How To Broach Square Holes

Created: June, 2008
Click on photographs to view in more detail



Today, most prop drive washers are fitted to the crankshaft with a tapered split collet. In days past, it was not uncommon for manufacturers of model engines to use a square spline for this task. This is not a bad idea, but it is a bit labor intensive to produce both parts and prop drivers made this way have a habit of falling off when there is no prop and prop nut present. This can result in the dreaded lost prop drive washer and the need, years later, for an engine restorer to wonder how they can make a suitable replacement. This how-to describes a simple yet effective way to produce square holes in steel and aluminum using (or perhaps we should say, "abusing") nothing more than a standard center lathe and a piece of simple to make tooling.

Generally, the dimension across the flats of the square section of the shaft will be the same as the diameter of the threaded section. In some cases, the corners of the square spline will be rounded slightly. What we are going to do is use the saddle of the lathe as a shaper and nibble out the square hole one corner at a time, indexing the work through 90° for each series of cuts.

Our tool will be a piece of round tool steel the same diameter as the shaft end which has had two flats at 90° to each other ground onto it for a short distance. You could also use a piece of drill rod allowing the flats to be filed in before hardening and tempering. To assist cutting, the end of the tool should be faced off at about 85° to the axis and the flats should be relieved at about 3° to reduce rubbing as the tool passes through the work. After forming (and heat treating if using drillrod), stone the two faces and the end of the tool to make the two cutting edges as sharp as you can.

This photo shows what we need to start broaching. On the left, we have the prop driver blank which is finished and has a hole drilled through its center the diameter of the shaft end (and our broach). Below that is a "pot chuck" which will hold the driver in the lathe chuck. The outer flange will rest against the chuck jaws preventing it from being pushed into the chuck by the shaping forces. The inner flange prevents the prop driver from being pushed into the pot chuck by these same forces. The saw-cut allows it to grip the driver tightly for indexing. The last part is our finished broaching (or shaping) tool, fitted to a carrier that can be gripped in the lathe tool post holder. I'm actually going to use a key-way slotting device that bolts to the lathe saddle, but the same job can be accomplished by mounting the tool in the tool post holder and winding the saddle back and forth with the carriage hand wheel. This is perhaps a bit unkind on the rack, but If we make the cuts no nore than 0.002" to 0.003" per pass, all should be well. You will feel what is an easy cut and what is a bit excessive easily enough.

After setting the job in the chuck we need some way to index it fairly accurately through 90° three times. One way is to use the four jaw chuck and place a block of wood under each jaw and the lathe bed, thus using the chuck itself as an index wheel. Another is to accurately scribe two diameters at right angles to each other onto the back of the driver and align these with the corner tip of the tool. Each of these requires that the chuck be prevented from rotating after each angle has been set. As there is no radial force applied to the job, usually simple belt (or gear) tension will be sufficient. The third way is to use a headstock dividing attachment which is it's own anti-rotation device. All will provide sufficient accuracy, but I'll use the latter as I happen to have made one which gets little use, so I'm always happy to justify its existence!

The last job is to align the cutter so that the axis of the tool is aligned with the lathe centerline, as is the corner of the cutting edges. This photo shows this job being accomplished with the help of a lathe dead-center. Of these, getting the axis of the tool precisely on the plane of the lathe centerline is the most important alignment as otherwise, the nibble will be skewed. But rotation is important too as we will index the cutter towards us with the cross slide and a moment's thought should reveal that the result will be a bit graunched in the center of one flat if the cutter tip is not on centerline.

We also need to know how much to advance the cutter to form the corner correctly. Good old Pythagoras will tell us this. Let's say the hole for the shaft is 1/4" and that that is also the distance across the square flats. The corner to corner distance will be the square root of two times 0.25 squared, or 0.354" We subtract the diameter, then halve the result to get the depth of each nibble: 0.052". I suggest going a little shallower, say 0.048" allowing us to clean up the last bit with a needle file, but you can go all the way if you want.

The photos above show the progressive result, left to right, as the corners are nibbled out. Start by jiggling the tool neatly and cleanly into the hole and set the cross slide dial to zero in this position as a reference. Withdraw the tool and wind the cross slide two or three thou towards you, then wind the saddle towards the headstock. The resistance will feel like almost nothing, but the amount of metal being shaved off will increase as the depth of the cut increases, progressively increasing the push required. If you make the cut too shallow, the tool will probably just deflect and rub. Too deep and it will jam, and deflect as well. The tool is thin, so it will spring back no matter what you do. This is why it is important that the edges be sharp. A bit of trial end error will give you the comfortable numbers. As you near the end, make several strokes at the same setting until the tool stops cutting and is just rubbing. Repeat three more times, indexing 90° between cuts and you are done.

Another good reason for not cutting to the theoretical depth is you can't do a trial fit of the shaft until the last corner is shaped. If your hole is too big across flats, it's all a bit late by then. But if it's too small, you can decide if a few more strokes are needed on each corner, or if it can be done better and easier with a bit of filing. A sloppy fit may be ok and probably be close to what original equipment manufacturers produced, but as Craftsmen, we can not tolerate anything less than perfection!

This shot shows the setup used to produce the hole in the drive washer at the top of the page (for the Morton M-1). Both the key-way slotting attachment bolted to the lathe saddle and the indexing attachment fitted to the lathe bull-wheel were made from kits bought from Hemingway Kits in the UK. The indexing attachment is a GHT design with many of the parts removable and common to his saddle mounted Universal Dividing Head. He sure designed great tools and while making the device is quite a project in itself, the end product is worth it as both a tool and a piece of pure model engineering. But neiter are necessary and perfectly acceptable results can be achieved with the lathe alone.

That's all there is to it. If you are making a replacement for a part with a blind broach (the back end entry is square, but the front end exit is round), just broach all the way through. It will really make no difference. If the corners of the shaft spline are rounded, don't grind your tool to a point and your hole will match the spline. Finally, prop drivers fitted in this way result in comparitively high torque concentrations and rely on the driver being pushed against the end of the larger diameter end face of the crankshaft by the prop nut. This is still a rather small area of contact, so if the driver is small, and is also the cam for ignition points, finish it by case hardening all over. The result will be as good as the original.



Back to the Construction Techniques Index Page


This page designed to look best when using anything but IE!
Please submit all questions and comments to