Two ballscrews on both X and Y axis?

On my router the distance, measured parallel to the X-axis, from the Y-axis (across gantry) ballscrew and spindle centre line is about 300mm. This allows the tool to deflect in the Y direction since the support is effectively 300mm away and the aluminium in between and the screw will bend since the linear bearings offer little support...

Attachment 4368

The obvious solution to this is to put the screw closer to the spindle (ideally intersecting the axis of rotation of the spindle) but realistically it's still going to be at least 120mm away.

So why not put another ballscrew on the other side of the Z-axis and add a second gantry cross piece on which one linear bearing will rest? Then the distance makes little difference and you've also pretty much eliminated the problem of the gantry twisting and helps with racking:

Attachment 4369

(Bad drawing I know... add one ballscrew mounted beside each of the Y-axis rails.)

Clearly the added cost of a second ballscrew is prohibitive (£65 for me), but I think it's worth the extra. I would probably link the two screws with a timing belt as adding another stepper and driver is expensive.

Tricky bit is mounting the Z axis ... I could use the same argument to have 2 ballscrews on Z, but that's getting rather silly.

Johathan, it seems to me that you've just outlined the classic comparison between bed mills verses a gantry mills. For any specific machine frame's total mass, more well-selected boxing-in material allocations make more rigid structures than "open C-shapes" like bed mills. Extending ideal mass allocation concepts to their extreme, as Buckminster Fuller explained ad nausium, triangular enclosures are more rigid than 4-sided box structures. Yet when we examine the most widely selected industrial machines, open C-shape configurations are more popular than fully-boxed double-upright gantry designs. Open bed mills can provide wonderfully satisfying human work/play space experiences because they minimize components "in the way" blocking work space access. Just off hand, I don't know of any triangularly supported mills, despite the fact that vertical space is often available which could accommodate upward projecting triangle intersections.

Perhaps these comments will trigger some inventive souls to design even more rigid mill configurations for any selected total mass. I have not seen conversations about rigidity/mass ratios related to machine tools comparable to aircraft and human-powered vehicle design conversations. Yet that kind of elegance seems appropriate. Lessons we've learned about frame design strategy in every venue should be applied to every other venue, including mill frame design. If one's goals include minimizing cost/performance ratio, maximizing rigidity/mass ratio, fitting machines within available environmental space, traditional design paths probably have not yet been optimized. I'd love to see a community of clever souls sharing thoughts and developing insights which evolve into a new and better high-performance mill class. With effort comparable to what many spend working cross-word puzzles, which in 5 decades effort generates zero gain to the world, we might make a wonderful contribution toward making human populations more productive so they can lead better lives.

If a pole-supported tent's center joining fixture supports a couple dog-chains leading down to a swing seat, that tent's support structure's apparent rigidity can instantly be increased many fold by adding person's weight to that center-supported swing seat. Similarly, top loading a box-like supported mill could greatly improve its effective rigidity. That top loading material need not be expensive. Don't be constrained by popular model design strategies.

Since I first studied and compared various rotary to linear motion conversion designs, I have always remembered that classic screw drive's high internal friction converts most input energy into waste heat, so we see lots of 30%-efficient range screw drives. But that same comparatively high friction acts as an effective one-way clutch. You can drive all screws by rotating them, but the driven load can't drive the screw unless its super steeply tapered.

Compare stepper motors driving ball screws verses driving acme screws. Highly energy-efficient (often around 90%) recirculating ball drives can't perform as a one-way clutch, so the motor MUST resist load-induced driving forces, unless you add an electrically released spring-loaded brake. By comparison, a stepper motor driving an acme screw can't be pushed about by the load because that force can't spin the screw. More expensive stepper motor electric drivers commonly allow the person setting up the system to select one of several motor de-energizing routines triggered by motor command inactivity. When a motor is not being commanded to spin, one choice would be to hold motor driving electric current constant, wasting power and keeping motors hotter than the work requires. That's how low-priced drivers behave. Another choice is to taper driving current down to about 30% after 100 milliseconds inactivity. Another choice can be dropping driving current to zero when triggered by inactivity. But if you're using micro-stepping and don't happen to stop on a full step position, dropping current to zero would let the stepper motor's permanent magnets rotate it to the magnetically-closest full step position. Inactivity-triggered electric unloading enables stepper motor driven devices to become much more electrically efficient than low-end drivers cause. Motor heating functionally depends on average electric load. So if your driver reduces or disconnects power during inactive periods, you motors are running cooler so you can usually briefly over-drive them with higher than "rated" voltage during peak acceleration periods without turning motor windings into a melted fuse. If you're running recirculating ball drivers on your vertical Z axis, I'd definitely suggest adding one counter weight if its an open-C configuration, or two opposite-side counter weights if its a gantry configuration. The goal is to make net average vertical force "seen" by the ball screw(s) approach zero without introducing parallelogram-distorting forces.

Just some thoughts triggered by your inquiry.
Enjoy the "mind candy."
John

Hi Jonathan

I've only had a very small think about this, but the ball screw moves it, the bearings hold it. You say the linear bearings ar not supporting? They should though? They look like they are hanging free from either end? Anyway, can you can mount a dial gauge on the bearing rod, or the bearings themselves and pick up exactly where the flex is. I bet it's the x bearing rods flexing? It's probably not the screw at all, in fact, moveing the screw closer may allow the bearings to flex more (or the flex at the tool will be greater because the screw forms a triangle?). Mine has an ali extrusion that supports the rod, the bearings have a break in them, they cover about 270 degrees of the rod. So the bearing is supported all the way across. You could upgrade and squeeze some rails in there but ££££s. To me, if you dont want to spend much, and your relying on the screw as an anti twist device of sorts cuz the bearings have play, move it away from teh bearings if you see what I mean?

You could add a third (or fourth) bearing, pref at the bottom, other side of ball screw so although the top still deflects, this will be minimised.. Whats the travel? I think there is a ground rod about a meter long and bearing (bearings if I can find them?) on the floor of my LR (Land Rover) (about an inch dia) (don't ask why!). You can have it for the postage mate if it's of any use?

Tom