Author Topic: Hirth Coupling  (Read 1160 times)

Offline petertha

  • Full Member
  • ****
  • Posts: 352
Re: Hirth Coupling
« Reply #15 on: January 02, 2021, 08:56:20 PM »
MUCH appreciated!

Offline Elam Works

  • Full Member
  • ****
  • Posts: 69
  • Glen Mills, PA, USA
Re: Hirth Coupling
« Reply #16 on: January 03, 2021, 05:44:47 AM »
I have been considered a Hirth joint for a two-throw crankshaft project, albeit not scale model. So I gave quite a bit of thought and some experimentation to the idea, some of which might relate to your project. The goal was to come up with something I felt would be accurate enough without specialized precision grinding of the Hirth joint that I do not have, or resorting to grinding the journals true after assembly.

First, the CAD geometry. I took a more empirical approach and did not use formulas. You already have a CAD model. What you will find is that for a given number of teeth, increasing the number of serrations or increasing the depth will just cut away material at the ‘crest’. The crest ends up being a resulting edge. You do not actually want that as it is a sharp edge which will foul any root radius you will end up with. So I just set my root and crest trajectory to be the same angle, and just kept increasing that angle until cut away nearly all of the crest, leaving just a small flat (actually a conical surface.) Since the CAD model had a sharp root, but a flat at the crest, I did have to use a small offset for the vertex. Else both root and crest would end up sharp. In reality you can just rub the crest surface with some emery cloth to quickly knock a flat (or more flat) on it, though a somewhat arbitrary method.

Cutting tool. There are two approaches that I am aware of. Using a double angle cutter such as you purchased, or using an engraving bit. Both have their pros and cons. What I found was the overriding consideration was the logistics of setup and what machine tools you have available, rather than the cutting tool type.

With a dividing head, the axis of rotation is the center of the Hirth pattern. If that is on a mains journal, fairly simple to indicate the workpiece true to the dividing head. If it is something like a crank throw, maybe not so simple. You will need something to indicate. This could be the hole through for the bolt or differential stud used to clamp the Hirth joint together. Just remember to cut the threads after and not before. You will want to bore those holes so you have a nice cylindrical surface to indicate, and your ability to step off the hole centers, or distance from the main to the crankpin, will determine the precision of your assembly. If all your throws end up having slightly different strokes, the crank will run out of true quite a bit. The alternative is some sort of fixture you can mount the workpiece in that does have reference surfaces to indicate, and the ability to register the work in it repeatable and accurately. However correcting the location (changing the stroke) pretty much means loosening the clamps and bumping the workpiece (or fixture) around on the face of the dividing head and hoping nothing moves when you re-clamp. Once all that is achieved, a dividing head makes spacing off the serrations pretty simple. Accurate too, if the dividing head is accurate of course! Used with the double angle cutter it makes for a clean cut.

Once you have sorted out the workpiece, there are a few challenges with the cutter. You need to put the apex of the angle on center with the axis of the work. Normally this is done like you would for a gear cutter; touch the side off of a known surface and move the machine axis the request amount. So you will need to measure the distance from the tip of the cutter to the apex of the angle; do not assume it is precisely half the cutter width. Then touch off the work. Maybe you feel when it bumps up against the work, pinch a paper shim, or observe when it scrapes away a wipe of layout dye. Whatever your method, and the accuracy associated with it. Being off a thou’ here, is not as bad as being a thou’ off in positioning the workpiece.

The alternative is to use an engraving bit. If you have a CNC milling machine, you do not even need a dividing head with this method. On a manual machine it is a bit more complicated as you have to tip the workpiece out of plane to the spindle (serration trajectory). That makes indicating and stepping off the stoke more difficult (though not impossible with fixturing) since it lies on a different plane, so I am not sure if it is then worthwhile. At that point you might as well use the double angle cutter and that setup.

I went the CNC route! The biggest negative with the engraving bit is the cutter velocity at the tip is zero no matter how fast your milling spindle turns. Yet they work, so nothing to lose sleep over. The second is you cannot get a sharp corner to the root of the cut. But you can get pretty close (0.010 to 0.005 radius engraving bits are commercially available), and anyway a sharp corner is a stress riser! The double angle cutters are more manageable for the cutting velocity and that velocity is more uniform over the flank of the serration, but the tips blunt on those too, so do not expect dead sharp corners. It is easy enough to flatten the crest of the machined serrations, so providing some clearance for the root radius. If you are doing very small Hirth joints then yes the sweep of the double angle cutter could collide with teeth on the opposite side. They do make thread mills in very small diameters, if you accept a 60 degree included angle. I suppose someone make Whitworth and BA thread mills if you want to use those angles.

Anyway, I digress. In my mind, the biggest plus was the simplification of the setup when using an engraving bit combined with CNC, and that is what I chose it to experiment with.  With the engraving bit, centering the cutter is easy; it is the axis of the milling spindle! Even if your collet or cutter holder runs out a little, the axis is still the spindle. The cutter just orbits a little and you are going to get a slightly larger root radius than intended. Hopefully your collets (and spindle!) run true, but remember there is no such thing as absolutely accuracy. As for centering the work - well you put the indicator in the spindle and clock the reference surface just like centering up of any other job. The positional error is as good as you dial indicator, the circularity of the feature indicated, surface roughness, etc. Conventional stuff.

The workpiece just sits there on the table (ha-ha), and you use the table axis to both generate the serrations, but also set the position of the pattern. Need more or less stroke, just move the table over and define that as the center of your Hirth pattern. No loosening clamps and bumping things about. But, the accuracy of positioning is now dependent on the accuracy of your machine tool. If you step off a desired distance and the machine tool moves something more or less than that and you have no way of checking that (DRO that you trust, gauge blocks, etc.) then it is a lost cause. Say you can get the position very accurate, but you are not sure the machine accuracy is good enough to generate the Hirth pattern? Perhaps your CNC machine has lead screws rather than ball screws or it otherwise has some backlash in the screws or ways. That, surprisingly, does not matter as much. Do not get me wrong, it will not work on worn out machinery, but the positional error of the Hirth pattern will be something smaller than the positioning error of the machine. How so? The Hirth pattern is made up of multiple serrations in a radial array, each one of which has a positional error (nothing is perfect). But the axis of the pattern is the summation of all those errors. Think of it as a shot gun blast at a paper target. Each pellet flies off at some trajectory of random error, but the center of pattern is where the barrel was pointed (hopefully at the center of the target!) When you clamp up the Hirth joint, it is going to register and settle on the average of all the serrations, not just any one serration that might be out of position this way or that. We are talking small deviations. I call this error averaging; I do not know if there is an official name for it. Of course, if you have one serration that is completely out of place, it is going to mess up your average. Just like a number well outside of range will skew a statistical average.

Do all the serration cuts the same way. By that I mean start from the outside and mill in to the center. Or do them all from the center out. Do not start from the outside, mill to the vertex, and then climb out the far side. If you have backlash, doing that could shift your pattern off center. But if the opposite serration was cut in an opposing direction (converging to the vertex, or vice-versa), the backlash error is more likely to cancel each other out. Also allow a little travel for the backlash to take up. Your cutter approach distance ought to be sufficient. If not, you definitely have a backlash problem!

The other thing I liked about his setup was I could do everything I needed with the work on the table and spindle vertical. Set of the stroke, cut the serrations. I could machine the fixture, leave it on the table, and install the workpiece, etc. without tearing down or moving the setup. Every time you move the work or setup, you introduce a chance of positional error. Actually you always introduce a positional error, you just hope it is so infinitesimal that you cannot measure (detect) it! All the critical moves were in the X & Y axis, with Z basically being just the depth of cut (serration trajectory) and not quite as critical insofar as positional accuracy. With a dividing head and double angle cutter you have three axis that need precision (2 linear, 1 rotary) and a forth that is not quite as paramount (centering the cutter) and usually locked. The more axis in motion or needing precision, the more chance for error.

Fixturing. A problem with doing the work on the table with a vertical spindle, and also with the dividing head, is if you need to reference something on the other side of the work. For example, crankpins on either side of the center web, or a main journal on one side and the crankpin on the other of a throw. Fixturing can help. While not a Hirth joint, two of the attached images show a center web of a two throw crank mounted in a fixture bolted to a lathe face plate; then the bare fixture. The fixture has two holes (A & B) bored for the center distance of the stroke. I did not have a CNC mill at that time (picture taken back in the days of 35mm film!) and never had a jig bore. So I bored the two holes on a knee mill using gauge blocks and a dial indicator to get the distance as accurate as possible. Then when setting up in the mill or lathe, it was a matter of using my best dial indicator to clock the hole true. “C” indicates tapped holes for four leveling screws (rather than skimming the face plate). “D” are cone point set screws that dogged in holes in the corners of the center web that would eventually be machined away when it took on a more lozenge shape. They were offset so that running them in tended to pull the center web down tight against the face of the fixture. A potential point of error was when doing the second side, the fit of turned crankpin needed to be a snug fit if the bore. If it could rattle around in the bore, then positional location was lost. The hole for the first crankpin (B) was oversized for clearance, the center distance was critical, not the fit. For the Hirth joint, it is in some way a little better situation. You mill a Hirth joint on a fixture similar to this, step over, and set your zero for the other Hirth pattern. Or if you do not trust your mill to step off a larger distance, have a hole pre-bored in the fixture (like this one) that you can clock off from. The throw with one Hirth joint milled is going to register on the fixture accurately; or just as accurately as it would in a crank assembly. No worrying about your cylindrical fit, or locking the workpiece against rotation. The Hirth joint takes care of that. If you are doing the throw with a main journal on one side and the Hirth joint on the other (facing up towards the spindle), then your fixture is going to revert to something with a hole bored in it that you clock first before installing the workpiece. Then you are back to how accurately it fits in the hole. It could be something like a collet chuck.

Gauging. I do not know of a ready method to reliably gauge the pitch plane of the Hirth joint. However the pitch plane is not as critical as the other positional challenges, as it just alters the overall length of the crank assembly rather than causing it to run out of true. But in one picture you will see a simple fixture, lower right, for mounting a single ‘crankpin’ where the first Hiirth joint has already been machined. (The ‘plain’ fixture on the left has a conical face to match the pre-turned blank.) One of these, or a pair, could be used for gauging by measuring over the outside. Sort of like thread triangles. You could use a pair to verify relative dimensions; get all your crankpins the same length, verify you have taken an additional few thou’ off, etc. 

Anyway, a few brief thoughts…  ::)

-Doug







« Last Edit: January 03, 2021, 05:48:20 AM by Elam Works »

Offline petertha

  • Full Member
  • ****
  • Posts: 352
Re: Hirth Coupling
« Reply #17 on: January 03, 2021, 05:46:24 PM »
Very good information there Elam Works. This is exactly the sort of experience/detail I was hoping to receive. Looks like you are making good progress. Like Vixen (Mike) was mentioning, there may be other options, but pros & cons & trade-offs to each based on the application at hand.

When I was drawing up the 6-cyl CS I overlooked a dimension detail where Hirth vee starts at a specific clock angle in intermediary component. I think that's related to getting the proper throw angles, so another setup complication.

I have lots of pondering to do. Thanks for the useful information this far!

Offline dieselpilot

  • Full Member
  • ****
  • Posts: 298
Re: Hirth Coupling
« Reply #18 on: January 03, 2021, 10:53:41 PM »
The geometry is quite simple, all faces converge to a single point in the center. https://grabcad.com/library/hirth-serration-1 shows this and the formula. For CAD, the formula is not necessary and can be fully defined with sketch geometry and the number of grooves. The hard part for production would seem to be setting the beta angle and making sure the teeth pass through the center. Tolerances in profile and position will determine how much contact area there is. For a part like a turret coupling this would influence repeatability and wear. For a static joint, this means contact stress. If you look at the geometry, it should be apparent that the 60°(or 90°) profile is not perpendicular to the Z axis of the part, but the beta angle. As such, milling with a 60° tool and the part flat with Z feed to follow the beta angle will not generate surfaces which are coplanar when the joint is assembled. This can be corrected by finding the appropriate angle for the tool and the particular design, be certain it's small ~0.5° for the model I drew, but it is there.

Offline petertha

  • Full Member
  • ****
  • Posts: 352
Re: Hirth Coupling
« Reply #19 on: January 04, 2021, 02:20:13 AM »
Thanks Dieselpilot I actually saw that same link. I didn't download the PDF showing the equation but maybe I should sign up & do so. (I cant quite make out the terms just looking at the screen).

Re the 'converge at the center' - I initially assumed that was the case too, but I'm pretty sure when I joined 2 identical parts & sectioned them, something wasn't right. The facets weren't co-planar. I've deleted those models now. But I noticed the Jung drawing showed a hidden line with the same reference angle going right across the span of the cylinder face so I tried that. In this manner a vee never intersects any another. The online calculators compute the same angle he shows based on OD & number of teeth. I took some screen grabs of my steps. I'm not saying its right, its just what I did. (This is why I posted the question to begin with).
« Last Edit: January 04, 2021, 02:38:59 AM by petertha »

Offline petertha

  • Full Member
  • ****
  • Posts: 352
Re: Hirth Coupling
« Reply #20 on: January 04, 2021, 02:21:29 AM »
Then the tooth crown facet bit (my own method, again not 100% sure here)

Offline petertha

  • Full Member
  • ****
  • Posts: 352
Re: Hirth Coupling
« Reply #21 on: January 04, 2021, 02:37:10 AM »
I realize my center hole was masking the 'cut straight across the entire face' method. So in this drawing I omitted & it.
The result is that it the teeth do appear to converge at the center. Or maybe that is an equivalent way of drawing it vs how I did it?

Offline dieselpilot

  • Full Member
  • ****
  • Posts: 298
Re: Hirth Coupling
« Reply #22 on: January 04, 2021, 03:10:58 AM »
The Hirth coupling by definition results in coplanar surfaces when assembled. If this isn't the case, the model is wrong. The easiest way to prove it is by mating the cylinders concentric and one pair of what should be mating flats coincident. If all flats are coplanar, the model is correct. This can easily be checked with interference detection using coincidence.

I did mine without calculating anything. The key is that ends of a line at the half height are on the cylinder. The rest falls into place. I just used an equal distance chamfer on the crest which gave a constant width flat. Geometry is my "thing" in SW, I use it vs. calculated or dimensioned anything whenever possible. The only dimension to define the grooves in my model is the angle at half height to set the number of grooves.

Offline dieselpilot

  • Full Member
  • ****
  • Posts: 298
Re: Hirth Coupling
« Reply #23 on: January 04, 2021, 02:13:16 PM »
We had some PM about this. If the beta angle is calculated with reasonable precision(no rounding in CAD), it will result in accurate geometry. I was set on solving this problem with geometry as the half height and beta angle are directly related.

Offline nj111

  • Full Member
  • ****
  • Posts: 126
Re: Hirth Coupling
« Reply #24 on: January 04, 2021, 06:06:31 PM »
Hello Petertha

Mercedes Benz finally went over to built up crankshafts with Hirth couplings in the 1950's and used them in the world beating W196 driven by Moss and Fangio.


Mike

That's interesting Mike,   recently I've been researching the 1950's Guzzi V8 500cc race bike. An idea that went from concept to reality on just 5 months.  That bike was faster than anything at the time (circa 180mph) but suffered from poor reliability, very often crankshaft related.  Guzzi tried all sorts of crank configurations and construction methods for a couple of seasons and eventually went to Hirth in Germany who solved that problem for them.   Shortly after this the demand in Italy for small motor cycles dried up - this regular production had provided Guzzi with their race budget, and so they withdrew from racing. In that year the V8 had very nearly won a Grand Prix, it was leading by a huge margin, but near the end of the race the battery lead came off.
Nick

Offline Vixen

  • Full Member
  • ****
  • Posts: 1924
  • Hampshire UK
Re: Hirth Coupling
« Reply #25 on: January 04, 2021, 07:04:48 PM »
Hello Nick

Maybe they should have used a Hirth coupling on the battery terminals as well.  :lolb: :lolb:

I have seen the Guzzi V8 at Sammy Millers Museum and also at Goodwood. An incredibly complex engine for 500cc but maybe too complex for it's own good.

Mike
It is the journey that matters, not the destination

Offline Niels Abildgaard

  • Full Member
  • ****
  • Posts: 159
Re: Hirth Coupling
« Reply #26 on: January 12, 2021, 11:46:27 AM »
Hello Petertha

As you well know, I have a great interest in the pre-war 'Silver Arrows' Grand Prix race cars built by Auto Union and Mercedes Benz.

All the pre-war Mercedes Benz crankshafts were machine in one piece with split roller bearing for both the Main and Big End bearings


Mercedes Benz finally went over to built up crankshafts with Hirth couplings in the 1950's and used them in the world beating W196 driven by Moss and Fangio.



Perhaps the Auto Union/ Mahle way of finish grinding the journals would also help you and me achieve the required accuracy in our model crankshafts.

Stay safe

Mike

Many of  earlier Messerschmidt 109 aircrafts had DB 601 engines with solid crankshafts and divided roller connecting rods as well.
Later they got DB605 motors with slidebearings
Hirth made a lot of trainer aircraft engines HM 504 fourcylinder engines .
Crankshaft was assembled from 36 Hirth splines without roller conrods .
Ground as a single crankshaft.
Dismantled again and put inside a single undivided crankcase with connecting rods.
Some are still flying today untouched.
Nobody dares for fear of not working again

https://www.homebuiltairplanes.com/forums/media/wp_20190822_001-1.79558/full?d=1566463915

Mr Schilling mentions that he worked for a short while at a firm that made Hirth crankshafts for racing motors.
« Last Edit: January 12, 2021, 05:23:16 PM by Niels Abildgaard »

Offline Vixen

  • Full Member
  • ****
  • Posts: 1924
  • Hampshire UK
Re: Hirth Coupling
« Reply #27 on: January 12, 2021, 05:56:29 PM »

Hirth made a lot of trainer aircraft engines HM 504 fourcylinder engines .
Crankshaft was assembled from 36 Hirth splines without roller conrods .
Ground as a single crankshaft.
Dismantled again and put inside a single undivided crankcase with connecting rods.
Some are still flying today untouched.
Nobody dares for fear of not working again

https://www.homebuiltairplanes.com/forums/media/wp_20190822_001-1.79558/full?d=1566463915


Niels,

Your photo of the Hirth crankshaft components is very interesting. You can clearly see the large diameter sleeve that pulls the component parts together. The threaded sleeve (bolt?) has differential threads. One side has a course thread while the other has a fine pitch thread. The result is the ability to apply very large compression forces on the splined coupling joint to produce a built-up crankshaft, equally as strong as a solid one piece crankshaft.

Mike

« Last Edit: January 14, 2021, 11:39:41 AM by Vixen »
It is the journey that matters, not the destination

Offline JonC

  • Full Member
  • ****
  • Posts: 66
Re: Hirth Coupling
« Reply #28 on: January 13, 2021, 07:40:58 PM »
Hi everyone, watching along with interest.

I just wondered if anyone has seen the location used on smaller cutting tools nowadays. The attached photo shows a 10mm diameter shank groove milling tool and insert. Because of the light tooth loads used on these, they have to run very true and cope with some significant loads.

I just wondered if the type of Hirth couple being considered is over complicated for the application?

Jon
It's not what you know, it's not who you know.
It's what you know about who you know

Offline Vixen

  • Full Member
  • ****
  • Posts: 1924
  • Hampshire UK
Re: Hirth Coupling
« Reply #29 on: January 13, 2021, 10:09:26 PM »
Jon,

The PH Horn mini grooving tools and tool holder look remarkable like a three groove version of the Hirth coupling. They seem to work the same way to achieve both accurate alignment and good torque transfer.

I have had no luck trying to find drawings of the geometry for this coupling, but suspect it would need another specialist cutter or EDM to form the grooves. At least the Hirth coupling can be cut using a more or less standard 'V' cutter.

Mike
It is the journey that matters, not the destination