1/3 Scale Ford 289 Hi-Po

[mayhugh1]

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

Kimmo, et. al.
Thanks for the interest and comments.

What happened in my tests is that one half of the joint was held fixed in a vise while the other half was twisted until something broke. What broke in all my cases was that the facing flats of two halves opened up and applied enough pressure to the sleeves to break them. If the sleeves had been thick enough, one of the shafts in the joint would have sheared off instead which would have been the failure mode of a solid crankshaft.

Keep in mind the Loctite isn't gluing the flats together. Its purpose is to fill in any gaps inside the sleeve (all test parts were machined for a one thou fit-up except at their ends) so the Loctite's compressive strength can eliminate any relative movement between the parts inside the sleeve after curing. It's other purpose is to keep the sleeve from spinning over the joint, and its more than adequate sheer strength keeps this from happening.

As a thought experiment, if you ignore the need for spun bearing protection, and visualize an essential zero gap fit-up, the Loctite can actually be left out. Then the net loss in strength by these joints is to reduce the shear strength of the shaft since the diameter of the shaft was reduced to make room for the sleeve, and then the cross-section of what was left was milled in half to give the two flats.

The shaft diameter starts out at 5/8" and tests with the 1/16" wall thickness sleeves showed the joint can still handle up to 90 ft-lbs which is stratospheric for a model engine. The three inner bearings will hold the three sleeves in place and limit their radial motion during operation, and so flex to deteriorate the cured Loctite (which is designed for much sloppier fit-ups) should be minimal. The advantage of using splices at these three points is that the Loctite can cure and fill in any gaps while the crank is laying dead straight in its alignment fixture. - Terry

 

[stanstocker]

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​

The methods shown in his videos were completely new to me when I first saw them. Not being a motor cycle guy I knew zero about pressed crankshafts, quite an eye opener. Enjoyable videos as well, quite a lot going on.

Cheers,
Stan

 

[nj111]

The methods shown in his videos were completely new to me when I first saw them. Not being a motor cycle guy I knew zero about pressed crankshafts, quite an eye opener. Enjoyable videos as well, quite a lot going on.

Cheers,
Stan

Also no loctite required as the pin Allen inserts through the middle of the pressed joint ensures no future movement. It's certainly a well proven alternative, and Allen's engines see far more loading than our models are likely to.

 

[GreenTwin]

I see the built-up crankshaft tests as static, wereas when the engine is running, it seems like it will be dynamic forces, more like what happens with an impact wrench, which is why impact wrenches work so well to remove nuts and things.

Again coming from an armchair observer with zero crank experience.

I know the motorcycle engines have pressed cranks, but there is a lot more beef there, and quite a bit more contacting surface area too.

.

 

[Kimmo]

Thanks Terry,
Looking forward to see the crank running!

 

[nj111]

If you prefer an assembled crankshaft but not using press fit then there's always the well proven Bugatti way. Rather than explain it myself I'll just leave this fascinating insight....

 

[peterl95124]

that Kawasaki crankshaft video is to die for, "to my surprise the TIR was .001", I'm well pleased, that saves a lot of hammering" --- LOLROTF, he got lucky, because what every machinist always needs is "a bigger hammer" !!!

 

[nj111]

that Kawasaki crankshaft video is to die for, "to my surprise the TIR was .001", I'm well pleased, that saves a lot of hammering" --- LOLROTF, he got lucky, because what every machinist always needs is "a bigger hammer" !!!

Not sure I understand your comment? Millyard didn't machine anything, Kawasaki did. When assembling this type of crank, you pretty much will always need to tap the web with a hammer to rotate on the crank pin and achieve final aligment within specification. All builders do that with this type of crank. Finally it's pinned to prevent further movement.

 

[mayhugh1]

Continuing on with the pieced-together crankshaft ...

The first step was to bandsaw four workpieces from a length of Stressproof purchased specially for this crank from Speedy Metals. The crankshaft's o.d. was designed around precision ground 1-5/8" diameter material sold by Speedy. The additional cost of the ground finish was negligible compared with the shipping charges, and it simplified machining some.

The workpieces were faced in the lathe leaving an extra quarter inch material on each end, and then a reference flat was milled along their entire lengths. With the workpieces standing vertical in a vise and their flats against its fixed jaw, the ends were indicated and center-drilled for the main and rod journals.

The flats were used to reference the orientations of the workpieces in the mill vise while the rod journals were rough machined into hexagons. The setup was rigid and repeatable allowing the inside walls of the counterweights to be finished at the same time. Since the rod finishing operations were going to be tedious, the distance across the hexagonal flats was set for just .020" more than the finished rod diameter in order to minimize the machining.

The asymmetrically removed material caused the parts to spring open an additional .003" directly across from the journals. The workpiece TIR's measured on either side of the journals correlated almost exactly with these spreads. I expected less distortion from Stressproof, and if I'd had some other 1-5/8" steel on hand, I'd have machined it identically just to see how it compared.

The journal offsets were too large for round workpieces in a four-jaw, and so a rectangular adapter was machined to provide friendlier holding surfaces. The adapter was bored for a close sliding fit with the ground workpieces, and a 5/16" grub screw against their flats held them in place. A spacer was used behind the shorter workpieces to stand them sufficiently out from the adapter. The rod journal locations were indicated with their counterbores.

The rod journals were finished using a .158" wide Kennametal A2040N002F02 carbide insert in a Kennametal A2BNSN32M04 .134" wide parting blade. The stick out was .9", and the insert was bifurcated and honed extra sharp on a diamond plate. A dial indicator run against the insert's cutting edge was used to align it to within tenths of the lathe's axis. The finishing operations used a .005" depth of cut with the lathe spinning at 80 rpm. My cutting technique was to slowly plunge the tool to its depth of cut in the center of the journal and then side mill to the left and then right wall before returning to center. The manual feed rate was SLOW with each journal requiring an hour of machining time. The same tools and machining technique were used on the solid cranks in my last two engines.

So far, the most satisfying part of this construction technique was the consistency of the finishing operations. Without the typical flex of a full size crank, the tool was able to cleanly work with as little as .002" depth of cut, and it wasn't at all difficult to wind up with four diameters within .0002" of one another without a lot emory time. Only 15 seconds each with 600g and 1000g paper cleaned up the light machining scratches in the turned surfaces for near perfectly round and polished finishes.

On the other hand, the workpiece distortion created by the rod journal roughing operations is a concern. In a solid crank's construction this error would have gone unnoticed among all the rest since it would been somewhat balanced out in the pairs of opposing throws. I need to think about how I'll compensate for them as well. - Terry

 

[Steamchick]

Thanks!
Interesting how the part deformed after machining.
The use of a bifurcated "parting tool" is exactly what I have used, however, mine is just high-speed steel, and I would have ground the tool's corners to a small radius to stress-relieve the corners of the journals. I'm sure it matters "not a jot" on these models, though.
Well done!
K2

 

[mayhugh1]

As mentioned earlier, I'm probably trading one set of problems for another ...

The distortion left in the workpieces by the rod journal roughing operations wasn't surprising, but since I'd used Stressproof its magnitude was disappointing. In any event, the rod journals were finished with respect to the inboard side of the workpieces (the side mounted inside the rectangular adapter), and so the journal axes presumably wound up aligned with them. To verify this, the inboard halves were indicated in a set-true chuck and a dial indicator run longitudinally across the journals in several places. The maximum error was just a couple tenths. The outboard halves, however, were left wobbling with respect to the inboard halves with a runout of some .004".

These outboard errors were correctable, but there was a price to pay. To correct them, each workpiece's inboard side was mounted and indicated in a set-true chuck while its outboard side was skimmed back into alignment with its inboard side. The workpiece was then flipped around in the chuck so its inboard side could be skimmed to the same diameter. The end result was the workpiece halves on either side of the finished rod journals were again running true to each other, and the rod journals were parallel with their axes. The price for this was that the outboard center drills were no longer on the axes of the workpieces, and so the tailstock couldn't be used during the finishing passes of the remaining turning operations. This was really only significant for the front section which has a long 3/8" diameter snout.

While working with these unbalanced workpieces I ran into unexpected difficulties keeping my set-true chuck running true. After indicating them using the radial adjusting screws and then tightening the axial mounting plate screws, I discovered the runout changed a couple thousandths during a 1 krpm turning operation. Over-tightening the radial screws after the axial screws were tightened seemed to help, but the TIR's continued to shift up to a full thousandth. Since they occurred in offset pairs and will likely cancel during assembly, they're more of an irritation than a problem. In retrospect, my 4-jaw chuck even with its much shorter jaws might have been less problematic, or perhaps I should have reduced the spindle speed.

The next steps included the major turning operations on the ends of each crankshaft section. Lots of material was removed, but since it was symmetrically removed from around the workpieces' longitudinal axes distortion wasn't an issue. The tailstock was used to stabilize the longer workpieces during their roughing passes but had to be loosened for the final finishing passes. As best I could measure, each of the four sections wound up with a final TIR of .0015" or less.

The next step was to verify the final dimensions of each section with respect to those in my model and remove the excess stock from the ends of each workpiece. Sanity checks on each section were performed in the completed block using some machined aluminum disks temporarily standing in for the main bronze bearings. (I'm not sure why I didn't just go ahead and machine the actual bearings.) The measured TIR's of each section's unfinished counterweights running in the dummy bearings inside the block was a thousandth, indicating the earlier truing process actually worked. The worst runout was .0015" at the tip of the long skinny nose of the front section.

The rod journal roughing distortion was a bump in the road, but so far I'm very happy with results. The next steps will include machining the tenons, key slots, and counterweights. - Terry

 

[mayhugh1]

Wrapping up the crankshaft ...

The sections were clamped horizontally in the vise so the alignment flats on each end could be machined. For these operations the rod journals in the front and rear sections were oriented straight up at twelve o'clock, and a gage pin between the journal and the fixed jaw of the vise was used to locate them. The journals on the two center sections were set at either three or nine o'clock for their machining. The measured heights of gage pins inserted in holes drilled and reamed earlier through the centers of the rod journals were used to orient these sections.

The flats were carefully machined for a maximum of two thousandths total clearance inside their sleeves. Loctite 638 used earlier for these fit-ups is too viscous (and its fixturing time too short), and so old school 640 which was designed for close (but not press) fits will be used instead.

A trial assembly of the crank was built up as each sections' flats were machined, and its fit in both the alignment fixture and the block were verified. With its ends supported on a surface plate the clearance stack-up allowed 3-4 thousandths sag at its center. This clearance is just enough to allow the assembly to 'self-align' and cure in the alignment fixture without the need for clamps.

With the sleeves running in the dummy bearings inside the block, the trial assembly rotated free with no tight spots. The TIR's measured at each counterweight (while they were still mostly round) as well as at the front and rear tips of the crank was on the order of .0015".
As mentioned earlier, a hole was drilled/reamed longitudinally through each rod journal. A pair of cross-drilled holes was added as well. With the current plan being to splash lubricate the bottom end, the purpose of these holes is aid oil flow to the rod bearings.

The next steps were to machine a pair of key slots in the nose of the front section and single larger slot in the end of the rear section. The front keys locate the camshaft chain drive sprocket and the crank damper. The key in the rear section will lock the ring gear assembly to the crank.

The final machining operations shaped the counterweights. Ford's balancing scheme includes wedges of material removed from either side of the rod journals. These wedges were manually removed with the help of a shop-made 22 degree angle block. The remainder (and most) of the counterweight machining was done on the Tormach using six different profiling operations. I was concerned that with so much material being removed from the two center sections I'd run into another distortion issue and spoil the runouts. However, another set of TIR checks in both the alignment fixture and block showed no change.

Finally, the bearing surfaces and flats were masked off for bead blasting. The entire crankshaft was cleaned with acetone and assembled with Loctite 640. After a couple hour setup in the alignment fixture, the excess Loctite was removed from the exterior surfaces with an acetone soaked cotton swab and assembly allowed to cure for 24 hours.

Final runouts were measured inside the block with the crank supported only by its outer ball bearings. The runout of the crank's rear end and its two inner rear bearings measured under a thousandth while front inner bearing measured two thousandths (darn!). The TIR of the crank's front nose measured .0015".

Epilogue ...

I might have gotten similar results from a single solid workpiece, but it would have been at the cost of more risky work and frustration and probably at least one scrapped attempt. If I do another one these cranks, it will be similarly assembled. Although only time will tell for sure, I'm one hundred percent confident it will stand up to anything a model engine can 'throw' at it. - Terry

 

[Steamchick]

Well done sir!
K2

 

[mayhugh1]

Before starting on the main bearings, a few loose ends need to be tied up on the block including the machining of the cylinder liners. Once they're installed, the decks can be finish machined and the main bearing work will begin.

Since the backup material purchased for the crankshaft wasn't needed after all, it was repurposed for the liners. A better choice would have been tubular cast iron, but I tend to limit my work with cast iron to just piston rings. The Stressproof o.d. finishes were flawless and a joy to turn, but roughing out the liners' i.d.'s on my lathe was tedious and required nearly an hour of drilling and boring each.

In order to simplify the piston ring machining later on, a major goal was to wind up with a common lapped i.d. (within a few tenths) for all the liners. Since I like to complete the lapping before the liners are installed, their o.d.'s were machined for a Loctite'd slip fit rather than a press-fit which might deform them. As it turned out, my barrel lap wouldn't have cleared the bottom end of a liner after it was installed anyway.

Eleven liners were started. A first article used to determine feeds, speeds, and a best order of operations wasn't likely to find its way into the engine. Another liner will be used later outside the engine to light-test the piston rings leaving eight for the engine and a spare to cover a possible outlier after lapping.

The first step in my process was to machine a trial part from Delrin and verify its fit in each cylinder. Minor variations in the block bores were noted so the liners can be installed in their best fit locations to minimize chances of coolant leaks. A .002" radial clearance was left for a gap filling Loctite.

The band-sawed starting workpieces included 5/8" work-holding spigots. After truing up an end, the workpieces were through-drilled to one inch diameter using a succession of five drills ranging from 1/2 inch to one inch. This was followed by four roughing passes with a 3/4" boring tool that opened up the i.d. to 1.165". After profiling the liner o.d.'s, the spigots were parted off, and the inside corner under the top flanges were slightly undercut.

The very last machining operation was a bore finishing pass that was run on all liners in succession using a brand new insert in the exact same setup (except when I accidentally bumped the cross-feed handle). Since the piston rings will be shop made later, the exact liner i.d.'s weren't important, but the messy lapping process on a large batch of parts will be minimized if the starting i.d.'s are very close to one another. In my experience, lapping a thousandth is reasonable. Lapping a couple thousandths makes for a long weekend, but if I have three thousandths to remove, I'll likely rebore the whole lot.

The pre-lapped i.d.'s wound up at 1.1745" plus/minus a half thousandth or so. An identifying number was scratched on each liner so its lapping progress could be tracked in a worksheet. A simple setup with a dial bore gage and a couple blocks of wood allowed quick and consistent i.d. measurements at the top and bottom of each liner with a repeatability on the order of a couple tenths. My lathe cut a half thousandth taper over the length of the liner, but I was careful to plan for the larger diameter to wind up at the bottom of the liner. Even though the entire taper can be lapped out, the most important thing is to obtain identical i.d.'s at the top of all liners. A couple extra tenths at the bottom of a liner isn't going to affect performance.

A brass Acro barrel lap and Clover 280 grit grinding grease was used for all lapping operations except for the final pass which was done with 600 grit grease. Since the stock lap was intended for a standard 1-3/16" bore, it had to be turned down a bit to accommodate my weird size bore. My lapping technique was to chuck the lap into a battery-powered drill running at 200-300 rpm and manually oscillate the liner over the lap spending extra time over any snug areas. After 30-45 seconds, the liner was washed in kerosene and its dimensions recorded. After each barrel adjustment, the liners with the smallest i.d.'s were lapped first and the whole herd moved slowly toward the finish line as a group. Short lapping times, small incremental changes, and frequent measurements are important. The price for impatience and big diameter changes can become wasted time chasing one's tail among a big batch of parts.

After a half dozen or so hours, all diameters (and tapers) were within an acceptable window. Lapping was declared done and a final pass made using 600 grit grease to improve surface finish without noticeable metal removal. - Terry

 

[peterl95124]

Richard, to avoid crush from chuck jaws I first bore and lap the inside of an oversized rod, then put it on an expanding mandrel to turn the outside. Also I use "helical lap" (name of company that makes them), so nearly impossible to get taper/barrel/flare that so frequently happens with "acro lap". I always make another extra liner to use as a lap for the rings, as I've tried all the usual methods and found that none make a true seal (no "light") without lapping. Peter.

 

[mayhugh1]

Houston, we have a problem ...

The first photo shows the highest outdoor temperature I've experienced in my entire 75 years, and that includes three years spent in the Mojave desert. For you metric guys, that's 45.5 celsius. When this photo was taken, it was 99F inside my shop, and I was setting up to paint the block.

In preparation for paint, the block was prepped for bead blasting. The coolant passages and several machined surfaces inside the block needed shielding from the glass beads, and so a few days of throwaway work were spent making screw-on cover plates to protect those areas. Every threaded hole was filled with a screw. The lifter valley between the decks was also bead blasted even though it would remain unpainted. An easily removable top plate was machined to cover it. The lifter bores and oil drain-back holes in the valley were also plugged with rubber stoppers to keep the beads out of the cam bore and the bottom end of the engine. After bead blasting, the cover plates were removed and the entire block scrubbed with Simple Green followed by Dawn dish detergent and plenty of warm water. After a thorough drying with compressed air, the cover plates were reinstalled for painting. The block was air brushed with Gun Kote's 'Anno Blue' and then oven cured at 300F.

The liners which had been machined to protrude .003" above the unfinished decks were installed next. They, as well as their mating surfaces inside the block were pre-cleaned with acetone. Loctite's instructions for slip fits recommends installing circular parts with a twisting motion to insure full coverage of the adhesive during assembly, and I've learned through bitter experience to follow those recommendations. A rubber stopper shoved into the top end of the liner acted as a temporary aid to assembly them to the block using Loctite 620.

During assembly, Loctite was unavoidably smeared across the surfaces of the liners inside the coolant jackets. Left on those open surfaces it wouldn't cure and would eventually contaminate the coolant. After a 24 hour room temperature cure plus an extra 12 hours at 140F for good measure, the coolant jackets were flushed with acetone.

Acetone leaked out the bottoms of three of the liners in bank one carrying uncured Loctite with it. I've no explanation why the Loctite on those three liners didn't cure since as far as I knew I'd met all Loctite's requirements: 1) all machined surfaces were smooth and bright and had been cleaned with acetone, 2) their .002" radial gaps were well within 620's .015" max spec, and 3) steel met the requirement of having at least one active metal in the bonded pair. In fact, in order to avoid this very problem, a generous half inch wide sealing flange had been provided at the bottom of each cylinder at significant cost to coolant capacity.

A third cure cycle, this one at 190F for 1-1/2 hours didn't seem to solve the problem. After cool down, a second flush was performed but this time using Loctite's 7471 activator. This yellowish liquid is a volatile copper ion-carrying solvent intended as a surface primer when bonding together two 'inactive' metals such as aluminum. The hope was for the additional free ions to kick off a cure during a fourth heat cycle since the yellowish liquid seeping from the bottoms of the three liners was obviously in contact with the uncured 620.

In retrospect, I should have immediately started another heat cycle to see if the copper would have activated the 620. However, after some solvent flash time, Loctite 290 was injected into the coolant jacket feed port above the leaking sealing flange and allowed to also seep out the bottoms of the leaking liners. This thin wicking variation is intended to thread-lock existing assemblies. The plan was to allow capillary action to draw it into the leaking areas and cure. The block was then returned to the 140F oven.

Four hours later, a 25 psi coolant jacket air test showed the leaks to be sealed. Unfortunately, I'll never know if the activator had finally kicked off a cure of the 620 or if it was the 290 alone that plugged the leaks. In any event the block was returned to a 140F oven overnight. An acetone flush performed the next day confirmed the liners were still sealed.

Since I really don't want to revisit this nasty problem after the engine is running and the liners are exposed to oil and coolant, I decided to add some additional insurance. The bottoms of the liners protrude slightly through the surface of the block, and these were dropper'd with Loctite 290 which formed menisci. Very little adhesive was drawn in which was a good indication that the Loctite above it was cured (either 620 or 290). Just in case, though, the block was allowed to cure for another 8 hours at 140F before wiping away the menisci. For good measure I injected a second dose of 290 into both coolant jackets allowing it to puddle on the floors of the jackets and then put the block in the 140F oven overnight. The jackets were thoroughly flushed with acetone the next day with still no sign of leaks.

This is the second time I've used Loctite 290 to rescue steel liners sealed in an aluminum block with Loctite 620. (I had a similar experience with the Offy engine.) I also had a different but similar issue with 620 during the Merlin build.

Be aware ...

A Loctite applications engineer once told me that Loctite products have a very long shelf life and will continue to work so long as they're able to flow. The 620 I used was purchased five years ago. (Be aware there are a slew Chinese copies in nearly identical red bottles being sold on Amazon.) I've recently noticed that Henkel is now marking their bottles with the date of manufacture and a two year 'use by' date. And so, for a hobbyist, small bottles are probably recommended. In the future I'll likely start using activator on all slip fit inactive surfaces (aluminum, stainless, etc) regardless of what metal they're being bonded to. The issue with using activators and why I've tended to avoid them when theoretically possible is that they tend to decrease Loctite's ultimate bonding strength some 20%.

In any event, some 60 hours after installing the liners, work on the block was wrapped up when the angle plate and machining fixtures made last year were located and the deck surfaces finally finish machined. - Terry

 

[Vietti]

I too have been disappointed with locking grades of loktite sealing. I used your idea for the wicking grade to seal brass valve cages in an aluminum head after they leaked from a locktite job. . Thanks for the tip. This wicking grade is amazing-make sure it isn't even close to something you don't want sealed or kept from turning!!

 

[gunna]

Jeez, I don't like the small display on the right hand side of your weather station which shows "Feels like 122". That's getting a bit warm.
Ian.

 

[peterl95124]

great work Terry, very much enjoy and appreciate your posts.

I pretty much only use two loctite grades now, 1) wicking grade, because I doubt I've ever machined a model engine part with enough clearance for standard grade, and 2) "sealant" for things I might need to disassemble in the future, especially because without heat is almost always desirable.

folks are well advised to flush excess loctite out of their coolant passages, I didn't when I built my Merlin blocks, then 10 (ten!) years later I finally finished and ran the engine, then at the end of that summer decided to blown out all the coolant for over-winter storage. well, it didn't blow, turns out the excess loctite got carried by the coolant to a copper and brass fitting and hardened there from all the copper ions and plugged it up. this probably explains why a plastic tube from the water pump to the block blew off when I was running it at a show that summer (at least I know the water pump works !).

 

[mayhugh1]

The insanely high temperatures and daily warnings about our shaky (Texas) power grid limited my shop time to a handful of midnight hours during the past few months. Long CNC operations are still pretty risky, and so for the next few weeks I'm focusing on loose ends and simple parts that don't require a lot of precision or machining time.

The machining on the heads was completed some time ago, but I've been procrastinating over painting them. So, the heads were finally bead blasted and another throw-away cover plate fabricated to mask off their top surfaces. As with the rest of the engine's 'castings', the heads were Gun Kote'd 'Anno Blue' and oven cured.

The Ford small block had its mechanical fuel pump bolted to its lower left front end. Since the model's working pump will be electric and hidden from view, its mechanical pump will be purely cosmetic. Several pump styles were available - some with and some without an integral fuel filter canister. My own '65 Mustang had the canister pump, but I opted for the non-canister version on the model. It was machined in two halves, bead blasted, and permanently assembled with a 4-40 stud. The base was left as-is 'cast' aluminum, but the lower plumbing section was alodine'd to simulate the original's cad plating.

The model's fan and its fabrication closely emulates Ford's four blade riveted assembly. Construction began with contour machining a bolted-together stack of .030" and .063" inch aluminum blanks (I couldn't make up my mind about the fan's thickness). A die set was then machined to press a pair of stiffening beads into each blade. In order to keep the dies simple but usable with both blank thicknesses, the blades were pitched manually using hand tools after the pressing operation. The rivets were set with a hammer and drift against a shop-made back-up plate. Both .030" and .063" fans were assembled and painted, and they're still waiting for me to make up my mind. A spacer was machined to offset the fan a proper distance from the water pump flange. - Terry