I can turn out sufficient accuracy with great repeatability and efficient time in a dedicated press.
I also make my own overhead cranes (just bought my first two factory-built ones last year, will go in new shop) but there are things there where accuracy is not just nice to have, it is deadly critical to safe use. To get final numbers that fit, I design most joints with shims to allow for adjustment. It is so difficult to get large weldments that accurate, nor are 20' measurements that easy to land on the button it solves the problem of getting fits and alignments on the button. I thought I was cheating until I met a company that makes a very large quantity of parts I will not identify. Their competitors bolt final pieces together on the as welded pads from their production tooling (talking about the LARGEST manufacturer of same in the world). These guys put EVERY final part (stressed and dynamic) on with a shim stack - and find that they really need to do a high-tech digital laser alignment on EVERY part to get the numbers they need. I felt vindicated in my "cheating".
