1/3 Scale Ford 289 Hi-Po

[mayhugh1]

A pulley turned from wood was bolted to the crankshaft flange on the tester. What I was about to do next deserved a little more thought, but I was anxious to see the starter in action. With the tester clamped in my workbench vise and the Nichibo motor installed, a five pound barbell weight was hung from the pulley on a 24 inch cord. As soon as contact was made to a 12V battery, the weight shot several feet up in the air nearly wiping out the tester on its way up and me on its way back down.

With new respect for these little motors I realized the measurements I'd been planning were going to be more difficult than I thought. The combination of a five pound weight, 4.4 inch pulley, and 24 inch lift height turned out to be a poor first choice that left me with no time to react.

Hanging the tester from a rafter in my shop added some reaction time, but it also increased the energy with which the weight would fall. A bump stop bolted to the bottom of the fixture provided it some protection, but could have used a hard hat and steel-toed boots.

A few more tests made it clear the weight would need to be significantly increased in order to slow the test down to a manageable pace. Instead of collecting measurements to fully characterize the motors, I decided to focus on a single operating point at my original crankshaft target of 2.6 ft-lbs @ 200-600 rpm.

Fifteen pounds hung from the pulley is equivalent to 2.75 ft-lbs of actual crankshaft torque. When divided by the 11.5 gear ratio, this torque referred back to the motor's shaft is .24 ft-lbs (or 3.3 kg-cm) which is close to the maximum specified output of the Nichibo 775-8511FDS. To meet my original torque spec the motor needed to 'wind the 15 pounds up on this pulley, and neither motor had a problem doing this.

The corresponding rpm can be calculated by measuring the time taken to wind the weight up a known distance. A 60 inch travel would require 4.3 pulley revolutions, and in order to meet a minimum 200 rpm cranking spec this needed to happen in 1.3 seconds. This measurement was hard to make in my setup because of the insane pace of the test, but short videos using both motors and a fully charged battery showed the 60 inch travel times coming in under two seconds.

I was glad to find both motors were essentially identical, but I also learned the inexpensive Nichibos are no longer available from Jameco. Its part number though is available from other manufacturers and can still be found on eBay. My only reservation with either of my two motors is their 70 amp current draw at my expected operating point. This high current results from running the motors at operating points of extremely poor efficiency. The roughly 73W of mechanical output power comes at an electrical power cost of 735W for a net efficiency of only 10%. Additional gear reduction would improve the efficiency but would likely also reduce the cranking rpm.

A survey of currently available 775 size motors for which I could find data sheets showed high torque 12V dc brush motors are becoming hard to find probably because applications for them are moving toward brushless and/or high voltage battery applications. The 775-9011F-R-NF, 775-7013F-R, and 755-9510-R should perform much better than my two current motors, but I haven't been able to locate a source for them in small quantities. The Mabuchi RS-775WC-8514 looks promising, and a couple have been ordered through eBay.

The Mabuchi motors will be tested when they come in, but until then I'll continue with the Nichibo. The bell housing and both motor adapters were finally cleaned and Gun Koted. Gun Kote's 'brushed stainless' is a close match to aged cast aluminum and will protect the bell housing from grease and oil stains. Finally, a flywheel bottom cover was fabricated from 24 gage 304 stainless, and this finally wrapped up work on the bell housing. - Terry

16 Jun 2023 Edit:

The Mabuchi RS-775WC-8514 motors arrived. Except for a larger shaft diameter their form factor is identical to the John Deere motor, and so its adapter was reused. The larger shaft diameter required a 13 tooth pinion to be used rather than the 12 tooth (addendum modified) pinion used on the John Deere motor. However it mated perfectly with original 18 tooth idler.

Testing showed the Mabuchi motor wound the 15 pound weight up in my test rig in about the same time as the Nichibo motor, but it used 10 amps less current. Additional testing showed serious voltage drops in my test set due to insufficient wire size that should have been accounted for when calculating the operating point efficiency. The Nichibo's terminal voltage was actually 6.5V rather than the reported 10.5 volts at its 70 amp current draw, and so its efficiency was actually 16%. Curiously, with its terminal voltage of 7.5V and 60 amp current draw the Mabuchi's efficiency was essentially the same. - Terry

 

[gbritnell]

As always Terry, just truly outstanding work and documentation!

 

[mayhugh1]

While wrapping up a number of 'minor' details related to the design of crankshaft so work can begin on it, the Mabuchi RS-775WC-8514 motors arrived. Other than a larger shaft diameter, their form factor is identical to the John Deere motor, and so its adapter was reused. The larger shaft diameter required the use of a 13 tooth pinion rather than the 12 tooth (addendum modified) pinion used on the John Deere motor. However it mated perfectly with the adapter's 18 tooth idler.

Testing showed the Mabuchi motor wound the 15 pound weight up in my test rig in about the same time as the Nichibo motor, but it required 10 amps less current. Further testing revealed voltage drops in my setup that should have been accounted for in the efficiency calculations. In my setup the Nichibo's terminal voltage was actually 6.5V at its 70 amp current draw rather than the reported 10.5 volts, and so its efficiency was actually 16%. Curiously, with the 7.5V terminal voltage at its 60 amp current draw the Mabuchi's efficiency was essentially the same. - Terry

 

[mayhugh1]

A handful of small parts still remained in the way of starting work on the crankshaft. A model of the crank (less its counterweight details) was created last summer and verified in a SolidWorks assembly that included the block, pistons, rods, and camshaft. Placeholders were used for the nose and rear end of the crank since the parts associated with them hadn't yet been designed. With the flywheel assembly finally completed, the crank's rear end could be finalized and the remaining associated parts created.

The first step was to machine the caps that secure the outer ball bearings to the block. A commercial lip seal was added for rear oil control and a custom bronze retainer attached to the rear of the block to control the flywheel's forward thrust. As mentioned in an earlier post, the flywheel's rear thrust is limited by a retainer on the clutch sleeve. The final design of the crank's rear end was verified using all the actual rear end components trial fitted to a test rod in the block with the bell housing and oil pan installed.

A retainer secured to the front of the block locates the front main bearing. An accessory pulley attached to a keyed damper will be secured to the nose of the crankshaft, and behind that inside the timing cover is the keyed drive sprocket for the camshaft. This sprocket must align with the camshaft's driven sprocket, and so moving on to the camshaft ...

Similar to the crankshaft, the camshaft's outer bearings are shielded ball bearings while the inner bearings are polished journals spinning directly in the 7075 block. Slots machined into the block beneath the camshaft will promote splash lubrication for both the bearings and lobes.

The distributor will be driven from the camshaft through a shop-made 45 degree helical gear set. These gears have already been designed and the block machined for their pitch diameter. The driving helical gear will live between the ends of the camshaft which in my shop means the camshaft will have to be machined in two pieces and spliced together. The camshaft's tenon'd ends will be Loctite'd and joined together inside the helical gear. I first ran across this splicing technique in George Britnell's inline six camshaft, and this will be my second use of it. I'm considering applying it to the crankshaft as well.

The camshaft's short front section includes a hub for the front bearing and the bolted-on driven sprocket/timing adjuster. The rear section has all the lobes and bearing journals. The cam's front section was machined and trial-fitted in the block's front cam bearing so the driven cam sprocket could be positioned and the crank's driving sprocket located to it.

I wound up painted into a corner with my camshaft design, and the front ball bearing and its retainer will wind up captured between the spliced sections. A few helical gear blanks were also machined so I could practice assembling a dummy camshaft to make sure there weren't any more surprises lurking.

In order to finalize the crank's nose design the crank damper, pulley, and water pump pulley were machined and their alignments verified using the crankshaft test rod in a trial assembly of the block, timing cover, water pump, and their gaskets. Keyways were machined into the test rod so the parts could be assembled using their actual keys. Along the way, the timing cover required minor modification to clear the timing chain which hadn't been a component in my modeling. The one-third pound damper was machined from steel, and the pulleys were machined from aluminum. All were Gun Kote'd black. A not yet installed shaft seal inside the timing cover will provide front oil control. - Terry

 

[mayhugh1]

I hate making crankshafts ...

The odds of machining a ten inch five bearing solid crankshaft that spins freely in a block are stacked against a builder without a dedicated grinder. I've adapted two different tool post grinders to my lathe, but when it comes time to using them on real crankshafts the lathe is always in its own way.

The alternative is an extended reach parting tool. But even when bifurcated, the cutting pressures of its requisite wide blade create part deflections that result in oval'd journals and maddeningly inconsistent results. Precision filler blocks help stiffen the crank during machining but significant deflections still occur. For me things get even more interesting since the bed on my lathe has a twist that I've never been able to correct that will add a 2-3 thou error of its own to a ten inch pat.

Most of the material on a crank's workpiece is removed asymmetrically from around its axis during machining, and the internal stress changes create unintended dimensional changes. Stressproof is the material of choice but not a total cure. And then there's the frustrating issue that a micrometer typically won't fit down into the main journals so they can be properly measured. When it's all said and done, for me a 10" crankshaft is going to be many hours of frustration and many more hours scraping a set of custom bearings.

End of rant ...
Terry

 

[mayhugh1]

Although it may be a swap of one set of problems for another, the current plan is to machine the 289's crankshaft as four separate pieces. Its tenon'd ends will be coupled together inside Loctited sleeves doubling as the main journals. A custom alignment fixture will be used, with each bearing providing a 're-straightening' opportunity.

A parting tool will still be needed to machine the rod bearings, but the deflections and lathe errors will be better controlled on short sections. With the option of pre-fitting each section inside the block, the entire crank won't have to be scrapped because of a single machining error.

Joint integrity is a question though. The ultimate strength of the joints are limited by the shear strength of the tenons and the tensile strength of the sleeves. Their failures include the shearing of a tenon or the breakage of a sleeve. The Loctite prevents the sleeve from spinning on the tenons, but it doesn't resist the joint's ultimate failure. Its real purpose is to fill the tiny (one thou) internal gaps between the tenons and the tenons and sleeve to lock in the alignment of the joined parts.

Some joint sanity checks were in order. Two test parts were made up using Stressproof rounds joined together inside drill rod sleeves. The main journal dimensions .625" o.d. by .400" long have already been established in the finished block. The test parts' tenon diameters and sleeve bores were selected around a .572" reamer that I had on hand. This left the sleeves with a wall thickness of only .026" which in retrospect was a bit thin. The parts making up the splices were carefully machined for a one thou fit-up.

A corner radius was left on each tenon to reduce localized stresses, and the sleeves were beveled to slip over them. According to its data sheet Loctite 680 reaches its ultimate bond strength after a one hour cure at 200F or after a seven day cure at 72F. All internal surfaces of the test part were coated with 680 and cured for 3 hours at 200F.

After curing, the ends of the first test part were machined so it could be held in a vise and twisted to failure with a torque wrench. Failure occurred at 60 ft-lbs with breakage of the sleeve. A crude calculation in one of the photos done to reconcile the failure shows the sleeve actually failed at a 2x higher torque than expected. The calculations also show a wide separation between the failure of a tenon and that of a sleeve, and so a thicker sleeve would be in order. Actually, 60 ft-lbs of torque is equivalent to some 35 hp @ 3000 rpm which is approaching the crankshaft requirements in a small automobile.

The plan for the second test part was to check its TIR after applying a number of twisting cycles below its total failure point. However, I got too aggressive with the machining of its end for the torque wrench, and the second part failed in the crash with a broken sleeve.

A third test part was made. This time Stressproof was used for the sleeve which was bored in a set-true chuck to .496" nearly tripling the wall thickness. While set up, I made a number of identical sleeves for later use with the crankshaft. The third test part was assembled using a slightly higher strength Loctite (638) and cured as before.

This part was torture tested with a hundred +/- twisting cycles each at 20, 30, and 40 ft-lbs. The sleeve's half thou TIR measured in a v-block never changed. The third part was finally taken to failure and the sleeve broke at something over 90 ft-lbs.

At this point I feel good enough about the splicing technique to attempt a pieced crankshaft. I made the alignment fixture which along with the already completed sleeves leaves just the 'fun' parts to machine. - Terry

 

[peterl95124]

Terry, yes I agree crankshafts are model machining hell. nice experiment, but I'd stick with one piece. I use two separate long reach carbide lathe bits, one left and one right, and both of them have sharp points so they don't chatter, and don't cause much if any deflection because with a sharp pointed bit there's little contact area and little being taken off per revolution. then of course the downside is having to polish out the machining marks left by a pointed bit, working through the grades of wet-or-dry paper (with oil), glued to a *wooden* popsicle stick so that when it catches on a crank web it doesn't break anything else :-( !!!. make your crankshaft first, then if any journals are off size its trivial to make a matching bronze bearing (so far I've been lucky and not had to do that, but I still stamp my bearings with what crankcase slot they go in anyway). I did have trouble with the first couple crankshafts where I didn't use a fully pointed bit, and the journals didn't come out perfectly aligned and concentric, but a bit of garnet lapping compound on the bearings and a few turns of the shaft (had to do that twice on the extra long Merlin shaft, but only once on the 4-cyl test engine) and everything was silky smooth after cleaning off the lapping compound and using light oil.

YMMV, but I still think a built-up crankshaft is an accident waiting to happen, also even if you engineer with a large margin to your materials YS, you haven't done an analysis that shows how these parts deflect under load; I'm guessing it will be different from a solid one piece C/S, and probably not in a good way. again just my wild-ass-guess, but possible food for thought.

(BTW, I always first lap my bearings with a separate solid rod (ie not the actual C/S) so they are perfectly aligned and concentric, because being able to move the lapping tool both in-and-out as well as rotating makes for a proper lapping, when lapping with the actual C/S you can't get the in-and-out AKA end-to-end motion and its not "proper" lapping, but works anyway if you're pretty close to start with)

(PPS, I've been using 1200 grit alumina for lapping recently, instead of 800 grit garnet, and it is a noticeable improvement, probably never going back, again YMMV, would like to hear what other folks have experienced)

 

[GreenTwin]

I would also stick with a one-piece crankshaft, but this is coming from an armchair observer who has never machined a crankshaft before.

I would not go with the grinder method, but would use the extended tooling method, and perhaps make a very beefy tool holder somehow.

To each their own, but that would be my approach; one piece at all costs.

Good luck with whatever method you use.

.

 

[petertha]

I'll be very interested to see what direction you go, Terry. Just to provide some added food for thought:

I asked some questions about Hirth coupling on the other forum, link below. That joint method is employed on a few of Jung's engines. I have the geometry figured out in SW & even bought the 60-deg cutter, but have not made any prototype joints yet. You will see some FS engine reference to Hirth in the link which is rather interesting.
https://www.modelenginemaker.com/index.php/topic,10107.0.html
Maybe you already know, but the lap joint has some precedence in model engines. OS used it on the Pegasus. Bruce Satra used in on his opposed 4-cyl opposed engine. I have some plan snips I could share. There might be others but they escape me right now.
https://craftsmanshipmuseum.com/exhibit/satra-0-440-4-cylinder-opposed-model-airplane-engine/
In the MEB link above is also some reference to Schillings methods. He used a few different variations of segmented CS: bolt-together/coupled, straight shaft throws with external thin bearings, a pinned joint method somewhat equivalent to the lap joint. The (German text) book is quite obscure but there are a few pics & sketches here & there to ponder.

The Porsche 917 work in progress project on MEM has an assembled CS.
https://www.modelenginemaker.com/index.php/topic,10504.0.html

 

[minh-thanh]

A parting tool will still be needed to machine the rod bearings, but the deflections and lathe errors will be better controlled on short sections. With the option of pre-fitting each section inside the block, the entire crank won't have to be scrapped because of a single machining error.

Terry

View attachment 148382

Hi Terry !

I have a question, you have made several crankshafts from one piece and you have succeeded - the crankshafts you have made them are amazing - Is there any other reason for you to decide to make crankshaft from many pieces besides the above reason ?
Thank you !

 

[mayhugh1]

Hi Terry !

I have a question, you have made several crankshafts from one piece and you have succeeded - the crankshafts you have made them are amazing - Is there any other reason for you to decide to make crankshaft from many pieces besides the above reason ?
Thank you !

Minh-thanh,
It's not my only reason. You might have missed my 'rant' post #125 just above my post with the photos. - Terry

 

[Kimmo]

Hello,Terry
Could you explane why torque goes through the sleeve? I´d thought the flats would take the torque.

 

[Steamchick]

Hi Terry. I think you are doing some amazing work! Well done.
But on the built-up crank assembly I have doubts.
First, I understand your machining limitations - accuracy, due to the twisted lathe etc. Been there and still try to fix mine!
Second, I don't know much about fatigue, except seeing some Weibul analyses... which were on log paper.
I.e. Putting it as simply as I can... halve the stress and the failure will occur at twice the number of cycles.
So considering your failure was at 60Nm. ... = 1 cycle. Equivalent to proof stress of the joint?
A stress-fatigue consideration would be:
30Nm = 10 cycles, 15Nm = 100 cycles, 7.5Nm = 1000cycles... etc.
Guessing you'd like say more than 10 hours running at - say - 3000rpm-ish average speed? = 1800000 cycles? Call it 2 million cycles.... = 1 Nm max stress.
Or that order of magnitude?
You could estimate the stress when running (peak, not average) and gestimate life = number of cycles from that?
This is very crude!
Do not believe it, as I am really trying to help by suggesting you do some fatigue-life estimates based on proper fatigue theory.
It would be a real shame to get it wrong and destroy many hours of excellent workmanship for selecting a design that fails in service.
I don't know enough to say it will or won't fail.
Just expressing a few words of caution.
K2

 

[Steamchick]

Kimmo, - just a guess.... The failure occurs because the applied torque is not purely rotational.... There is some bending moment applied as well. The combined torque and bending stress has (probably?) sheared the flat surface joint, when the thin sleeve has been cut by the two bits of shaft???
Just a guess...
K2

 

[Steamchick]

Or maybe.... the torque has caused the loctite to fail, as the sleeve is then expanded by one bit of shaft levering against the other, the sleeve has simply burst as the hoop stress is exceeded?
But I favour the shafts being twisted out of alignment by bending stresses affecting the joint...
K2

 

[Charles Lamont]

Interesting. Generally it sounds OK to me and definitely worth trying. I think a test in modes other than torsion might be worthwhile. One idea: make a joint with a shaft extension, put one end in the lathe chuck and a gash ball bearing on the outboard end, spin it and apply a radial load with the toolpost. Also, on test piece one, the Loctite is too thick in parts, but maybe that is just on the semicircular ends. Personally I would aim for less than 0.001" thickness of the adhesive.

PS. Thinking a little more, a press fit, still with Loctite, might be better. Some preload hoop stress in the sleeves could be a good thing.

 

[Steamchick]

Doubling the sleeve thickness may double the failure torque... ?.
Making the sleeve an interference fit on the pair of half shafts may pre-stress the sleeve, but help resist play that could permit the shearing of loctite?.
Or a silver soldered or brazed assembly?
The half-shaft and sleeve joint is designed to give the required assembly accuracy, so a stronger sleeve, and better jointing "Bond", may well be adequate?
Can you make the sub-cranks longer/fewer? Say 3 pieces, a 3 journal timing end piece, 5 journal middle and 3 journal drive end piece? Fewer joints, less risk of misalignment?
K2

 

[Steamchick]

Better still, 2 pieces of crankshaft, with a larger journal at the joint main bearing? Depends on how long you can make a crank "within spec."?
K2

 

[nj111]

Terry,
I'm wondering what the downside is of a simple pressed construction, as per motorcycle crankshafts? You could easily make all the parts in the home shop. For example see the genius British engineer Allen Millyard adding cylinders to a kawasaki crankshaft in link below. (Hope its ok to paste a link?) Allen has done this for decades in his small home shop and in the last couple of years has shared how he builds his engines on his you tube channel. Eg, a 4 cylinder becomes a six etc. His engines develop a lot of power and are often used to the max without issues. Just might be another approach for you to consider.



or search for

Millyard Kawasaki S1 Four Cylinder Crankshaft - How its Made - Episode 2​

 

[Kimmo]

Thanks Steamchick,
Bending could be the reason. How about needle roller bearings inner sleeves on the joint? They could double as a main journal for plain bearing.