New forum member

Moderators: HopefulSSer, admin

DLB
Platinum Member
Posts: 1985
Joined: Fri Nov 08, 2019 11:24 am
Location: Joshua Texas

Re: New forum member

Post by DLB »

The version of the procedure that RFGuy attached above is one of several variations published by SS in manuals and/or videos. It has some assumptions built in that I reckon will lead to unquantifiable errors. Two things I dislike compared to the OP's suggestion:

1) There's no correction or measurement for angular runout on the face of the blade as assembled. If I pick a different line on the blade to use as a reference I'm going to get a different alignment. Neither of those lines is likely to be an accurate representation of the plane of the cut. If I use a blade or arbor with a lot of runout (relatively) I get more error than one with little. Bottom line: This, IMO is a poor reference to align to. Many of us do not use this reference, at a minimum selecting a line on the blade that is predicted to produce better results. An alignment plate seems to be preferred by those that have them.

2) We use a square of varying quality and unknown accuracy along with human error in its use. It's probably pretty good, but the typical woodworker's method for squaring a square is not extremely accurate.

These are both nits, I get it. My point is that the simple method introduced here eliminates them. I don't have a particularly good 1/2" reference cylinder so I'm not using this method, but I do see its merit. I think that table tilt is meant to be referenced to the axis of rotation.

IMO the alignments in the manual are designed to achieve reasonably close alignment using only equipment that came with the Mark plus a combination square. We generally accept that introducing better equipment, including references, results in an improvement that is in the eye of the beholder. If we had an accurate way to measure results we could evaluate the two methods, perhaps someone with an alignment plate and precise square could test the theory and see which method introduces more error referenced to the alignment plate.

- David
HopefulSSer
Gold Member
Posts: 398
Joined: Sun Oct 03, 2021 6:07 pm
Location: NC

Re: New forum member

Post by HopefulSSer »

DLB wrote: Thu Nov 18, 2021 10:44 am I don't have a particularly good 1/2" reference cylinder so I'm not using this method, but I do see its merit.
I'm not even convinced that's necessary if one's rod is stiff enough (apologies for the double entendre ;) ). If one rotates the quill shaft, runout caused by the rod or chuck will be amplified at the free end -- the longer the rod, the greater the amplification. So rotate the quill such that the maximum runout is positioned at TDC (top dead center), then rotate another 90º. It should be easy enough to eyeball it very close. Now all the runout is confined to the front-back plane where it doesn't matter. Then align the table parallel to the rod.

Intuitively it seems to me that this eliminates a lot of variables inherent in the "standard" procedures. As you note David, the standard procedure involves establishing a reference 90º to the axis of the quill (with error possible in the blade, the arbor, or both), then aligning the table 90º to the first 90º (with error possible in the square itself, the size of the square, or its use). So there are multiple opportunities for error, whereas the proposed method eliminates all these and aligns the table to the quill axis directly. The opportunities for error are not shifting the runout into the front-back axis, droop in the rod, and measurement errors between the rod & table. The blade, the arbor, and all the 90º-off-of-90º-off-of-the-axis translations are removed from the picture.

(...says the guy who can't get two 90º crosscuts :D viewtopic.php?f=10&t=27242 )
Greenie SN 362819 (upgraded to 510), Bandsaw 106878, Jointer SS16466
RFGuy
Platinum Member
Posts: 2740
Joined: Mon Dec 31, 2018 8:05 am
Location: a suburb of PHX, AZ

Re: New forum member

Post by RFGuy »

HopefulSSer wrote: Thu Nov 18, 2021 11:11 am If one rotates the quill shaft, runout caused by the rod or chuck will be amplified at the free end -- the longer the rod, the greater the amplification. So rotate the quill such that the maximum runout is positioned at TDC (top dead center), then rotate another 90º. It should be easy enough to eyeball it very close. Now all the runout is confined to the front-back plane where it doesn't matter. Then align the table parallel to the rod.
To a 1st order, this "might" work. It assumes that if the shaft is non-straight that it has a simple bend or deflection to one side ONLY. What if it is more complicated than this, e.g. a bend and a twist as just one example. I guess my point is now we are eyeballing stuff in the alignment procedure and hoping that it is straight along one axis and perfectly precision ground and flat. This introduces a bit more human error, in my mind, than simply eyeballing any gaps between a precision machinist square and the table/sawblade. JMO. What I don't like with this procedure is it assumes a dead flat main table from Shopsmith which is seldom the case with this aluminum table. I much prefer a machinist square right next to the sawblade and contacting the table surface nearest to the blade as reference. The most critical reference surface in this procedure, in my opinion, is the main table section right next to the blade because this is what sets a square edge for a saw cut. So, I want this small section nearest the sawblade to be the main reference surface, not the section of the main table farther away from the blade.
HopefulSSer wrote: Thu Nov 18, 2021 11:11 am Intuitively it seems to me that this eliminates a lot of variables inherent in the "standard" procedures. As you note David, the standard procedure involves establishing a reference 90º to the axis of the quill (with error possible in the blade, the arbor, or both), then aligning the table 90º to the first 90º (with error possible in the square itself, the size of the square, or its use). So there are multiple opportunities for error, whereas the proposed method eliminates all these and aligns the table to the quill axis directly. The opportunities for error are not shifting the runout into the front-back axis, droop in the rod, and measurement errors between the rod & table. The blade, the arbor, and all the 90º-off-of-90º-off-of-the-axis translations are removed from the picture.
What about errors between the drillpress chuck and the sawblade arbor? Runout errors, concentric errors and orthogonality errors would seem to exist to me when comparing an alignment done using the drillpress chuck versus the sawblade arbor. They might be small, but without data I am having a hard time saying they are negligible. This procedure assumes essentially no errors exist between a drillpress chuck and a sawblade arbor in terms of runout, orthogonality, concentricity, etc.
📶RF Guy

Mark V 520 (Bought New '98) | 4" jointer | 6" beltsander | 12" planer | bandsaw | router table | speed reducer | univ. tool rest
Porter Cable 12" Compound Miter Saw | Rikon 8" Low Speed Bench Grinder w/CBN wheels | Jessem Clear-Cut TS™ Stock Guides
Festool (Emerald): DF 500 Q | RO 150 FEQ | OF 1400 EQ | TS 55 REQ | CT 26 E
DC3300 | Shopvac w/ClearVue CV06 Mini Cyclone | JDS AirTech 2000 | Sundstrom PAPR | Dylos DC1100 Pro particulate monitor
HopefulSSer
Gold Member
Posts: 398
Joined: Sun Oct 03, 2021 6:07 pm
Location: NC

Re: New forum member

Post by HopefulSSer »

RFGuy wrote: Thu Nov 18, 2021 12:05 pm
To a 1st order, this "might" work. It assumes that if the shaft is non-straight that it has a simple bend or deflection to one side ONLY. What if it is more complicated than this, e.g. a bend and a twist as just one example. I guess my point is now we are eyeballing stuff in the alignment procedure and hoping that it is straight along one axis and perfectly precision ground and flat. This introduces a bit more human error, in my mind, than simply eyeballing any gaps between a precision machinist square and the table/sawblade. JMO. What I don't like with this procedure is it assumes a dead flat main table from Shopsmith which is seldom the case with this aluminum table. I much prefer a machinist square right next to the sawblade and contacting the table surface nearest to the blade as reference. The most critical reference surface in this procedure, in my opinion, is the main table section right next to the blade because this is what sets a square edge for a saw cut. So, I want this small section nearest the sawblade to be the main reference surface, not the section of the main table farther away from the blade.
I think the assumption that the rod has only a simple bend or curve is a pretty safe one, but if the rod really pretzeled then you'd find another rod. ;) If your rod is, say 24" long and there's a 0.125" runout on the free end (easily visible), then that's an angle error of less than 0.3º.

And I wouldn't say it assumes a dead flat table; rather I'd say it averages out any irregularity in the table over the table's entire width (given a rod that spans the table). I get your concern about the area nearest the blade, but I think that depends too on the size of the stock you're cutting. If you're halving a 2x4 sheet of plywood then table irregularity is probably not a concern. If you're cutting a piece that's only 3" long then obviously it's a big concern. So, yes, that's a big caveat. Which reminds me, I need to lay something resembling a straight edge on my table.....
What about errors between the drillpress chuck and the sawblade arbor? Runout errors, concentric errors and orthogonality errors would seem to exist to me when comparing an alignment done using the drillpress chuck versus the sawblade arbor. They might be small, but without data I am having a hard time saying they are negligible. This procedure assumes essentially no errors exist between a drillpress chuck and a sawblade arbor in terms of runout, orthogonality, concentricity, etc.
They're independent of each other. Runout in the chuck can be compensated for as I outlined above. And arbor runout, the way I understand it, can't give you a non-perpendicular cut can it? Axial runout (i.e. blade wobble) in the saw arbor should give you a wider kerf but a cut still perpendicular (orthogonal) to quill axis, yes? (Extreme example is a wobble-blade dado cutter) Radial runout (i.e. lack of concentricity) would make the blade hop up & down but should not change the kerf and couldn't make the cut non-perpendicular.

I appreciate you going into the weeds & geeking out with me on this. :) Thanks! Good discussion!
Greenie SN 362819 (upgraded to 510), Bandsaw 106878, Jointer SS16466
RFGuy
Platinum Member
Posts: 2740
Joined: Mon Dec 31, 2018 8:05 am
Location: a suburb of PHX, AZ

Re: New forum member

Post by RFGuy »

RFGuy wrote: Thu Nov 18, 2021 12:05 pm What about errors between the drillpress chuck and the sawblade arbor? Runout errors, concentric errors and orthogonality errors would seem to exist to me when comparing an alignment done using the drillpress chuck versus the sawblade arbor.
HopefulSSer wrote: Thu Nov 18, 2021 1:04 pm They're independent of each other.
This is exactly my point! Any errors inherent to the drillpress chuck are independent from any errors inherent in the sawblade arbor. This is why I have been arguing that using a "1/2" dia ball bushing shafting" in drillpress mode to set the 0°stop is not a good idea because it may not compensate for these errors. Let's say that the drillpress chuck is perfect, but perhaps there is a 0.2° off-axis on the sawblade arbor perhaps. Setting the 0° stop using the drillpress chuck and the "1/2" dia ball bushing shafting" will still result in a 0.2°error to square. Using the manufacturer alignment procedure would have corrected for this 0.2°error to square. Hopefully the off-axis error is far less than this, but Shopsmith doesn't provide any specs on their arbors or chucks and I don't have equipment to measure this.
📶RF Guy

Mark V 520 (Bought New '98) | 4" jointer | 6" beltsander | 12" planer | bandsaw | router table | speed reducer | univ. tool rest
Porter Cable 12" Compound Miter Saw | Rikon 8" Low Speed Bench Grinder w/CBN wheels | Jessem Clear-Cut TS™ Stock Guides
Festool (Emerald): DF 500 Q | RO 150 FEQ | OF 1400 EQ | TS 55 REQ | CT 26 E
DC3300 | Shopvac w/ClearVue CV06 Mini Cyclone | JDS AirTech 2000 | Sundstrom PAPR | Dylos DC1100 Pro particulate monitor
HopefulSSer
Gold Member
Posts: 398
Joined: Sun Oct 03, 2021 6:07 pm
Location: NC

Re: New forum member

Post by HopefulSSer »

How does arbor runout cause an error to square? Given some amount of axial runout, the blade would be wobbling but the wood would see both the high and low sides which would cancel each other, albeit creating a wider kerf. Or said another way the line of the cut would be defined by the low (near) side of eccentricity, not the high (waste) side. Yes? I'm not picturing any way a rigid spinning saw blade can create anything other than a cut orthogonal to the blade's rotational axis....
Greenie SN 362819 (upgraded to 510), Bandsaw 106878, Jointer SS16466
DLB
Platinum Member
Posts: 1985
Joined: Fri Nov 08, 2019 11:24 am
Location: Joshua Texas

Re: New forum member

Post by DLB »

Though I'm critical of the method in the manual I use a variation of it. Cuz I have all the needed stuff. But, I can rotate the blade and readily see that this method is unreliable. Tilt measures 0 degrees only to the specific place that I used, and only on that blade. There are two obvious ways to improve on this. One is to use a blade with less runout. The other is to identify the part of the blade with maximum runout and use a part of the blade 90 degrees from that for the alignment. Both improvements can be used together. This is the same logic, and the same reason, that you use the same tooth on the blade when setting the miter slot parallel to the plane of rotation. Why would we accept that uncorrected saw blade runout would create an unacceptable error for setting the table square to the blade in one axis but the same error is acceptable in the other axis?

- David
User avatar
SteveMaryland
Gold Member
Posts: 187
Joined: Sat Nov 06, 2021 3:41 pm
Location: Baltimore, Maryland

Re: New forum member

Post by SteveMaryland »

Thomson ball bushing shafting specs attached.
Attachments
THOMSON SHAFTING SPECS.pdf
(90.12 KiB) Downloaded 146 times
Mark V, Model 555510, Serial No. 102689, purchased November 1989. Upgraded to 520
User avatar
JPG
Platinum Member
Posts: 34610
Joined: Wed Dec 10, 2008 7:42 pm
Location: Lexington, Ky (TAMECAT territory)

Re: New forum member

Post by JPG »

Gotta back up on this. The original post(top of this thread) stated a method for adjusting the "table level" aka tilt 0° stop.

Somewhere it got moved 90° to a drill press positioning doing essentially the same thing.????

BTW I would use a 1/2" router chuck/adapter for this rather than a drill chuck.
╔═══╗
╟JPG ╢
╚═══╝

Goldie(Bought New SN 377425)/4" jointer/6" beltsander/12" planer/stripsander/bandsaw/powerstation /Scroll saw/Jig saw /Craftsman 10" ras/Craftsman 6" thicknessplaner/ Dayton10"tablesaw(restoredfromneighborstrashpile)/ Mark VII restoration in 'progress'/ 10
E[/size](SN E3779) restoration in progress, a 510 on the back burner and a growing pile of items to be eventually returned to useful life. - aka Red Grange
RFGuy
Platinum Member
Posts: 2740
Joined: Mon Dec 31, 2018 8:05 am
Location: a suburb of PHX, AZ

Re: New forum member

Post by RFGuy »

.
Last edited by RFGuy on Fri Nov 26, 2021 6:21 am, edited 1 time in total.
📶RF Guy

Mark V 520 (Bought New '98) | 4" jointer | 6" beltsander | 12" planer | bandsaw | router table | speed reducer | univ. tool rest
Porter Cable 12" Compound Miter Saw | Rikon 8" Low Speed Bench Grinder w/CBN wheels | Jessem Clear-Cut TS™ Stock Guides
Festool (Emerald): DF 500 Q | RO 150 FEQ | OF 1400 EQ | TS 55 REQ | CT 26 E
DC3300 | Shopvac w/ClearVue CV06 Mini Cyclone | JDS AirTech 2000 | Sundstrom PAPR | Dylos DC1100 Pro particulate monitor
Post Reply