There are a bunch of benchy speed runs out there, and I find it fascinating. This thread got me thinking. We have all sunk a bit and been amazed at it plowing through really deeply just fine. How about we intentionally push things?
What sort of speed runs could we do?
I was thinking a specified depth (12mm?) in MDF since it should be very similar in all regions? We of course would need to factor in accuracy. So time alone does not count accuracy would have to factor in. Scoring algorithm ideas?
I could make a specific sized DXF (the v1 logo?) and all else is fair game. Bit size, shape, type, router, etc.
No way that is part of the variables. I’m excited to see how a $4 bit does against a $40.
I used the router by hand (GASP!!!) with a expensive 1/4" single flute yesterday…I didn’t even know I was cutting anything. I thought I forgot to lower the bit past the guard. I planned on testing the current single flutes by hand today against that expensive one to see how they “feel”.
In the end, I just want to see how fast this can go.
I think the logo is a good test as there are no flat spots to measure. So all measurements are diagonals (invlolve multiple axis), and you have multiple to measure. So if you measured outside corners you not only have “how close to x.xxmm are all three measurements”, but how much does each measurement vary from the next from the same machine during the same cut!? Speed, precision, and accuracy all in one. I am open to another shape though.
For a speed run, you should mostly care about speed, but you should post some dims. It’s just for bragging rights, right? So people posting their total time and some dims. Bad dims still count, in my book.
In school when we did things like this we had scoring equations. Speed would be the largest factor but if two people got the same speed, accuracy added in. Just need to decide how much accuracy is a factor.
Think of it this way, if you scaled the cut a tiny bit you would cut less material=faster. I know no one is in it for cheating but it helps to eliminate the people who build is 20% off for some reason, or a accidentally scaled dxf.
So like 60% speed+40% accuracy. In CNC what is the point of speed without accuracy? Should you use that finishing pass or not…
Also to, like fighting or racing, should there be a class hierarchy? I’m sure a bone stock 30x30" C build won’t fare well against a heavily modified thick-wall 18x18" stainless build with NEMA 34s.
Total score = Total time * (1 + abs(error/expected size))
So if you cut something 95mm that should be 100, in 10.00 mins would be a score of 10.0 * (1+0.05) = 10.5min? You could square the error, but honestly, is anyone going to go 5mm over or under and submit the score?
This is the sort of math I am not good at, I am not sure how to think about the boundary cases other than checking them. I am not being picky I think this is just an interesting math problem, Your equation works.
I guess this all sort of depends on actual cut times and part size how heavy the error weighs in.
If we did Score=Time(s)+20*abs(error(mm)) By adding as time comes down error matters more, change the multiplier to increase the weight of the error.
So 600s+30*.4=612.
If we did 3 measurements we can weight accuracy.
Score=Time(s)+20(abs(error(mm)+abs(error(mm)+abs(error(mm))
600s+20(.2+.4+.3)=618
Or precision
Score=Time(s)+20(error(mm)+error(mm)+error(mm))
600z+20(.2-.4+.3)=602
Or both…
Score=Time(s)+20(abs(error(mm)+abs(error(mm)+abs(error(mm))+20(error(mm)+error(mm)+error(mm))=620
Sorry, fun with math, who knew I would ever really say that…dam my high school self would slap me right now if he was here.
Idk, I think weighing too much on dims could leave out what a race might indicate… like how fast you can rough cut. In that regard cuts can be off quite a bit. You can have a roughing race where it is pass/fail with a simple limit on leftover material >0. Maybe another race for finishing cuts, with a specified DOC and step over, where accuracy is factored in.
Edit… maybe even break it down in classes like auto racing… different classes depending on volume rate.
Honestly, the unit conversion is what is making that look funny to me.
If there’s no real “skin in the game”, and the benefits are just bragging rights and s, then I say keep it simple. As long as it looks like the original, if it was super fast or deep, I would respect it.
Is there a way to do a visual comparison? If we took a picture of it on top of a ruler/object of reference, could a graphical tool be used to measure degree of similarity, and use it in an equation?
I just did a safe cut, the time to beat is 15:33 seconds. I think I can cut it in half pretty easily. I just did a run at what I think are safe recommended values. Now…time to play!!
I think I will start a new thread. I will post it after this one.
.
s, then I say keep it simple. As long as it looks like the original, if it was super fast or deep, I would respect it.