Profile cover photo
Profile photo
Dale Dunn
Dale's posts

Post has attachment
In response to a request from +Shauki B for examples of useful printed things: Farkles for my motorcycle.
6 Photos - View album

Post has shared content
Faster Marlin coming soon?
Great news  GRBL v09 is out!!!! (

Update Summary for v0.9 from v0.8
**IMPORTANT: Default serial baudrate is now 115200! (Up from 9600). And your settings will be over-written! Make sure to have a backup.
**NEW Super Smooth Stepper Algorithm: Complete overhaul of the handling of the stepper driver to simplify and reduce task time per ISR tick. Much smoother operation with the new Adaptive Multi-Axis Step Smoothing (AMASS) algorithm which does what its name implies (see stepper.c source for details). Users should immediately see significant improvements in how their machines move and overall performance!
**Stability and Robustness Updates: Grbl's overall stability has been focused on for this version. The planner and step-execution interface has been completely re-written for robustness and incorruptibility by the introduction of an intermediate step segment buffer that "checks-out" steps from the planner buffer in real-time. This means we can now fearlessly drive Grbl to it's highest limits. Combined with the new stepper algorithm and planner optimizations, this translated to 5x to 10x overall performance increases in our testing! Also, stability and robustness tests have been reported to easily take 1.4 million (yes, million) line g-code programs like a champ!
**(x4)+ Faster Planner: Planning computations improved four-fold or more by optimizing end-to-end operations, which included streamlining the computations and introducing a planner pointer to locate un-improvable portions of the buffer and not waste cycles recomputing them.
**Compile-able via Arduino IDE!: Grbl's source code may be now download and altered, and then be compiled and flashed directly through the Arduino IDE, which should work on all platforms. See the Wiki for details on how to do it.
*G-Code Parser Overhaul: Completely re-written from the ground-up for 100%-compliance to the g-code standard. (* Parts of the NIST standard are a bit out-dated and arbitrary, so we altered some minor things to make more sense. Differences are outlined in the source code.) We also took steps to allow us to break up the g-code parser into distinct separate tasks, which is key for some future development ideas and improvements.
**Independent Acceleration and Velocity Settings: Each axes may be defined with unique acceleration and velocity parameters and Grbl will automagically calculate the maximum acceleration and velocity through a path depending on the direction traveled. This is very useful for machines that have very different axes properties, like the ShapeOko's z-axis.
**Soft Limits: Checks if any motion command exceeds workspace limits before executing it, and alarms out, if detected. Another safety feature, but, unlike hard limits, position does not get lost, as it forces a feed hold before erroring out. NOTE: This still requires limit switches for homing so Grbl knows where the machine origin is, and the new max axis travel settings configured correctly for the machine.
**Probing: The G38.2 straight probe and G43.1/49 tool offset g-code commands are now supported. A simple probe switch must be connected to the Uno analog pin 5 (normally-open to ground). Grbl will report the probe position back to the user when the probing cycle detects a pin state change.
**Tool Length Offsets: Probing doesn't make sense without tool length offsets(TLO), so we added it! The G43.1 dynamic TLO (described by and G49 TLO cancel commands are now supported. G43.1 dynamic TLO works like the normal G43 TLO(NOT SUPPORTED) but requires an additional axis word with the offset value attached. We did this so Grbl does not have to track and maintain a tool offset database in its memory. Perhaps in the future, we will support a tool database, but not for this version.
**Improved Arc Performance: The larger the arc radius, the faster Grbl will trace it! We are now defining arcs in terms of arc chordal tolerance, rather than a fixed segment length. This automatically scales the arc segment length such that maximum radial error of the segment from the true arc is never more than the chordal tolerance value of a super-accurate default of 0.002 mm.
**CPU Pin Mapping: In an effort for Grbl to be compatible with other AVR architectures, such as the 1280 or 2560, a new cpu_map.h pin configuration file has been created to allow Grbl to be compiled for them. This is currently user supported, so your mileage may vary. If you run across a bug, please let us know or better send us a fix! Thanks in advance!
**New Grbl SIMULATOR! (by @jgeisler and @ashelly): A completely independent wrapper of the Grbl main source code that may be compiled as an executable on a computer. No Arduino required. Simply simulates the responses of Grbl as if it was on an Arduino. May be used for many things: checking out how Grbl works, pre-process moves for GUI graphics, debugging of new features, etc. Much left to do, but potentially very powerful, as the dummy AVR variables can be written to output anything you need.
**Configurable Real-time Status Reporting: Users can now customize the type of real-time data Grbl reports back when they issue a '?' status report. This includes data such as: machine position, work position, planner buffer usage, serial RX buffer usage.
**Updated Homing Routine: Sets workspace volume in all negative space regardless of limit switch position. Common on pro CNCs. But, the behavior may be changed by a compile-time option though. Now tied directly into the main planner and stepper modules to reduce flash space and allow maximum speeds during seeking.
**Optional Limit Pin Sharing: Limit switches can be combined to share the same pins to free up precious I/O pins for other purposes. When combined, users must adjust the homing cycle mask in config.h to not home the axes on a shared pin at the same time. Don't worry; hard limits and the homing cycle still work just like they did before.
**Optional Variable Spindle Speed Output: Available only as a compile-time option through the config.h file. Enables PWM output for 'S' g-code commands. Enabling this feature will swap the Z-limit D11 pin and spindle enable D12 pin to access the hardware PWM on pin D12. The Z-limit pin, now on D12, should work just as it did before.
Additional Compile-Time Feature Options: Line number tracking, real-time feed rate reporting.
SLATED FOR v1.0 DEVELOPMENT Jogging controls and feedrate/spindle/coolant overrides. (In v0.9, the framework for feedrate overrides are in-place, only the minor details to complete it have yet to be installed.)

Post has attachment
I'm thinking sorta seriously about adapting this for the Z axis. It's probably cheaper than the Ø10mm bearing rods, and maybe easier to deal with (aluminum can be cut to length much more easily). With custom long carriages, it might even be more rigid.

A question from a stepper noob: I have steppers on hand from Massdrop, with current, voltage, and holding torque specs. For the purposes of designing and selecting related mechanical components, I need an estimate of the largest torque the stepper will deliver at common 3D printer voltages and current. Is it as simple as multiplying the holding torque by the ratio of rated electrical power to planned electrical power?

Post has attachment

Post has attachment
Here's a design study of a tool changer based on the old HP pen plotters. They didn't seem to need an extra motion control of any kind to pick or stow a pen on the carousel. I had never had an opportunity to dismantle one to see how they work, so a bit of Googling led me to the HP online museum where a user's manual showed me enough of the mechanism to figure it out. The gist of it is this: Spring-loaded fingers on either the carriage or the carousel grip the tool. During a transfer, the carriage is forced against the carousel, and the empty side fingers force the other fingers off of the tool, gripping it in turn. Every time the carriage and carousel come together, the tool is transferred from one side to the other.

I adapted that concept to an E3D V5 Bowden hot-end. The hot end is gripped on a modified fan shroud, instead of by the groovemount. In my models, the green side is the carriage, and the purple side is the station in the tool magazine (if the wiper wasn’t a dead giveaway). Clearly a lot of detail is missing, such as screws, and springs in the obvious places. I only put enough effort into the design to see if it pans out.

The tool-changer’s advantage is the lowest possible moving mass for multiple tools. Lower mass allows higher acceleration, which tends to allow higher quality or speed. The tool changer also allows the opportunity to design for large numbers of tools, including things that aren’t extruders. Things like digitizing probes, swivel-knives, pens, and maybe small machining spindles. 

The primary disadvantage is complexity. A printer with a single travelling beam (H-bot, CoreXY, etc.) can give up about 55 mm of travel on one side of the machine to mount a row of tool stations. So no other controls need to be added than are necessary for the tools themselves. Anything else will need to be able to move a magazine of tool stations into position for the carriage to pick tools from.

Whether it's worth the trouble is not an easy question to answer. The carriage-side parts are about half the mass of a 2nd E3D V5. The Kraken appears to be amazingly mass-efficient for 4 extruders. I think it will still add significantly more moving mass than this tool-changing carriage, but I don’t have good enough numbers for the Kraken’s mass to be sure the tool-changer complexity is worthwhile. It does seem obvious that the simplicity of the Kraken makes it an obvious choice as the development platform for multi-extruder printers.

I think something like this will become necessary in the pursuit of utmost speed, but for now, I think I’ll need to be content with mounting a 2nd V5.

Images and some models files in the Drive folder linked below.

Post has attachment
Here's my idea for a Z axis drive for an Ingentis or derivative. Again, all three axes are shortened by 200 mm to keep everything in view.

The M6x1.0 threaded rod gives 0.005 mm per whole step, but will be admittedly slow. The threaded rod is supported by 696-ZZ bearings. I didn't want to support it at both ends, but with the Z stroke I think I can get, it will be 425 mm long. The green piece carries the nut and isolates the carriage (blue) from any wobble in the screw. It's a derivative of this piece: The nut carrier will twist slightly on the screw with any wobble, leading to small, cyclic Z position errors. If I have any noticeable banding, and I can't straighten the screw, I'll have to come up with a nut carrier that doesn't twist with wobble.

But all that's beside the point. The point is, driving the Z axis with a screw like this gives you a simple Z axis drive that won't fall when power is lost, and there should be room for 300 mm of Z axis stroke within the standard Ingentis frame dimensions. 

The pulleys for the Spectra line are 624-UU U-groove bearings. The square tube is more 10x10x1 aluminum.

It just occurred to me that the bearing at the motor end of the screw is redundant. And I have an idea now for a way to decouple screw wobble from the carriage without inducing any Z error. I'll implement that as the rest of the design matures.
Ingentis Derivative
2 Photos - View album

Post has attachment
OK, here's the idea I got based on the work +Shauki Bagdadi has been doing on his Quadrap. There's still a ton of work to be done here, but I've at least worked out that it all fits together without interference. Clearance is a little tight. There's 2 mm between the beams and 0.5 mm between the Ø5 mm shafts.

Right now, the moving mass is a few dozen grams more than his, but I have much smaller drive pulleys. So he should have higher travel speed, and I should be able to accelerate faster with the same steppers.

I'm still planning the same overall frame dimensions as Ingentis. I have the frame scrunched together to keep everything in view at the same time. Compared to Ingentis, this is using 19 mm less vertical space and perhaps 45 mm less horizontal space. I'll have a better number on that when I get an extruder carriage worked out. I don't intend to mount a Kraken. 

The drive pulleys are 15 tooth, 2 mm GT2, and all the bearings are 5x10x4 MR1052RS. Happily, the Ø10 mm bearing rods and their cost are gone.
Ingentis Derivative
4 Photos - View album

Post has attachment
Not my work, but I thought it worth passing around

Post has attachment
Oh, so that's why it was pulling to the left.
Wait while more posts are being loaded