They are in mm measured in from the anchor points but that’s not an intuitive number and it’s not a number that is right for every frame making a default value hard to come up with that works for everyone out of the box.
What if we made those into percentages instead something like:
I like the idea… I’ve not got my Maslow yet but I’d assume it would be easier to physically measure distances in mm than try to work out that same distance as a percentage? Again, I’ve not got my Maslow yet so I’ve been unable to test so I’m going purely off assumptions.
I think the % is confusing. I like the earlier suggestion that you do a basic calibration in the middle of the board and then drive it to the corners manually to define the corners of the board in belt coordinates. When calibration finishes it can calculate the corners in real world coordinates.
I have been thinking about making a dialog that allows you to define all the parts of your frame in a representative diagram, that just populates the maslow.yaml values… i.e. a picture of a “frame” with a “spoilboard” that you just plug in values for the sides of your frame and your spoilboard (calibration area). This dialog could then just populate those values in the configuration and save the file/ reboot like you do when entering values. not sure how to represent the Z values or the others but this seems more intuitive than entering values in a yaml or even in a list.
I would like ultimately to be able to use the Maslow to ‘teach’ the three or four corners of the plywood to be cut so that I can make use of the factory edges and in particular to use the Maslow as a panel ‘saw’. As a panel saw it can cut up a sheet in ways that a panel saw cannot and that make some projects fit into a single sheet that otherwise would not.
To me this points us towards teaching the corners rather than setting offsets, especially since the work surface may be (almost certainly will be) rotated a little relative to the machine coordinate system.
If you aren’t going to cut to the edge, you could just teach it once and leave those as defaults.
Interesting way to use Maslow. Not sure though if this would be very effective. A panel saw (or any circular saw) rips a sheet in a second or two. Maslow would have to do multiple passes at much, much lower feed rates. And as you point out already, the work surface (the surface that needs to be cut) will be rotated relative to the machine. Rotated differently with every sheet you put on the spoil board. So cutting at a precise angle (e.g.90 degrees) relative to the edge of your sheet will not work.
I have been thinking about exactly this too. It seems like 100% the way to go. I think that we can visualize the calibration process better too, something more like this:
It seems like percentage isn’t getting a ton of love, so here’s another idea. What about defining the size of the grid IE the default grid size is 3.5x7 feet or something like that so no matter what size your frame is the grid will fit on a sheet of plywood.
There’s often a key difference in software that’s been written by actual engineers (electrical, electronic, mechanical, fire, civil, sewerage, etc. - with due apologies to everyone I’m about to offend) vs. professional software developers (without some Eng qualification).
The software designed by engineers has every permutation of option available in some inscrutable form, because they’ve actually designed for someone with a shared background to themselves. Which includes assuming that the users have the required knowledge to understand the interactions/relationships/ratios between those options and parameters. And that information will be front and centre, because that is what they want to see first.
Professional software engineers (other) will generally try to hide as much of that as possible. And if they really know their stuff, they’ll provide a tool to help get all the options aligned correctly with an explanation of their inter-relationships, but only the minimum amount of information for users to make an ‘informed decision’.
I think that is an INCREDIBLY important point and that it is very important that we move in the direction of simpler / less options exposed to make it easier.
If we ask users to make five choices in a row and they get them right 90% of the time (pretty good) that still only works out to it working less than 60% of the time. It’s especially true since calibration is the very first thing that you do which makes it harder.
Ok but that assumes you’re going to be going full size from the outset and a few users (me included !) are going for table top sized horizontal set ups.
I think what @dlang came up with is a good start - let the Maslow find a 3x3 or so grid in the centre then motor out to your corners of the spoil board or your desired cutting area, fix the 4 points then calibrate from there.
The problem is that I don’t have a panel saw, nor do I have room for both the Maslow and a panel saw. I was originally going to try to design a frame that would serve as both, but then I realized that the Maslow could function as a panel saw, with the added plus that it can cut up a panel in ways that a panel saw cannot.
You are quite right about the rotation issue, and that is why I’ve advocated for the ability to set three corners of a workpiece to rotate the user coordinate system used for cutting. This is also needed for anyone who wishes to use the factory edge, and maximize the utilization of the workpiece. Plywood is expensive these days.
The increased time doesn’t bother me. I can do other things in the shop while it runs.
Plus, since I work alone and rarely have someone available to help, manhandling a full sheet of say 3/4" ply onto sawhorses horizontally and then cutting it up with my circular saw and guide is a real pain.
Professional software engineers (other) will generally try to hide as much of that as possible. And if they really know their stuff, they’ll provide a tool to help get all the options aligned correctly with an explanation of their inter-relationships, but only the minimum amount of information for users to make an ‘informed decision’.
Absolutely right. “Nominal” user experience should be as simple as possible, but no simpler. However, I really dislike UI experiences that ignore the fact that there ARE expert users and complexities that they need access to. Ideally the user interface is simple enough to be easily understood by both advanced and beginning users, and make it somewhat obvious to those that need access to the details have a path to that, but for most users, they don’t need to care, and its obvious that they “don’t need to click that thing”
This is by no means easy to accomplish, and takes iterating on not only designs, but also software, and talking through this with people that have different needs.
I think your problem is easily fixed, I’m assuming you’ll have a vertical frame?
After calibration, screw a strip of 9mm ply on the spoilboard, under and slightly overlapping the work area. Let the maslow do a horizontal cut on it right at the lower edge of the work area. The strip will now function as a ledge to put the factory edge of your sheet on, and any horizontal cut will be parallel with the factory edge.
Hope this helps,
That’s a really good idea, and perhaps should be part of the standard setup process when you build a new frame. Perhaps also do the same thing with a strip on the left edge so that the location of the work piece is set in both X and Y.
That would only really help if you are using full panels. I very often am using smaller pieces centered in the workspace for the most accuracy (carryover from my Maslow1 experiences but may not be as big a deal with Maslow4).
I’m instead planning to put a 60deg v-bit into the router and then have it carve a very shallow set of horizontally and vertically spaced lines across the entire spoil board so that I have reference lines across the entire work area when I’m trying to line a piece up.
Another possibility would be to use the Maslow to ‘drill’ a series of holes on a grid in the spoilboard. You could set the small piece onto pins inserted into the holes and square it up in x and y. Of course the pin holes would slowly get destroyed over time.
Not only is this not easy to do, it’s especially hard for the system designer
who knows all the details and knows why they may be needed. It takes iteration
with new users (and once the new users have a bit of experience, you need new
‘new users’ as the older ‘new users’ start having asumptions. Heathkit used to
consider anyone who had build 4 kits to be too experienced to be a valid tester
for their beginner level kits)
The current process is a good beginning, but as we get more people involved, we
can do better to simplify things, and only present what they need to answer (and
can know how to answer) at each step.
ideally not requiring them to measure and type in numbers where we can avoid it.
Based on the discussion here, let me make sure I am making the right assumptions by asking some questions, because I’m not sure I 100% understand all the items at play here. I think there are two subjects being discussed.
First, “calibration” - I see this as a defined area within the anchor points for maslow to use to run its routine to figure out where the corners of the frame really are. You would always want this to have a support surface under it so maslow does not wander off that and get tangled up.
Is this bounding box later used as limits for future cutting operations, or does the maslow allow you to go outside that bounding box? If it will allow you to go outside, then great. make the initial calibration “safe” and well inside your cuttable area but large enough to get a good measurement. If not, then maybe a suggestion of building strips of material outside where the cut area is to keep maslow from falling into unsupported areas during calibration? This one is dear to my heart as my maslow did wander off the spoil area during calibration.
Second, “workpiece setup” (I’ll call it). Once “calibration” is done, how to define the size, direction and location of the workpiece such that its aligned correctly.
I don’t think that calibration has to be run again at all for this. This is simply a matter of alignment within the defined area, is that correct?
