#linuxcnc Logs

Oct 20 2018

#linuxcnc Calendar

01:01 AM gloops: quiet
01:04 AM XXCoder: shh
01:19 AM XXCoder: gloops: how goes your project
01:21 AM gloops: i got the saw frame rolling and the motor working (been outside a while) need to fit motor and get a blade
01:21 AM gloops: the bandsaw prject that is
01:23 AM gloops: this weekend though i might test dovetailing jig on cnc router
01:26 AM XXCoder: awesome :)
02:04 AM gloops: well, nobody here building simple machines with steppers and chinese screws so nothing much i can help with atm
02:04 AM XXCoder: yeah pretty quiet
02:04 AM XXCoder: its 12 am here
02:05 AM gloops: 8am here, looking a pretty typical october day
02:05 AM Deejay: moin
02:05 AM Deejay: 9 am here
02:05 AM gloops: morning
02:05 AM XXCoder: yo
02:05 AM Deejay: hi gloops
02:05 AM Deejay: & XXCoder
02:05 AM Deejay: ;)
02:56 AM gloops: https://www.ebay.co.uk/itm/Adcock-and-Shipley-Horizontal-Milling-Machine/323497961435?
02:59 AM XXCoder: wow nicely decocated with oxide
03:01 AM gloops: soon clean that up
04:13 AM jthornton: morning
04:14 AM XXCoder: welcome back
04:15 AM jthornton: pretty slow night in here lol
04:15 AM XXCoder: yeah
04:16 AM jthornton: it's getting to be a pia with this linux mint pc not waking up from sleep mode
04:24 AM gloops: i had that, never found any cure - i dont think there is one
04:27 AM jthornton: odd that it just started recently before one shake of the mouse and it would wake up
04:27 AM XXCoder: sounds like theres something growing, like one directory getting full of files, or log getting too long
04:27 AM XXCoder: did you turn on log of any kind?
04:28 AM jthornton: no
04:28 AM XXCoder: is ahrd drive getting full?
04:30 AM jthornton: no it's 2TB IIRC
04:31 AM XXCoder: yeah doubted that as hd is cheap
04:33 AM XXCoder: new clickspring
04:33 AM XXCoder: 56 minutes ago, literally
04:34 AM jthornton: new coffee time just heard the grinder go off
04:41 AM fragalot: https://www.youtube.com/watch?v=kHBcVlqpvZ8 <= brilliant.
04:41 AM jthornton: hmm the wife forgot to put coffee beans in the pot...
05:06 AM merkatorix: Ich überlege, ob ich ein wenig Zeit habe, um mit CNC-Dingen zu spielen. Fertige CNC-Maschinen sind entweder astronomisch teuer oder als China CNC teuer und machen nicht genau das, was ich will.
05:07 AM merkatorix: Da dachte ich, kaufe ich mir einfach die Einzelteile bei Amazon
05:07 AM merkatorix: 2 Schrittmotoren mit Parallel-Schnittstelle: https://www.amazon.de/Pre-SPS-ACT-Motor-23HS8430-Automation/dp/B01KI0UAQU/
05:07 AM merkatorix: Würde das als Elektronik für eine CNC reichen?
05:08 AM merkatorix: Ah, sorry,
05:09 AM merkatorix: Everytime when booting up hexChat my client switches to automatically connected channels.
05:09 AM merkatorix: So I didn't notice that I am in the wrong channel.
05:09 AM merkatorix: But actually I am in the right one :D
05:09 AM merkatorix: I thought about starting to toy aroudn with linuxcnc
05:10 AM merkatorix: And I thought if, the amazon article I linked above woult actually be sufficient, if combined with an atom PC and linuxcnc
05:10 AM XXCoder: cool. yeah linuxcnc can run virtually any cnc machine that dont use reprap style controller board
05:10 AM merkatorix: Aside from the spindle it is all I need, isnt' it?
05:11 AM merkatorix: But I was askind in a hardware channel, because I was already quite confident about linuxcnc. I just wodered, if it is the right hardware.
05:11 AM XXCoder: tb6560 ehh not very good experence
05:11 AM XXCoder: tb6600 cost little bit more money, and buy BOB with it, and you got cheap electrics that work well
05:11 AM merkatorix: BOB?
05:11 AM XXCoder: not best, but about the best for very low cost
05:12 AM jthornton: Break Out Board
05:12 AM XXCoder: yeah breakout board for parallel port
05:12 AM merkatorix: Ah, so instead of an all in one combination a breakout board + standalone drivers?
05:12 AM XXCoder: tb6560 has serious issues you'd have to fix board to work better
05:12 AM XXCoder: chinesium ones anyway
05:13 AM merkatorix: Which kind of issues? Over current problems? But yes, although I would consider me quite experienced in electronics and I plan to build my own brushless ESCs soon, concerning the CNC I wanted results and no endless tinkering.
05:14 AM merkatorix: Are there standard breakout boards, which are robust and save to not damage the parallel port?
05:14 AM XXCoder: unfortunately im far from expert, but I had issues all over then evenually I said screw it and get tb6600s and BOB
05:14 AM XXCoder: worked way better
05:16 AM SpeedEvil: merkatorix: PCI parallel port card is probably cheapest
05:16 AM jthornton: merkatorix: the Mesa 7i96 is the best way to go once you leave the cheap level
05:17 AM SpeedEvil: Though simply connecting 1K resistors in parallel with all your parallel port wires except ground will pretty much protect from a lot
05:17 AM XXCoder: yeah merkatorix https://acassis.wordpress.com/2014/03/15/be-aware-of-issues-with-these-tb6560-boards-from-ebay/
05:18 AM XXCoder: that was what my experence was
05:18 AM XXCoder: it kept stalling or messing up
05:18 AM * jthornton gets back to chicken door programming
05:18 AM XXCoder: my recommandation: dont bother, tb6600 is maybe couple bucks more and little bit more wiring work
05:19 AM merkatorix: I planned to buy a 120€ intel atom board, computer + power supply and chassis. I think Mesa is too expensive and I was told the onboard parallel ports work better than the PCI ones?
05:19 AM XXCoder: AND you can stright up swap one tb6600 if it die for some reason
05:19 AM XXCoder: pci is swappable
05:19 AM merkatorix: Just googled and some people even recommend Leadshine M542, but that is a level more expensive, I guess?
05:19 AM XXCoder: my pc has onboard but I bought decent parallel card and latency test was very good
05:19 AM jthornton: what is too expensive for you in $$$ your time or?
05:20 AM merkatorix: More wiring work, because I don't buy it as a driver + breakoutboard comparsion?
05:20 AM XXCoder: he wants it to toy with linucnc so I guess cheapest possible
05:20 AM jthornton: parallel port then
05:20 AM XXCoder: merkatorix: 3 drivers tb6600s is not single board so yeah more wires, but very easy
05:24 AM XXCoder: if even I can do it.. you, the electrics expert can do it
05:24 AM merkatorix: Mh, I just starded my new job, but as long it is <100€ it is definately the time, which is expensive, I guess.
05:24 AM SpeedEvil: If you google round, you can get very cheap linuxcnc capable systems on ebay for $10ish
05:24 AM SpeedEvil: That would take some time to find old reports of latencies for lots of boards, but still.
05:24 AM jthornton: how much is that in Dollars?
05:24 AM merkatorix: XXCoder, well, even if I should know what I am doing, I am a mathematician by trade and might be too distracted to wire it right ;-)
05:24 AM XXCoder: yeah my linuxcnc pc was my main pc, and 14 years old. latency was very good
05:24 AM SpeedEvil: jthornton: 100ish
05:24 AM jthornton: thanks
05:24 AM XXCoder: anyway soldering fix tb6560 takes way more time than tb6600 wiring
05:28 AM merkatorix: Mh, I thought the mesa is more expensive. But buying the mesa it will only work with linuxcnc, will it? On the other hand I could instal Linuxcnc on a USB-Stick and use an old laptop to get the data to the MESA board, all I needed would be the T6600 drivers, a power supply and the stepper motors?
05:28 AM jthornton: yea quite a change from the 5i20 7i33TA 7i37TA days for sure
05:29 AM XXCoder: MESA is more exensive but not huge jump
05:29 AM merkatorix: I guess I was a little scared, because it is not aviable on amazon and I have to handle the toll and all
05:30 AM jthornton: toll?
05:30 AM merkatorix: customs
05:31 AM merkatorix: sry, it's a topic I don't talk a lot in English
05:32 AM merkatorix: But I will have to think about to how to get it in Germany.
05:32 AM jthornton: I ship to Germany often
05:32 AM jthornton: did you see my pm?
05:34 AM merkatorix: Ah, thx, not used to pm in irc. You produce the mesa cards?
05:35 AM jthornton: no, I'm a Mesa dealer
05:35 AM jthornton: pcw_mesa: produces them
05:35 AM rmu: i ordered some from mesa about 2 years ago (to austria), no problem with customs
05:36 AM merkatorix: Ah, cool. As there is the firmware options, where do I need the PWM for?
05:36 AM jthornton: if you have a PWM spindle
05:37 AM jthornton: err PWM spindle driver
05:38 AM merkatorix: That reminds me of the problem how to tell my own spindle how fast it should go. Which other options do I have? I guess it is on/off and PWM?
05:38 AM merkatorix: Or are there I2C/SPI Spindles?
05:39 AM rmu: merkatorix: there are VFDs you can talk to via modbus (RS485)
05:39 AM rmu: but those usually also take 0-10V input (that will be generated with the PWM output)
05:40 AM rmu: and of course there a numerous other possibilities
05:40 AM merkatorix: But I guess PWM is one of the most common? And I guess there are easy ways to make my own spindle understand the PWM.
05:41 AM rmu: what spindle do you have?
05:41 AM jthornton: merkatorix: I recommend to start you download the LiveCD and boot from that or install the pure simulator on some old PC and learn a bit about LinuxCNC for $0
05:41 AM merkatorix: Is there a common frequency? A very low frequency low pass + adc might already do it.
05:42 AM rmu: PWM frequency with mesa can be configured, usually you would use something around 20khz i think
05:42 AM merkatorix: At the moment I don't have anything, but I plan to combine an attiny + Brushless outrunner + 3.17mm collet https://images-na.ssl-images-amazon.com/images/I/61Thfi1Hf%2BL._SY355_.jpg
05:42 AM merkatorix: I want to mill PCBs at first
05:43 AM merkatorix: Oh, I have to get go to the wastedump, which closes in 15 Minutes. BRB
05:43 AM rmu: attiny as interface to ESC?
05:43 AM merkatorix: Attiny as ESC
05:43 AM XXCoder: jthornton: yeah seems best, just install linuxcnc and use sim
05:44 AM XXCoder: I dont know how to have the fancy 3d rendering though lol
05:44 AM jthornton: aye you need to start somewhere
05:44 AM jthornton: what do you mean?
05:44 AM XXCoder: theres videos of linuxcnc running a virtual machine
05:44 AM XXCoder: it had spindle and table and stuff
05:46 AM jthornton: ah yea the Vismach http://linuxcnc.org/docs/2.7/html/gui/vismach.html
05:46 AM XXCoder: yeah'
05:46 AM jthornton: you should make one work and tell me how you did it so I can update the documents lol
05:46 AM XXCoder: lol
05:47 AM XXCoder: there should be generic mill and generic router
05:47 AM XXCoder: script at end of article is interesting
05:48 AM merkatorix: Mh, but thinking about it, I should postpone my esc project and use a generic one. I guess many of them take pwm as input anyhow, don't they?
05:51 AM XXCoder: jthornton: some elements of it is unknown like capture()
05:51 AM jthornton: I tried to make one once but could not figure it all out
05:52 AM XXCoder: saved link for now. my movation is very very low so I doubt I will work on uit but yeah
05:52 AM jthornton: same here my motivation is to get my new chicken pc working as I want :)
05:53 AM XXCoder: sadly I wish I had sucj excuse
05:53 AM XXCoder: well reading it some parts is very olvious
05:53 AM XXCoder: creation of 3d model and assigning names
05:53 AM XXCoder: and positioning it
05:54 AM XXCoder: have you managed to connect virtual machine with real linuxcnc?
05:55 AM jthornton: it's been quite some time since I messed with it but the simulation works fine if you want to look at that
05:56 AM XXCoder: so import or make, translate it to position, then add HAL motion range
05:56 AM XXCoder: then all together
05:56 AM XXCoder: Collections
05:57 AM XXCoder: so it makes machine named comp
05:58 AM XXCoder: but what is unclear is "meta" part. once machine is "made", how do I configure linuxcnc HAL to connect to it, so on
05:59 AM XXCoder: https://www.youtube.com/watch?v=1JzuyvvDWzc
06:00 AM jthornton: dunno, you might look at the simulator to see
06:01 AM XXCoder: there is only 2 videos and both has virtual machine already running
06:01 AM jthornton: that's a pretty cool one
06:01 AM XXCoder: second and last one https://www.youtube.com/watch?v=JbPgHUxsEBA
06:02 AM jthornton: ah ha that is hazzy now I know someone that has done vismach
06:02 AM XXCoder: youtube search found more
06:03 AM XXCoder: lame. https://www.youtube.com/watch?v=l5ucJukbZ1w maybe theres something spoken
06:04 AM XXCoder: um this one broke the tool lol https://www.youtube.com/watch?v=46LZGWQUSXo
06:05 AM XXCoder: and broke tool 3
06:16 AM merkatorix: When using a mesa card, I can run linuxcnc on ubuntu, can't I?
06:17 AM XXCoder: I'd recommand use distro that linucnc comes with.
06:17 AM jthornton: you can but it involves more work than using debian and many use linux mint
06:17 AM XXCoder: far less work all around
06:17 AM merkatorix: I thought mint and ubuntu is practially the same, when it is about internals.
06:17 AM XXCoder: just never update certain os aspects I think kernel is only one that should never upgrade?
06:18 AM merkatorix: So I still need a realtime kernel? I thought ne mesa card is then taking over a lot of stepping work.
06:18 AM XXCoder: when you hit e-stpp you want it to stop now.
06:18 AM merkatorix: True
06:18 AM jthornton: the ethernet cards require the preempt RT kernel and LinuxCNC Uspace
06:18 AM XXCoder: not 10 seconds later when pc is laggy or something
06:20 AM merkatorix: How much active control can I actually introduce into linuxcnc? I would like to integrate features like this autoleveling https://www.autoleveller.co.uk/
06:21 AM jthornton: you can do almost anything you want with LinuxCNC...
06:21 AM XXCoder: yeah linuxcnc is pretty amazing
06:21 AM merkatorix: So I don't have to take many steps before milling a pcb. When (if it ever comes that far) I finish the cnc, I would like it to automatically probe and mill.
06:22 AM XXCoder: theres someone who made program to probe the surface and contour engrave gcode to it
06:22 AM XXCoder: engraved a mouse which unless you live in 1980s, are very curvyu
06:22 AM merkatorix: Ah, I just noticed autoleveler already works with Linuxcnc ... cool :D
06:32 AM merkatorix: Actually I wanted to do the automatic probing by using the conductivity of the spindle and the pcb... just noticed that ball-bearings are insulating depending on the layer of grease :/
06:33 AM SpeedEvil: merkatorix: replace spindle oil with mercury
06:33 AM jthornton: just use a probe and look at the example G code file grid probe
06:33 AM jthornton: lol
06:33 AM XXCoder: you could build a probe using some plastic, 6 precision bearing balls, 3 precision rods, some circuit and soldering
06:33 AM XXCoder: and some kind of tip
06:34 AM XXCoder: https://www.youtube.com/watch?v=KOew0RzdaYE
06:34 AM SpeedEvil: A contact that is adequate to provide sense current - running on the spindle - and a PCB at an elevated voltage would in principle work too
06:34 AM SpeedEvil: A proper contact tip would be much more flexible and not only work on conductive things, and allow you to re-probe the partially done board though
06:35 AM XXCoder: lol! that guy? uses gold plated piercings
06:35 AM XXCoder: easy way to do precision I guess
06:35 AM SpeedEvil: Actual gold wire is pretty inexpensive, and can make easy contacts.
06:36 AM SpeedEvil: (for narrow diameters)
06:36 AM XXCoder: it needs to be rod its for interior contacts not touching part and such
06:36 AM gregcnc: but Renishaw...
06:36 AM XXCoder: costs hundred bucks or more
06:37 AM XXCoder: merkatorix: other way is quite literal contact switch
06:37 AM XXCoder: just keep in mind those has backlash of certain range but may be accurate enough for engraving
06:37 AM XXCoder: limit switch that is
06:38 AM XXCoder: variant https://www.youtube.com/watch?v=prb60JoxQSQ
06:39 AM XXCoder: guy did it hard way, even using bearing ball in lathe
06:39 AM XXCoder: but wow does look nice
06:41 AM XXCoder: its even adjustable nice
06:42 AM gregcnc: robovoice
06:42 AM XXCoder: huh
06:42 AM gregcnc: what material ws used for the contacts?
06:43 AM XXCoder: first video guy used percing that is gold plated
06:43 AM XXCoder: second one not known
06:45 AM XXCoder: second video is pretty amazing really
06:50 AM Tom_L: busy morning i see
06:50 AM XXCoder: hey
06:50 AM jthornton: morning Tom_L
06:53 AM Tom_L: jthornton, you used v-carve recently didn't you?
06:53 AM Tom_L: err f-engrave i mean
06:53 AM jthornton: yea made the harley thingy
06:53 AM Tom_L: is it able to do pocketing?
06:54 AM Tom_L: like for raised lettering instead of carved letters
06:54 AM jthornton: yea it can do inlays which mean it makes the hole for the letter and the positive for the letter
06:55 AM jthornton: and I think it can do positives as well
06:57 AM Tom_L: ok
06:59 AM jthornton: XXCoder: yippe I have the spring test working now to consolidate it all
06:59 AM XXCoder: :)
07:03 AM XXCoder: laters
07:04 AM jthornton: night
08:36 AM merkatorix: Well in case of the pcb the advantage is that I don't have to think about the hight of the actual tool, because the tool is the switch. I guess I'll just think about getting a better contact first (maybe a brush or I will just clamp the contact to the tool and unclamp it before starting. It's one step more, but that doesn't sound too hard)
08:49 AM rmu: i am entertaining the idea to get such a "pcb-probe" manufactured... just draw everything as PCB, get it manufactured for USD2 for 10 pieces ;-), add bearing balls and carbide shafts, done. small µC, probably stm32 because of familiarity. power it with lithium coin cell, and use a rfm69 or similar for wireless communication
08:50 AM rmu: just for fun
09:04 AM miss0r|office: I wonder what accuracy you could expect from that setup
09:04 AM miss0r|office: I mean; PCBs are pretty accurate, but not micro level
09:07 AM rmu: with a commercial tip i think it would not be that bad
09:07 AM rmu: the switches don't have to be that accurate, i think what matters more is repeatability
09:10 AM miss0r|office: hmm. Interresting thought
09:11 AM rmu: dimensional stability of FR4 also is very good usually, low thermal expansion coefficient
09:12 AM Rab: merkatorix, be aware that electrical contact creates a small spark that wears away the contact surface. This can degrade the cutting tip of your tool, and it's definitely not good for the bearings if those are part of the circuit. I would use a removable clip to the tool for probing, and keep the voltage and current low.
09:12 AM Rab: And that's another reason to use a "real" probe instead of the tool (although that can indeed be a handy shortcut).
09:22 AM rmu: Rab: never had a problem probing aluminium with typical HSS bits, input is some kind of 3.3v cmos schmitt trigger without too much protection i think, perhaps some resistor in series about 100k and a ferrite bead.
09:22 AM rmu: funny thing was that probing would only work if spindle was not running
09:23 AM rmu: (kress 1050)
09:24 AM rmu: don't think the energy involved is large enough to cause real problems
09:50 AM Rab: rmu, that sounds like a good setup. I believe that technically any voltage sufficient to wet the connection will ablate the contact points, but if it's on such a micro scale that it doesn't compete with normal tool wear, then it must not be a problem.
09:51 AM burkelfoo: Renishaw probes sense the contact resistance and reduce the current before the contacts open
09:51 AM Rab: I wouldn't use e.g. 12V straight from the PSu with no current limiting though.
09:52 AM burkelfoo: electrical erosion even at mA levels is a problem
09:54 AM Rab: 3.3V / 100K = 33uA, pretty small. ;)
09:56 AM burkelfoo is now known as pcw_home
09:56 AM pcw_home: you might take a look at the Renishaw patents
09:57 AM pcw_home: I think I would use sapphire balls and magnetic pickup (basically a differential accelerometer)
10:39 AM fragalot: Hey
10:49 AM merkatorix: Rab, thx, appart from the fact that the grease on the bearing balls stopps my idea. I really forgot about ruining them, especially when I ramp up the voltage
10:52 AM merkatorix: But I just thought some more about my setup, which seems to grow in expense, it is like 100€ for the Nema23+ TB6600, 100E for a Meanwell power supply and 120E for a Mesa Ethernet breakout board. Now I just noticed I forgot about endstops. So I have to search for good switches. And it seems like the TB6600 is actually the most basic you can operate a stepper. What is the cheapest way to operate a stepper motor and sensing the curren
10:52 AM merkatorix: t and lost steps? Does linuxcnc handle such data? Could you recommend a driver? I guess the price will make me feel OK with TB6600
10:53 AM merkatorix: rmu, :D The resistor is an important part in your setup. Good point to limit the current.
10:54 AM merkatorix: :D I just have to imagine a bit weldet to the copper, because of intense currentflow :D
11:22 AM Rab: merkatorix, I like Geckodrives and think they're a pretty good value, but they're not super cheap. Uh, I don't know about detecting lost steps unless you have axis encoders and are probably running closed-loop.
11:23 AM Rab: I believe the best cure for lost steps is preventative: a well-designed machine with appropriately sized and tuned components.
11:28 AM Rab: If your machine is just gonna lose steps by itself, that needs to be fixed...if it loses steps through some horrible event outside its performance envelope, that's probably not recoverable anyway.
11:30 AM merkatorix: But it detects obstacles? The problem is that gecko is almost 10 times the price of TB6600 and closed loop doubles it again, doesn't it?
11:30 AM merkatorix: Well, I think about detecting machine failues.
11:30 AM merkatorix: Or the object fell over and clamps somewhere.
11:31 AM merkatorix: But I guess I am back to the original design I planned.
11:32 AM merkatorix: As a power supply I plan a 480W 24V 20A ; MeanWell, DRP-480-24
11:33 AM merkatorix: Are there 24V supplies, which don't look as if you constantly are in danger of getting a shock?
11:34 AM merkatorix: I would preferr a schuko plug (earthed plug)
11:34 AM Rab: merkatorix, either TB6600 doesn't fail and you're happy, or it does fail and you can channel your rage into saving for Geckodrives...win/win
11:34 AM merkatorix: Yep, that's my plan ^^
11:34 AM merkatorix: And I have to check where to get the Mesa board
11:35 AM Rab: merkatorix, MeanWell is fine. I dunno about power supplies with integrated plug. That type of frame PSU is meant to go inside an enclosure to mitigate shock hazard. You could make a plastic shield over the terminals if you're concerned.
11:35 AM JT-Shop: if speed is not a issue then 24v is ok if it is then more voltage = more speed and acceleration with steppers
11:36 AM merkatorix: Mh, true even the spindle would need more. But I would like to be save to not kill the TB6600
11:37 AM Rab: Some PSUs have an IEC socket, which would work with a cordset for your locale.
11:38 AM Rab: Separate smaller PSUs dedicated to your drivers and spindle might be cheaper and more flexible.
11:39 AM merkatorix: Do you just take a PSU for everything and connect the GNDs?
11:40 AM merkatorix: I am somehow concerned about potential differences. ^^ I have to admit that I am used to power everything I build by batteries, as that is save to be probed by Oscilloscopes
11:40 AM Rab: That's one way to do it, and that happens to be how I'm set up...some people use monolithic PSU for everything though.
11:41 AM merkatorix: I was already thinking about a step down + lm7805 to power the Mesa
11:41 AM Rab: Max drive voltage for the TB6600 is 42V. I would consider a 36V PSU if you can find it, but 24V@20A is probably fine.
11:41 AM Rab: The Mesa what?
11:42 AM Rab: Like a breakout board?
11:43 AM Rab: A 7805 is only good for about 1A, with good heatsink...I would check the supply voltage/current specs for the Mesa product.
11:43 AM merkatorix: jthornton recommended the 7196 Step/IP an ethernet breakout board.
11:44 AM merkatorix: I did, it seemed like it wouldn't go over 2A and actually likes to be current limited. Therefore I would have combinded the LM805 and a buck converter.
11:47 AM Rab: Looks like 7i96 draws 250mA without an external load. You mean a buck converter from 24V to =>7V and LM7805 for regulation?
11:47 AM merkatorix: Yes
11:48 AM merkatorix: Just because I don't trust this cheap converters.
11:48 AM andypugh: Why not just buck-convert straight to 5V?
11:48 AM merkatorix: But I think you convinced me to use multiple supplies
11:49 AM Rab: merkatorix, I agree not to trust (especially, cheap) buck or other switching converters...they are often poorly regulated and can generate a lot of noise.
11:49 AM andypugh: My milling machine has a single 24V 2kW supply and then three buck-converters to make 5V for mesa cards, 5V for the LED lights and 12V to power the PC Motherboard.
11:49 AM andypugh: Works fine...
11:50 AM merkatorix: Now I would use this https://images-na.ssl-images-amazon.com/images/I/81coj3oIlEL._SL1500_.jpg 5V 2A supply
11:50 AM rmu: pcw_home: do you have links to any relevant patents? regarding this "contact resistance measurements"?
11:50 AM Rab: The one in the pic is 12V, but yeah, that's probably OK. ;)
11:51 AM merkatorix: ... Maybe I should search for a different article
11:51 AM merkatorix: But something like that.
11:51 AM andypugh: I just realised that the 24V PSU is of no current use. It was there for driving the A-axis servo on a 7i39, but I no longer used that.
11:51 AM andypugh: Might as well leave it, in case I want a 5th axis I suppose.
11:51 AM rmu: re. probing, I also tried glueing a stylus to a piezo speaker. don't know how stable or repeatable it is, but you can measure impact when probing ;)
11:52 AM andypugh: I have some souch probes that include a pieze element, so the technology is sound
11:52 AM merkatorix: Well, when it is about probing I need a certain place to calibrate my z-Axis, don't I?
11:53 AM Rab: merkatorix, if you're going to use a wall wart like that, it's good to source a name-brand OEM PSU e.g. CUI Inc or similar. I buy wall warts from Mouser/Digi-Key or other electronics distributor for critical applications rather than, like, eBay.
11:53 AM merkatorix: ... So I actually have 2 probes. One connected to the base, which detects the tool and one, which replaces the tool, to probe the working surface
11:53 AM andypugh: https://www.zeiss.co.uk/metrology/products/sensors/rst-p.html is the ones I mean, combined 3-pin mount + piezo
11:54 AM Rab: rmu, interesting approach. Seems like you could also use a phono cartridge.
11:54 AM merkatorix: Rab: Good point.
11:55 AM rmu: andypugh: the piezo speaker approach probably isn't ;)
11:55 AM merkatorix: I thought about it during the afternoon and forgot about it, as buying at amazon is so easy
11:56 AM rmu: pcw's suggestion with ruby balls and magnetic pickup amounts to phono cartridge
11:56 AM Rab: Amazon will happily vend total crap, though...you gotta do research to trust that route.
11:58 AM merkatorix: I went to the page of an local shop and they even have a cheaper Meanwell power supply which is less likely to be a fake.
11:59 AM Rab: Seems like piezo/cartridge/other stylus approach would be very sensitive, but might have issues with reproducability...like homing with limit switches. What are the thermal expansion characteristics of a phono stylus arm, and is it within a thou/tenth mil? Beats me.
12:00 PM rmu: Rab: that doesn't really matter, if temperature of you measuring arm is changing the temperature of the rest of the machine is changing too, so you have bigger problems
12:02 PM rmu: steel is about 1e-5/K
12:02 PM rmu: CFK could be in theory around 0 or even negative
12:02 PM rmu: i.e. shrink when being heated
12:03 PM Rab: The needle is mounted on some type of suspension, though, which is designed to be super sensitive and low inertia.
12:04 PM Loetmichel: the problem with using MM/MC pickups for measuring: they are meant to do AC
12:04 PM Loetmichel: not DC
12:05 PM rmu: they will give a peak at contact
12:05 PM Loetmichel: yes, but depending on speed the peak may be of different height
12:05 PM rmu: but you will scratch your work with the needle ;)
12:05 PM Rab: Ceramic cartridges are piezo-based.
12:05 PM merkatorix: Loetmichel, :D Didn't realize you joined over here, too.
12:06 PM Rab: Hmm, the good thing about a phono needle is that you could probe in 3D.
12:07 PM Loetmichel: rmu: usual pickups need about 10mN of force for optimal deflection
12:07 PM Loetmichel: i doubt that coud scratc steel or alu surfaces even with a diamond needle
12:11 PM rmu: probably not
12:11 PM rmu: if it would scratch steel it probably would ruin proper records immediately
12:12 PM Rab: Diamond/sapphire will certainly scratch steel...and wear records with every play.
12:13 PM Rab: Differentiating XYZ from two stereo outputs might be a trick.
12:17 PM sync: Loetmichel, rmu scratching is a real problem
12:17 PM sync: even at lower forces
12:21 PM Loetmichel: i can imagine. depends on how deep a scratch you can tolerate.
12:28 PM rmu: yeah, i wouldn't use a diamond needle
12:38 PM andypugh: Here’s an idea for a probe based on the design of one of those Zeiss probes.
12:38 PM andypugh: https://photos.app.goo.gl/h2Y8L3WZAufaRtd97
12:39 PM andypugh: Rather than balls you suspend the tripod between 6 hardened steel dowels
12:40 PM andypugh: It looks like a complicated shape to make, but if you look, I have arranged the picture carefully to show that the dowel holes are just drilled straight in pairs with a 120 degree rotation between pairs.
12:40 PM andypugh: So you can make that with nothing more exotic than a hex collet block.
12:41 PM andypugh: (Ie one of these: https://www.arceurotrade.co.uk/Catalogue/Collets/ER-Collet-Fixtures/ER32-Collet-Blocks
12:42 PM fragalot: the other good thing is that they don't even have to be all that precise
12:42 PM andypugh: Indeed, you get the symmetry from the rotation.
12:42 PM andypugh: And it doesn’t even matter if the dowels are pushed in different distances
12:43 PM andypugh: I ought to make one to prove it works :-)
12:43 PM fragalot: :P
12:44 PM rmu: I think i saw a construction like that in one of the patents
12:45 PM andypugh: What struck me was how very simple it actually is to machine.
01:38 PM fragalot: andypugh: have you made one yet?
01:47 PM gloops: would says thats a good buy https://www.ebay.co.uk/itm/Three-LIN-Engineering-2-8A-NEMA23-Stepper-Motors-with-Optical-Encoders/253934648773?
01:58 PM Roguish: LIN motors are pretty good. Local Silicon Valley company, now owned by a Chinese corp.
02:12 PM gloops: well, steppers with encoders are usually 100+ new
03:06 PM nallar is now known as Ross
03:11 PM Tom_L: threaded and the tips could be adjustable as well
03:11 PM Tom_L: andypugh ^^
03:11 PM Tom_L: for centering the probe
04:29 PM andypugh: Tom_L: Doesn’t help.
04:30 PM andypugh: No mater how far in or out the dowels are, the tripod sits in exactly the same place.
04:32 PM flyback: https://www.youtube.com/watch?v=RGs-9w91XM0
04:32 PM flyback: also ve7it BMCC
04:34 PM andypugh: We had a bunch of NiFe caving lamps when I was a student (in the caving club)
04:34 PM XXCoder: andypugh: dunno flat ends of rod inserts are those adjustable so all sides equally respond to touch?
04:35 PM andypugh: You seem to be misunderstanding where the tripod sits
04:35 PM flyback: I love watching them cnc vent adaptors
04:35 PM XXCoder: gloops: that seems nice though so long stem dunno lol
04:35 PM flyback: hehe
04:35 PM flyback: well not cnc
04:35 PM flyback: lathe
04:36 PM XXCoder: andypugh: maybe I was thinking that theres 3 rods sitting on those
04:37 PM XXCoder: im still thinking how to make super shallow probe but its diffult lol
04:37 PM XXCoder: even if its just for Z
04:46 PM andypugh: https://photos.app.goo.gl/wutDWGmUQResyGuE8
04:47 PM XXCoder: ah i see!
04:47 PM XXCoder: uses more precise sides or rod not ends
04:47 PM XXCoder: *of
04:47 PM andypugh: Indeed
04:48 PM Deejay: gn8
04:48 PM XXCoder: is those adjustable?
04:48 PM XXCoder: so can have zero all way around on tip
04:48 PM andypugh: You would need to mount it with adjusting screws. The Renishaw heads have a spherical seat and 4 adjusting screws, I think any spindle probe needs that
04:49 PM andypugh: That design is (deliberately) not adjustable itself.
04:50 PM XXCoder: interesting
04:51 PM andypugh: I would make one, but I have more probes than I need. I found 10 of the Zeiss ones in a skip
04:51 PM XXCoder: nice lol
04:52 PM XXCoder: I dont know if I can ever use a probe on my router
04:52 PM XXCoder: clearance is too small
04:53 PM XXCoder: wow pretty dumb. people now claiming drinking milk is "purity" for white people
04:53 PM SpeedEvil: I wonder how many of them know most of it comes from brown and black cows?
04:53 PM XXCoder: lol
04:54 PM XXCoder: I guess thinking goes like this "i am pure white. this is white so I should drink it"
04:54 PM XXCoder: I wonder if someone could troll em into requiring that everything eaten must be white
04:55 PM SpeedEvil: There aren't that many white foods.
04:55 PM XXCoder: exactly
05:18 PM XXCoder: if cows formed a country, they would be third most largest greenhouse gas pollutor
05:18 PM XXCoder: below china and usa
05:19 PM XXCoder: china 10.3 gigatons co2, usa 5.3, republic of cows 5.0, india 2.5
05:21 PM andypugh: I am not white. I am a sort of pinkish beige. Definitely not the colour of milk
05:22 PM andypugh: If I spend long enough in the sun I suppose I can take on the hue of strawberry mik.
05:23 PM XXCoder: my family tans easily and are light olive colored.
05:23 PM XXCoder: i stayed indoors nearly whole summer yet i look like I live on rock and not under it.
05:31 PM andypugh: Calling people “olive coloured” is even dafter than calling them white. Are you a green or a black olive?
05:33 PM XXCoder: no idea why its called that
05:33 PM XXCoder: specific shade of pink-tan'
05:40 PM SpeedEvil: We should only specify with pantone.
05:41 PM XXCoder: lol
05:43 PM andypugh: I think I about RAL 1015
07:24 PM Tom_L: doesn't look like the angle you drill those would be all that cricical either
07:25 PM XXCoder: Tom_L: as long as its all same I dont think so no
07:25 PM andypugh: No, the whole thing looks to be easy and non-critical as long as the collet block is good.
07:25 PM XXCoder: though if closer to parallel it'd have harder time losing contact
07:26 PM Tom_L: i got a feeling you may build one just because :)
07:26 PM andypugh: I am thinking about it
07:27 PM XXCoder: document it
07:27 PM XXCoder: youtube it]
07:27 PM Tom_L: are the switches each half of the 'triangle' ?
07:27 PM andypugh: You need to wire the other ends of the dowels together.
07:27 PM Tom_L: or the whole triangle set and the tip
07:28 PM XXCoder: andypugh: you might be able to make it adjustable
07:28 PM XXCoder: somehow need to push rods in or out
07:28 PM Tom_L: wire to the 3 tip legs and another to all 6 triangle pieces
07:28 PM XXCoder: that would alter where contact point is
07:28 PM andypugh: Current flows onto one dowel, across the tripod pin, to the next dowel, onto a short wire to the next dowel and so on.
07:29 PM Tom_L: so each dowel pair is a switch
07:29 PM andypugh: Yes
07:29 PM andypugh: But you connect them all in series
07:29 PM XXCoder: hmmmmm
07:29 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/probe_base.jpg
07:29 PM andypugh: Carbide rod might be a good choice, it is conductive (enough) and doesn’t get an oxide film
07:29 PM Tom_L: like that but wires
07:30 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/probe_contacts.jpg
07:30 PM andypugh: Like that but dowels. How is the electrical connection done on that?
07:30 PM XXCoder: plenty of carbide tools to make em from
07:30 PM Tom_L: carbide rod or something plated with gold or platinum
07:31 PM XXCoder: andypugh: if its usual, then theres pcb under it
07:31 PM Tom_L: i thought i had pics of it...
07:31 PM XXCoder: balls are soldered to em'
07:31 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/emc/small_probe8_small.jpg
07:32 PM andypugh: Basically my design works exactly the same, but the geometry is a lot less critical. (If you put one ball a bit too deep on the ball designs it messes things uo)
07:32 PM Tom_L: yeah
07:33 PM Tom_L: so the body would still be some sort of non-metalic
07:33 PM andypugh: Yes, probably Delrin, pressed into a metal tube
07:33 PM pfred1: ballpoint pen tips are carbide
07:33 PM Tom_L: is this your design?
07:34 PM XXCoder: andypugh: think your design depends more on holes spacing
07:35 PM XXCoder: though easier to be more precise with that than ball placement
07:35 PM andypugh: XXCoder: the hole spacing isn’t critical as long as it is symmetrical.
07:36 PM andypugh: So if you can index accuratly with a hex collet block, you can leave the vice and drill in the same place.
07:36 PM Tom_L: XXCoder the angle couldn't be too great or the rods would cross each other on the far end
07:36 PM XXCoder: tom yeah thats the other end of range limit
07:36 PM XXCoder: andy yeah was talking about that :)
07:36 PM Tom_L: you'd have to do opposite on 3 wouldn't you?
07:37 PM andypugh: I reckon you could make it with a drill press...
07:37 PM andypugh: Tom_L: The other half of each pair can be at a different angle and position (within reason) and it still works.
07:37 PM Tom_L: right
07:38 PM andypugh: I might have to prove that a drill press will do the job.
07:38 PM Tom_L: the order you drill them may prove critical
07:39 PM Tom_L: you may do the rod holes before the larger ones
07:39 PM XXCoder: so label pairs 1 2 3 and drill em in that order?
07:39 PM andypugh: Yes, definitely. And using end-mills not drills I reckon too.
07:39 PM Tom_L: i'm not sure. either way the bit may tend to wander some
07:39 PM XXCoder: endmill wanders less
07:40 PM Tom_L: you actually may want to use and endmill on the larger holes after drilling the rods
07:40 PM XXCoder: spring cut snot out of it also, so no taper inside also
07:40 PM Tom_L: might cooperate better than a drill point
07:40 PM andypugh: XXCoder: No, set up the drill of centre and the vice at about a 10 degree angle. Drill a hole, index, drill a hole, index, drill.
07:40 PM andypugh: Then move the drill across to (approximately) the same distance the other side of centre, and drill three more holes.
07:41 PM Tom_L: just use a center cutting endmill
07:41 PM andypugh: Then drill the clearance holes.
07:41 PM Tom_L: probably cut the center out last
07:42 PM andypugh: Anyway…. Sleep
07:42 PM Tom_L: later
07:42 PM XXCoder: later
08:41 PM XXCoder: guys dishwasher at home is functional now. it was a long and complex repair