Gearbox

Last Updated: 3 January 2008

The model ML saw uses a simple 2-speed transmission, pictured in its as-purchased state (sans pulley). Note the amount of accumulated sluge. It's hard to tell from the photo, but the box itself is leaking (albeit very slowly) from the seal where the case splits in half, at the bottom. It's also leaking from the lower bearing cover plate.

The leaks had the unfortunate effect of making an already disgusting mess inside the drivetrain housing an order of magnitude worse. Instead of a dry mess that could be cleaned with a vacuum, I have a sticky goo that must be washed out.

The oil fill is simple 3/4" black iron pipe.

Here's Doall's cutaway drawing of the gearbox. This design was in use through 1947, and the last serial number to use it was 4712809. Starting with the 12810th Metalmaster (produced in 1947), some minor details were changed, but the overall mechanical design was never significantly altered from this first design.

My box is missing part 17-13521, elbow, which acts as a breather. It wouldn't be a problem to fabricate one, except that I'd have to cut a hole in the top of the gearbox as well. My box doesn't quite match the drawing in a few respects, and the breather at the top is one example. Instead of a pipe fitting as a breather, mine has a small (1/8") hole in the case at the top. The flaw with the hole is that it permits debris to enter the gearbox, which is likely the reason DoAll later added the elbow.

To dismantle, I began by removing the front bearing cover. Here the box is sitting in a large drain pan. Notice the fluid is red - at first I thought automatic transmission fluid was used. Turns out the inside of the box is painted red, which leeches into the oil.

So far, so good. Note the four slotted screws around the perimeter of the case - these are holding the case together. There were also 2 locating pins (one at the top, one at the bottom) that had to be pressed out. I was worried the bearing outer races might be pressed into the case, but they weren't.

How to tell if your gearbox is bad without disassembly
This was a big moment for me. Since the gearbox made a terrible metal-on-metal banging noise when it was run in high gear, I was very apprehensive about what I might find inside. Remember that the gearbox is the single most expensive component on the saw, and if it's no good the saw is, for the most part, worthless.

Of course, if you're buying one of these saws there's no practical way to check for a bad gearbox other than listening for bad noises (like mine had). I bought mine in spite of the noise because although it sounded bad, I felt it wasn't something that meant the gearbox was irrepairable.

The good news is there are very few catastrophic failures with this type of unit. I figure there are only 2 ways to kill one: either run it for a long period without oil, or break a gear tooth. Obviously a broken tooth will grenade the unit in short order. Running without oil might only destroy the bearings, which are all easily replaced.

It should be fairly simple to tell if either of these failures is present in a saw you're considering for purchase. If the bearings are shot, it will probably make a grinding noise when run. If the saw can't be operated, removing the gearbox belt (simple to do even without tools) and spinning the gearbox by hand in each gear will reveal a bad bearing as a strong binding.

If a gear tooth has been broken, the box will probably make a nasty racket when run. If it can't be run, then spinning the gearbox by hand should reveal the broken gear as either an inability to turn the shafts by hand, or severe "crunching" or grinding.

Step 1: Crack Open the Case
Again, this looks pretty good. These gears are in good shape. Other than a lot of old oil gunk this assembly looks fine. In fact, even the bearings seem to be in good shape. The problem with this transmission, however, is a severe binding through roughly 1/2 rotation (easily felt by manually turning the input shaft). At this point in the disassembly I hadn't yet located the source of the binding.

Note there's little sludge from the oil, which suggests perhaps it was changed periodically.

Step 2: Remove Secondary Shaft Ball Bearing
Here's the pressing operation to remove the secondary shaft ball bearing.

That's a 5-ton puller, and the bearing was difficult to press off. It's a 6204 bearing, which is for a 20mm shaft. The secondary shaft measures 0.791" at the bearing mount, representing an interference of 0.004" with the 20mm 6204. According to Machinery's Handbook, the acceptable range of shaft diameters for this bearing is +0.011 to +0.002 for an interference fit. The condition of this bearing and the "modern" specifier "6204" stamped in the shield makes me think it's not original.

Step 3: Remove output shaft assembly
The output shaft ball bearing is in pretty good shape, but has no numbers stamped in it. To make matters worse, the parts manual lists 4 separate Doall part numbers for the bearing, depending on the manufacturer (SKF, Norma-Hoffmann, NDWC, and Hoover).

To complicate things even more, there are 4 different bearing covers, depending on which bearing manufacturer was used. The bearing on this unit happens to be an SKF. After searching online for interchanges for those numbers I came up empty-handed. I contacted Doall for help, but they never returned my call. This is the first problem encountered with the box. It will be very difficult to remove due to close clearances with the output shaft gear (can't get a puller in there to grab it).

The bearing measures out with an OD of 62mm and a width of 19mm. Some time searching the NTN catalog indicated an NTN number 8506 single-side sealed bearing should be an exact match. I bought an 8506 and although it's dimensionally correct for this application, it doesn't have the same outer shield on it. That means it won't quite fit the bearing cover plate, but I think it would work fine nevertheless.

In the end, I didn't replace the bearing because after cleaning this part up it seems to work just fine.

Step 4: Remove Shifting Assembly and Intermediate Gear
With the secondary shaft rear bearing removed I was able to slip out the output shaft and gear. Here's what the gearbox looks like with it removed. Note my nomeclature for the shift fork and intermediate gear.

The shift dog (called a "clutch" in the Doall cutaway view) slides on the output shaft to affect the high/low gear change.

When the shift fork moves the dog to disengage the output shaft from the input shaft, torque travels through the secondary shaft to the output shaft, lowering the saw speed and greatly increasing the torque.

Conversely, coupling the output shaft to the intermediate gear via the shift dog sends torque through the intermediate gear, then directly into the output gear at a much lower ratio, giving higher speed but lower torque. This mechanism is in good working condition (other than the layer of gunk).

Note that this input shaft gear is NOT attached to the shaft! It spins freely from the shaft, making its use dependent on the position of the shift dog. In low gear, the shift dog is engaging the output shaft gear (the one shown in these photos). In high gear, the shift dog is engaging the intermediate gear (see photo above) and the output shaft gear is rotating independently of the output shaft.

Step 5: Remove Input Shaft To remove the input shaft, the transmission is flipped over and the rear bearing cover is removed. This permits pressing out the input shaft.

All this stuff was in good shape, and the binding turned out to be the secondary shaft. The secondary shaft gears lie on opposite sides of a big cast iron plate (cast into the center of the gearbox; you can see it in the cutaway drawing).

The binding was occuring because the secondary shaft wasn't axially correct, causing the big gear to rub slightly on the center bearing for the input shaft (not depicted here). You can see the "shiny" part on the inner gear in the picture above where it was rubbing. By pressing the secondary shaft a little toward the input side of the transmission I was able to free up this bind.

Problems with the Transmission
According to the drawing, these gears use a simple key to lock them to the shaft. Unfortunately, this gearbox doesn't quite match the drawing: the gears have been welded to the shaft. Because of this, there's no way to remove that center shaft. Based on the quality of the welds, it's apparent the gears were welded in place after the box was assembled. It's not really necessary to remove the secondary shaft...unless you want to replace the secondary shaft ball bearing! Of course, I would like to replace that secondary shaft ball bearing. I tried a few times to press the center shaft in such a way as to "squirt" out that bearing to no avail. It is probably possible to remove the bearing by destroying it. The problem is that there's nothing to "grab on to" on the bearing - it has a smooth inner and outer race (as most bearings do). Some "creative" cutting on the bearing might yield a means of removal, or might irreparably secure the bearing where it is.

There are other inconsistencies between the gearbox and the drawing. Notice in the above picture the output gear on the secondary shaft (the small one closest to the camera) runs the entire length of the shaft down to the "outer gear". In the drawing, however, that gear is only slightly longer than the output shaft gear is wide.

Although it makes me nervous to do so, I'm going to leave that secondary shaft ball bearing in place rather than mangling it in order to try to remove it. It's clearly worn (you can feel some slight binding when rotating the secondary shaft by hand), but I fear any attempt to remove it may cause more harm than good.

Finally, the center bearing (not pictured until reassembly, below) is an SKF labeled "FL-20". A search on SKF's site indicated that part number has been supersceded by "FL 20-RS1". A search on that bearing yielded no results. Fortunately FL-20 can be replaced with a 204-sized bearing.

So I'm glad to see the gear teeth are in good condition, the shift mechanism is good, and the shafts are apparently straight and true. But I'm frustrated by the inability to completely dismantle the unit to replace all the bearings. I believe 4 out of the 5 bearings were replaced at some point in the saw's history. When that overhaul was performed, I think the gears were welded in place (although I don't understand why this was done). I think the output shaft bearing is original equipment from 1941.

Lubricant
Doall specifies an SAE 40-weight oil for the transmission. These days 40-wt gear oil seems to be a little hard to come by. Most oils are offered in 30 or 50 weight, but not 40. The relevant specifications for SAE 40 weight oil are (from Machinery's Handbook):

Viscosity @ 100°F (~40°C): 173 centistokes
Viscosity @ 212°F (~100°C): 14.5 centistokes

Assuming this transmission will run in the lower end of this range 99% of the time, the theoretical lower limit of viscosity should be 173 centistokes. There seems to be some disparity among other sources as to the real viscosity of SAE 40 at 100°F; I've seen it as low as 134 centistokes. After some study I decided to try Mobilgear 629, which has the following specifications:

Viscosity @ 100°F (~40°C): 150 centistokes
Viscosity @ 212°F (~100°C): 15.8 centistokes

I figure oil technology has come a long way since 1941, and although the 629 is lower viscosity than specified by DoAll, the strength and additive package on the 629 will make up for it. Plus it's cheap and easy to get.

Bearings
After much research and careful measurement I figured out what each of the bearings in the gearbox are. The center bearing, secondary shaft ball bearings (at both ends), and the input shaft ball bearing all have 20mm ID and 47mm OD, making them all 204-sized bearings.

As for the output shaft bearing, that one measures out with an OD of 62mm and a width of 19mm. Some time searching the NTN catalog indicated an NTN number 8506 single-side sealed bearing should be an exact match.

After all that work, I ordered up a full set of new bearings...and then decided to keep them as spares and continue using the existing bearings. Once they were all cleaned up, the existing bearings looked too good and worked too well to justify the expense required to replace them. They all show very little wear, so I'm going to reassemble the gearbox with all the old bearings and see what happens.

Reassembly
With the bearing issue resolved (for now) the gearbox was ready for reassembly. I cleaned everything using either Gunk parts cleaner or Simple Green (in a 10:1 mix with water). Parts cleaned in the Simple Green solution were liberally sprayed with WD-40 after removal to drive off water; this prevented rust formation. Notice in the proceeding how clean and shiny things are relative to the disassembly photos.

Whereas most of the disassembly was done using a hammer to remove press-fit parts, for the more "delicate" task of reassembly I used a 3-ton arbor press. I prefer this method, since you can feel exactly what's going on and apply pressure in a very controlled manner.

Step 1: press the input shaft (with input shaft ball bearing already installed) into the gearcase.	Setup to press
	Press: 1-1/2" socket used to distribute load to the outer race only.
	Finished

Step 2: press in center bearing. This one's a little more complex because it uses a tight press fit on the input shaft and a light press fit in the dividing plate in the gearbox. As a result, it's important to support the input shaft during the press in order to prevent pressing the input shaft back out of the case (the input shaft ball bearing is also a tight press on the input shaft and light press in the case). I cut a block of wood to serve as support for the input shaft as shown.	Block installed to support input shaft during press
	Inside of gearbox with center bearing removed to show nice clean gears
	Center bearing in place before pressing
	Finished. Note the same 1-1/2" socket was used during this pressing operation as in the previous press to apply load only to the outer race.

Step 3: install intermediate gear and output shaft/gear assembly. No pressing was required: these parts literally drop in place.

First, slide the shift dog into the output shaft.
Then drop the intermediate gear spacers, intermediate gear, and output gear with shift dog and shift fork into the case (in that order). Note the end of the output shaft is concentric with the input shaft at the center bearing (a bronze bushing acts as a bearing surface, since the input and output shafts rotate at different speeds).

The strange thing about the intermediate gear is it's pretty much just sitting in there without any axial support. There are two spacers under it (between the gear and the center bearing; 5-04307 and 5-04306 in the cutaway drawing), but they aren't going to provide much support for axial loads resulting from the helical gearset. A thrust bearing would have been better at that location. I suppose the axial loads are small.

First, slide shift dog onto output shaft

Step 4: install secondary shaft ball bearing. Had a little trouble with this one. As I commented in the disassembly, this bearing was a very tight interference fit. It made the pressing operation almost impossible - I couldn't get the bearing to press on straight. I ended up taking a smooth mill file to the secondary shaft and removing a thousandth or so, which made the pressing operation a piece of cake. In fact, it was so easy (maybe 500-700 lb) there was no need to support the opposite end of the secondary shaft to prevent it from moving axially during the press. This is convenient because supporting that shaft on my arbor press would've been a real challenge.

At this point most of the mechanicals for the gearbox are finished. Next up is gaskets. I bought a sheet of Buna-N (1/16") to cut gaskets for the front and rear bearing covers and the case flange. This stuff is easy to cut and relatively cheap.

Press. No need for a socket to distribute load.

Finished

I also built a simple test stand for the gearbox. My plan was to eventually fill the box with oil and run it for a little while using my 1/2" drill. This required the test stand and a shaft adaptor (because the input shaft is 3/4" diameter, so it won't fit in a 1/2" drill chuck).

This picture shows the test stand (made from scrap lumber) and the gearbox with shaft adaptor attached. The shaft adaptor is a piece of 1" square aluminum bored 3/4" on one side and 1/2" on the other. A 1/4-20 bolt acts as a set screw to retain a piece of 1/2" dowel rod that the drill chuck holds. This works, but the holes in the adaptor aren't quite concentric with the gearbox input shaft (owing in large part to my primitive tooling).

I have since filled the gearbox and it's been sitting there full of oil for several months without even the slightest leakage. I've also spun up the input shaft with an electric drill to get everything coated with oil. Everything works perfectly.

Problem 1: secondary shaft bearing
Two important problems were encountered when trying to reattach the bearing covers. Both involve the secondary shaft. As I commented earlier, I had to press the secondary shaft toward the input side of the transmission to relieve a binding that was occuring as the big first gear on the secondary shaft contacted the center bearing during roughly 1/2 its rotation. In so doing, I pushed the secondary shaft ball bearing out of the case by 10 or 20 thousandths.

This is a problem because it prevents the steel cover from fitting flush with the surface of the case. To fix this, I have to make a riser plate out of 1/8" aluminum. The riser will move the cover plate up away from the bearing and permit a good gasket seal. A picture of the riser plate is below. I cut it out with a jig saw (ironically, a bandsaw would've really come in handy for this). I adhered it to the gear case using silicone gasket maker.

Problem 2: axial movement of output shaft
The second problem lies at the opposite end of the transmission. The output shaft bearing is not press-fit into the case. As a result, once the unit is assembled the output shaft is free to move axially.

Normally this wouldn't be a problem, except that when the shaft is fully extended away from the case the output shaft gear makes contact with the secondary shaft ball bearing, causing a bind. I'm not talking about a lot of movement - maybe 0.050" or so.

To fix this problem, I cut a 1/32" piece of PTFE sheet gasket into an annulus properly dimensioned to occupy the space between the cover plate and the output shaft bearing. With the cover plate in place, then, the PTFE ring makes contact with the steel plate and prevents any axial movement of the output shaft.

Notice that axial loading from the helical gearset will tend to pull the output shaft inward, toward the back side of the transmission (which is good).

There is a potential problem with this solution. The output shaft must transmit the blade tension to the output shaft bearing, then into the case. If the blade tension causes an axial component of force pulling the output shaft out of the transmission, all that axial load will be handled by the bearing cover plate (next picture). That bearing cover plate isn't strong enough to take much load. I hope, therefore, that axial loading due to band tension is very small.

This works remarkably well. Here's what it looks like with the cover plate and new gasket installed. You can't actually see the PTFE ring with the cover in place.

I spent some time straightening up that front cover - it was mangled pretty badly.

Next up is the testing phase. I filled the unit with oil, then installed the fill plug with a couple magnets. I've been running the gears for awhile (manually and with an electric drill) to get the oil into all the bearings and flush out contaminants that were introduced during reassembly (the magnets should pick up any ferrous particles).

Reassembly

The box sat for months as above without leaking a drop. Then, after installation, it was drained and refilled with oil, at which point the bearing cover plates began leaking at the bottom. To remedy this, the lower bearing sections were coated with Permatex No. 2 gasket maker.

On the motor side of the box, the Permatex worked well. On the wheel side, however, it was ineffective. I believe the issue lies in the waviness of the front cover. Although I straightened it as far as possible during the rebuild, it remains somewhat distorted (not flat). I expected the relatively thick gasket (1/16" Buna-N) to compensate for the irregularities, but there was an unforseen problem (one of which I might have been aware if I had more experience with gasket design). Buna-N is an exceedingly plyable material, and it compresses very easily. As a result, it wasn't possible to tighten the cover bolts sufficiently to remove some of the uneven-ness of the plate. The gasket, therefore, was not thick enough to overcome the waviness, and allowed leakage. Even with Permatex No. 2.

At the same time, I noticed that with the cover plate in place, the pressure exerted by the combination of the PTFE ring and the gasket on the face of the output shaft bearing was sufficient to cause additional drag on the gearset.

Both of these problems (that is, drag on the gearset and leakage from the cover plate) were remedied by removing the Buna-N gasket and applying silicone RTV gasket maker to the plate. It had been my intention from the beginning to avoid any such adhesives on the gearbox, in order to ensure future disassembly (if required) would be easier. But paramount to that requirement is the necessity of zero-leaks. The elimination of the gasket material, however, reduced pressure on the output shaft bearing, which eliminated the extra drag.

At right are photos of the front (wheel side) and back (motor side) of the gearbox re-installed.

Of particular note here is the new plumbing for the gearbox oil. The old arrangement used a simple iron pipe nipple and elbow with a plug. A line cast into the gearcase indicates where the oil level should be. The problem with that, however, is it's impossible to see where the oil level actually is. Oil level must be inferred roughly based on where it sits inside the iron pipe, which is basically impossible to do since there's no room to get your head in there to look. As a result, I assume most people simply fill the box to the top of the pipe elbow, with makes the box over-full.

My new arrangement eliminates this whole problem by using a slight glass plumbed into the drain hole as shown. A ball valve is also included to facilitate easy drains. I've marked the proper oil level on the sight glass.

I found the gearbox capacity is 24 oz.

	Before
	After. In this photo the shift lever locking ring is installed backwards. I fixed it after this picture was taken.

Band Tracking Issue
In the first picture at right, we see how the lower wheel was installed when I bought the saw. Notice the boss at the center of the wheel is roughly 3/4" tall.

After I put the saw back together (exactly as it was when I took delivery of the saw), I had a severe misalignment of the lower wheel with respect to the upper. The second picture shows a plumb bob supended from the center of the upper wheel. Notice it points beyond the front edge of the lower wheel (note that the saw is nearly level, so the position of the bob cannot be dismissed as being due to unlevelness in the saw itself). Obviously, when the blade was mounted and tracked to ride near the center of the upper wheel, it rode beyond the front edge of the lower wheel!

This caused me no small amount of frustration. I mean, I had the saw reassembled precisely the way it was when I received it!

I thought I had the problem solved when, upon close examination of the cutaway drawing of the gearbox (second picture at the top of this page), I noticed it shows the tall boss positioned toward the gearbox, as in the lowest picture at right. With the lower wheel in that position, the band tracks perfectly on both wheels.

Unfortunately, with the lower wheel riding that far out, it's impossible to install the lower wheel housing door, which has a 1.5" lip at the top. That lip strikes the lower wheel before the door can be fully closed (the wheel needs to be at least 1/4" farther back to permit proper operation of the door).

Of course, the door lip could easily have a relief ground in it to prevent contact with the wheel. But after mounting the table, I discovered that it's also impossible to tilt the table front/back the full 10° it's rated to, because part of the underside of the table strikes the wheel.

A friend with a similar saw, whose blades track correctly on the lower wheel, measured the spacing between the back side of his wheel and the shell and found it to be roughly 3/4". My saw has the same measurement if installed as in the first photo.

So, with the lower wheel installed as in the first photo, the bands track (1/2" band) with the saw teeth protruding out the front of the wheel (just slightly beyond the wheel flange), but everything else (the table, and the wheel housing door) works properly. With the lower wheel installed as in the bottom picture, the band tracks perfectly on both wheels but the table and the wheel housing door won't function properly.

By all my measurements, the upper wheel is properly positioned and the lower wheel is also properly positioned. Why the heck is the lower wheel 3/4" out-of-plane with the upper wheel when everything is properly installed? If a 1/2" wide blade puts the teeth just beyond the lower front wheel flange, a 1" wide blade (which the saw is rated to) will be fully 1/2" beyond the lower front flange! That can't be right!

I'm beginning to wonder whether these are not the original wheels.

Belts
There are 4 belts required for the unit:
1. Motor to air pump (my saw has no air pump, so I don't need this particular belt)
2. Motor to Variable Speed Drive (VSD)
3. VSD to gearbox
4. Gearbox to speed indicator driver

A member at Practical Machinist contacted DoAll and they provided the following Gates belts:

Application	Gates Number	Standard Size	Length (in)	Circumference (in)	Max. Width (in)	Height (in)
Motor to air pump	2260	4L260			1/2	5/16
Motor to VSD	3370	5L370	36.92	37	21/32	3/8
VSD to gearbox	3510	5L510	50.94	51	21/32	3/8
Gearbox to speed indicator	1290	3L290	28.68	29	3/8	7/32

But I found some of these numbers are not accurate for my machine! Specifically, the main drive belts - motor to VSD and VSD to gearbox - were wrong. After some careful measurements, I'm using the following:

Application	Gates Number	Standard Size	Length (in)	Circumference (in)	Max. Width (in)	Height (in)
Motor to VSD	3340	5L540	53.94	54	21/32	3/8
VSD to gearbox	3540	5L340	33.91	34	21/32	3/8

I'm using only Gates Truflex belts.