I have an old Delta Homecraft 8-inch Bench Model Tilting Arbor Saw that
my father bought used in the 1950s. This saw is a patchwork quilt of
components, the core being made by Delta, the wings having been
fabricated by somebody with a sheet metal shop and a welder, the motor
being a GE model 1E161, which is an industrial 115 Vac (60 Hz, single
phase) 3/4 HP capacitor-start 3450 rpm continuous-duty unit with
built-in overload protection (the original motor is 1/2 HP), all on a
shop-built wooden base that doubles as a sawdust collection bin.

The problem is that the arbor ball bearings are contaminated, and turn
roughly to the hand. It feels as if some sand has gotten into the
bearing grease. This is actually due to the bearing grease having
completely dried out. Time to replace the bearings.

First chore is to obtain a pair of replacement bearings. The bearings
are New Departure model ND-87503. While the New Departure company went
out of business long ago, many firms still make these bearings, which
were very widely used. On the web, there are many model 87503 bearings
on offer with wildly varying prices and quality claims, and many of the
offered bearings are reported in multiple user reviews to have very
short service lifetimes. The solution turned out to be a local
industrial bearing house (Action Bearing Company, 201 Brighton Avenue,
Boston, Mass 02134), which had ND87503 bearings made by KSK in Japan,
for US $10.27 each.

In the following description, Delta part and figure numbers are
referenced, based on ŒHomecraft 8² Bench Model Tilting Arbor Saw *
Operating and Maintenance Instructions¹ (Delta document PM-1722 dated
9-19-49, 16 pages).

PM-1722 is available: http://vintagemachinery.org/pubs/1141/3643.pdf.

It is assumed that the reader will have a copy of PM-1722 in hand
while
reading the following.

Now that I have the bearings in hand, time to install them. First issue
is how to get at the arbor. My initial approach was to unbolt the
cast-iron table top (TAB-101) and remove it from the sheet-metal housing
(TAB-110-S). This approach was abandoned because it was too much
trouble to get the two hand-wheel shafts disentangled from the sheet
metal housing because the tension pins (SP-2711) are difficult to remove
undamaged without a special tool to prevent the pin from splaying and
jamming as it is driven out.

The second approach was to unbolt the entire cast-iron table plus
sheet-metal housing assembly from the wooden base and place the table
plus housing assembly as a unit upside down on my workbench, after the
motor plus motor bracket assembly (and the saw blade) were removed and
set aside.

Removing the motor and motor bracket (TAB-109 and TAB-117 and Figure 5)
assembly as a unit is easily done, as shown in Figure 4. Disconnect the
power cord at the motor (after recording which wire goes where), loosen
the two square-headed set-screws (SP-301 in TAB-109), and slide the
complete motor bracket assembly off the two half-inch posts (Figure 3,
TAB-130 protruding from TAB-105).

With the saw table assembly upside down on the workbench, one can remove
the arbor assembly (TAB-106-A) by loosening the 1/4² square headed set
screw (SP-301) in the front trunnion (TAB-104 in Figure 3), and tapping
the half-inch diameter steel hinge shaft (TAB-125) out with a short
drift punch and a small hammer. One rotates the hand wheels (TAB-165-S,
Figures 9 and 10) for blade height and blade tilt angle as needed to
maneuver and extract the arbor. A pair of wide black steel wave spring
washers (Figure 3, TAM-184) will also come out as the arbor is removed.

Now that I¹ve removed the arbor once, I could probably do a subsequent
remove and replace without removing the table plus housing assembly from
the wooden base, but having the assembly on the workbench and easily
seen was essential the first time, as I was not able to fully visualize
how things were arranged in three dimensions from the figures in PM-1722
alone.

Once the arbor assembly (Figure 7) is liberated, one has the problem of
taking it apart. There are two nuts to remove, both at the pulley side
of the arbor. The first is a 7/8² hex nut (BM-23) on the arbor shaft
(TAB-137-S) * this requires a deep 6- or 12-point socket to reach down
into the arbor while clearing the arbor shaft stub. The arbor shaft was
clamped by its hog ring (against which the saw blade is clamped using
nut BG-23 and dished washer LCS-8) in a big vise with copper jaws, but
the shaft tended to slip in the vise, and a little impact was required
to break the nut free.

The second is a custom spanner (ring) nut (BG-12), with two 0.092² wide
radial slots spaced 180 degrees apart to accept a special spanner
wrench. Attempts to undo this nut by driving it with a drift punch in
the slots only served to mangle the slots, the nut being well recessed
inside the arbor casting (Figure 3, TAB-106-A), and so use of the punch
was abandoned.

Using a lathe and a vertical mill, fabricated the needed special spanner
wrench from a piece of 12L14 steel left over from another project. The
business end of the wrench is a cylinder 1.568² OD and 1.270² ID, with
one end milled across the diameter such that there are two protruding
ears, each ear being 0.090² wide and 0.100² high. These ears fit neatly
into the radial slots on the ring nut. The cylinder portion must be at
least 0.250² deep (not including the two ears) to reach down into the
arbor casting and fully mate with the ring nut within. There must be a
central hole through which the arbor shaft (TAB-137-S) protrudes (a
1.000² ID hole is generous), and a way to apply torque without
interfering with the arbor shaft. I drilled a pair of 0.257² ID holes
parallel to the arbor shaft to accept the pins of an Armstrong model
34-154 3-Inch Adjustable Face Spanner Wrench that I already had. With
this newly-fabricated spanner wrench, the ring nut was easily removed.
While there are many ways to fabricate a ring spanner wrench, the point
is that such a wrench is essential.

We now have both nuts removed. Now to remove the arbor shaft from the
arbor casting. This was accomplished by tapping on the pulley end of
the shaft with a lead hammer. The components that emerged are as
follows, on the saw-blade end: hog ring (an integral part of the arbor
shaft TAB-137-S), ball bearing (ND-87503), two narrow bearing-loading
wave springs (LTA-516), and a machined steel spacer (TAB-124). The
spacer is 2.018² long. Note that while PM-1722 (written in 1949)
specifies that one wave spring should be used, in fact a stack of two
spring washers was installed. One assumes that this is a later change,
to increase the bearing preload. At the pulley end, one ND-87503
bearing remained, stuck in the arbor casting. This bearing was removed
using an impromptu puller consisting of some half-inch all-thread, two
nuts and a big washer. The bearing was not pressed in place, but had
been glued in place by old dried oil. Everything was cleaned with
acetone and/or brake cleaner to remove all the dried oil.

One bearing is still on the arbor shaft, very close to the hog ring, and
couldn¹t be pried loose with screwdrivers and the like. A large gear
puller will make short work of this, except that the puller jaws are too
large to fit between bearing and hog ring. A large steel washer (2-1/2²
OD, 1-1/16² ID, 3/16² thick) was purchased from the local hardware
store, and a roughly square piece was removed to yield a C-shaped
washer. This was slipped between hog ring and bearing, allowing the
gear puller jaws to apply force to the old bearing, which promptly slid
free under the urgings of the puller.

The arbor shaft was then cleaned up with crocus cloth and solvent, ready
for installation of the first new bearing using the puller and a big
steel washer. This is a very light press fit. The lubricant used is
traditional for press fits, Castor Oil (which can be bought in small
quantities from drug stores, where it is sold as a mild laxative).
Mobil-1 synthetic grease (available at auto parts stores) would also
work.

The arbor was then reassembled with the new bearings in place. ND87503
bearings are asymmetrical, where the inner race protrudes beyond the
outer race on only one side. Note that while it¹s hard to be sure from
PM-1722, the bearings are installed back-to-back with noses protruding
outward, with the machined spacer between the backs.

It all took 12-14 unhurried hours, including the fabrication of the
ring-nut spanner wrench and the C-washer, the unsuccessful attempt to
take the cast-iron table off the sheet-metal housing, and some machining
of the left wing for a better fit to the table.

The above text and the photos listed below are in the Metalworking
Dropbox. Google for "Delta_Homecraft_Bearing_Replacement" without the
quotes.

Photo 1: Front view of the saw, for context.

Photo 2: Perspective view of the saw tabletop and blade.

Photo 3: Various components mentioned in the text. At the top is the
Armstrong Adjustable Face Spanner Wrench. The middle row is a thick
washer used for pushing a bearing up against the hog ring, a thick
washer with a piece removed for pulling a bearing away from the hog
ring, the shop-made spanner wrench for the ring nut, and a six-inch rule
for scale.

Photo 4: Closeup of the shop-made spanner wrench for the ring nut.

Photo 5: The back of the saw, showing the motor and motor bracket
attached to the rear trunnion. There is a trouble light in the saw
housing. The orange belt is a link-belt; this was a great improvement
over a standard V-belt.

Photo 6: Closeup of the saw blade (this is a 7-inch blade, although the
saw will accept an 8 inch blade). The rind shaft at the top of the
photo is the arbor hinge pin.


Joe Gwinn