Looking for some scraping tooling tips

Please keep in mind that I'm just a hobbyist who barely knows what he's talking about at the best of times. I'm a long way from an expert! Please read all of the following accordingly,
Disclaimer noted! If you are a hobbyist, then I have no idea what that makes me :p


The angle of the bevel on a straightedge isn't important, you just want a narrow enough angle that it clears going into dovetail ways. You only use one side or the other as the flat reference to mark up the ways. You want both sides of the angle scraped flat because the bulk of the straightedge will interfere with the work in one orientation or the other, but the actual angle just needs to be smaller than the dovetail angle. You only print and scrape one side of the dovetail at a time.

Even the exact angle between the mating dovetails on the work doesn't really matter. Who cares if it's off a degree or two as long as they are mating perfectly and nothing gets tighter or moves off-axis as the table slides? In practice, you usually scrape the moving part to match the fixed part, first getting the fixed geometry right with a straightedge, then bluing up the fixed part to mark up and scrape the moving part.
Interesting... I'll have to think this through. I assumed your straightedge would be used to print both surfaces at the same time thus defining flatness and angular accuracy. I guess, without knowing better, I thought you would scrape both mating sides of the dovetail and the gib separately and then assembly. Perhaps a bit of light fitting would be needed then, but the flatness and dovetail angle would be set on the bench.

If the exact angle of the dovetail is unimportant, then it would be on the gib to take up for the discrepancy in mating angles. I guess thats OK.

Is there a standard to scrape the male or female dovetail first, or just what is convienent?

Getting the actual geometry right for sliding prismatic or dovetail ways is far too complex to write about in a paragraph or three, refer to the Connelly book mentioned above for a full explanation (or if you just need a nap).

I have that book in my eBay watch list, but ouch! They are expensive!

Despite my joke above about needing every measuring tool, the absolute best tool to own (I don't, sadly) is the one Richard King's dad designed, the King-Way alignment tool. It's a clever arrangement of precision levels, sliding joints, indicator mounts, and hemispherical and ball feet to assist in measuring any misalignment of machine tools. It's not a mandatory tool, but you'll be cobbling together custom jigs without one.

The King-way was designed with sound kinematic principles. Not every jig I've made has been!

Looked it up. Pretty interesting. I immediately see the use in lathe ways, but don't yet see how it measures dovetails. Need to think more.

I think I'm replying to multiple people here, but in my opinion, the only truly mandatory metrology tools are a (hopefully calibrated) surface plate, a decent surface-gauge/indicator-stand, a 0.0005"/graduation dial test indicator, and a precision machinist's level (on the order of 0.0005"/12in per graduation). I'd really want to add a 0.0001"/graduation indicator, too, but it's only useful for the particularly persnickety stuff. Other things like feeler-gauges/shim-stock, parallels, 1-2-3 blocks, and jo-blocks/gauge-pins can also come in very handy.

Got most of that. Short the gage blocks and pins.

Lastly: either a 20mm or 25mm blade is fine for scraping flat, unobstructed surfaces. In general, the wider the blade the easier it is to handle, but the more likely you'll have trouble fitting into tight spaces. The compound on a small lathe will require a much narrower tool to get into the dovetail than 20mm. I use a 20mm blade for almost everything, but a 10 or 15mm blade when I need to get into tight spaces. A 10mm blade is much more difficult to handle well — it's all too easy to tip the blade accidentally and catch a corner. (It's worth rounding the corners of all your blades for exactly this reason.)

Gotcha. Do you have a source for the smaller tools or is it more of a braze your own kind of deal?
 
I assumed your straightedge would be used to print both surfaces at the same time thus defining flatness and angular accuracy

I had the same misconception when I started. The order of operations depends on the specific circumstances of the work, but a straightedge is not intended to be an angular reference, simply a "portable surface plate" you can take to the work when it's not feasible to bring the work to the plate.

Without attempting to write a full-blown "how to scrape machine ways" treatise (Richard King's 5-day class is really the gold standard and best way to get an introduction to the topic), here's a mini-overview:

If I were scraping in the cross slide of a lathe. I'd first look for an original, unworn horizontal way surface (the mating cross-slide usually doesn't ride on the full surface, so there's usually some unworn areas). This gives you a clue to the original geometry when the machine was new.

I'd then scrape one of the two horizontal surfaces on the saddle flat and parallel to the original plane (most likely the way on the other side of the saddle since scraping will destroy the original surface). Once you get it flat, you check for parallel with an indicator: reference the base on the scraped surface (parallels and 1-2-3 blocks below the base come in handy here) and place the indicator tip on the original surface. If there is tip in either the X or Y plane you'll need to "step scrape" to correct the error. (Step scraping is too long a topic to cover here, but is doubtless described in at least one YouTube video somewhere.)

Once one horizontal surface is parallel to the original surface, I'd scrape the other horizontal way to be coplanar (not just parallell) to the way I just scraped.

Then I'd start in on the dovetail on the saddle. I'd start on the fixed side before dealing with the gib side. Again, I'd look for any unworn surface for clues before diving in, with a goal of minimal metal removal in the most worn areas. I'd then use a pair of dowels or gauge pins to mic across the dovetail from front (operator side) to back, writing down readings every inch or two. The gib side should be completely unworn (any wear should be on the gib itself) but it's worth a sanity check with the straightedge to ensure it's also reasonably flat.

Note that my lathe uses flat steel, not tapered cast iron, gibs, so I used the gib-side as a rough reference. With tapered gibs, you'll need to first eliminate any bend in the gib and then scrape the outer surfaces flat (but tapered). You'd then mic across the dovetail with the gib in place, scraping as necessary until both sides are flat and parallel in X (across the bed ways).

The exact angles do not matter and they don't even need to be the same, you just want to ensure that there is no pinching when the cross-slide moves fore or aft.

Once I mic the same all the way across the four saddle way surfaces (two horizontal and two dovetail sides) with good PPI and truly flat surfaces, I then apply spotting ink to those surfaces and use the saddle itself as a reference for the cross-slide.

You then scrape in the cross-slide to match the saddle.

First, you'd scrape the two bottom surfaces flat at the same time (without the gib in place, of course, when you mark up). Once both sides are flat, you want to check the geometry to ensure the top surface where the compound rides is parallel. If there is any discrepancy in Z (along the bed ways) or especially in X (across the bed ways) you need to step scrape to correct the error.

Once the bottom surfaces are finished, you start scraping the dovetails on the cross-slide. First the fixed side, then the gib side as before. Since this time you need to mic an interior dimension, I use a stack of fixed and adjustable parallels between the dowels, and mic across them to ensure they are scraped parallel.

Once the saddle, gib, and cross-slide are all fully scraped in, lube, assemble and do some system-level verification/calibration to ensure everything operates as expected. Everything should slide smoothly with no pinching or significant misalignment as you move things from one extreme to the other. Connelly does a very good job of explaining how to do this.

If things aren't exactly aligned as expected, choose the surfaces that will create the least amount of work to make the correction. For example, if the cross-slide wasn't moving exactly perpendicular to the spindle axis, it's smarter to correct the saddle-to-bed-ways surfaces than re-scraping the cross-slide because the contacting surfaces are small and far apart from each other (giving you much finer control and creating less work).

I have that book in my eBay watch list, but ouch!

It's about $100 new from Dapra. Textbooks are always expensive, but I consider mine a tool purchase. Whatever you do, don't buy from any of the evil people on Amazon trying to scam people out of several hundred dollars for a copy.

Got most of that. Short the gage blocks and pins.

I consider a set of gauge pins up to 0.500" indispensable tools. You'll find them FAR less expensive than gauge blocks and they suffice for >99% of any tasks a hobbyist might face. I use them for setups at least as often as I use them to measure bore diameters. I pretty much never need finer than 0.001" granularity for setups or measurements (gauge blocks let you go down to 0.0001"). One surprisingly obvious tip: you can measure a bore up to 0.999" diameter with a set of pins up to 0.500" by inserting pairs of pins into the bore. Add two more pins (1.000" and 1.500") and you can measure up to 1.999" in one thou increments.

Do you have a source for the smaller tools or is it more of a braze your own kind of deal?
The Biax I bought on Ebay years ago came with a couple usable blades. I also bought a few new ones from Dapra (after searching the couch cushions for a long time to find some spare change). It's absolutely possible to braze carbide chips onto pieces of mild steel, shape them on a green wheel, and hone them with a diamond wheel, but I'm fortunate enough to be at a point in my life where I can afford to buy rather than build most of the tooling I need. As you've discovered it's also possible to use HSS scrapers, but you'll spend a lot more time sharpening.
 
Last edited:
a long time to find some spare change)
Don't let him kid you, Rex is loaded :)
He is also a very established machinist and rebuilder.
I've seen his work. The Logan he scraped in is perfect.
 
I bought the Sandvik hand scraper for Richard King's class.
Based on my limited experience, the Sandvik in it's original form, enabled me to scrape in the test block Richard gave us to learn on. It only took a few hours.
I may remove some material from the shaft before I scrape my Rucker 9".
Maybe put a channel down the center to allow for a little more flex.
Once Richard let me loose with a Biax, I never went back to that hand scraper. It's too bad they run $3,500 and up!!
I think if I ever got the idea of scraping in a machine I wanted to save, I would invest in a Dapra-Biax.
For small stuff, the hand scraper is adequate IMHO.
 
Don't let him kid you, Rex is loaded :)
He is also a very established machinist and rebuilder.
I've seen his work. The Logan he scraped in is perfect.
Hmm...

I can't help but point out how easy it is to mistake impulse-buying for wealth, brash confidence for competence, a single rebuild for experience, and pretty surfaces for perfection.

Jeff's shop is full of several big-boy machines I can only lust after, and he knows how to use them (despite his pretenses)!
 
I had the same misconception when I started. The order of operations depends on the specific circumstances of the work, but a straightedge is not intended to be an angular reference, simply a "portable surface plate" you can take to the work when it's not feasible to bring the work to the plate.

Without attempting to write a full-blown "how to scrape machine ways" treatise (Richard King's 5-day class is really the gold standard and best way to get an introduction to the topic), here's a mini-overview:

Huh. That's really interesting. So other than the inaccessibility of rotating the straightedge upside down in some instances, you don't "need" both scraped surfaces on the straight edge?

I'd then scrape one of the two horizontal surfaces on the saddle flat and parallel to the original plane (most likely the way on the other side of the saddle since scraping will destroy the original surface). Once you get it flat, you check for parallel with an indicator: reference the base on the scraped surface (parallels and 1-2-3 blocks below the base come in handy here) and place the indicator tip on the original surface. If there is tip in either the X or Y plane you'll need to "step scrape" to correct the error. (Step scraping is too long a topic to cover here, but is doubtless described in at least one YouTube video somewhere.)

Once one horizontal surface is parallel to the original surface, I'd scrape the other horizontal way to be coplanar (not just parallell) to the way I just scraped.

Following that so far! Have seen step scraping and understand what it trying to accomplish.

Note that my lathe uses flat steel, not tapered cast iron, gibs, so I used the gib-side as a rough reference. With tapered gibs, you'll need to first eliminate any bend in the gib and then scrape the outer surfaces flat (but tapered). You'd then mic across the dovetail with the gib in place, scraping as necessary until both sides are flat and parallel in X (across the bed ways).

OK so always working with reference to another surface. In this case, the fixed way which you just scraped flat.

Once I mic the same all the way across the four saddle way surfaces (two horizontal and two dovetail sides) with good PPI and truly flat surfaces, I then apply spotting ink to those surfaces and use the saddle itself as a reference for the cross-slide.

Interesting! So I figured you would scrape the cross slide against the surface plate too, but now that you've made a master surface on the other side, there is no reason to.

Is there a way with dealing with how much material you are removing? I'm sure eventually you will open it up enough that the original gib does not fit. Just make a new one? Or do Turcite?

or two. The gib side should be completely unworn (any wear should be on the gib itself) but it's worth a sanity check with the straightedge to ensure it's also reasonably flat.

Gotcha. So again, not really trying to correct the dovetail to a particular angle, but rather develop a planar surface. Probably doesn't matter if the two dovetails are angled the same either then!

It's about $100 new from Dapra. Textbooks are always expensive, but I consider mine a tool purchase. Whatever you do, don't buy from any of the evil people on Amazon trying to scam people out of several hundred dollars for a copy.

OK for that price, I think I can stomach it. I'm not looking at going full blown scraper here, but it seems like a good skill to pick up. I own import equipment which most people would suggest is not worth bothering to scrape at all, but it also makes me more willing to give it a try. I don't tend to rush projects and can be meticulous, so hopefully I can try a few easy things and maybe make my machines better than factory.

I consider a set of gauge pins up to 0.500" indispensable tools. You'll find them FAR less expensive than gauge blocks and they suffice for >99% of any tasks a hobbyist might face. I use them for setups at least as often as I use them to measure bore diameters. I pretty much never need finer than 0.001" granularity for setups or measurements (gauge blocks let you go down to 0.0001"). One surprisingly obvious tip: you can measure a bore up to 0.999" diameter with a set of pins up to 0.500" by inserting pairs of pins into the bore. Add two more pins (1.000" and 1.500") and you can measure up to 1.999" in one thou increments.

I love measuring stuff. My first job was in metrology and I just want it all! Gage blocks and pins are definitely in my "next to be purchased" list.

The Biax I bought on Ebay years ago came with a couple usable blades. I also bought a few new ones from Dapra (after searching the couch cushions for a long time to find some spare change). It's absolutely possible to braze carbide chips onto pieces of mild steel, shape them on a green wheel, and hone them with a diamond wheel, but I'm fortunate enough to be at a point in my life where I can afford to buy rather than build most of the tooling I need. As you've discovered it's also possible to use HSS scrapers, but you'll spend a lot more time sharpening.

I do have an eBay search for a biax, but A) I haven't started scraping so I feel like I need to be initiated to hand scraping first, and B) I bet the rest of you guys will snatch them up!
 
Thanks again for taking the time to write such detailed responses. As of today I am the proud owner of a Sandvik scraper, and after a few more days of hemming and hawing over it, I'll probably end up with Machine Tool Reconditioning too.
 
Brilliant !
Hardly, but I had the same slap-on-the-forehead moment when I first heard of the technique. Now I use it regularly. :)

I use gauge pins almost daily in the shop for all sorts of things. Literally today I needed to compare the widths of two trepanned channels to get them to the same width (exactly 3/32"). I often find it useful to place a hole or undercut feature on a part a precise distance above the mill vice jaws — much less fiddly to use a gauge pin from the top of a hole than a depth-mic from the top of a part. Have two features you are turning on the lathe precisely 0.279" apart axially? Find the right pin and insert it between the carriage-stop and the carriage.

I prefer to use 123 blocks, gauge pins, and shim stock for setups or measuring whenever possible. I still screw up measuring with calipers or micrometers or even scales on occasion. The only tricky thing with gauge pins is to ensure you read the label from the correct row (the label below the pin, not above!) — the bigger pins are marked, fortunately.

Less useful (at least I've not had occasion to try it yet) but I read somewhere about inserting two hex keys that are half the size of the one you need if you can't find one large enough.
 
So other than the inaccessibility of rotating the straightedge upside down in some instances, you don't "need" both scraped surfaces on the straight edge?

Correct. The bevel is for clearance and access, it's not intended to be a precision angle reference. If the top body of the straightedge didn't get in the way occasionally, you wouldn't need to scrape both planes flat.

So I figured you would scrape the cross slide against the surface plate too, but now that you've made a master surface on the other side, there is no reason to.

It's often useful to double-check against a plate occasionally, and I'd probably start with a blue-up on the plate for simplicity, but as long as you are confident that the saddle is truly flat there should be no surprises scraping the cross-slide to match. Note that it IS possible to get two curved surfaces to mate perfectly. That's why it's important to get the "secondary reference" (the horizontal saddle ways) truly flat first (and to occasionally double-check against the primary reference, your plate).

Is there a way with dealing with how much material you are removing? I'm sure eventually you will open it up enough that the original gib does not fit. Just make a new one? Or do Turcite?

With a steady hand, you should be removing roughly the same amount of material each pass. It's important to measure how much you remove each pass on average in order to step scrape. You can measure the depth of an individual scrape with a tenths indicator, but I find it more useful to scrape one full pass over an entire area and determine the average depth you scrape each pass. That is: scrape a test bar completely flat. Then do another pass on just the end few inches of the bar. Next, place an indicator stand on the original surface with the indicator tip resting on a gauge block or parallel that averages out the points on the original surface and zero out the indicator. Finally, move the gauge-block over to the freshly scraped surface and measure how much lower it's gotten. Even better, do the same after five or ten passes and average the result.

Other than that, the only trick is what I think Rich calls "tip scraping." Let's say a blue up shows the two ends of the work are high and it's low in the middle. The obvious thing to do is to scrape off the blue at the ends. Less obvious is that you can leave one end alone and scrape down the other end (and eventually a bit in the middle) until it's flat. Tilting the plane this way when you know you need to remove more "meat" from one end rather than the other is often quite useful.

Re: Turcite: I'm reminded of the old joke. If you cut a rope too short you can always splice some more on, but if you cut it too short there's nothing you can do about it! The best answer is not to scrape too much (or wait until there is too much wear to correct the problem with scraping). When you have no choice but to add material, epoxying something like Turcite is a great solution (it's such a terrific way material that some machines are designed to use it from the start).

I'm with you on metrology. Fascinating topic. One of my favorite engineering expressions is "if you can't measure it, you can't improve it." Sometimes figuring out how to measure things is half the battle.

The thing about a Biax is that once you've used one it's hard to go back. I really want a power flaker myself, but I do have an old blue-bullet scraper. You're right, they've become much more expensive and less common thanks to all the YouTubers making the skill valued again.

hopefully I can try a few easy things and maybe make my machines better than factory

Depends on the factory, of course! :) But better than current is usually a realistic goal. It does take a lot of patience, cleverness, and meticulous measuring but cast iron is usually pretty forgiving. Rule 1 if you find yourself getting deeper into a hole is to stop digging. Fortunately, the digging goes quite slowly with a scraper!

Practicing is an excellent idea. Once you're confident you can make something flat, practice making the top and bottom of a test piece perfectly parallel in two axes. Once you've got something with top and bottom parallel, make one of the ends perfectly flat AND square to those planes.

Once you can do those three things confidently you're ready to start rebuilding, which is mostly an exercise in figuring out what are the reference planes, how to indicate/measure deviations from the ideal, and figuring out an efficient plan of attack. The hardest thing about rebuilding is that everything you change tends to affect everything else, so the order of operations really matters.

Somebody mentioned something to the effect of their offshore machines not being worth scraping in. I couldn't disagree more strongly. If it's made out of cast iron, it can be scraped and it's often the cheaper offshore machines that are most often in need of it. The ways on lathes and mills are the most critical things to machine things accurately. Get those right and the rest of the bells and whistles are far less important.

That said, the quality of the castings on WW-II and earlier machinery is often outstanding. If you've got a LOT of free time, you can sometimes turn a very cheap, clapped out, and rusty heap into a precision machine.
 
Back
Top