My battle vs Router Table Sag

[alex grams]

So I used to have a mast-r-lift on my sawstop extension wing. It sagged over 3-4 years to an unacceptable level. I would loved to have found a 30" cast iron router wing, but they are all in 27", so I opted to build one.

It is made from 1.25" thick side grain hard maple, with 3/16"x1"x1" angle iron on its udnerside. The end boards and breadboards are 2.75" tall, and the breadboard pieces are lag bolted with 2.5" lag bolts into the table.

The finish is BLO, then 2-3 coats of Arm-R-Seal, followed by about 4 coats of wax for a nice slick, low friction finish. There are two 3" casters on the underside of the cabinet which allow it to move when the saw is lifted on its base.

http://www.sawmillcreek.org/attachment.php?attachmentid=211027&d=1319431362

http://www.sawmillcreek.org/attachment.php?attachmentid=211026&d=1319431350

http://www.sawmillcreek.org/attachment.php?attachmentid=211024&d=1319431330

http://www.sawmillcreek.org/attachment.php?attachmentid=211025&d=1319431340

[pat warner]

Photgraphs well.
In spite of your yeoman's job, you have no immunity to cup, twist, or bow.
Don't be heart broken if a corner sticks up or here or there.

[Mike Zilis]

That is a sweet setup. I love the router fence attached to the table saw fence. Mike

[Michael Weber]

Purty! I considered attaching a RT fence to my TS fence but worried that not locking down on the back side would lead to flexing if stock was pressed against the fence very hard. Has that been a problem? I have a unifence and it does flex if you press on the outfeed side.:mad:

[alex grams]

Pat, that was always a concern of mine in the design and build. If the table bows, it will be doing so against some significant securing to the table and the guide fences. It is bolted about every 6 inches along its connecting faces with the fence guide and the cast iron top.

Another precaution I took was in securing the angle iron to the top. I realized I could always have an upward bow in the table, as long as the router was at its apex. I sanded down each end of the table about 1/16" lower than the apex. Then when I bolted the angle iron to the underside, I shimmed the centerbolts with a few washers, and the next outer bolts with 1 washer, and finally the last bolt with no washers. This puts the angle iron in a bending moment which pushes up on the center of the table at the router lift.

If anything were to move it would be expansion of the table longitudinally (perpendicular to the cut direction of the tablesaw). This is fine to expand, as it has no effect on the cut of the router).

[pat warner]

Ordinary maple moves ~3/16"/ft/season. No amount of fastening will change that, expect surprises.
You have my condolences.

[alex grams]

Pat, bolt holes through the skirts are oversized to allow for tangential movement.

By my calculation, the maximum movement I will see is about .2" over the length/tangent of the center board wood grains. The wood can expand that direction as much as need be with no affect on the cutting path for the router.

Not sure where you got 3/16" per foot, but I have not seen anything nearly that drastic in my workbench (assembled via same method). At most on the breadboards of my workbench throughout the course of a year I get about 1/16" movement over a 36" perpendicular grain surface.

For reference, here is the wood movement calculator I used (http://www.woodworkerssource.com/movement.php)

[pat warner]

Whilst I do read the engineering numbers with respect to time, temp.,%RH, EMC, dimensionality/seasonality and so on, I rely essenitally on what I've done, seen, and fixed myself.
I am no stranger to slip/slide and allowance thereof. I have made a study of it, the hard way.
Let me just say one thing.
Precision routing, if that is of concern, requires precise material prep and a router table that is not mishapen.
Now if your router table is flat (+ or - .002" is generous), & your fence is straight and square to the table top, you have a good chance with respect outcome. Ignore the dimensionality or metro-analysis (measurement) of either the the table or the stock: Expect surprises.
And it would not surprise me to see square/straight/ground straight edges rock & roll or your table top on a weekly basis.
One week across the diags. the next across the width, then splits spontaneoulsy.
Not looking for a fight here just presenting the truth, and hoping my woodworking brothers & sisters suffer less than I did.

[Peter Aeschliman]

Pat, I've spent some time on your site.

No intention to hijack this thread, but since this is a discussion about router table flatness, I thought it would be okay to discuss here.

In particular, I've taken interest in your approach to the router table top and how simple you've kept it- one piece of MDF with hardwood webbing underneath, no router plate, etc. My current router table- 2 face-glued pieces of 3/4" MDF- has sagged and needs replacing.

Given this discussion about how hardwood expands and contracts, causing problems with router table flatness, why doesn't your hardwood support structure cause your mdf top to lose flatness as well? The same question could be asked of any torsion box as well, if hardwood is used for the webbing.

[Greg Mann]

Whilst I do read the engineering numbers with respect to time, temp.,%RH, EMC, dimensionality/seasonality and so on, I rely essenitally on what I've done, seen, and fixed myself.I am no stranger to slip/slide and allowance thereof. I have made a study of it, the hard way.Let me just say one thing.Precision routing, if that is of concern, requires precise material prep and a router table that is not mishapen.Now if your router table is flat (+ or - .002" is generous), & your fence is straight and square to the table top, you have a good chance with respect outcome. Ignore the dimensionality or metro-analysis (measurement) of either the the table or the stock: Expect surprises.And it would not surprise me to see square/straight/ground straight edges rock & roll or your table top on a weekly basis.One week across the diags. the next across the width, then splits spontaneoulsy.Not looking for a fight here just presenting the truth, and hoping my woodworking brothers & sisters suffer less than I did.Pat,You know more about routing accuracy than just about anyone but I would like to point out a flaw in your presentation here that I see made quite often by others. Flatness is expressed as a uni-directional tolerance, that is, from zero to a plus number only, with zero representing perfection. In that context, when you use plus or minus .002, are you actually referencing .004 total variation or .002? This may seem like nitpicking but the difference between the two values is significant to someone trying to achieve your recommendations for accuracy.Put another way, it really doesn't make sense to say the flatness is minus .002 ( or straightness, parallelism, or squareness for that matter). If one wants to talk about the desirability of flatness in terms of convex versus concave, that can be another matter, but it is still flatness from zero and should be further described as either concave or convex. Would you care to comment on which condition is more acceptable, assuming perfection (zero) isn't achieved?

[Bruce Page]

Very nice Alex, I don't think you'll ever see any sag with that setup. Did you ever get quotes on the steel fab?

[Harvey Pascoe]

Plus or minus is the right way to express it when a dual axis is involved so that -.002 equals downward or concave variance from perfectly level. Here the variation remains only .002 because the deviation can be in one direction only. Contrasting this with variance in a rotating shaft a +/- variance of .002 would be for a total of .004 variance. So there is no flaw in Pat's statement.

I just wonder if .002" isn't a bit extreme and what kind of work you do that demands that tight tolerance normally reserved for precision metal work? Mine certainly doesn't. .002 is the thickness of a fine plane shaving.

[JohnT Fitzgerald]

That looks pretty.........pretty awesome!The table on my SS sagged significantly even without a router lift. I wanted to install a lift but wanted to get a flat table first. I went back and forth with SS customer service a few times describing the problem, sending photos, etc. The table design is fundamentally flawed IMO - far too little bracing underneath; some stretches secured only to each other, not the table; soft wood (could be pine) for the stretcher material. SS sent me a new table, which is still waiting to be reinforced before I'll install it. I am curious as to whether or not they have improved it.

[Jeff Monson]

Looks great Alex, if your profiles are off .002 from seasonal movement....who cares in my book. Enjoy the table and very nice work.

[Howard Acheson]

>>>> In particular, I've taken interest in your approach to the router table top and how simple you've kept it- one piece of MDF with hardwood webbing underneath, no router plate, etc. My current router table- 2 face-glued pieces of 3/4" MDF- has sagged and needs replacing.

Let me add some to Pat's comments.

I have always advocated NOT constructing a router table with a miter gauge track. It serves no purpose except to attach some types of devices that hold the workpiece against the fence. There are other options for that. However, the miter slot creates a weak area in the table. I challenged members of another forum to identify any router table operation that required a router table with a miter slot. There a number of other ways to do end of board operations. Nothing was presented. Keeping the router table simple is the best approach.

>>>> why doesn't your hardwood support structure cause your mdf top to lose flatness as well? The same question could be asked of any torsion box as well, if hardwood is used for the webbing.

Almost all the change of size of solid wood is across the grain. This means the a web support system for a torsion box type support must have all the grain running parallel to the top surface. The box will and can expand in thickness but all parts will move equally and flatness should not be compromised

[Keith Hankins]

That is a sweet table love the finish. I built my router table over six years ago and my top is two 3/4 mdf pieces with formica on top wrapped in oak. My plate is a woodpecker PRL and a PC 7518 in it which aint a feather. The cabinet was Norm's plans and I have to say 0 sag in all that time and I've hogged a lot of sawdust through it.

[pat warner]

"why doesn't your hardwood support structure cause your mdf top to lose flatness as well? The same question could be asked of any torsion box as well, if hardwood is used for the webbing."

A torsion box it is not. Torsion boxes are stressed on all 6 surfaces. My assembly (http://patwarner.com/images/selecting4.jpg) is essentially frame and 1 skinny (5/8") skin. There are 6 beams under the 2' x2' top, ~ 1.25" x 2.37" in section that are in the same plane. The top surfaces of all stix were granite sanded flat (to<.002"/L) individually and slip dovetailed into the same plane on 4 rails & legs. After assembly, some sanding and scraping was required anyway. The MDF is held down with 12 1/4-20 FH machine screws through the top and tapped into the Ash skeleton below with no allowance for movenment. Strong? Indeed, but still subject to deformity.
The 4 compartments making up the spaces below the top are pretty well divided up so their seasonality/dimensionality changes are small/increment and don't screw up the top x much.. Notwithstanding, whilst the MDF has a low K of expansion, it still will change shape. It is not perfect but it is simple on its face. Anybody could reproduce it without heroics.
For awhile (after assembly) I used a granite flattened hardback sanding block to flatten the booger (MDF) out.
It is still not flat but all 10" radians aound the cutter hole are within ~.002". And that is close enough for the limited work I do. Amazing that something so cheap & simple could stay flat and last this long (>8 years now).
So bottom line: Most router table tops made from wood, ply and man-made panels will change shape.
Whether stressed x a 14 pound router hangin', or change in humidity, the SOB's change shape.
Flatness critical? Then consider seasoned, ground cast iron.

[Greg Mann]

Plus or minus is the right way to express it when a dual axis is involved so that -.002 equals downward or concave variance from perfectly level. Here the variation remains only .002 because the deviation can be in one direction only. Contrasting this with variance in a rotating shaft a +/- variance of .002 would be for a total of .004 variance. So there is no flaw in Pat's statement.

I just wonder if .002" isn't a bit extreme and what kind of work you do that demands that tight tolerance normally reserved for precision metal work? Mine certainly doesn't. .002 is the thickness of a fine plane shaving.


Sorry Harvery, but I still stand by my statement. It is not always true that a surface can deviate in one direction only. In reality, flatness measurement where it is most often done, on a co-ordinate measuring machine, will average out all data points taken and create a best-fit average that runs through the middle, with the highest and lowest being the same value, regardless of where they fall in the pattern. But they are not express as plus and minus but as a total positive value.

There is no standard that says -.002" represents a concave condition. Flatness is a profile condition and is always expressed as a positive value with zero representing perfection. Why? Because a surface can be concave in one area and convex in another, .002 up and .002 down, making it .004 out-of-flat.


Variance in a rotating shaft is called run-out, and it is never described as +/- either. It is often referred to as 'Total Indicator Runout' and it makes absolutely no difference where the zero point of the indicator falls. You are correct that readings on an indicator of +.002 to -.002 is .004 runout, but the same would be true if the numbers were .002 and .006.


FWIW, I wasn't the one that was worried about the flatness and I agree that .002, or even .004 is probably just fine for a router table. Very few of us really have any good way of checking this condition with any certainty anyway. The thing that really counts is the quality of the work being done. I often chuckle at some of these concerns about how flat router tables are or aren't but I do agree with Pat that the flatter the better and one can solve some mysteries in their work by starting off in a good place. And finally, precision metal work is how I make my living, not that is important to this conversation.

[Peter Aeschliman]

Nerd Alert!!! :p;)

[Chris Parks]

I have never understood why woodworkers think that thickness makes a router table either flatter or less susceptible to bending. If you want a table that does not move don't build it using wood based material and when you do build it it might be a good idea to do a bit of thinking as to structural integrity. Someone here somne years back was building a torsion box aluminium table but I don't recall the outcome. It would be easier and better to use ferrous materials, glue rather than weld and then machine the top surface. Mind you flat for the purpose of the exercise is most probably like the old saying of near enough is good enough. I bet no wood based router table is flat in the true sense of the word and they work so why get all anal about it.

[Chris Tsutsui]

I like your router table top as well. Mine is just thick MDF.

I use my router table for various tasks and if the routing was off by 1/64th then I'd be happy.

I think some times my work piece won't be flat to begin with... That's what sandpaper is for. :)

[Greg Mann]

Read the whole post, Peter. I totally agree the flatness issue often goes over the top. My original intent was to ask Pat to clarify what he meant by +/- .002. Was he saying .002 or was he saying .004? That is a difference of 100%. My goal was clarification according to a standard way of expressing flatness so that it could be clear to everyone. Unfortunately, it begins to sound a little more technical than one would like. Then Harvey chimed in with his interpretation, well intentioned, but still misleading. Letting that go meant nothing gets accomplished toward clarification.

I would still like Pat to comment regarding what he originally meant (.002 or .004). I would also like his opinion about flatness of RTs in general. FWIW, my personal opinion is that they only need to be flat enough to give you the results you are looking for. If someone isn't happy with their outcomes then flatness may be ONE of several things they might want to explore. I am willing to bet if we polled everyone here what the flatness of their table is the most common answer would be, "I have no idea."

[Neil Brooks]

No position on the argument (I love a good geek-fest, but ... I got nothin', on this one ;)), but ...

a) that's a beautiful RT, and

b) when I built the table for my RAS, I did the "Mister Sawdust" method, using steel box sections, epoxied, in grooves, between two sheets of marine plywood. So far ... it's been fantastically stable, but ... I wouldn't dare put numbers to that :)

Nice job !

[alex grams]

Wow, my thread that spawned a monster and keeps coming back! Must be Halloween or something

In regards to flatness: My goal wasn't to be flat to the hundreths or even thousandths of an inch.

As I said, the table apex's at the router plate was the goal, and to prevent the significant sag experienced by hanging a heavy router + lift on the factory sawstop extension table.

In regards to the comment:

I have never understood why woodworkers think that thickness makes a router table either flatter or less susceptible to bending.
I am not trying to start an argument on engineering, but the thicker the table, the less it will bend from loads/forces/gravity. It is as fundamentally related as the RPM on your tires are to how fast your car is going. Though thickness is completely independent of flatness.

I realize the wood will move, and have taken that into consideration, given the annual humidity gradient where I live and its change throughout the year. I think even in the worst of move movement scenarios (short of a catastorphic failure of a joint/glue) as long as the axis of feed over the router bit does not sag or twist, then I will be happy.

As a final note (and not to sound like I am criticizing anyone) but I find the woodworkers focus on machining precision to be kind of peculiar. I can understand if we are building an engine and tolerances are in the hundredths and thousands of an inch, but for a medium such as wood, i really don't see the return on spending so much time checking that last thousandth.

I think[/URL] Chris Tsutsui[URL="http://www.sawmillcreek.org/member.php?46583-Chris-Tsutsui"] (http://www.sawmillcreek.org/member.php?46583-Chris-Tsutsui)'s post summed it up pretty well for 90% of the creekers.

[Matt Cook]

Wow, that table is beautiful!

One thing I didn't notice in your post is how the work surface itself is supported over the length of the work surface and how the support of the work surface is supported to something capable of supporting the weight of the work surface.

I have extended experience building metal fabrication tables and, in most cases, sag is due to a lack of support over a distance and, ultimately, the lack of support of the weight. In your case, a 1 x 1 x .188 piece of angle iron, though heavy in your hand, isn't really going to add any strength for you over a distance. The working support of that angle iron, when attached to the base of your work surface, is really only a single 1" x .188 leg of the angle iron. The second leg, though useful for attachment, doesn't add much strength, only weight and that leg is even reduced in strength each place you drill a hole for attachment. The larger the hole, the more it reduces it's strength. If nothing is attached/supporting this angle iron, any flex in the table will surely bend the angle and defeat it's ability to support the work surface. If it were me, I'd be thinking overkill - 4" x 4" x .250 or .375 angle iron supported through the base every 12 inches. Unless you can mathematically calculate the forces at play, it may be difficult to spec a support that will truly keep the table straight(er) and you have to lean towards overkill to just get close.

Also consider that the rails of the Sawstop (I have a similar one to you so I've seen it in pieces) attach to your work surface with bolts in the horizontal way. These bolts offer minimal support and, given the factory bolts are countersink style with a root hole that's much larger than the fastener itself, can easily move within the rail and wood. They almost don't count for anything other than keeping the work surface where you put it at assembly plus the weight of material being cut. They're certainly not the right fastening method for supporting weight.

Looking at your pictures I just noticed that, though the butcher block design looks awesome, it may have worked to your favor to assemble the boards 90* to how they are right now. I would think the length of the board is stronger than a glue joint (with or without fasteners).

I'll let the more experienced comment on wood deviations over time, I have minimal knowledge in that area but I do think that the structure to which the table attaches must do it's part to reduce any movement or it will just compound the problem.

[pat warner]

"I would still like Pat to comment regarding what he originally meant (.002 or .004). "

Let's not get too academic; nothing is black & white. E.g. I have seen flatness expressed as +/-, net, deviation/unit length and so on. Wood, ply and most organic stuff made x man or machine changes shape.
And incidentally, the worst specs I've seen on substrate sheet stock include extruded aluminum and plastic laminate, (whether paper, linen, jeans or whatever). Moreover, Nylon, Polyethylene, PVC and polycarbonate flatness specs are outrageous.
Ground aluminum and steel can't be beat.
For clarification, on MDF, the surface can pretzel, thus the +/- values. Medite, the stock I use, claims they thickness sand to a thickness spec of .002". I.e., a 3/4" slab varies from .749 - .751". Not flatness but thickness. So I use it for those surfaces that are not stressed. Put a couple of 150 + pound toggles on MDF and you've bent the hell out of it. A router table surface is another matter; it's passive and unstressed.
Now about the deformity in MDF.
Most any surface that can suck up water is going to change shape. Most of the time, surface stock is glued, screwed, bolted or clamped down on some assembly you hope won't change shape. One face of the slab is open to the air, the other side less so. The upside is usually the drier of the 2 and therefore shorter and therefore it cups up. Notwithstanding, the slab could cup on the downside if the cabinet is hot inside & highly ventilated. It can go ether way.
So what? Well for close work the question is how much do want to sand? A + or - .002" (.004" net) surface is too much deformity for joinery. A cope & stick cut could be displaced x .008", a country mile. Glue joints, tenons, and most joinery cuts that remove stock from both sides could be out from 0 to .008". I would not want to sand 8 mils of anything. Decorative profile cuts can also divot or step if your table is not flat. Feed poorly milled stock to the cutter, and anything can happen.
The deviation on my table when I made these cuts (http://patwarner.com/images/dovetailed-lap.jpg) was .0015".

Maybe you can tolerate it, but somewhere (usually @ the C/L) along your stick you'll find a step that might bother you & require sanding. I don't want to sand that out either. So for me, & I cut wood, metal + plastic, I shoot for a .002" total deformity or less.
As an aside, you folks, with plastic, aluminum or steel router inserts already know what I'm taking about. You're used to adjusting for these changes. Your stock or sled always (often) bangs into the insert, right?

[alex grams]

Matt,
I had considered running the boards along the other axis, but I wanted the boards to run perpendicular to the direction of sag, and have their length against that force.

I thought about your comment on the steel reinforcing and its sum effect on the strength of the table. and ran some bending moment calculations. Alltogether, assuming the axis of its bending moment resistance is coplaner with the surface of the table, the 2x1"x1"x1/8" pieces of angle iron add about 50% more strength to the tabletop against the bending forces. For the sake of ease of the calculation, I assumed the router lift + router load was 100lbs, and that the bolts holding the table to the fence rails are free to rotate and offer no moment arm resistance against the loads.

[Carl Beckett]

I am an engineer (mechanical). I know all about GDT, old school tolerancing (prior to CMM), structural design, flatness, creep, twist, deflection, etc etc.

I think its a great table! And will no doubt be capable of producing many fine projects for years to come.

I have a router top that is two layers of MDF with shellac on top. A heavy Bench dog lift with a heavy PC router in it. Its supported underneath pretty well by the cabinet structure, but its been great for several years (no noticeable sag - but I havent bolted it up to a CMM either)

Congrats Alex - for building a fine piece of shop equipment and for providing a source of lively discussion on SMC!! Nearing in on 1000 views - cant say I have ever posting anything that generated that interest.

:)

-C
(like to keep this forum as positive a tone as possible - with open debate but positive!)

[Matt Cook]

...the 2x1"x1"x1/8" pieces of angle iron add about 50% more strength to the tabletop against the bending forces. For the sake of ease of the calculation, I assumed the router lift + router load was 100lbs, and that the bolts holding the table to the fence rails are free to rotate and offer no moment arm resistance against the loads.

Interesting. I'm not classically educated in mechanical engineering however shouldn't the error factor be more than 50% considering the factors don't include a force being applied to the work surface other than the surface itself? Examples would be you, yourself, leaning on it while feeding material, a stack a material resting on it during other ops (obviously I don't have knowledge of your processes), or any other "thing" that finds it's way onto the table top?

Another thought - I have a lift too and when I cut the hole for the lift in the middle of the table I was curious as to how a hole like that would impact the strength of the table.

And from way out there - does the vibration of the table saw and/or router create a harmonic? Seems I remember discussing an engineering failure where a walkway that was built based on the standard engineering formulas didn't take into account the environment and the harmonics from a band playing below caused it to fail. Twas a long time ago in school so I may have it a little wrong.

I might have missed this...but where is the sag in the top? Is it dead middle or on an end? I've been assuming it's the middle but knowing otherwise, I'd probably change the way i'm thinking about this.

Matt

[alex grams]

Matt, the removal of the material from the table to mount the router will of course weaken the table, but I was trying to roughly gauge the added strength from adding the steel reinforcement.

The router weighs about 15lbs, and the lift weighs about 15lbs (rough values taken from shipping weights of the products from Amazon). That leaves another 70lbs for the weight/force exerted on the table for me pushing.

It was a rough calculation of an ideal scenario of the forces being applied to the table and its resistence to bending under those forces. Some factors which will affect the calculation (positively and negatively) are:

-Weight of the Table
-The bolts on the ends securing the table provide some bending moment resistence (though probably a minimal amount)
-The bolting pattern/spacing of the steel to the table and its effective transference of moment/load to/from the table.

There is no sag at the moment in the new table. It actually hogs by about 1/16" of an inch in the center over the front/back.

[Peter Aeschliman]

Read the whole post, Peter.

Just teasing you man. I meant nothing by it!

[Matt Cook]

There is no sag at the moment in the new table.

I apologize, I thought were were trouble shooting this table. Somehow I missed that part...now that I'm caught up, I now understand. I thought were trying to figure out why this table was sagging and, based on your design, I didn't get it but thought we were gifted with something to ponder.

To unwind and start over, I'll only add that that is one beautiful router extension. Nice work! ;)

[Greg Mann]

Just teasing you man. I meant nothing by it!Ok, I got it.

[Homer Faucett]

Alex,

Fantastic looking router table. After reading all the hubbub your table has caused, I am only left to ponder how on earth woodworkers with maple benches were ever able to make a single worthwhile piece of furniture! ;)

But, then again, I know they were well beyond our time, as they were almost exclusively using cordless tools. Thank goodness I'm a woodworker and not an engineer . . . I might not ever get anything done.:D