@@kentvandervelden for sure, even a small miscalculation or bug in the code would mean end of the Haimer. It's so cool and for us its like video of the day!, I showed it to everyone here. Now we probably need a camera input for MASSO :) LOL
I was looking for info on analogue to digital gauge conversion. You are the MAN! Info on the Haimer sensor is good too! Double thanks for the GitHub page, I have to stare at some code.
Glad to help :) Let me know if you need if need something. There are also academic papers on this too, generally trying to read banks of analogue gauges, maybe at power plants.
@@kentvandervelden I am trying to adapt your code to run on an openMV H7 cam IOT device. It has find circle, and find line micropython example code using Hough Transform. Could be a usable device for legacy gauges that need monitoring. I just need the time. I have another more powerful chip, a K210 board with camera and screen. It has their version of tensor cores for machine learning speedup. Thanks again for your video, a drill down on the code would make a good companion video, no pressure!
Thank you! A followup video adds a second camera for identifying holes. Wouldn't be cool to have measurement verification in the mill, with accuracy that's reasonable for DIY CNC conversations?
@@kentvandervelden That's totally what I was thinking. Like, put a GoPro on your DTI, hook up to it with the app to view it wirelessly, and then you could see the dial clearly from every angle without paralax. The best use (that I can think of) would be when the DTI is connected to the spindle and facing the wrong way. It could make videos more under'stan'dable too. (I think I came up with the idea watching Shaden HKW's (Bar Z) videos. :) ) With GoPros (and other cameras, I'm sure) you can unscrew or screw-in the lens to change the focus, so you could possibly have the lens real close to the dial, reducing the how much bulk the camera and mount would add. The delay would be annoying, but managable, I think.
I know this is 2 years old but I am pretty sure some digital test indicators have the same 4 solder pads under a cover that many digital calipers have. It's usually a USB connection which you could use to read it. I'm pretty sure even my cheapo Accusize dial indicator has them. You can google to see how they reversed it on a set of digital calipers. I have no idea how accurate it just came to mind after watching this. Pretty slick how you did it with camera and Python. Huge python fan here, been using it since V1.4
I have parts for trying this: a digital dial indicator and the serial cable. But, the dial indicator is sensitive to movement along one axis, while the Haimer is sensitive along three axes. Are you thinking that the dial indicator would be a direct replacement for the Haimer or for another application? Thank you
@@kentvandervelden No, I was just thinking if there is a digital version that works like the Haimer it may have the USB connection to read directly for tinkering with. I'm not familiar with the Haimer, first time I saw one was your video. It's pretty cool.
@@dev-debug There are, but not in my budget :) I did get an electric probe, which tells if contact is made but does not tell distance. I should redo this video with that. Thank you for support! I started with Python at 2.6, after using C++ for years. I've not figured out what to do with it, but I enjoy it a lot :) The people who code the Python language and libraries are very gifted.
@@kentvandervelden I just looked up the digital version, yeah $820 USD is mighty expensive. If the probe you got is 3D as well it should be pretty easy code a script to measure the stock or even find x/y center and Z offset. I've never used the Linux CNC but plan on switching to it from Mach 3. Hopefully it accepts python plugins.
Hey! Just ran across this older vid of yours while searching for the vid where SMW totally trashed a Haimer. Seriously, one of these days we need to talk, as a lot of our idea paths cross so very closely. My idea a long while back was to create a probe that used a single axis movement via any axis direction input. This is accomplished with a hemispherical dome. The initial thought was to use an optical switch or some other overly complicated method, when I realized how simple it would be to just add something like a nice Metrol switch on top of it. 3D printed a few versions and initial tests were awesome, but way too course on a 3D printer. Then I ran across a video from NYCCNC (SMW now) and he took apart a Haimer he'd destroyed.....bingo...the light went on. So then began my search for someone who had trashed a Haimer via starting the spindle...LOL...ouch. Well, fast forward to a few days ago and I finally found someone and it's on the way to me. The probe is in tact other than a trashed dial. So now begins the fun of designing an internal adjustable mount for a Metrol switch to create a digital Haimer. And with the great benefit of NOT having all the issues associated with those 3 contact point probes all over the market. I'll put it out on my channel (yeah, I know I've done nothing in forever....should change soon). Take it easy Kent.
Sounds cool, I'll be looking for your creation. I took a glance at the 3D Taster patent, I can see how the switch could work. How about an absolute encoder on the dial shaft? Might be easier while maintaining the overtravel. CUI makes some nice small encoders with various readout protocols. If you are interested, I make a followup to this video with an additional camera. I do machine vision work for hire, so that's the hammer that I have :) Since then, I picked up a touch probe that I assume uses microswitches from Drewtronics, but have not tried it yet. I'm sure it'll be accurate enough to the limits of my little mill. I'm always interested in talking with people about ideas. Let me know what works for you.
@@kentvandervelden Yup, very familiar with CUI as I use their AMT-102 on my spindle for position. Excellent parts for their relative cheap price. Though I'm not sure how I'd use that within a probe scenario. I could easily grab those results and display them, but then it's just an extension of the face you can already read. You probably have way more use within LinuxCNC than me with UCCNC. I think my first experiment would still just be a method of turning the very linear probe ability of the Haimer into a simple digital input for standard probing. To me that would also be way more interest to the average user since then it's just a simple NO switch being used (maybe put a plug adapter on the side of the Haimer). Need to get it in my hands before I can rough it out. As for the Drewtronics, it's the same 3 point contact deal as everyone else. They have their secret sauce like others (yeah, grease to supposedly get contact to break consistent). My issue with these probes is always been that angle of probe can/will cause different contacts to lift and Z could be any combo of the three. If it's made well, then sure it's within the tolerances of the machine. But my thought was always why not make one better, that is ultimately more simple in the end. More to come! Keep on creating!
Thank you Danilo. Have you looked at MachineKit? I've been wondering what the differences with LinuxCNC are, and which one is best for what. I know MachineKit is derivative and I see some of folks contributing to both projects.
@@kentvandervelden Couple of years ago when it started I ordered Beaglebone Black and tried using machinekit with it, but gave up. Centroid did a better job of utilizing it as a black box controller with external interface
That's pretty cool Kent! As a home gamer I haven't been able to justify a $500 "edge finder" but I see a lot of folks are now using these so I feel like I'm missing out. Hoping there's some competition soon so the prices are more reasonable lol
Listening to the video later, I too thought the comment about the electronic probes being too expensive hypocritical when using a $500 edge finder (and $50 tips). I didn't mean it to be. It's been a long time since I bought the Haimer and the joy of using it with pre-measured tools in a library, compared to when I used collets and touched off each tool, I'd forgotten the cost. I tried searching Alibaba for an alternative. There are many inexpensive tool lengths setters, but I didn't find something like the Haimer. Perhaps there will be in 2021? patents.google.com/patent/US7024786B2/en
Part 2 video: ua-cam.com/video/7mBQwIzaiaQ/v-deo.html When the nine dials are shown, the displayed values for the red and black hands are identical. Fortunately, this was only a display bug and does not impact the final measurement. Also, while the camera is 720p, the captures are done at 640x480 to keep frame rate and sampling rate high. The shown FPS is ~10 because the screen is being recorded, saving three images per screen update, in the same thread.
Food for thought.. The current Datron machines use a large camera to point 'about' the x/y zero probing points (you move the machine by swiping the screen) and select on the screen where it should probe. Probe extends, and probing is done. I'm currently using a cheap USB microscope for x/y zeroing, with a known offset to the spindle. It's also well within 0,015mm accuracy. But you'll have to jog manually, not as spiffy as your solution! There are intelligent camera's available nowadays with some recognition in it as well, ESP32 cams. They have face and object recognition as well. Awesome job on the programming Kent.
Thank you Rene, I'll look at this; the Daltron machines always very impressive. I want to tinker with making the machine aware of it's environment, but my ideas quickly turn out to be bad ideas once one looks at them closely :) Hope all is well with you and your family!
This is really cool maybe with some software wizardry you could use use this as a pressure sensitive cmm for reverse engineering complex 3d parts look up the zeiss contura nyc cnc showed that head being able to trace his hand just by telling the cmm to measure a contour
Great. What about recognition of what is on a dial once tip touched the edge of the blank. In case if the software knows how much is on the dial - we should know how much steps should be run to get the 0 on a dial
That is how it works, assuming I understood your question correctly. The additional movements that may be causing you to ask is just in case the recognition is wrong. So, first move 95% of the read about, and so on. In this or the there other video it's also pointed out that there's no need to move to 0 once the measurement on the dial is known. It's only done for emulating how a person using the probe. Other optimizations are possible too. See the notes in the code on GitHub. Let me know if this didn't answer your question. Have a great weekend! Kent
@@kentvandervelden the major thing that I like here is a usage of powerful technology to simplify things... I guess that instead of making full dial photo there should be a way to do a macro photography to read dial precisely. When 2 arrows are down (no matter how far it is from 0 ) - next step is to add/substract a value, calculated. Could also be more interesting to read from digital gage. They are very cheap nowadays and have precision up to a micron. But we don't need absolute value. Most important to grab relative value, diff. Once we have a diff - there is no need to go to 0. Having 2(3) diffs in 120°(not necessarily) directions could provide extreme precision
Reading a digital gauge would be easier and some higher camera resolution would help. This was only 640x480. Your other points make perfect sense. Did you see part 2 of this, where holes are probed? Someone mentioned that the Haimer probes loose concentricity but I've not found good information on this. Ideally, the second camera of part 2 and a good touch probe would seem a good combination.
Yes. Someone commented about the concentricity, and if you are looking for high precision check that out. For me, I suspect the Haimer exceeds accuracy exceeds my skill. There are some cool uses too that are probably more sketchy such as dragging the probe to measure surface variation, but for relative measurement even that's probably fine. It's a fantastic tool.
Thank you, that's a neat video. The tool offset steps shown in the video must be done regardless of orientation. Looks like the original Haimer patents have or are nearing expiration in the US. I started to compare the Haimer to the probe from the video. This appears to be the related patent. Digging into the cited and family patents could be interesting. Best wishes patents.google.com/patent/US20140366393A1/en
this is so cool stuff Kent, great job and what an awesome idea!!
It's great to hear from you Jatinder! Not shown was my hand on the e-stop the whole time :)
@@kentvandervelden for sure, even a small miscalculation or bug in the code would mean end of the Haimer. It's so cool and for us its like video of the day!, I showed it to everyone here. Now we probably need a camera input for MASSO :) LOL
"Video of the day" I love that :) As for the other, maybe so ;)
I was looking for info on analogue to digital gauge conversion. You are the MAN! Info on the Haimer sensor is good too! Double thanks for the GitHub page, I have to stare at some code.
Glad to help :) Let me know if you need if need something. There are also academic papers on this too, generally trying to read banks of analogue gauges, maybe at power plants.
@@kentvandervelden I am trying to adapt your code to run on an openMV H7 cam IOT device. It has find circle, and find line micropython example code using Hough Transform. Could be a usable device for legacy gauges that need monitoring. I just need the time. I have another more powerful chip, a K210 board with camera and screen. It has their version of tensor cores for machine learning speedup. Thanks again for your video, a drill down on the code would make a good companion video, no pressure!
Wow I always discover amazing videos like that years after you publish them.
I'm pretty low-key. There's a follow-on video that probes holes that you might be enjoy.
This is easily the coolest project I've seen
Thank you! A followup video adds a second camera for identifying holes. Wouldn't be cool to have measurement verification in the mill, with accuracy that's reasonable for DIY CNC conversations?
I've thought about putting a camera/GoPro mount onto dial indicators, but I never thought of using a computer to actually read the dial. Very cool!
Thank you! Would the camera be for remotely displaying the dial indicator when mounted in an awkward location? That would be interesting too.
@@kentvandervelden That's totally what I was thinking. Like, put a GoPro on your DTI, hook up to it with the app to view it wirelessly, and then you could see the dial clearly from every angle without paralax. The best use (that I can think of) would be when the DTI is connected to the spindle and facing the wrong way. It could make videos more under'stan'dable too. (I think I came up with the idea watching Shaden HKW's (Bar Z) videos. :) )
With GoPros (and other cameras, I'm sure) you can unscrew or screw-in the lens to change the focus, so you could possibly have the lens real close to the dial, reducing the how much bulk the camera and mount would add. The delay would be annoying, but managable, I think.
I know this is 2 years old but I am pretty sure some digital test indicators have the same 4 solder pads under a cover that many digital calipers have. It's usually a USB connection which you could use to read it. I'm pretty sure even my cheapo Accusize dial indicator has them. You can google to see how they reversed it on a set of digital calipers. I have no idea how accurate it just came to mind after watching this.
Pretty slick how you did it with camera and Python. Huge python fan here, been using it since V1.4
I have parts for trying this: a digital dial indicator and the serial cable. But, the dial indicator is sensitive to movement along one axis, while the Haimer is sensitive along three axes. Are you thinking that the dial indicator would be a direct replacement for the Haimer or for another application? Thank you
@@kentvandervelden No, I was just thinking if there is a digital version that works like the Haimer it may have the USB connection to read directly for tinkering with.
I'm not familiar with the Haimer, first time I saw one was your video. It's pretty cool.
@@dev-debug There are, but not in my budget :) I did get an electric probe, which tells if contact is made but does not tell distance. I should redo this video with that. Thank you for support! I started with Python at 2.6, after using C++ for years. I've not figured out what to do with it, but I enjoy it a lot :) The people who code the Python language and libraries are very gifted.
@@kentvandervelden I just looked up the digital version, yeah $820 USD is mighty expensive. If the probe you got is 3D as well it should be pretty easy code a script to measure the stock or even find x/y center and Z offset. I've never used the Linux CNC but plan on switching to it from Mach 3. Hopefully it accepts python plugins.
Very good video!!! Regards from Austria ;)
Good idea...Good Job...Thank you Kent...
Kent, this is genius! Well done. Such a cool idea.
Thank you Gavin, it's great to hear from you!
Thank you kent that is super cool very impressive
Hey! Just ran across this older vid of yours while searching for the vid where SMW totally trashed a Haimer. Seriously, one of these days we need to talk, as a lot of our idea paths cross so very closely. My idea a long while back was to create a probe that used a single axis movement via any axis direction input. This is accomplished with a hemispherical dome. The initial thought was to use an optical switch or some other overly complicated method, when I realized how simple it would be to just add something like a nice Metrol switch on top of it. 3D printed a few versions and initial tests were awesome, but way too course on a 3D printer. Then I ran across a video from NYCCNC (SMW now) and he took apart a Haimer he'd destroyed.....bingo...the light went on. So then began my search for someone who had trashed a Haimer via starting the spindle...LOL...ouch. Well, fast forward to a few days ago and I finally found someone and it's on the way to me. The probe is in tact other than a trashed dial. So now begins the fun of designing an internal adjustable mount for a Metrol switch to create a digital Haimer. And with the great benefit of NOT having all the issues associated with those 3 contact point probes all over the market. I'll put it out on my channel (yeah, I know I've done nothing in forever....should change soon). Take it easy Kent.
Sounds cool, I'll be looking for your creation. I took a glance at the 3D Taster patent, I can see how the switch could work. How about an absolute encoder on the dial shaft? Might be easier while maintaining the overtravel. CUI makes some nice small encoders with various readout protocols. If you are interested, I make a followup to this video with an additional camera. I do machine vision work for hire, so that's the hammer that I have :) Since then, I picked up a touch probe that I assume uses microswitches from Drewtronics, but have not tried it yet. I'm sure it'll be accurate enough to the limits of my little mill. I'm always interested in talking with people about ideas. Let me know what works for you.
@@kentvandervelden Yup, very familiar with CUI as I use their AMT-102 on my spindle for position. Excellent parts for their relative cheap price. Though I'm not sure how I'd use that within a probe scenario. I could easily grab those results and display them, but then it's just an extension of the face you can already read. You probably have way more use within LinuxCNC than me with UCCNC. I think my first experiment would still just be a method of turning the very linear probe ability of the Haimer into a simple digital input for standard probing. To me that would also be way more interest to the average user since then it's just a simple NO switch being used (maybe put a plug adapter on the side of the Haimer). Need to get it in my hands before I can rough it out.
As for the Drewtronics, it's the same 3 point contact deal as everyone else. They have their secret sauce like others (yeah, grease to supposedly get contact to break consistent). My issue with these probes is always been that angle of probe can/will cause different contacts to lift and Z could be any combo of the three. If it's made well, then sure it's within the tolerances of the machine. But my thought was always why not make one better, that is ultimately more simple in the end.
More to come! Keep on creating!
Very interesting. Thanks for lettings us know and for sharing your results :)
Thank you Stefan, it's nice to hear from you :)
Cool program!
Thanks for this video.
Thank you for the support :)
Kent this is magnificent. Linuxcnc all the way !
Thank you Danilo. Have you looked at MachineKit? I've been wondering what the differences with LinuxCNC are, and which one is best for what. I know MachineKit is derivative and I see some of folks contributing to both projects.
@@kentvandervelden Couple of years ago when it started I ordered Beaglebone Black and tried using machinekit with it, but gave up. Centroid did a better job of utilizing it as a black box controller with external interface
Great project Kent! So that's been keeping you busy the past couple of hours ;-)
That's pretty cool Kent! As a home gamer I haven't been able to justify a $500 "edge finder" but I see a lot of folks are now using these so I feel like I'm missing out. Hoping there's some competition soon so the prices are more reasonable lol
Listening to the video later, I too thought the comment about the electronic probes being too expensive hypocritical when using a $500 edge finder (and $50 tips). I didn't mean it to be. It's been a long time since I bought the Haimer and the joy of using it with pre-measured tools in a library, compared to when I used collets and touched off each tool, I'd forgotten the cost. I tried searching Alibaba for an alternative. There are many inexpensive tool lengths setters, but I didn't find something like the Haimer. Perhaps there will be in 2021? patents.google.com/patent/US7024786B2/en
Wow! Kent, you rock!
Part 2 video: ua-cam.com/video/7mBQwIzaiaQ/v-deo.html
When the nine dials are shown, the displayed values for the red and black hands are identical. Fortunately, this was only a display bug and does not impact the final measurement. Also, while the camera is 720p, the captures are done at 640x480 to keep frame rate and sampling rate high. The shown FPS is ~10 because the screen is being recorded, saving three images per screen update, in the same thread.
Food for thought.. The current Datron machines use a large camera to point 'about' the x/y zero probing points (you move the machine by swiping the screen) and select on the screen where it should probe. Probe extends, and probing is done. I'm currently using a cheap USB microscope for x/y zeroing, with a known offset to the spindle. It's also well within 0,015mm accuracy. But you'll have to jog manually, not as spiffy as your solution! There are intelligent camera's available nowadays with some recognition in it as well, ESP32 cams. They have face and object recognition as well. Awesome job on the programming Kent.
Thank you Rene, I'll look at this; the Daltron machines always very impressive. I want to tinker with making the machine aware of it's environment, but my ideas quickly turn out to be bad ideas once one looks at them closely :) Hope all is well with you and your family!
@@kentvandervelden Making it more aware is a good idea! Thanks for your reply Kent! all is fine on this side of the pond.
Now that's just the coolest thing I've seen in a while! Nice job!!
Wow Chris, that's awesome for you to say, thanks!
Absolutely brilliant Kent
Many thanks Vik :)
Can you share the 3D Printable assets?
This is really cool maybe with some software wizardry you could use use this as a pressure sensitive cmm for reverse engineering complex 3d parts look up the zeiss contura nyc cnc showed that head being able to trace his hand just by telling the cmm to measure a contour
That's an awesome idea! I have a mic with readout that could work well for your suggestion. Thank you
well that is pretty cool !!
I was thinking of using a load cell (strain gauge) for touch off. I know others have tried piezo elements, but maybe strain gauge is the way to go?
That would be fantastic! I don't have experience with either method, but would like to try someday. And a huge thank you for the longtime support!
That is cool as hell man!
Great.
What about recognition of what is on a dial once tip touched the edge of the blank. In case if the software knows how much is on the dial - we should know how much steps should be run to get the 0 on a dial
That is how it works, assuming I understood your question correctly. The additional movements that may be causing you to ask is just in case the recognition is wrong. So, first move 95% of the read about, and so on. In this or the there other video it's also pointed out that there's no need to move to 0 once the measurement on the dial is known. It's only done for emulating how a person using the probe. Other optimizations are possible too. See the notes in the code on GitHub. Let me know if this didn't answer your question. Have a great weekend! Kent
Also optimizations have some cost. Mainly, I was careful to avoid breaking Haimer probe tips :) Still, was great fun to write
@@kentvandervelden the major thing that I like here is a usage of powerful technology to simplify things...
I guess that instead of making full dial photo there should be a way to do a macro photography to read dial precisely. When 2 arrows are down (no matter how far it is from 0 ) - next step is to add/substract a value, calculated.
Could also be more interesting to read from digital gage. They are very cheap nowadays and have precision up to a micron.
But we don't need absolute value. Most important to grab relative value, diff. Once we have a diff - there is no need to go to 0. Having 2(3) diffs in 120°(not necessarily) directions could provide extreme precision
Reading a digital gauge would be easier and some higher camera resolution would help. This was only 640x480. Your other points make perfect sense. Did you see part 2 of this, where holes are probed? Someone mentioned that the Haimer probes loose concentricity but I've not found good information on this. Ideally, the second camera of part 2 and a good touch probe would seem a good combination.
Can the Haimer 3D sensor be used horizontally?
Yes. Someone commented about the concentricity, and if you are looking for high precision check that out. For me, I suspect the Haimer exceeds accuracy exceeds my skill. There are some cool uses too that are probably more sketchy such as dragging the probe to measure surface variation, but for relative measurement even that's probably fine. It's a fantastic tool.
I'm sorry, I might have misunderstood. Yes, the Haimer can be mounted horizontally, e.g. on a lathe. Peace
@@kentvandervelden : Thanks was just wondering, because of this link, ua-cam.com/video/kacnj-kpErw/v-deo.html
Thank you, that's a neat video. The tool offset steps shown in the video must be done regardless of orientation. Looks like the original Haimer patents have or are nearing expiration in the US. I started to compare the Haimer to the probe from the video. This appears to be the related patent. Digging into the cited and family patents could be interesting. Best wishes
patents.google.com/patent/US20140366393A1/en