Floyd Mah
05-17-2009, 8:39 PM
Riving knives are required in new machines, but I have a Delta contractor's saw that's over 20 years old. I've been improving it over
the years, ever since I discovered that the original fence guide bar was in the shape of an arc once it was assembled. The value of
the add-ons have more than doubled the price of the saw (Unifence, overhead blade guard, removable splitter, and a Grizzly sliding
table). (That doesn't include the Bridgecity Jointmaker that I turned it into, at a savings of over a thousand bucks.) Anyway, in
the interest of continuing to improve my saw on a budget, I designed an built a riving knife attachment for the saw. The total cost
is under $60, as rough guess. The following is a moderately complicated project for the average woodworker, but a simple project for
a beginning metalworker (like me).
The advantages of a riving knife is improved safety. There are many articles on their advantages, especially over fixed splitters
which are the most common in older American saws. I already have an easily removable Delta splitter which came with the overhead
blade guard, but I tackled this as an interesting problem. Now that I have created the knife, I realized that the problem is not
unlike making a military weapon. It had to be durable and sufficiently accurate, but also had to work under adverse circumstances. I
haven't had it on the saw for a long time, so I don't know how it will hold up, but I've tried to address the problems of durability
and accuracy. Because a saw blade will typically travel vertically about 3" (3.125" in mine) and the riving knife has less than 1/32"
clearance on either side of the cut, a respectable tolerance to achieve is less than 1/64" or 0.016" deviation (about 0.29 degrees) in
the movement of the riving knife relative to the plane of the saw blade. That is why my first attempt was a
prototype: it was not sufficiently robust or adjustable to be usable as a safety device. This is a description of my second effort. I used the prototype to get the usable dimensions.
The principal reason making a riving knife retrofit for a Delta contractor's saw a difficult process is that the blade is cantilevered
off a 6" arm. This arm is attached to a trunnion, which consists of front and rear cast iron plates, that pivots. The plates are
joined by two steel 1" rods running front to back, parallel to the blade plane, and below the blade center. The blade center moves in
an arc and no other parts of the trunnion moves with the blade except for the arm. I realized that the total arc swept out by the
blade center was only 4" out of a 31.4159" circumference, as did any point on the periphery of the blade. In other words, it
approximated a straight line for the most part. If I drew a tangent to the blade at the middle of its' travel, a riving knife moving
along the tangent could stay quite close to the blade. If I did not require that the knife tilt, the distance between the knife and
the blade would vary about 0.25". A knife support, tilted to be tangential to the midpoint of the blade travel (on the arm) would
guide the knife close to the blade. If this support were fastened to the trunnion, it would also tilt laterally along with the blade.
Thus the only requirement for implementation was to make a linear bearing that was parallel to the plane of the blade and
approximately parallel to a tangent to the midpoint of the blade's travel.
The original support for the blade guard, splitter, anti-kickback pawls for the Delta saw is a 5/8", fine threaded bolt, screwed into
the rear trunnion plate. This is a through hole, accessible from the front and back of the plate. A 6" bolt, with about 1.5" of
thread can be bolted to the back plate. It can be used to support the linear bearing portion of the riving knife assembly. Because
the linear bearing had to be parallel to the saw blade by the 0.29 degrees tolerance I calculated above, the linear bearing must be
adjustable so that it can be positioned in a repeatable fashion. In aircraft terms, this would be the "roll". The rigidity of the
5/8" bolt constricts the pitch and yaw variables.
Once the linear bearing guide is in place and aligned to the saw blade plane, a carriage has to be designed to move along the linear
bearing guide, but also constrained to not "roll" while it travels along the guide. I implemented this by using a 3/16" x 3" x 10"
plate attached to the 5/8" bolt and also a 13" long 1/2" diameter steel rod. They meet at a 106.5 degree angle.
This is a sagittal view:
this is the front of the machine ---->
bolt
____
. \ rod
. \
. \ The backslashes represent the guide rod.
^ These periods represent the guide edge of the plate
The plate edge runs parallel to the steel rod. The side of the plate away from the rod is the guide edge used to prevent "roll" of
the carriage. (Parts of the plate are cut away to reduce the weight of the assembly. This facillitates bending the plate.) After
assembly, the guide edge and the steel rod must be made co-planar. I did this by clamping the top end of the rod (and the attached
part of the plate) in a metal working vise and using a large wooden screw vise on the guide edge to bend the plate (that portion
connecting the guide edge and the rod). It's very important to keep the guide edge and the rod straight--just the intermediate
portion of the plate can be bent. I used a dial indicator on a very flat surface to measure the small changes I was making.
The carriage of the riving knife consists of a linear bearing, a support arm extending from the bearing, a guide pulley riding on the
guide edge of the plate described above and the attachment for the riving knife (part of the support arm).
The linear bearing is a piece of 1/2" pipe, length 4". Two brass end caps were drilled to fit the 1/2" rod and tightened onto the
pipe. This slides on the rod. A tighter fit along the 4" would make this bearing at risk for binding from the dust and debris inside
the table saw, so, by using the end caps, there is limited bearing contact. A more refined ball-bearing linear bearing will probably
fail from the harsh conditions inside the table saw. There are two attachments to this bearing.
The support arm is attached to the bearing. The rear end has an upright extension to hold the actual riving knife. The support arm
is made from a 10" strap hinge. This is the type of hinge used to mount gates on a bolt screwed into a fence post. I enlarged the
hole in the hinge to accommodate the 1/2" pipe (approximately 3/4"). Be careful to not weaken any side of the resulting hole by
removing too much metal. The pipe can be placed inside and fixed in place by screwing the legs of the hinge around it.
A hook in the shape of an "F" is attached to the support arm at the bearing. The "C" part of the "F" is 2.125" and fits over the
blade arm where the saw arbor is attached. The leg of the "F" is attached to the support arm. This causes the linear bearing to move
up and down with the blade arm. It is open so that the slight front-back motion of the blade arbor is not tracked, just the vertical
motion. By drilling and tapping into both the bearing and the support arm, two small screws can be inserted and fastened with
Lok-Tite, the heads cut off and used to mount and dismount the "F" hook (brass thumbnuts). In the adjustment process, this attachment
goes on and off many times, so that ease of removal is important.
I opened up a CD drive to see how they kept the tolerances on the diode carrier. The carrier rides on a rod and the roll is
controlled by constraining the opposite rod in a U-channel, with a leaf spring maintaining pressure of one side of the U-channel
against the second rod.
I machined a pulley from steel to fit on the guide edge of the plate above. This part would be somewhat difficult to do for the
average woodworker, but I guess this post if for someone who actually has the tools to tackle modification of major piece of
machinery. This pulley is attached to the support arm with a shaft that must be adjustable so that once the pulley is in place on the
guide edge, moving the pulley relative to the support arm alters the parallelism of the riving knife blade relative to the saw blade.
This is a crucial adjustment. I positioned it using feeler gauges and a caliper, but using an adjustable stop with a screw would have
been preferable. Besides the pulley I used a wavy washer to force the pulley to be in a tighter tolerance in its' range of movement.
Also, I mimicked the action of the leaf spring by unwinding part of a coiled spring, mounting it on the shaft of the pulley and
aligned the free end to press against the plate. Again, this proves to be crucial. When the trunnion is tilted for a bevel cut, the
weight of the assembly will force the pulley against the guide edge, but in the blade upright position, the spring is necessary to
create this force. In my final adjustments, this proved to be cause of at least .02" of variablility in the position of the riving
knife, way beyond the acceptable variation. I adjusted the shaft by creating a clamp for the shaft (hole, slit on one side of the
hole and a screw squeezing the jaws of the clamp).
Because the plate incorporating the 5/8" bolt can roll on the axis of the bolt, I stabilized it by using a U bolt on the right 1"
trunnion rod to mount a 1/4" x 20 toilet bolt so that the free end of the bolt projects radially out from the rod. The plate has a
corresponding mounting hole and the degree of roll is limited by clamping the plate between two nuts on this screw. The nut on the
rod side of the screw is a lock-nut, so that it is relatively stable. The nut holding the plate on the other side has a lock washer
to keep it from loosening in the saw. The hole in plate should be slightly oversize and relieved at the top outside edge and lower
inside edge to allow the plate to be swung out of the way when adjusting the locknut. Turning the locknut in or out will alter the
roll of the plate. The sequence is to unfasten the large nut, tilt the plate out of the way, adjust the lock nut, reverse the
operations to tighten things down. The carriage for the riving knife needs to be removed for this operation. If you have access to a
good machinist's protractor, you can preset the settings. I guess that this was the most time consuming part of adjusting the riving
knife.
The 5/8" bolt is mounted using a lockwasher and a washer. Because all of the apparatus keeps you from freely screwing in the plate
assembly, you need to screw in the bolt, then fasten the plate onto the bolt, and then tighten the nut holding on the 5/8" bolt. In
my final adjustment, I found that a deviation of 0.3 degrees made a visually detectable shift in the position of the riving knife. My
stop on the trunnion was probably off by this amount, so that just adjusting the linear bearing assembly to be perpendicular to the
table surface was not enough, since the blade was not actually perpendicular to the table surface. I would caution against overtightening the large nut, because breaking the casting would be catastrophic and it's not necessary.
Once this last adjustment can be made and the pulley adjustment can be made, then you can finalize the dimensions of the mount for the
riving knife and upright support. It's preferable to be able to adjust the position of the riving knife using shims, since the
ability to add or subtract a shim allows you to accommodate different blades. If you machine the holder for the riving knife to
exactly the dimension that you think is final, seemingly small tweaks to the adjustment can shift the riving knife position by easily
1/16" or 0.0625".
So, the plate adjustment is intended to align the linear bearing with the plane of the saw blade. The pulley adjustment rolls the
plane of the riving knife on the linear bearing so that it is coplanar with the saw blade. The shimming of the riving knife shifts
the riving knife so that it is centered on the saw blade.
Another note: Bigger isn't better. There is limited space in the box of the table saw. If the pipe for the linear bearing is too
long, you need more of the steel rod, and when the blade is tilted, it can catch on wiring inside, on the bottom of the cabinet or the
sides. Using too much of the rod to attach to the plate also lengthens the rod. I found that sometimes to fit the carriage assembly
required me to tilt the saw blade by 5 degrees.
As usual, sorry for the iPhone photos and no disassembled pictures, but I think anyone capable of doing this project can figure out
most details from my description.
the years, ever since I discovered that the original fence guide bar was in the shape of an arc once it was assembled. The value of
the add-ons have more than doubled the price of the saw (Unifence, overhead blade guard, removable splitter, and a Grizzly sliding
table). (That doesn't include the Bridgecity Jointmaker that I turned it into, at a savings of over a thousand bucks.) Anyway, in
the interest of continuing to improve my saw on a budget, I designed an built a riving knife attachment for the saw. The total cost
is under $60, as rough guess. The following is a moderately complicated project for the average woodworker, but a simple project for
a beginning metalworker (like me).
The advantages of a riving knife is improved safety. There are many articles on their advantages, especially over fixed splitters
which are the most common in older American saws. I already have an easily removable Delta splitter which came with the overhead
blade guard, but I tackled this as an interesting problem. Now that I have created the knife, I realized that the problem is not
unlike making a military weapon. It had to be durable and sufficiently accurate, but also had to work under adverse circumstances. I
haven't had it on the saw for a long time, so I don't know how it will hold up, but I've tried to address the problems of durability
and accuracy. Because a saw blade will typically travel vertically about 3" (3.125" in mine) and the riving knife has less than 1/32"
clearance on either side of the cut, a respectable tolerance to achieve is less than 1/64" or 0.016" deviation (about 0.29 degrees) in
the movement of the riving knife relative to the plane of the saw blade. That is why my first attempt was a
prototype: it was not sufficiently robust or adjustable to be usable as a safety device. This is a description of my second effort. I used the prototype to get the usable dimensions.
The principal reason making a riving knife retrofit for a Delta contractor's saw a difficult process is that the blade is cantilevered
off a 6" arm. This arm is attached to a trunnion, which consists of front and rear cast iron plates, that pivots. The plates are
joined by two steel 1" rods running front to back, parallel to the blade plane, and below the blade center. The blade center moves in
an arc and no other parts of the trunnion moves with the blade except for the arm. I realized that the total arc swept out by the
blade center was only 4" out of a 31.4159" circumference, as did any point on the periphery of the blade. In other words, it
approximated a straight line for the most part. If I drew a tangent to the blade at the middle of its' travel, a riving knife moving
along the tangent could stay quite close to the blade. If I did not require that the knife tilt, the distance between the knife and
the blade would vary about 0.25". A knife support, tilted to be tangential to the midpoint of the blade travel (on the arm) would
guide the knife close to the blade. If this support were fastened to the trunnion, it would also tilt laterally along with the blade.
Thus the only requirement for implementation was to make a linear bearing that was parallel to the plane of the blade and
approximately parallel to a tangent to the midpoint of the blade's travel.
The original support for the blade guard, splitter, anti-kickback pawls for the Delta saw is a 5/8", fine threaded bolt, screwed into
the rear trunnion plate. This is a through hole, accessible from the front and back of the plate. A 6" bolt, with about 1.5" of
thread can be bolted to the back plate. It can be used to support the linear bearing portion of the riving knife assembly. Because
the linear bearing had to be parallel to the saw blade by the 0.29 degrees tolerance I calculated above, the linear bearing must be
adjustable so that it can be positioned in a repeatable fashion. In aircraft terms, this would be the "roll". The rigidity of the
5/8" bolt constricts the pitch and yaw variables.
Once the linear bearing guide is in place and aligned to the saw blade plane, a carriage has to be designed to move along the linear
bearing guide, but also constrained to not "roll" while it travels along the guide. I implemented this by using a 3/16" x 3" x 10"
plate attached to the 5/8" bolt and also a 13" long 1/2" diameter steel rod. They meet at a 106.5 degree angle.
This is a sagittal view:
this is the front of the machine ---->
bolt
____
. \ rod
. \
. \ The backslashes represent the guide rod.
^ These periods represent the guide edge of the plate
The plate edge runs parallel to the steel rod. The side of the plate away from the rod is the guide edge used to prevent "roll" of
the carriage. (Parts of the plate are cut away to reduce the weight of the assembly. This facillitates bending the plate.) After
assembly, the guide edge and the steel rod must be made co-planar. I did this by clamping the top end of the rod (and the attached
part of the plate) in a metal working vise and using a large wooden screw vise on the guide edge to bend the plate (that portion
connecting the guide edge and the rod). It's very important to keep the guide edge and the rod straight--just the intermediate
portion of the plate can be bent. I used a dial indicator on a very flat surface to measure the small changes I was making.
The carriage of the riving knife consists of a linear bearing, a support arm extending from the bearing, a guide pulley riding on the
guide edge of the plate described above and the attachment for the riving knife (part of the support arm).
The linear bearing is a piece of 1/2" pipe, length 4". Two brass end caps were drilled to fit the 1/2" rod and tightened onto the
pipe. This slides on the rod. A tighter fit along the 4" would make this bearing at risk for binding from the dust and debris inside
the table saw, so, by using the end caps, there is limited bearing contact. A more refined ball-bearing linear bearing will probably
fail from the harsh conditions inside the table saw. There are two attachments to this bearing.
The support arm is attached to the bearing. The rear end has an upright extension to hold the actual riving knife. The support arm
is made from a 10" strap hinge. This is the type of hinge used to mount gates on a bolt screwed into a fence post. I enlarged the
hole in the hinge to accommodate the 1/2" pipe (approximately 3/4"). Be careful to not weaken any side of the resulting hole by
removing too much metal. The pipe can be placed inside and fixed in place by screwing the legs of the hinge around it.
A hook in the shape of an "F" is attached to the support arm at the bearing. The "C" part of the "F" is 2.125" and fits over the
blade arm where the saw arbor is attached. The leg of the "F" is attached to the support arm. This causes the linear bearing to move
up and down with the blade arm. It is open so that the slight front-back motion of the blade arbor is not tracked, just the vertical
motion. By drilling and tapping into both the bearing and the support arm, two small screws can be inserted and fastened with
Lok-Tite, the heads cut off and used to mount and dismount the "F" hook (brass thumbnuts). In the adjustment process, this attachment
goes on and off many times, so that ease of removal is important.
I opened up a CD drive to see how they kept the tolerances on the diode carrier. The carrier rides on a rod and the roll is
controlled by constraining the opposite rod in a U-channel, with a leaf spring maintaining pressure of one side of the U-channel
against the second rod.
I machined a pulley from steel to fit on the guide edge of the plate above. This part would be somewhat difficult to do for the
average woodworker, but I guess this post if for someone who actually has the tools to tackle modification of major piece of
machinery. This pulley is attached to the support arm with a shaft that must be adjustable so that once the pulley is in place on the
guide edge, moving the pulley relative to the support arm alters the parallelism of the riving knife blade relative to the saw blade.
This is a crucial adjustment. I positioned it using feeler gauges and a caliper, but using an adjustable stop with a screw would have
been preferable. Besides the pulley I used a wavy washer to force the pulley to be in a tighter tolerance in its' range of movement.
Also, I mimicked the action of the leaf spring by unwinding part of a coiled spring, mounting it on the shaft of the pulley and
aligned the free end to press against the plate. Again, this proves to be crucial. When the trunnion is tilted for a bevel cut, the
weight of the assembly will force the pulley against the guide edge, but in the blade upright position, the spring is necessary to
create this force. In my final adjustments, this proved to be cause of at least .02" of variablility in the position of the riving
knife, way beyond the acceptable variation. I adjusted the shaft by creating a clamp for the shaft (hole, slit on one side of the
hole and a screw squeezing the jaws of the clamp).
Because the plate incorporating the 5/8" bolt can roll on the axis of the bolt, I stabilized it by using a U bolt on the right 1"
trunnion rod to mount a 1/4" x 20 toilet bolt so that the free end of the bolt projects radially out from the rod. The plate has a
corresponding mounting hole and the degree of roll is limited by clamping the plate between two nuts on this screw. The nut on the
rod side of the screw is a lock-nut, so that it is relatively stable. The nut holding the plate on the other side has a lock washer
to keep it from loosening in the saw. The hole in plate should be slightly oversize and relieved at the top outside edge and lower
inside edge to allow the plate to be swung out of the way when adjusting the locknut. Turning the locknut in or out will alter the
roll of the plate. The sequence is to unfasten the large nut, tilt the plate out of the way, adjust the lock nut, reverse the
operations to tighten things down. The carriage for the riving knife needs to be removed for this operation. If you have access to a
good machinist's protractor, you can preset the settings. I guess that this was the most time consuming part of adjusting the riving
knife.
The 5/8" bolt is mounted using a lockwasher and a washer. Because all of the apparatus keeps you from freely screwing in the plate
assembly, you need to screw in the bolt, then fasten the plate onto the bolt, and then tighten the nut holding on the 5/8" bolt. In
my final adjustment, I found that a deviation of 0.3 degrees made a visually detectable shift in the position of the riving knife. My
stop on the trunnion was probably off by this amount, so that just adjusting the linear bearing assembly to be perpendicular to the
table surface was not enough, since the blade was not actually perpendicular to the table surface. I would caution against overtightening the large nut, because breaking the casting would be catastrophic and it's not necessary.
Once this last adjustment can be made and the pulley adjustment can be made, then you can finalize the dimensions of the mount for the
riving knife and upright support. It's preferable to be able to adjust the position of the riving knife using shims, since the
ability to add or subtract a shim allows you to accommodate different blades. If you machine the holder for the riving knife to
exactly the dimension that you think is final, seemingly small tweaks to the adjustment can shift the riving knife position by easily
1/16" or 0.0625".
So, the plate adjustment is intended to align the linear bearing with the plane of the saw blade. The pulley adjustment rolls the
plane of the riving knife on the linear bearing so that it is coplanar with the saw blade. The shimming of the riving knife shifts
the riving knife so that it is centered on the saw blade.
Another note: Bigger isn't better. There is limited space in the box of the table saw. If the pipe for the linear bearing is too
long, you need more of the steel rod, and when the blade is tilted, it can catch on wiring inside, on the bottom of the cabinet or the
sides. Using too much of the rod to attach to the plate also lengthens the rod. I found that sometimes to fit the carriage assembly
required me to tilt the saw blade by 5 degrees.
As usual, sorry for the iPhone photos and no disassembled pictures, but I think anyone capable of doing this project can figure out
most details from my description.