![]() |
#1
|
|||
|
|||
Just How Accurate and Repeatable is the MechMate?
After building my MechMate I made the usual axis calibration adjustments. The typical formulas to find the steps per rev, per inch were used. Later, I used Mach3’s axis calibration to dial in the axis a little tighter. To finalize the build I wanted to document the accuracy and the repeatability of my machine.
In the end, I successfully ‘Dialed In’ all three axis of the MechMate. The machine is very accurate and very repeatable. The MechMate design is rock solid, and as far as my machine goes, I have documented and photographic results to prove it. Please read through my process below. I’m sure there are other ways to calibrate the MechMate. Below is the procedure I have chosen. You will find photographs attached below. They are pretty self explanatory. Notations are made where necessary. Gerald, Thank You again for the design of the MechMate, and all the help you have provided me and many others here. Goal: The goal is to define the Accuracy and Repeatability of the MechMate. Success/Failure: The success or failure will be determined by successive data points gathered from all three axis. Successful data points for all axis will be those points found to be plus or minus .001 inches (per foot travel) of the desired length. One error data point (greater than plus or minus .001 per foot travel) in the data set is allowed. Because the ‘Z; axis travel is short, 4.5 inches, only plus or minus .001 deviation is allowed. Tools: 1) An accurate digital measuring device will be attached to each axis. 2) Temporary attach brackets 3) Mach3 DRO’s 4) Mach3 controls 5) Silicone spray 6) Clean rag Calibration: Each axis will be driven from a ‘Zero’ position (approximately 1 inch from the home switch) using the Axis calibration tool located on the ‘Setup’ page. Each axis will be driven back to the ‘Zero’ position using the ‘Go To Zero’ button found on the home page under the DRO displays. The ‘X’ and the ‘Y’ axis will be driven 27 inches forward and backward, documenting the readout of the remote display. Does the axis move the required 27.000 inches? And when driven backward, does the axis return to the ‘0.000’ position? The ‘Z’ axis will be driven 4.5 inches forward and backward, documenting the readout of the remote readout display. Does the axis move the required 4.500 inches? And when driven backward, does the axis return to the ‘0.000’ position? The digital display of the measuring tool and the DRO readout of the MACH3 will be compared. The ratio will be adjusted using the ‘Set Steps Per Unit’ tool found on the Setup page to ‘Dial In’ the machine. Three consecutive agreements with the measuring tool and the Mach3 DRO, at plus or minus .001 will be considered, ‘Dialed In’. Once the axis is ‘Dialed In’ the calibration standard will not be changed. The axis will be driven forward and backward a minimum of 40 consecutive times, 20 each direction and the data points will be logged. Procedure: I purchased a remote router measuring kit with a 31 inch track extension from Wixey Development (wixey.com) and attached it to the Y axis. Procedure note: It is important to set a ‘ZERO’ position away from the axis home switch. Homing the axis, or bringing it back to the home position can introduce error in this procedure. When you ask Mach3 to ‘Home’ an axis, it drives the axis as far as it needs to, until the Home switch is found. When the switch circuit is ‘opened’ Mach3 halts, then reverses direction and drives forward until the switch is closed. At this point Mach3 resets the axis DRO to 0.0000. The mechanical switches are prone to small inaccuracies; therefore you cannot assure the axis will always be returned to the same ‘Zero’ position as before. These tests require Mach3 to drive the axis in both positive and negative direction. Remember, Mach3 DRO is a calculated result, not a measured one. And, the true measured result is found on the remote display. Therefore Mach3’s DRO result will always be compared to the result found on the remote display unit. Likewise, when asked to return to ‘Zero’, Mach3 will calculate the required steps needed and will drive back to the ‘Zero’ position. Again, both DRO’s will be compared and any error found will be seen on the remote display. ‘Y’ Axis Wixey Setup: Drive the ‘Y’ car off the home switch approximately 1 inch. Zero all Mach3 DRO displays Install the 31” scale extension track, parallel to the lower edge of the ‘Y’ axis rectangular tube. Allow enough room for the sensor to glide freely along the scale extension path. Spray a small amount of silicone lubricant on a clean rag and apply a thin film of silicone to the edges of the aluminum extrusion. The lubricant will allow the shuttle sensor to glide smoothly along the travel path. This will provide more consistent data result. Install the scale sensor to the sensor track. Install a temporary bracket from the ‘Y’ car and extend it down to within approximately 1/8” from the top of the sensor scale. Drill a 1/8” hole in the bottom of the temporary ‘Y’ car bracket, approximately over the centerline of the scale extrusion. Attach the supplied magnet to the temporary ‘Y’ car bracket. Insure the magnet does not touch the sensor scale throughout the travel length. Slide the scale sensor back to the magnet. It will attach itself firmly to the magnet. Install the CAT cable to the sensor and the readout. Attach the Remote Readout where convenient. Power up the Wixey Readout and zero the display. ‘Y’ Axis Data Procedure: Insure ALL Mach3 DRO displays are set to 0.0000 Insure the remote display readout is set to 0.000 From the Setup page, drive the ‘Y’ axis forward 27 inches. Compare the readings of both displays Log the reading of the remote display From the Main page, drive the ‘Y’ axis back to the ‘Zero’ position using the ‘Go To Zero’ button under the DRO displays. Only the ‘Y’ axis can move because the other axis should be at their ‘Zero’ position. Compare the readings of both displays Log the data ‘X’ Axis Data Procedure: Repeat the same procedure as above ‘Z’ Axis Data Procedure: Repeat the same procedure as above, but install the short scale to the ‘Z’ axis. Drive the axis only 4.5 inches ==================== Photos: The first photo is the Wixey digital remote, scale, sensor, and cable. The kit comes with some attach hardware, screws, etc. There are no batteries so you have to provide you own, 2-AAA The next four photos are for the 'Z' calibration. The small scale and hardware is attached to the top of the 'Y' car. A small bracket with the magnet is attached to the 'Z' slide assembly. As the 'Z' axis moves up and down, the 'Z' bracket with the magnet attached move the sensor up and down the scale. The next two photos are the setup of the 'Y' axis. The following three photos indicate three consecutive 'Y' axis drives at 27 inches And the final three photos indicate my poor penmanship as I document the results from the axis moves. You'll notice that I did not capture the returns to Zero data. I felt it was not necessary to capture that data unless the indication was beyond the plus or minus threshold. |
#2
|
|||
|
|||
Brilliant! - thank you.
Link to the Wixey site: http://www.wixey.com/index.html UK stockist: http://www.machine-dro.co.uk/wixey-r...-300mm-12.html Last edited by Alan_c; Sun 03 March 2013 at 14:42.. |
#3
|
|||
|
|||
Wow man ! great stuff ! really extensive testing
What combination of drives and motors do you have? Motors look like those from Keling |
#4
|
|||
|
|||
Before placing an order for the Wixey remote, please note that there are different models with varying resolutions. Look their website over first before placing an order.
These things are used on routers, fence tables, etc. To get the resolution I wanted I had to purchase the shorter Router Remote unit WR525. The WR525 (my model) has a resolution of .001 inches / .05mm. I also had to purchase the WR7002 track extension for the ‘X’ and ‘Y’ axis. So if you want to be sure you get the right stuff with the resolution you want, send Barry a note and he can clarify the models and resolution for you. Barry Wixey is at digital@wixey.com. |
#5
|
|||
|
|||
Danilo
I purchased the motors and controller from Campbell Designs. The motors are nema 34-640oz and I believe Bob ordered the Keiling motors. Each axis has a 4.5:1 reduction unit. Yes, the nema 34 is overkill on the ‘Z’ axis, but I wanted to keep all the drives identical. Last edited by domino11; Sun 03 March 2013 at 17:02.. Reason: correction |
#6
|
|||
|
|||
Nice effort/info and a good documentation on your process !
Can you post some more specs/images of your machine ? I do not see a build thread, and I think I am I looking at your modified Z. I see you used bought V profiles on all axis ? Beltdrives ? Bolted or welded table ? taped or screwed racks ? etc. |
#7
|
|||
|
|||
Build Notes
Fox,
Thanks for your questions. I will update my build log to address your questions. The build log is here: http://www.mechmate.com/forums/showthread.php?t=1227 |
#8
|
|||
|
|||
Hi William,
thanks for posting the specs, I am just back from my business trip and gearing up to start working again on the MM. Currently designing my Z, and comparing different options. Do you care to share some pictures thoughts in your build thread ?! |
#9
|
|||
|
|||
When Stuff Goes Wrong
I was asked by a friend whether the MM is capable to cut precision parts. I told him that given the correct material, the machine with proper tuning could produce good result. We started with Delrin.
Step 1 - Cut blanks to see if the machine can produce the same shape repeatedly. Step 2 - Cut bores and pockets to see if they align properly. Step 3 - Assemble the parts and test for proper fit. Photo One shows the machine is capable to produce the same perimeter cut time after time. Photo Two shows the blanks are not true and the center bores are not centered. Photo Three shows the rough, partial assembly. Discussion: The first blanks were cut all together from a single program. The second blanks were cut individually from individual programs. I found much of my problem shown in photo two from the use of Home Switches. Home Switches have been discussed here many times. Some of you guys may have success with them. But for this application, I have lost a lot of time and material to continue using them. The switches have been dismantled and replaced with temporary stop blocks. Now, before every program run the X and Y axis are pushed up against the stop blocks and then zeroed in Mach3. Later, I decided to precisely calibrate all the axis, thus the Real Reason for starting this thread. And I thought it best to look over the drive units as well. The axis calibration was very successful as shown above. And a look over the drives netted more problems. The A-axis drive assembly was not working properly. Play was found in the drive shaft bearings of the reduction unit. Further inspection found the grub screws on the 72 tooth pulley loose too. These repairs have been made and photos will be posted later. And the final problem to solve is, to find a way to hold the Delrin tightly to the spoil board. This stuff is really slick. This project is ongoing with more notes and photos to follow. @Fox, On my build thread, I’ll break down the Z-axis in a few days, and photograph it to show you what I got. |
#10
|
|||
|
|||
A question on the way you decided to machine the part in the lower left picture.
I wonder why you didn't make the center hole at the same time you cut the outer shape. When making this part I would have never made the blank and then made the center hole in another operation. The time required to indicate that center hole, would excessive and without a very accurate holding fixture almost impossible to make all the parts concentric in that assembly. |
#11
|
|||
|
|||
Regarding the lower left photo.
Good observation, Steve. My goal was not to prove tool path efficiency, but to prove machine performance. Please let me clarify. The parts of the first photo were cut from a nest routine, whereby a tool path was generated for the tool holes, and another tool path was generated for the perimeter cuts. The perimeter cuts came out identical because they were from a single tool path, where one offset from the origin was used. For the second photo I cut parts from scrap. Since I’m using scrap, I need different offsets from the origin to take advantage of the material. So for these five parts I need, five tool paths for the tool holes, and I need five tool paths for the bore holes and the part cut outs. Building parts from scrap is not efficient. The second photograph illustrates a huge performance problem with the machine. Regardless of any tool path efficiency, identical part cuts would never be possible. The root of the problem was found in four places: 1) Home switches, not reliable to define an exact origin for X and Y axis 2) Axis calibration, X and Y axis not calibrated precisely 3) Failing A-axis drive, not providing harmonic movement with the X-axis because of slipping gear 4) Inefficient holding fixture method I believe machine performs regardless of origin offset, should produce identical parts in the calibrated envelope. |
#12
|
|||
|
|||
In your scenario I have reproduced multiple cut routines on parts in different times.
the plastic blanks were provided by a manufacturer, then depending on the application, a different hole pattern was used, Thus different parts with same perimeter blank. The key was just building a surface jig on the table with absolute home that the machine was reset to when needed. The problem I was having that occasional power failures would cause the machine to loose the home thus make me reset the machine. Adding an absolute reference point in the file with tool offsets let me achieve this daily. 6000 parts later, over a period of 6 weeks. All was perfect. |
#13
|
|||
|
|||
I know it's an old post ... but Sean... did you use a probe/edgefinder for finding the absolute home again ? Or just drove the machine by hand (without power) into the stops, and powered on to set home ?
|
#14
|
|||
|
|||
they way I checked accurancy and repeatability from homing
I placed two dial indicators (meetklok for the ducht and south afrikan (?) ![]() on the side of the C bar en end of the C bar when homing the machine I zero the dial indicator. the magnet holders are just small round ones, not the bigger square ones. I repeat the homing to see what the repeatebility is from my sensors. It's within 0.03 mm when finishing my router program I sent the x and y stage back to zero and check the dial indicators for position. since a couple of weeks it's within 0.03 mm for x and y and z (ball screw) is 0.01 I check this with a height set gauge ![]() I made a hole pattern in my machine bed in these holes I can place aluminium dowel pins for accurate and repeatable position from my work piece I use aluminium dowel pins to save my router bids. just in case ![]() |
#15
|
|||
|
|||
FOX!
Good to hear from you. I just checked in and see you revived an old thread. I did use a probe/finder before.....But realistically, I just put a engraving bit in the spindle and jogged it until I was perfectly x/y 0,0 over the reference hole in the bed. Reset all to 0,0,0 in Mach and off you go. |
#16
|
|||
|
|||
Thnx, if you use a jig to conform the angle of your piece stays the same along the axis, the hole centering would work, used it myself as well. Was just pondering if it was worthwhile to go for a pro tool, as I don't need it every day.
|
#17
|
|||
|
|||
Yes, it is Repeatable and it is Accurate
This thread started over two years ago. With the exception of the Delrin parts described above, the accuracy of the MM has never been a real concern for me. Whatever widget I have thrown to the machine it has cut well.
Over the last eighteen months this MechMate has been put to a real world test; doing what it was designed to do, to cut wood accurately and repeatedly. I was asked by a design team to provide high quality tooling necessary for a new high performance experimental aircraft to be introduced at the Reno Air Show in 2016. The MechMate has been busy creating all of the tooling dies needed to form the bulkheads and some of the internal aluminum flat parts needed for the structure. The real proof of accuracy and repeatability is found when the parts are assembled on the jig. Do they fit correctly, do they align correctly? The photos below tell the story. photo 1) Bulkhead parts as they were received from the press. These parts were derived from the form blocks cut by the MM. photo 2) Firewall looking aft. The fuselage begins life assembled in the jig, upside down like a canoe. photo 3) Aft section of the fuselage. The cutout in the third bulkhead set from the left is for the retractable tail wheel. Sandwiched between the double bulkheads is an aluminum .125 T3 hardened backbone plate. The MM cut the form dies and the back plates for the parts shown here. photo 4) The MM cuts a back plate for one of the double aft bulkheads photo 5) This is the Wing Spar attach bulkhead assembly. Once assembled in the aircraft, the wing attaches to this pair of bulkheads. Precision is important in all areas of the aircraft, but of all the assemblies this one has to be nearly perfect. The MM cut eight die blocks and one wing spar simulation blank for the assembly shown here. EVERYTHING FITS PRECISELY! Last edited by WilliamT; Mon 22 June 2015 at 06:23.. Reason: typo |
#18
|
|||
|
|||
Wow!
|
#19
|
|||
|
|||
Bill
How about a link or a picture of what the finish plane looks like? Thanks |
#20
|
|||
|
|||
Hey, Bill,
So happy to hear from you. Looks like you and your MechMate have been busy. Like Nils, I'd like to see more. It appears that the FINISHED plane may be a bit further down the calendar, but what about pictures of the jigs you made while we're waiting for the plane to be assembled. Also were the plane builders able to "shape" the metal parts directly on the wood jigs you provided or was there an intermediate step required? Also, this doesn't look like your shop. For one thing, the ceilings don't look high enough. Are these pictures of your customer/friend's shop or did you REALLY fill up your shop? |
#21
|
|||
|
|||
Nils, Steve and John, thanks...
First, before I answer Nils request; The reason I bring forward the latest work from this machine is to impress the point, that the design and engineering behind this machine is solid. There are over 122 or more numbered MechMates built and all are success stories. This is just one of them. And so I want to say to the guys viewing this post, and wondering, "Should I consider building one of these MechMates?" The answer is , "Yes, you should." And to the guy that has one under construction, "Yes, finish it already!! You will be happy when you're done." About the aircraft and a little history.. The aircraft is being scratch built in Taylor, TX at a friend's hangar. The fuselage design is 'inspired' by that of the Hughes H1 racer, residing at the Smithsonian, Air and Space Museum. You'll find the most current documentation of the H1 here. https://en.wikipedia.org/wiki/Hughes_H-1_Racer A replica of the H1 was built. Follow this link and you will find two large wallpaper photos of the replica about mid page. Unfortunately, this airplane and the pilot perished in a crash at Yellowstone Park a few years ago. http://www.google.com/imgres?imgurl=...4vP73zAM&itg=1 The owner of the plane we're building wanted us to use the organic shape of the Hughes H1 as the base fuselage shape for his aircraft. The organic shape has a lower drag coefficient than that of a factory built aircraft such as a P51 Mustang, but it is more difficult to build, especially to skin. We will be using a 9 cylinder radial engine, 450hp. So the fuselage and aft section of the aircraft will look identical to the Hughes H1, scaled slightly smaller though, and will be fitted with a Team Rocket EVO tapered high performance wing. Team Rocket F1 can be found here at DocThrock's page: http://docthrock.com/ And DockThrock's EVO wing can be found here: http://www.docthrock.com/Wings.shtml Should you want to know more about Team Rocket F1 just google it. Many examples of the aircraft to be found there. John- The finished parts like in photo 1 were pressed from a die block cut from MDF by the MM. The process is simple. The die block is placed on the surface of the press. A flat aluminum blank is placed over the die block. Tooling (alignment) holes and pins are used to align the flat aluminum to the die block. You can see two small 1/4" holes in the aluminum part, photo 1. A rubber pad is then placed over the two of them and a 30 ton press squashes the rubber pad around the die, thus forming the aluminum. This is all done with soft aluminum and the heat treated afterwards. Sorry, I don't have photos of the dies we built. We cut so many of them they all kind-of lookalike after a while. To cut the dies we burned through 20 sheets of 3/4" MDF, two spoil boards, and interestingly enough, some pretty heavy aluminum. I said above that the back plate was cut from .125" sheet. My error, we cut the back plate from .25" T3 hardened plate. I cut the aluminum dry, no cooling, at .050 depth and at 50ipm. The aluminum cuts better at a higher rate of speed. While building the tools for the aircraft, I became unhappy with the controls of my MechMate. Yes, it was accurate and it was repeatable, but it was not reliable. A number of times during a cut routine an axis would just stop responding to the controller and therefore, scrap the part we were cutting. I decided to upgrade my MechMate with a new controller and new drivers. My MechMate is now driven by a Hicon Integra controller, and four NEMA 23 servo motors. Yes, you heard that right!! Four NEMA 23 servo motors. They are 198 oz/in each and they propel the gantry and the 'Y-car' very quickly. I had them running one time at 700ipm but have since reduced them to 400ipm and may reduce them further. Regardless, so far I am impressed with their performance. The final test is to see if the NEMA 23's are capable to perform under load. We'll see soon, and after further tests I will begin a new thread on the upgrade. John, you are welcome by anytime for a test drive. |
#22
|
|||
|
|||
You have been busy I see... Servodriven me... the first I think!
|
#23
|
|||
|
|||
Yes - I am interested in seeing more on the servo setup for sure !!!
|
#24
|
|||
|
|||
Mike,
Please advise where to begin the thread. Discussed will be the complete upgrade process to include, servos, controller, cabling, power supply, Mach3 control, performance tests, etc. I am building the information stack and will be taking photographs today. BillT |
#25
|
|||
|
|||
What we lack with Mach3 is a true closed loop controller. Unless something had changed, Mach 3 and 4 still do not offer direct control of the servo end position and leave that up to the control downstream. It does appear that LinuxCNC has made some progress in this area. So any information is valuable but a software controller that actually displays and corrects for error is the ultimate end result. I am interested to hear more.
|
#26
|
|||
|
|||
Bill, WOW!
How I love to see a plan come together. You're now DOING what you were TALKING about doing over 6 years ago. What a great story. ![]() I've sent you a PM. |
#27
|
|||
|
|||
Working with Mach3/4 is a challenge. Sure, it does not have the closed loop intelligence. But it is inexpensive to purchase, and considerably less expensive that a commercial controller.
Linux CNC is an option for the future when it becomes a closed loop system and I have the time to spend with it and to learn it. But for now, I have Mach3, a very stupid Mach3 and a very smart closed loop Integra CNC controller, and some very, very smart, (almost too smart) servo motors. The motors receive step and direction pulses for the controller, and maintain their position and error tracking internally. Indeed, very, very smart motors. In the end they will all learn to play together. |
#28
|
|||
|
|||
Bill,
Thank you for the link to the photos of the plane. It looks like it is very fast. Nice lines. Very interesting. I will be looking forward to seeing your progress and photo. |
#29
|
|||
|
|||
Take a look at edingcnc
|
#30
|
|||
|
|||
build update
For those following this thread, Mike and others have asked for more information regarding my servo setup. Since the servo setup is now incorporated into my build configuration, my comments regarding the servos and the controller will be found in my build thread.
please follow Now has LOGOs #111 – Georgetown, TX http://www.mechmate.com/forums/showthread.php?t=1227 |
![]() |
|
|