where did I go wrong - repeatability..

[mound]

Trying to figure out where I went wrong. Making half-lapped square lattice for a fence - a lot of it. 
I was cutting 1-1/4" dados across 6"x8' boards (which would later be ripped into 1-1/4" strips)

I used a cross-cut sled with an 8' bed, and used my dado-stack at 3/4". Thus, each half-lap cut required two passes.

I made a wooden "key" which was exactly the width of my dado-blade and mounted it to the sled to the left of the blade.

For each 8' length, I carefully marked and cut the first 1-1/4" dado cutting to knife-marked lines and verified a snug fit. 

Then, much like cutting box joints, I simply moved my workpiece over to the left each time, indexing off my pin. 

But because each dado required two passes to get the 1-1/4" width, I'd first reference the left edge of the key against the left edge of the previous dado, then slide it over so the right edge of the key referenced the right side of the previous dado.

See image:



By all accounts that I can think of, this should have led to full repeatability. As long as my first dado was the right width, I'd expect all subsequent ones to be the same width, and same spacing.

But what I found was that somewhere, some small error was accumulating, and I'd get a few cuts into my 8' length and find that the width of my resulting 2-pass dado was getting greater and greater.. I'd adjust and keep going. (BIG MISTAKE- the first time I noticed this happening I should have done a full stop to figure out why.)

In the end, this progressive error meant that when it came time to fit together the half-lap lattice (4'x8' panels of it) they didn't just snap together like Lincoln Logs, rather, I had to tweak cuts and introduce gaps in the half-laps to make things fit. What a nightmare. 

I was able to more or  less correct the situation by ganging the parts together, and using the router and a dado-jig (which I probably should have done from the start, in hindsight) to cut new "best fit" and in the end, the panels look more or less fine. (it is a fence after all) and you really have to look closely for the gaps. 

But I just can't wrap my head around how the setup would lead to progressively wider dados.. If the first dado was exactly 1-1/4" and the pin was exactly the width of the dado-stack then why would they get progressively wider??

PSA - If you are ever building a fence and decide to make a bunch of 8'x4' lattice panels by hand, stop and consider buying them instead! 

[6270_Productions]

Cumulative error?

If your cut is off by 1/16 of an inch, after 16 cuts you will be off by 1 inch.

(Just a guess.) 

[mound]

That's what I'm getting at though - where is there any cumulative error coming in?

If my first "hand cut" dado were cut wrong, say 1/16" larger than my desired 1-1/4", I would expect all of my dado's to be off by 1/16".  I wouldn't expect that error to add from cut to cut though. 

I know there was an error, and it added, I'm trying to understand the source of that error.

[Gary G™]

The force of the blade can move the workpiece.
I think you needed two keys: one for the initial cut, and one to lock the 1-1/4" cut in place.

[mound]

Hmm hadn't thought about the blade moving the workpiece. Each cut was just my hands pushing the board into the sled. 
I had thought about two keys as I was getting started but couldn't see how I could place two keys on the sled without interfering with each other.

[CLETUS]

PSA - If you are ever building a fence and decide to make a bunch of 8'x4' lattice panels by hand, stop and consider buying them instead! 

Making lattice would have never crossed my mind!

[Alan S]

You set the width of each dado based on the width of the previous dado, so a minor increase at each stage adds up. 

If you made the key narrower than the blade by the amount you want the dado wider than the blade, and went through and made one cut for each dado always pushing the stock to one side, you would wind up with a set of fairly evenly spaced dados of the width of your blade.  Now you can go through and widen the dados by pushing the stock to the other side of the key, making sure to reference an unwidened dado.  You can widen all but one dado this way, and the errors will not accumulate.

While they won't accumulate, they will be a bit larger than they'd be if you made the widening pass referenced from the dado you are widening, rather than the next one over.  This can be done, but requires a change of key before the widening pass.

[mound]

ahh I see what you mean.. Yup I think that's the answer.

[Routerman]

Simply deflection in your jigging.
A common error. All text books dictate similar indexing.
All results stink and are variable.
Indexing, stops, crowders, hold-ins, etc have to be rock solid.
The work cannot move and all the parts must have a common pre-selected end and face.
A terrible lesson to learn the hard way.

[Steve N]

That's what I'm getting at though - where is there any cumulative error coming in?

Without being there to measure what you have I'd say your 3/4" dado isn't, or your spacer you are using to adjust the cut with isn't correct, possibly both. It only needs to be off a squiggle, remember accumulative error is what is happening.

---

Other possibilities is your saw is simply out of adjustment, or you have a slight angle on your dado blade. Make sure it is true 90* to the table.