Starter: grind, bind noise

[Canoe]

Found burs at each end of each tooth on the flexplate from when they were cut. Seem small, but as what all contributes to starter issues is unknown, I decided to address them.

I used a Dremel with a small stone head to "whisk" each tooth end until the burs were gone. And a light skim across the top of the tooth to smooth any roughness. And a touch up the height of each forward tooth end to smooth minor nicks. A few of the teeth seemed to have taken some small intent damage from the starter gears blowing apart, so I smoothed those too. Then brushed with a brass spinning brush driven by a drill. All done through the starter locator hole... So a lot of small hand-turns of the engine.

Runout of the flexplate looks good.

Next is bench test the start to ensure its Bendix gears extends fully. Then test fit the new starter and hand-turn the engine to check for clearance with the flexplate teeth.

No more trusting that the parts I receive are proper.

[Canoe]

Success - partial.

With the:

• 3/8" shank on the bolts,
• the matching 3/8" lower bolt hole in the flange on the starter, and
• the ~3/8" hole in the bell housing,

the starter bolts up in the only possible position - in spite of the enlarged hole for the starter in the locator plate. With a the minor difference in the axis of the threaded holes for the two bolts, there is no play when the bolts have the starter against the plate against the bell housing; both when just finger tight and of course when torqued tight. There is no other possible position.

Turns over very nicely.

Doesn't start... (but was running and shutdown to pull the broken starter).
So I don't yet have confirmation of my measurements that the flexplate teeth clear the resting starter's teeth, although I did turn the engine through 360 degrees by hand without any noise.

p.s.
To get the widest part of the slightly tapered shank in play, I didn't use a lockwasher, but liquid threadlocker.

[Canoe]

Turns over fine.
Nearly all of the time.
Occasionally, turnover stalls, then resumes. I suspect there's one part of the flexplate where the teeth are not cut quite right, resulting in a 'valley' that is not quite deep enough and the starter-to-flexplate binds. This turnover stalling seems to match what I recall just before the old starter's Bendix gears exploded.

I assume my easiest way to find this is to get setup to pull the starter, then keep starting it until I get the beginning of the stall/bind and quickly turn the key off. Hopefully that will leave the part of the flexplate with the issue at the starter locator hole in the plate.

I guess I need to make a template of what the valley cut should be, so I can place it tooth by tooth to find which has the issue(s).

[Canoe]

Here's the photo of the old & new starter, showing the difference in the lower bolt hole, which is significally smaller in the new starter, meaningfully closer to the 3/8" bolt size. There are differences in the dome too, but they both show the five ribs. Both are refurbs from Champion, and both (and the prior that got its dome destroyed by engine kickback) had astoundingly more power/speed turning over the engine than I've ever had in an Eagle. And this is on a rebuilt engine, 4.0 head and new/tight cylinder rings.

Also, the photo of the two pairs of 3/8" 1.5" stainless bolts, which have shanks. Both have hugely less play in the lower bolt hole than bolts that are threaded without shanks.

• The pair on the left have narrower and more tapered shanks, and have double or greater play in the bolt hole than the right pair. 
• The pair on the right have shanks that start at the head at so very close to 3/8", and have less taper - meaningfully closer to a shoulder bolt, and have significantly less play in the bolt hole. These 'tighter' bolts are the ones I used to install the new starter.

I attribute the success in spite of having an enlarged hole for the starter in the locator plate, to:

• the 3/8" size lower bolt hole in the new starter,
• the shank on the bolts so they act much like a shoulder bolt, and
• using thread-locker instead of a lockwasher, so more of the shank and the widest part of the shank is within the bolt holes,

so the registration of the starter to the flexplate is determined by the bell housing, not the hole in the locator plate.

I would expect for those starters with an oversized lower bolt hole in its flange, that an insert of solid rod of that hole's diameter could be cut to fit the flange hole, then drilled out dead-centre at 3/8", would provide the same or similar registration benefit that I got in the new starter with the smaller hole. This is similar to what I tried earlier with the stud I made with the threaded 'shank' on it, but would be meaningfully more precise.

I expect that the 'gasket' I was planning to correct the enlarged locator plate starter hole is NOT a good way to address this problem: it's a lot more work, and is dependant on an error-prone judgment to locate the correct size hole to provide starter location registration. Using bolts with shanks closest as possible to 3/8" to closely function like shoulder bolts is clearly a superior way to go.