#linuxcnc Logs

Jan 08 2018

01:36 AM gloops: 2 stepper motors, 1 axis, chinese BOB, 2 normal drivers, linuxcnc, any reason this will not work?
01:43 AM IchGucksLive: morning from germany
01:44 AM XXCoder: hey
01:44 AM XXCoder: dunno gloops
01:44 AM IchGucksLive: XXCoder: lowest temp in history in canada yesterday
01:44 AM XXCoder: fun
01:44 AM IchGucksLive: as the hotest ever mesurerd in sydney
01:45 AM gloops: a few times now archivist has said the big design flaw is going to show soon, in my machine, im just thinking it over again, cant think of anything
01:45 AM gloops: im sures hes spotted something haha
01:47 AM gloops: coldest temp in canada must be pretty cold
01:47 AM XXCoder: maybe? not sure
01:47 AM XXCoder: yeah recently flornia had colder temperate than alaska
01:48 AM gloops: this has been a cold winter here, last year was cold and the year before, i dont know how global warming is working
01:48 AM MrHindsight: gloops: what machine are you making?
01:49 AM XXCoder: gloops: its global climate change
01:49 AM XXCoder: and cold now is abnormal because it USED to happen much more often
01:49 AM XXCoder: but we havent had one for quite a long while by now
01:49 AM gloops: MrHindsight just a router
01:50 AM gloops: its almost made
01:50 AM gloops: i dunno XXCoder i can remember a period of warmer winters maybe 10 years ago where we never had a frost for a few years
01:51 AM XXCoder: heh it didnt even get cold here
01:51 AM XXCoder: those years.
01:51 AM gloops: i also remember old people talking about winters in the 40s and 50s which sounded cold
01:51 AM XXCoder: thing is this - ice in greenland is melting a lot now
01:51 AM XXCoder: which cooled water and messed with weather patterns
01:51 AM gloops: ups and downs are normal
01:51 AM XXCoder: we are talking about GLOBAL climate after all.
01:52 AM XXCoder: 2 degrees c global is huge
01:52 AM XXCoder: 2 degrees c seems small till you realize its 2c increase all around world. that is huge amount of heat.
01:52 AM gloops: well we are heading into an ice age anyway
01:53 AM XXCoder: we WAS going into ice age.
01:53 AM XXCoder: we [pushed it out and now too far
01:53 AM gloops: haha, we still area
01:53 AM gloops: are
01:54 AM gloops: MrHindsight did you do the coding on the axis/joints setup of 2.8?
01:54 AM XXCoder: sure you say that when greenland is melting at massive speed.
01:54 AM XXCoder: north pole ice may never return
01:54 AM MrHindsight: gloops: not me, I try to avoid code :)
01:54 AM XXCoder: south pole just lost HUGE chunk of ice, which broke off
01:55 AM XXCoder: chunk larger than some states of usa
01:55 AM gloops: ice ages work on mechanical cycles, the tilt of the earth, our orbit around the sun, they cant be changed
01:55 AM gloops: regular as clockwork
01:55 AM XXCoder: uh
01:55 AM XXCoder: earth isnt clockwork
01:56 AM XXCoder: otherwise we would never have leap seconds
01:56 AM gloops: over a long period the cycle is consistent
01:56 AM XXCoder: it was
01:56 AM XXCoder: also not quite, even pre-human era
01:56 AM gloops: this is not the highest co2 level
01:57 AM XXCoder: just that we humans is changing world at far faster rate than ever
01:57 AM gloops: volcanic activity was much higher, co2 from animals has been much higher, all kinds of factors, never stopped anything
01:59 AM XXCoder: when?
02:01 AM gloops: http://cdn.antarcticglaciers.org/wp-content/uploads/2013/10/Ice_Age_Temperature.png
02:01 AM gloops: we are somewhere near the top of the cycle
02:03 AM XXCoder: "Glaciers are shrinking world-wide, and the rate of recession is generally increasing. It is difficult to relate the behaviour of individual glaciers to climate change, but the general trend of accelerating recession is indicative of a world-wide response to global warming."
02:03 AM XXCoder: http://www.antarcticglaciers.org/question/evidence-show-glaciers-shrinking-due-global-warming/
02:04 AM XXCoder: same site you cited.
02:04 AM gloops: probably a minor variable
02:04 AM XXCoder: okay.
02:05 AM gloops: it is not going to change anything in the long term
02:05 AM XXCoder: okay.
02:05 AM XXCoder: honestly you are correct
02:06 AM XXCoder: if humans die off, earth will recover in just thousand years?
02:06 AM XXCoder: nothing compared to 5 billion years of earth history
02:06 AM XXCoder: nothing changes long term, in least till SOL eats earth in its red giant era
02:09 AM XXCoder: actually sun will make earth too warm to have life at 500 million years. but then who knows with evolution since change is so slow.
02:09 AM gloops: humans wont die off, we are well within historical climate extremes atm, could go a good bit higher temps and that wouldnt be unusual, but the coming trend is a big decrease
02:09 AM XXCoder: unlike human caused changes
02:11 AM gloops: i live alongside a valley that was cut by ice dragging boulders against the rocks
02:12 AM gloops: there is a prehistoric ice age shelter used by humans
02:12 AM XXCoder: region I live in rarely have soil deeper than 2 or 3 inches
02:12 AM XXCoder: less sometimes
02:12 AM XXCoder: also due to ice age
02:12 AM gloops: atm it is green and leafy (in summer) with a tinkling stream, but the ice will return
02:26 AM Deejay: moin
02:37 AM gloops: well i suppose i better go and have a look at it
04:56 AM thochguertel_ is now known as thochguertel
05:29 AM srk: https://wiki.base48.cz/File:Shaepoko_rebuild0.jpg
05:30 AM srk: XXCoder: rebuilt from smoothie to linuxcnc a wekk ago ^^ :)
05:30 AM XXCoder: nice
05:31 AM XXCoder: better than smoothie I guess
05:31 AM srk: yup, barely usable
05:31 AM XXCoder: whats difference>?
05:31 AM srk: we even got rs485 working for spindle control
05:31 AM srk: jogging for example :D
05:31 AM srk: interface
05:32 AM srk: basically
05:32 AM XXCoder: input/output serial?
05:37 AM srk: XXCoder: yup
05:37 AM srk: not for long ;)
05:39 AM XXCoder: planning to upgrade or?'
05:44 AM srk: yes, both the machine and electronics
05:44 AM srk: ballscrews, supported rails or linears and canopen 402 based servos hopefully :D
05:45 AM XXCoder: linears seems so cheap now
05:45 AM XXCoder: dunno if expecially better than sbr or nopt
05:46 AM enleth: srk: now that's a proper hackerspace - there's a club mate stack table
05:46 AM srk: :))
05:47 AM srk: more of them! :D
05:47 AM enleth: srk: it helps of you mill a piece of plywood with an inset that fits inside the crate, so that it doesn't slide off
05:48 AM enleth: *if
05:48 AM srk: yeah, sounds good, though we might replace this one with a shelf as we are running out of space
05:48 AM srk: no space left on device!
05:48 AM enleth: we have a couple at HSWAW, very popular as standing desks
05:49 AM srk: cool!
06:30 AM fragalot: steppers arrived \o/
06:31 AM jthornton: Yea
06:32 AM fragalot: They're smaller than I expected
06:42 AM archivist: said the actress to the bishop
06:42 AM jthornton: as in shorter?
06:43 AM fragalot: I just expected N34 8Nm steppers to be larger than they are
06:43 AM fragalot: they're not small by any means :P
06:57 AM jthornton: I picked up a pack of Sylvania LED bulbs 800 lumens 8.5w and dang they are brighter than the 60w they claim to replace
06:57 AM fragalot: that's because they are more directional
06:58 AM fragalot: a 60W incandsecent bulb produces about 800 lumens too, but since it's omnidirectional, most of it gets lost
06:59 AM jthornton: these are just really bright as in lights up the stairs like the sun
06:59 AM fragalot: :)
06:59 AM fragalot: could someone please tell me why the hell "(0 + ((4) * (1)))" works out to "9" according to this compiler ebcause it's driving me insane.
07:01 AM jthornton: must be new math
07:08 AM XXCoder: jthornton: how old is that 60w it replaced? because all lights dim with age
07:08 AM jthornton: wow got a 46g egg this morning... getting closer to a medium egg
07:09 AM XXCoder: next, 150g ;)
07:09 AM jthornton: the old ones were compact florescent bulbs but the LED's are brighter than a brand new 60w bulb
07:09 AM jthornton: next hen is in the nest box laying one now
07:09 AM XXCoder: ah yeah crfl dim with age even worse than old
07:10 AM jthornton: that's why I replaced them
07:10 AM jthornton: LED's are getting more cost effective now
07:10 AM XXCoder: this house and last house 100% led.
07:11 AM XXCoder: I love led, because color is much better.
07:12 AM XXCoder: i was one of earliest buyers of led lights lol, at 10 years ago or so
07:12 AM XXCoder: my first blub was $30, and next 2 was $30 total
07:12 AM XXCoder: and it was 40w lol
07:12 AM XXCoder: or rather, 11w/ 440lumens
07:13 AM XXCoder: it was worth it. I could easily see differece between gray and bley lego colors :) I used to sort em a lot back then
07:15 AM IchGucksLive: hi
07:16 AM fragalot: I'm just using LED lighting because I prefer heating my houes with gas rather than electricity :P
07:16 AM fragalot: hi IchGucksLive :) -- steppers arrived!
07:16 AM IchGucksLive: col
07:16 AM IchGucksLive: o
07:16 AM XXCoder: fragalot: insecents save heating money if you live in csuper cold places in winter
07:16 AM IchGucksLive: beer
07:16 AM the_wench: I am not your waitress!
07:16 AM XXCoder: I would change em out as soon as it warms up though
07:16 AM fragalot: XXCoder: electricity is more expensive than gas.
07:16 AM XXCoder: ah yea depends on where
07:17 AM IchGucksLive: venezuelas gas 0,14Cent a gallon
07:18 AM IchGucksLive: 14cent
07:40 AM IchGucksLive: hi Meduza
08:36 AM thochguertel_ is now known as thochguertel
08:38 AM IchGucksLive: thochguertel:
08:40 AM IchGucksLive: thochguertel:
10:54 AM thochguertel_ is now known as thochguertel
11:32 AM gloops: tonights job - find 2 missing packets of bolts
11:32 AM MarcelineVQ: good luck :O the thing about bits is that they're never missing, they're hiding
11:34 AM gloops: already looked everywhere, know theyre in the garage as ive already used a few, infuriating
11:34 AM gloops: thats the trouble with cracking on leaving everything where you drop it, eventually you hit a wall
11:35 AM IchGucksLive: hi
11:36 AM miss0r: gloops: I'm not sure my memory serves me right, but werent you the one that made a comment on the mixture of hex, allen and pz screws on my design? Well, I've changed it all to allen now :P
11:36 AM miss0r: I meant I had to buy 600 screws, to get the 8 I needed - but surely it was nessesary
11:37 AM miss0r: s/I/it
11:37 AM gloops: wasnt me miss0r but i did see the comment heh, i had bolts that were slightly too long and wouldnt cut them down, ordered some more, got them, tried a few, perfect, now comes to assembly - cant find them
11:39 AM miss0r: lol
11:39 AM miss0r: I quite quickly posted a video in here yesterday on a working trial run with the EDM, did you happen to see it?
11:40 AM Rab: gloops, perhaps you put them someplace safe, where they were sure not to be misplaced or discarded...that's how I lose everything.
11:40 AM gloops: i did see it, congrats, im probably not the best critic of an edm as no experience of them all, but seemed to working as it should
11:41 AM miss0r: gloops: Doesn't matter. I like having people to share a project with. I hope to see some photos of yours soon?
11:41 AM gloops: probably Rab, ive looked and looked, you end up just stood looking and wandering round not knowing what to do about it lol, theyll be somewhere where i thought put these where they wont get knocked over and spill everywhere
11:42 AM gloops: i hope i have some photos soon too miss0r, getting past the time it should have been working, but anyway bbiab
11:49 AM fragalot: hi miss0r :)
11:51 AM gregcnc_: miss0r, my fastener OCD is relieved even without a photo
11:53 AM fragalot: gregcnc_: what he hasn't mentioned is that it's part imperial hex & part metric
11:54 AM gregcnc_: regretfully I once threaded a part with metric diameter and imperial pitch, I still have to come up with a metric gear set for my clausing
11:55 AM fragalot: xD
11:55 AM fragalot: Dremel does that too
11:57 AM IchGucksLive: gregcnc_: it is as it is
11:58 AM fragalot: that is one thing that I like about imperial thread VS metric though, that it's a "teeth per distance" thing
11:59 AM fragalot: at least for threading on a manual lathe without needing to reverse.
12:49 PM IchGucksLive: hi ulli
12:54 PM Deejay: hi michael
12:54 PM Deejay: ^^
01:19 PM Tom_itx: MrHindsight, http://www.coldheaders.com/
01:19 PM Tom_itx: try them about the cap screws
01:19 PM Tom_itx: my guy didn't stock them
01:19 PM Tom_itx: but suggested those guys if anybody would have them
01:26 PM Tom_itx is now known as Tom_L
01:40 PM miss0r: Back. I had to do an engraving job on another PC sharing the screen with this one. so: hello fragalot
01:40 PM MrHindsight: Tom_L: thanks for checking
01:40 PM miss0r: gregcnc_ So it was you that played on my OCD? :) Anyway, i've changed it all out now
01:47 PM gloops: end of the day a bolt is a bolt, theres no room for bolt elitism when something is in the trial stages, you need to clamp it up to test it
01:49 PM jesseg: Later on I guess the need may arise however for a bolt that speaks with a distinguished posh accent, no? :P
01:50 PM gloops: if i were selling engineering services i wouldnt post pics of shoddy bolt work anywhere the general public could see
01:50 PM gloops: otherwise not important
01:53 PM gloops: the worst bolting ive come accross is the single 8mm stretch bolt used to secure the injectors on mercedes diesel engines - and many other engine
01:54 PM gloops: you cannot apply enough torque with a ratchet to seat the injector properly ever
01:57 PM sync: well, if you can stretch it it worked
01:57 PM gloops: the bolt snaps in the head first, the only way to do it is to get someone with a hammer and chisel on the injector, whack it while you tighten, then you can get the bolt down far enough, a touch at a time
01:58 PM gloops: sync it doesnt work it stretches before the injector is sealed
01:58 PM MarcelineVQ: :(
02:01 PM sync: when I changed the injectors on some om606 I had no issue
02:02 PM gloops: how did you tighten?
02:04 PM sync: put the new injectors in and then just used my torque wrench
02:04 PM miss0r: So now I have a working proto type of the EDM head. Now its time to design and build the real thing :)
02:05 PM gloops: you mean that isnt the real thing miss0r? lol
02:05 PM miss0r: no!
02:05 PM phiplii: JT-Shop, homemade relay board works - been running for 2 days now without issue
02:05 PM miss0r: Its just a proto type. The real thing needs to be alot more professional
02:06 PM gloops: sync they would have been blowing before he got to the end of the road
02:07 PM miss0r: like: the spindle drive, needs to be a step motor, so I can spark out threads with it too. Most of this is not build to be the best it can be, rather I used what I had laying around
02:07 PM JT-Shop: phipli: cool, a 5v relay board?
02:08 PM JT-Shop: Henry is free ranging now :(
02:08 PM sync: no, the engine still runs gloops
02:08 PM * fragalot is wondering who at autodesk decided it was a good idea to keep the "history" as a single icon row at the bottom of the screen
02:08 PM sync: it was not too much hassle, they came out suprisingly easy and went back in easy
02:08 PM MarcelineVQ: fragalot: yeah :<
02:11 PM fragalot: currently trying to think of creative ways to avoid using spacer blocks on the ballscrew :P
02:13 PM phiplii: JT-Shop, I used 12v relays because it was what I had - I was using a 12v supply regulated down to 5 for the MCU anyhow, so wasn't the end of the world
02:13 PM phiplii: If I make another one I'll use 5v relays
02:14 PM phiplii: Poor Henry
02:16 PM miss0r: gloops: ofcourse its not the real thing. This was made up as I go. Now I have learned alot of stuff. Also: I know the amount of work I've put into this could seem like a waste, but stuff like this - You just won't get any useable results with a flimsy makeshift setup.
02:17 PM miss0r: And I still need to figure out a good filtering system... The one I use is, as I mentioned before, makeshift :]
02:18 PM gloops: its good to see that something actually can work though miss0r, even if it does need making all over again
02:20 PM miss0r: Yeah. This thing as some issues that needs fixing :) And I think I've made the nodes & lessons to make that happen. Now Its time to see if that is just halfway true :]
02:20 PM fragalot: make proper motor mount, or jerry-rig one & cnc a proper one?
02:20 PM miss0r: Also, I could never show up at a client with this thing looking like the franken-EDM, they would probally not allow me near their equipment with it :)
02:21 PM fragalot: miss0r: that's why you paint it
02:21 PM fragalot: :D
02:21 PM miss0r: Theres just somethings you can't cover with paint :)
02:21 PM fragalot: it's amazing what a can of spraypaint and a few cable conduits can do
02:22 PM roycroft: whatever you cannot cover with paint you can wrap in duct tape
02:22 PM miss0r: Also, last night I found that I didn't have as tight tolerences on the axle seals as I thought. So I had to abort when it sprung a leak, and water came in contact with the bearings
02:22 PM miss0r: roycroft: Which closes the circle on the franken-EDM
02:23 PM fragalot: spraypaint the duct tape, say it's speedtape.
02:23 PM miss0r: woah! thats alot easier than what I had in mind :�
02:23 PM gloops: its a bit different making the real one though, the first one there was always some doubt, is all this for nothing? when youve seen it work the second one is a more confident job
02:23 PM miss0r: Indeed
02:24 PM gloops: when you look at some of the great engineering designs, they tried hundreds of times before it worked
02:24 PM miss0r: I hope this one proto type will do well enough thou :)
02:24 PM miss0r: let me upload a picture of the controller.... its good lols
02:25 PM miss0r: https://imgur.com/a/q6Gge
02:25 PM miss0r: This realy realy needs some tidying up hehe
02:26 PM fragalot: is that a space heater at the top?
02:26 PM miss0r: Its actualy a rig I made back in the day for testing soft starters
02:26 PM gloops: looks ok to me
02:26 PM miss0r: whats left of it anyway
02:26 PM miss0r: gloops: I don't want 'OK'.. I want wow :)
02:27 PM fragalot: ^ :)
02:27 PM gloops: i cant see anything wrong there lol
02:27 PM miss0r: I've added a picture of controller 0.1alpha on that link
02:28 PM gloops: its a machine, are you supposed to open the fuse box and say 'wow' haha
02:28 PM miss0r: So, as you can see i've move some stuff around... then moved to a new box hehe
02:28 PM miss0r: gloops: well.... yes? :)
02:30 PM gloops: nah, all thats usually said is 'can you see anything thats black'
02:30 PM miss0r: I have some serious OCD when it comes to control boxes, I know.
02:30 PM miss0r: That happened a few times.. AND it was inside that 'wind tunnel' at the top. Its a real pain taking it apart
02:32 PM fragalot: gloops: You are supposed to open a machine & not be intimidated to tear into it to troubleshoot it, without having to fear you'll never get all the wires back in :P
02:32 PM miss0r: I'm also going to replace the two PLCs with a single arduino. Thoes PLCs sucks reading the analog signal from the amp winding. The fastest they can update are once every 200ms. That is too slow for overcurrent detection
02:33 PM gloops: fragalot yeah it should be neat and properly wired up, miss0rs pic that looks fine, i wouldnt look at that and think any dodgy wiring was going on or anything
02:33 PM fragalot: gloops: so the loose wires flopping over the IGBT don't bother you? :)
02:34 PM gloops: no, they arent bare
02:34 PM skunkworks: miss0r, how fast do you need to see the over current?
02:34 PM fragalot: gloops: look closer
02:35 PM miss0r: skunkworks: I'm not sure yet. I will know alot more once I get the arduino in there. The best answer I have right now, is that 200ms is too slow.
02:36 PM fragalot: miss0r: bear in mind the arduino won't be able to deliver a higher performance than about 100µS out of the box
02:36 PM miss0r: fragalot: Thoes are lacquer coated motor winding copper. not bare ;)
02:36 PM fragalot: miss0r: IGBT2, L & N are disconnected
02:36 PM fragalot: :)
02:36 PM miss0r: fragalot: That is still twice as fast :)
02:36 PM fragalot: miss0r: micro. not milli
02:36 PM fragalot: i'm not taking into account the rest of the program, that' sjust the "analogread" function
02:37 PM miss0r: damn, yeah. I think that will be plenty fast
02:37 PM miss0r: and they are not as much disconnected, as they were used as a test point when the photo was taken
02:38 PM jesseg: miss0r, fascinating project. I've puzzled a bit on trying to make a microcontroller based EDM, but never did. EE is my background, and I might have some ideas if you're interested
02:38 PM fragalot: the only experience I have with EDM is when I made a doorbell powered one.. not the most elegant or efficient one, lol
02:39 PM miss0r: jesseg: I am. I will always be collecting as much info for this up to the point I build the real model
02:39 PM fragalot: BUT, it did work.
02:39 PM miss0r: I started with the exact same stuff :)
02:39 PM miss0r: horrible little thing. It hardly made a dent in a stanley blade. :)
02:39 PM gloops: i wouldnt go nitpicking over little bits and bobs like that fragalot
02:39 PM fragalot: haha
02:39 PM miss0r: This one cuts a 5mm hole through a stanley blade in about 8 seconds now
02:40 PM fragalot: gloops: not sure if it's obvious - but I'm not serious...
02:40 PM fragalot: miss0r: my original one did that in about 1 millisecond flat.
02:40 PM fragalot: until I learned that 600A was too much.
02:41 PM fragalot: :D
02:41 PM jesseg: miss0r, I think once the arc starts all the work is done that's going to be done because after a few microseconds the arc endpoint surface area spreads over too large an area, so I think you want to cut off current *really fast* once the arc is started. You may want to use a micro that is running plain C or even better assembly, and using a 1M sps ADC, or maybe even just use a threshold detector to detect when the arc is started
02:41 PM miss0r: hehehe. Mine let the smoke out 5 minuts into a trial run.. hehe, I never fed it more than 50 volts at 5 amps. but that did the trick. (it was a 6 volt dc winding :] )
02:41 PM fragalot: jesseg: there really is no real need anymore to write direct assembly with modern chips
02:42 PM fragalot: if you go to an STM32 based microcontroller, most of the "high speed" stuff has hardware support to work even faster
02:42 PM jesseg: fragalot, true, although sometimes interrupt latency on 32 bit micros is surprisingly long
02:42 PM fragalot: jesseg: latency I measured on ours was still in the nanoseconds terriroty
02:43 PM jesseg: but yeah, anything that can get you microsecond response times to arc initiation should work
02:43 PM jesseg: fragalot, excellent! What cpu and how fast was it running?
02:43 PM miss0r: jesseg: I am doing something simular actualy. I'm pulsing the output on and off at a variable frequenzy. Then I basically control the current by increasing or decreasing the sparkgap with the stepper. This is where I get some issues with the analog read function is being too slow.
02:44 PM fragalot: jesseg: STM32f7 based, somewhere around 200Mhz
02:45 PM sync: jesseg: why would it be long? the latency on stm32 is something between 7-9 cycles depending on how the busses happen to be aligned
02:46 PM jesseg: sync, 7-9 is not too bad. I'm not familiar with the STM32f7 myself, I just know that some of the 32 bit stuff is more around 30 instructions -- which surprised me coming from the 8 bit RISC world where it was a couple instructions
02:47 PM sync: most arms are that way
02:47 PM jesseg: are which way?
02:47 PM fragalot: jesseg: the 8bit RISC world only had about 2 or so interrupt levels, the STM32 series generally have a lot more than that
02:47 PM sync: taking 7 cycles to go into an interrupt
02:47 PM sync: at least with cortex M
02:47 PM jesseg: ahh gotcha
02:48 PM miss0r: We are already waay out of my comfort zone here :) This is why I started with something as simple as PLCs - something I know well :)
02:48 PM fragalot: I hate PLC's with a passion, personally :p
02:48 PM miss0r: I have an electrician background, so my skills with this stuff 'somewhat' limited to what I have been doodling around with over the years
02:49 PM jesseg: sync, what's a typical average Mhz/instruction cycle for the STM32f7?
02:49 PM miss0r: fragalot: I like the idea that someone took their time to 'harden' it for industrial use, so I don't have to.
02:50 PM sync: you get around 462dmips from an m7 at 216mhz
02:50 PM jesseg: miss0r, what sort of arc repetition rate are you getting?
02:50 PM miss0r: These MicroLogix 1100 are surprisingly sturdy. Also; I have a metric shit ton of them laying around
02:50 PM fragalot: miss0r: the issue I have with them is that they are too darn slow for what I need :P
02:50 PM sync: jesseg: you really need energy control in an edm application
02:50 PM fragalot: I need my applications to respond within microseconds, not seconds
02:50 PM sync: and you can get away with not touching the work
02:50 PM sync: by looking at the current and then backing off
02:51 PM miss0r: jesseg: with these IGBTs i'm using, I'm limited to around 4khz. but the future model will have better output switches. mosfets I'm thinking
02:51 PM jesseg: sync, hmmm.. They must 4X pll the clock and run the CPU actually at a much higher speed.. Instructions generally have to take several actual clock cycles
02:51 PM sync: no
02:51 PM sync: the core runs at 216MHz
02:52 PM jesseg: sync, well I don't see how two instructions could execute on the same clock cycle, unless they do it in parallel
02:52 PM jesseg: maybe it's a dual core :D
02:52 PM sync: no it is not
02:52 PM jesseg: most fascinating!
02:52 PM fragalot: jesseg: they do actually contain quite a lot of nifty gadgettery to work faster
02:52 PM fragalot: the f7 for example has a floating point coprocessor
02:53 PM fragalot: and my personal favorite, really good support for DMA
02:53 PM fragalot: the only downside is that the CPU's are evolving faster than the libraries & documentation >.>
02:53 PM jesseg: Ahhh...... so the mathco could bring up the dmips figure by emulating a much much faster pure software math library
02:53 PM sync: no, dmips is only integers
02:54 PM miss0r: jesseg: should you be interrested: Heres a video of me doing the first spark test with the rotating head: https://drive.google.com/file/d/1BrbwNgFe8D2m1Dmtb2OdeaanEL2sRaFV/view?usp=sharing It reaches the 'correct' electrode distance at 00:40 aprox.
02:54 PM sync: not float
02:54 PM sync: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0165b/I1028171.html
02:55 PM sync: ah shit that is A9
02:55 PM sync: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0165b/I1028171.html
02:56 PM jesseg: sync, all very fascinating. With an introductory understanding of FPGAs and discrete logic and CPU operation I am quite perplexed how multiple instructions could be executed on a single clock cycle, one at a time without *some* form of clock multiplication or subdivision
02:56 PM miss0r: Thats it for me for tonight. I have to get up real early tomorrow. See you 'round
02:56 PM jesseg: miss0r, really cool! (Video)
03:00 PM jesseg: sync, ahh thanks... Just as I thought, looks like the average is around 2.5 clock cycles per instruction maybe
03:00 PM sync: it is lower
03:01 PM sync: most of them take 1 cycle
03:01 PM sync: and it is a pipelined cpu
03:03 PM sync: which makes CPI>1 possible
03:03 PM jesseg: Well, most of them aren't 1. But are you saying then that due to pipelining they effectivelty are mostly 1?
03:04 PM jesseg: Doesn't pipelining mean that for a given cycle, you have one instruction executing, but you also have the last one writing and the nextone fetching or whatever?
03:04 PM sync: yes
03:04 PM sync: but they are also superscalar
03:05 PM sync: so they can run some instructions in parallel
03:05 PM sync: as opposed to a scalar cpu or a vector cpu
03:06 PM sync: and they do single cycle multiply/accumulate and simd
03:06 PM sync: they are quite modern cpus
03:06 PM jesseg: Ahhh, which on the dmips scale would be credited as having done it in plain software which would have taken a lot more cycles
03:06 PM sync: for something like an edm machine I'd probably go to something like an stm32f334 with its high resolution timer
03:07 PM fragalot: they also do the same cheat that intel did in their most recent "bug" reveal
03:07 PM sync: yeah they have a "predictive" pipeline
03:07 PM jesseg: does that execute two paths of code at the same time?
03:07 PM sync: two or more, yes
03:07 PM fragalot: jesseg: essentially, yes
03:07 PM fragalot: I believe the F7 does 6 or so
03:07 PM sync: yes
03:07 PM fragalot: but I could be wrong
03:08 PM jesseg: oh, so it almost does benefit from the concept of dual core :P
03:08 PM sync: cortex M7 is 6 stages of pipeline
03:08 PM jesseg: or 6 core :P
03:08 PM sync: not really
03:08 PM sync: it does not execute 6 threads
03:08 PM jesseg: well if two different bits of code is executing at same time, umm, it's not one core :P (semantics asside.)
03:08 PM jesseg: aside LOL
03:08 PM sync: it is the same core
03:09 PM jesseg: well, the core is made of logic gates.. A given set of logic gates can only execute one stream of code at a time, unless there's another set that's doing the other stream/path
03:09 PM sync: if you have an alu that can do 128x128 or 4x 64x64 ops still the same alu
03:10 PM sync: it still executes on linear stream of code, just that it already executed stuff that would otherwise be executed at a later date
03:10 PM sync: or would be waiting for some other stuff to execute
03:11 PM fragalot: probably would be executed*
03:11 PM sync: yeah
03:11 PM sync: hence "predictive"
03:11 PM jesseg: When it was already executing that code preemptively, did that take time away from it executing anything else?
03:11 PM sync: yes and no
03:11 PM jesseg: :D
03:11 PM sync: hopefully it did not
03:11 PM sync: but it can
03:11 PM sync: which is why compilers and people optimize their code that it will not cause the pipeline to flush
03:11 PM sync: because that really slows you down
03:11 PM jesseg: so it can, under some circumstances, execute two different paths at the same time without them stealing time from eachother?
03:12 PM sync: as it has to fill with instructions again
03:12 PM sync: it does all the time
03:12 PM sync: but it just does not execute two different bits of code
03:12 PM sync: you cannot run different code concurrently on it
03:12 PM jesseg: so if one core's registers are leaded with path A code, what registers get loaded with path B code?
03:12 PM fragalot: jesseg: basically, it assumes your "if" statement will be true, so it already executes that bit of code.. if your true ends up being false when you actually get there, it has to flush the cache & start running the else
03:13 PM sync: or it ran both and just takes the right one
03:13 PM fragalot: so you gain speed in case the "if" was true, but lose some if you hit the else
03:13 PM fragalot: sync: or that.. :P
03:13 PM sync: or it misses entirely and then rip
03:14 PM jesseg: Ahhh, so it doesn't really execute two paths, it gambles on which one is likely to be the correct one, to keep from having to flush the pipeline at the if?
03:14 PM fragalot: I just wanna say that I love 'm and it's a fresh breath coming from 8bit RISC chips :P
03:16 PM jesseg: that's all I really need to know I guess :D
03:16 PM * gloops loves steak pies
03:16 PM fragalot: jesseg: simplified, but yes. :)
03:17 PM fragalot: that's only one of the neat things about them though
03:17 PM fragalot: the level of control you have and the amount of things you can offload to hardware is amazing
03:17 PM fragalot: you want the ADC to sample at 6Mhz? sure. why not.
03:17 PM jesseg: cool
03:17 PM fragalot: want it to tell you when a certain value is reached? no problem.
03:18 PM jesseg: super cool!
03:18 PM fragalot: need to receive a MODBUS message but don't want to deal with the stupid 3,5 character timeout yourself?
03:18 PM fragalot: great news!
03:18 PM fragalot: IT DOES THAT FOR YOU
03:18 PM fragalot: :D
03:18 PM jesseg: In hardware? what, does it have TCP stack implemented in hardware too? :D
03:20 PM fragalot: No idea. I just use a compactCom to do that part for me
03:20 PM jesseg: Is there a nice free Linux IDE for STM32f7? (i.e. one that runs in Linux)
03:20 PM fragalot: eclipse is pretty well supported
03:21 PM jesseg: Ahh I used to use eclipse before Android abandoned it
03:21 PM fragalot: atollic truestudio is a nice package to start with
03:21 PM fragalot: based on eclipse
03:21 PM fragalot: and the prov version will become free soon as ST bought them
03:22 PM fragalot: (but the free version is very usable, pro just adds some additional diagnostic tools, but nothing you can't do manually)
03:23 PM jesseg: all very fascinating!
03:24 PM jesseg: I've been interested in learning more of the STM32 stuff but just assumed official SDK was windows only and therefore out of the question
03:24 PM jesseg: but not so
03:24 PM fragalot: Nope.
03:24 PM fragalot: ST has some really nice discovery boards too with a lot of variation of CPU's
03:24 PM jesseg: cool
03:25 PM jesseg: although I used to be a PCB designer so I don't mind throwing together my own discovery board :D
03:25 PM fragalot: it gets really fun when you start with dual ported RAM expansions on this cup
03:25 PM fragalot: cpu
03:26 PM jesseg: cool
03:26 PM fragalot: from a PCB design perspective I mean
03:27 PM fragalot: lots of nice high speed signals that need to arrive at the same time :P
03:27 PM jesseg: is openstm32.org trustworthy? i.e. I won't get a tonna spam or etc if I sign up?
03:28 PM roycroft: use a throw-away email address in case you do
03:28 PM fragalot: no ideq
03:29 PM gloops: use tor
03:29 PM jesseg: Well I have to sign up to get the dev tools, right?
03:29 PM fragalot: use signal
03:29 PM fragalot: xD
03:29 PM jesseg: What is tor? or signal?
03:29 PM jesseg: :P
03:29 PM fragalot: </meme>
03:29 PM fragalot: jesseg: atollic truestudio comes with an arm toolchain
03:29 PM fragalot: and they are trustworthy
03:30 PM fragalot: as for programmers, the ST-link is a nice cheap option. Not the fastest, but just for trying it out it's fine
03:31 PM fragalot: segger makes the more "professional use" programmer & debuggers.. MUCH faster, plus you can do a FULL trace log
03:31 PM fragalot: every instruction can be logged as long as your PC can keep up
03:31 PM jesseg: wow
03:31 PM jesseg: what's your favorite IDE?
03:31 PM fragalot: another really intersting thing is the "SEGGER RTT" library.. printf for debugging without much overheat at all
03:32 PM fragalot: so you can print strings practically in realtime
03:32 PM fragalot: jesseg: I use truestudio at work.. works fine
03:32 PM jesseg: that's pretty clever
03:33 PM jesseg: And their free version is full featured? or is it restricted?
03:33 PM fragalot: restrictions are a popup every few hours, and you can't use some of the more advanced diagnostic features like hardfault analyzer, or static stack analysis
03:33 PM fragalot: I use the free version mostly, as we only have 1 pro license for 3 people
03:34 PM fragalot: (but like I said, ST bought atollic, and will release pro for free soon)
03:34 PM jesseg: ahh cool
03:35 PM fragalot: segger also has a tool that you can use to convert the cheap ST-link to a segger-compatible "J-link" to use for example their RTT library or some other features
03:35 PM fragalot: ST has things pretty well figured out imho
03:36 PM jesseg: super cool
03:36 PM jesseg: Would you say its better than Arm has things figured out?
03:36 PM fragalot: more wide-spread support
03:36 PM jesseg: cool
03:40 PM jesseg: well I gotta take off here, but thank you all very much for the info!
03:40 PM fragalot: any time
03:48 PM Deejay: gn8
04:04 PM gregcnc_: lol used boring bar https://www.ebay.com/itm/401359151218
04:04 PM XXCoder: 50 bucks.
04:04 PM XXCoder: whats price of it, new? I guess 60 bucks lol
04:05 PM gregcnc_: it's a carbide bar with the head blasted right off
04:06 PM gregcnc_: I don't know what it cost to get that rebuilt, but basically carbide scrap at this point
04:06 PM XXCoder: yeah figured that out :P
04:06 PM XXCoder: some p[eople price stuff way too high
04:06 PM gregcnc_: people have no idea what they are selling
04:09 PM XXCoder: is it more than $50 in scrap? heh
04:10 PM gregcnc_: <10USD/Lb
04:14 PM SpeedEvil: Is that a boring bar to hold carbide, or a boring bar to hold carbide, made out of carbide?
04:15 PM gregcnc_: indexable carbide bars have brazed steel heads which has an insert pocket
04:18 PM roycroft: i should hope that a boring bar would not be made of carbide
04:18 PM gregcnc_: uhm?
04:19 PM roycroft: what would be the advantage of having the bar made of carbide?
04:19 PM MarcelineVQ: rigidity
04:19 PM roycroft: it would be a lot more expensive and a lot more fragile than a steel one, would it not?
04:19 PM gregcnc_: deeper boring, less chater, only the stuff machinists dream of
04:19 PM roycroft: hmm
04:20 PM gregcnc_: i can assure you carbide boring bars are very common
04:20 PM roycroft: i believe you
04:20 PM SpeedEvil: there are also things like segmented carbide bars with a steel core under tension
04:20 PM MarcelineVQ: there's a lot of kinds of carbide as well, there's no reason a carbide bar has to be more fragile than steel, but it also depends for what meaning of fragile a person is using
04:21 PM SpeedEvil: Usually for steel boring bars, they are not tempered glass hard.
04:21 PM SpeedEvil: And may bend a little. Carbide won't.
04:21 PM roycroft: yes, the increased rigidity makes sense
04:21 PM SpeedEvil: (bend permenantly under overload)
04:22 PM DaViruz: i'd like a material that doesn't bend period
04:24 PM archivist: impossible dream
04:25 PM roycroft: a quixotic quest
04:25 PM DaViruz: it does tend to go against everything known about nature
04:25 PM MarcelineVQ: neutronium-polymer mill
04:25 PM roycroft: even parallel lines converge at the edge of the universe
04:28 PM DaViruz: how about at the end of time?
04:29 PM sync: jesseg: you can just use the arm provided gcc
04:29 PM sync: and the st hal stuff
04:36 PM Tom_L: end of who's time?
04:37 PM Cromaglious_: Dr Who's of course
04:37 PM Tom_L: what time is it on mars?
04:37 PM jdh: 17:37
04:38 PM SpeedEvil: DaViruz: Well - it's quite doable for 'small' loads with fast servos.
04:43 PM Cromaglious_: Wow we are getting real liquid sunshine!
04:48 PM SpeedEvil: Todays random number. Best efficiency from electricity -> calories is about 6%. (potatos with good LEDs)
05:20 PM DaViruz: preferably LEDs driven by potato batteries
05:25 PM Cromaglious_: waiting, waiting, waiting, waiting, waiting, waiting, waiting, waiting, waiting, waiting, waiting, waiting, for the next 20 hours or so for money to show up...
06:03 PM phiplii: http://www.zachtronics.com/shenzhen-io/
06:05 PM phiplii: night folks
07:00 PM skunkworks_: http://www.machsupport.com/forum/index.php/topic,36287.0.html
07:15 PM Cromaglious_: there took out the shift fork and all the gears out of X2 head
07:19 PM Cromaglious_: hmmm think I have an incling of a path to saabalize the head on the round column 33686
07:21 PM Cromaglious_: take the gear rack, surface grind the edges to they are parallel. then screw it to the column, and add in gibs to locate on the sides
08:28 PM Cromaglious_: Bid on a 4" rotary grizlly vice withe a 4 jaw scroll chuck
08:28 PM Cromaglious_: vise
08:30 PM Cromaglious_: oops 4 jaw independent jaws wth etra jaws
10:45 PM Cromaglious: 5" raspberry touch screens showed up