#linuxcnc | Logs for 2013-09-10

[02:19:20] <DJ9DJ> moin
[03:32:22] <Loetmichel> hmm, coming to work, coffe, first work: make tools to do actual work... -> hmm, darfste erst mal dein wwerkzeug selber bauen damit de arbeiten kannst... *frääs* -> http://www.cyrom.org/palbum/main.php?g2_itemId=14454
[03:32:38] <Loetmichel> oh sprry
[03:33:49] <Loetmichel> ignore the german part please
[03:45:11] <archivist> but..you can buy radius guages
[03:47:23] <Loetmichel> archivist: not at stock here.
[03:47:30] <Loetmichel> and i needed one NOW..
[03:47:56] <archivist> but there is one in my draw
[04:35:57] <MattyMatt> MDF's last excuse for existing, it's flatness, has been exposed. 6mm toughened glass on 25mm MDF is unflat by 0.1mm in the middle. I've left it convex side up with weights on top, when it gets flat I'll flip it and pour concrete under it
[04:37:27] <MattyMatt> how much would a cast iron surface table warp if incorrectly supported?
[04:49:49] <Valen> fibreglass shinks when it cures
[04:50:00] <Valen> it'll pull your mdf out of whack
[04:50:30] <Valen> if its being used for a mill table you put the MDF in, put another sacrifical sheet on top then mill that sucker flat
[05:20:24] <MattyMatt> it's for a glass topped surface plate
[05:21:06] <MattyMatt> andypugh says your G-E spindle is really nice, got pics?
[05:22:32] <MattyMatt> G-E is still fantasy for me though, if I can't even make a flat surface to assemble the mould on :p
[05:42:05] <DraXem> anybody know JT-Shops webpage URL?
[05:43:14] <jthornton> gnipsel.com
[05:43:20] <DraXem> or maybe I'm confusing him with someone else...there is a guy in here who has a CNC webpage with some cooking recipes and beer and stuff
[05:43:36] <DraXem> jthornton: thanks =)
[05:43:49] <jthornton> that's him
[05:44:23] <DraXem> jthornton: love the gcode tutorial keep forgetting stuff and keep going back to it lol I should make a pdf version of it
[05:44:54] <jthornton> thanks
[05:51:05] <ReadError> link?
[05:52:28] <archivist> MattyMatt, just go to large DIY place and get a granite tile, look at reflection of roof for flatness
[05:53:55] <archivist> MattyMatt, also supplied as place mats for kitchen or posh dining
[06:26:26] <Tecan> http://i.imgur.com/UvYYjVi.png my wheel shadows are funky
[06:44:30] <MattyMatt> archivist, I've already located floor tiles. they even offer to send a sample tile
[06:45:44] <archivist> I have one in this picture http://www.collection.archivist.info/archive/DJCPD/PD/2012/2012_11_07_Drive_gear/IMG_1400.JPG
[06:46:59] <archivist> I did contemplate being evil and buying tiles and selling them in the model engineer section as poor mans surface tables
[06:47:44] <archivist> I got one from sainsbury and a few from morrisons
[06:47:56] <MattyMatt> http://www.wallsandfloors.co.uk/range/natural-granite-tiles/quartz-stone-mirror-tiles/ 12mm thick is the best they have atm. it was 15mm last year
[06:49:18] <MattyMatt> I'll still need it to stay flat while I fit ribs. 12mm will sag
[06:49:55] <archivist> for small work it wont sag much
[06:50:10] <MattyMatt> If a saw one into 4" strips then lap them straight against each other, then glue them on as ribs, that might do nicely
[06:50:55] <archivist> but if I want a real one...I use a real one
[06:51:58] <archivist> or the table on the nearest machine tool (milling machine)
[06:52:15] <MattyMatt> I'm waiting for a local one on ebay. I'm prepared to pay the £60 they're going for atm for a good 18x12 iron one
[06:53:13] <archivist> I got one from army surplus
[06:53:40] <MattyMatt> good point. my mill table is nicely restored. 20"x8"
[06:56:40] <MattyMatt> if I put the glass on the mill, it'll be flat enough to assemble a mould for granite-epoxy
[07:00:04] <MattyMatt> I guess those tiles will only be polished on one side, with no guarantee of parallel back, so glue-on ribs won't necessarily make it flat
[07:02:47] <MattyMatt> with any luck, silicon carbide ones will appear, and the price of iron ones will plummet
[07:03:06] <MattyMatt> iron is <£100/t on the LME atm
[07:19:57] <archivist> cost of an iron one is the casting and making flat
[07:21:13] <MattyMatt> charcoal is cheap atm :) summer's gone
[07:23:03] <MattyMatt> momma's pretty garden isn't mine to do casting in tho
[07:27:18] <MattyMatt> I'll do it in the road :)
[07:27:41] <MattyMatt> no one will be watching us
[07:28:26] <kengu> nobody but The Police
[07:29:01] <kengu> they say "i'll be watching you"
[07:32:37] <MattyMatt> they wouldn't notice if I dressed like a road worker
[07:33:45] <MattyMatt> http://www.ebay.co.uk/itm/stihl-petrol-leaf-blower-bg-55-easy-repair-/161097825124 charcoal fired and this baby, and I'd be set
[07:34:09] <MattyMatt> that's too far for me to pick up
[12:17:26] <FinboySlick> Wow, had you guys seen this? http://www.wimp.com/designtechnology/
[12:24:16] <WalterN> thats kinda slick
[12:25:02] <FinboySlick> Indeed.
[12:25:28] <WalterN> probably $8000 software or something
[12:25:38] <ssi> heh if I wrote it, I'd want $8000 for it
[12:25:38] <FinboySlick> I could see this being applied to quick reverse engineering of photographed parts.
[12:26:17] <FinboySlick> You just add dimension constrains and you get a pretty good approximate part for quick replacements.
[12:26:26] <CaptHindsight> nice to see some failed examples to show what it really does and its limitations
[12:27:10] <FinboySlick> The background replacement technique is something they added in photoshop last year.
[12:27:27] <FinboySlick> The narator called it patch match or something.
[12:27:49] <FinboySlick> Not useful for CAD, but nifty regardless.
[12:29:08] <FinboySlick> Actually, it might be a bit earlier than last year, it was for version 5 I think.
[12:31:12] <CaptHindsight> it's good enough to make a facsimile of something where dimensions and small details aren't important
[12:31:59] <CaptHindsight> perfect for glue gun printers
[12:32:31] <FinboySlick> CaptHindsight: It obeys constraints though. You could just come back after the initial shape and specify dimensions between elements.
[12:33:29] <CaptHindsight> it assigns constraints that can then be modified
[12:34:12] <FinboySlick> My point is that a dimension is a constraint that could easily be added on later in the process.
[12:34:13] <CaptHindsight> not sure if that application can but that could be the next step
[12:34:33] <CaptHindsight> yeah, you made sense :)
[12:36:36] <FinboySlick> I imagine refinement could also be added from multiple views, telling the software that the shape you defined in one picture is the same as in this other view.
[12:37:01] <FinboySlick> A bit like the obelisk example, but for the same shape rather than different parts of it.
[12:38:08] <CaptHindsight> http://en.wikipedia.org/wiki/Mathematical_morphology
[12:39:41] <CaptHindsight> it was fun trying to do this with film back in the 80's on PC and array processors
[12:42:13] <FinboySlick> Ah the good old days where several users would share a video card (SGI).
[12:46:35] <IchGuckLive> hi all
[12:47:08] <IchGuckLive> BBQ with homemande pizza from home made oven YAM
[12:48:21] <FinboySlick> IchGuckLive: Brick oven?
[12:48:39] <IchGuckLive> no plasma cut
[12:48:39] <FinboySlick> Or stone?
[12:48:47] <IchGuckLive> steel
[12:51:37] <CaptHindsight> "Cooking with CNC" Todays show will feature "Plasma cut-n-baked Pizza" and "Milling you veggies to perfection"
[12:52:35] <IchGuckLive> :D
[12:53:34] <CaptHindsight> time to go bake myself in the shop, 36C today
[12:53:56] <IchGuckLive> http://stores.ebay.de/Pizzaofenwelt/Pizzaofen-Holzbackofen-/_i.html?_fsub=2&_sid=74103980&_trksid=p4634.c0.m322
[12:55:40] <andypugh> Tomorrows show includes "Morse or Jarno taper for carrots?"
[13:07:25] <CaptHindsight> http://www.indiegogo.com/projects/rubicon-3d-scanner/
[13:07:25] <FinboySlick> CaptHindsight: I'll gladly import the extra 16 degrees, it's about 8 here.
[13:08:02] <IchGuckLive> most come with 3D data
[13:09:06] <CaptHindsight> http://www.wimp.com/designtechnology/ generates 3d from 2d images
[13:09:34] <IchGuckLive> 123d.com is better
[13:10:18] <IchGuckLive> from autodesk
[13:11:17] <FinboySlick> But autodesk is the spawn of Satan.
[13:11:34] <FinboySlick> They're like the Sony of CAD.
[13:11:53] <archivist> the etch a sketch of cad
[13:12:17] <IchGuckLive> http://www.123dapp.com/catch
[13:12:23] <CaptHindsight> IchGuckLive: the video shows them generating 3d constrains from a single photo
[13:12:47] <CaptHindsight> for 123 you might as well just use video
[13:13:46] <IchGuckLive> CaptHindsight: that only can be a height map aplication
[13:13:53] <IchGuckLive> you need at least 3
[13:14:02] <IchGuckLive> for a full 3D object
[13:14:20] <CaptHindsight> watch the vid
[13:14:24] <FinboySlick> IchGuckLive: Unlike 123d, they do not appear to just point-cloud either, they generate geometry from primitives.
[13:15:38] <IchGuckLive> CaptHindsight: that can CREO Proe PTC also
[13:15:52] <IchGuckLive> as Solidworks as well
[13:16:20] <CaptHindsight> it depends on how precisely you want to skin the cat
[13:18:21] <IchGuckLive> it is best tro start
[13:21:31] <IchGuckLive> CaptHindsight: do you kow the name of the design tool
[13:21:37] <CaptHindsight> I had trouble finding a 1/2" dia. diamond core bits at least 4" long
[13:22:26] <CaptHindsight> I ordered one on Amamzon with limited specs and it showed up today at 15.5"!
[13:23:58] <CaptHindsight> IchGuckLive: it's just a presentation at SIGGRAPGH, they refer to it as 3-sweep
[13:24:18] <IchGuckLive> ok
[13:26:04] <CaptHindsight> watching that video again you really see how they pretty much reverse how you would generate the models in cad
[13:26:38] <CaptHindsight> extrude, revolve a profile around an axis etc
[13:26:54] <IchGuckLive> the edgefinde algorythem must be good
[13:26:56] <IchGuckLive> or the faked it
[13:27:09] <IchGuckLive> by coloring 2 pixel around in a big picture
[13:27:28] <IchGuckLive> so as inkscape does on vectoring
[13:27:49] <CaptHindsight> they show how limited pixel depth does effect it, the cloudy bottle pic
[13:28:41] <CaptHindsight> it just needs enough contrast or you'd have to "interpolate" the fuzzy edges by hand
[13:30:26] <CaptHindsight> it's not exact but good enough for lots of applications
[13:32:13] <CaptHindsight> 30 years ago we were playing with this on old movies for restoration, preservation or colorization
[13:33:03] <IchGuckLive> this will fake alot of things
[13:33:47] <CaptHindsight> we even had computers as powerful as 1 Mips :)
[13:34:02] <IchGuckLive> O.O
[13:34:11] <IchGuckLive> ok im off by
[13:34:30] <FinboySlick> CaptHindsight: You scared the German.
[13:37:13] <CaptHindsight> FinboySlick: any idea what is used now for things like what Video Toaster used to do?
[13:37:47] <FinboySlick> CaptHindsight: Wow, you brought back great memories of the Amiga there.
[13:38:33] <FinboySlick> CaptHindsight: They use things like Adobe Premiere and what not, not much funky video hardware involved anymore, though they started offloading to GPUs.
[13:38:43] <archivist> was that just genlock and overlay
[13:41:06] <CaptHindsight> FinboySlick: I've only used premiere for non-linear editing. Can it track and copy/paste moving objects?
[13:42:21] <FinboySlick> CaptHindsight: I think Matrox does funky video live video processing stuff, but that's not a world I'm very familiar with.
[13:43:32] <FinboySlick> CaptHindsight: With storage nowadays, most of what I've seen was non-linear editing.
[13:45:05] <CaptHindsight> Charles on the ML designed the Video Toaster Flyer hardware
[13:45:28] <CaptHindsight> http://en.wikipedia.org/wiki/Video_Toaster#Video_Toaster_Flyer
[13:45:35] <CaptHindsight> small world sometimes
[13:45:47] <archivist> we did frame grabbers
[13:46:14] <archivist> and the toy in the garden is an outside broadcast truck
[13:46:30] <CaptHindsight> archivist: we used Imaging Technology's
[13:47:07] <CaptHindsight> I remember the big announcement for the first RS-170 real time grabber
[13:47:14] <archivist> we were grabbing digital monitor video for printing
[13:47:24] <CaptHindsight> but processing was another story :)
[13:48:22] <archivist> we then started developing an averaging grabber, but that never got out
[13:49:42] <archivist> board under the pile http://www.collection.archivist.info/archive/DJCPD/PD/2009/2009_10_15_PCB_design/IMG_0353.JPG
[13:49:48] <CaptHindsight> averaging and simple maths were real time, can't recall the DSP's we used
[13:50:18] <archivist> were were going to use TTL
[13:51:08] <archivist> I do have a broadcast averager, does a nice noise reduction
[13:51:31] <andypugh> After a bit of a wiki wander, I found the Quantel Paintbox. I recall getting a demo of that system in the microscopy department of Surrey Uni during a course early in my PhD. :-)
[13:51:31] <archivist> but all the old standards are history now
[13:51:34] <CaptHindsight> the layout looks really nice, was it by hand?
[13:52:30] <archivist> a lot by hand the autorouter got stuck after about... cant remember what %
[13:53:03] <archivist> andypugh, I have a little bit of Quantel in loose pcbs
[13:54:51] <archivist> i did manage to get an NEC Eflex but is is sick and needs fixing
[13:56:21] <CaptHindsight> just for fun I always run old PCB files though the latest autorouter package and hardware
[13:57:14] <CaptHindsight> what I recall taking overnight or days now runs in seconds
[13:57:34] <FinboySlick> Anyone familiar with the way modern pc hardware handles interrupts? I have a setup here where USB and RAID are sharing an interrupt and I'd like to see if it can be avoided.
[13:57:39] <CaptHindsight> stuck autorouters, don't remind me
[13:58:37] <CaptHindsight> FinboySlick: are they integrated into the MB?
[13:59:40] <FinboySlick> Yes and no. I can't move slots if that's what you were going to suggest.
[14:00:19] <Loetmichel> re @ home
[14:00:24] <CaptHindsight> FinboySlick: BIOS option if the chipset supports it
[14:01:27] <FinboySlick> CaptHindsight: Alright, I'll take a peek. I was mostly just wondering if the kernel would override that once it takes over.
[14:02:44] <CaptHindsight> FinboySlick: it's a combination of how they physically wired the lines and the interrupt controller registers
[14:03:37] <FinboySlick> The controller is in the 'main' 16x slot of the mobo, I'd imagine there's a way to get that slot dedicated to some hardware since this is a server board.
[14:03:37] <pcw_home> if its new enough all that is gone (MSI interrupts)
[14:03:56] <FinboySlick> It is bran spanking new.
[14:04:11] <FinboySlick> So new that I had to take a dremel to the 1U case and make my own backplate :P
[14:04:37] <pcw_home> should use MSI and therefore no IRQ lines anymore
[14:05:18] <CaptHindsight> FinboySlick: pre-atx 1U case? :)
[14:05:25] <FinboySlick> pcw_home: IO-APIC-fasteoi says /proc/interrupts. There are MSI interrupts for the NICs though.
[14:05:44] <FinboySlick> CaptHindsight: It is a bog standard ATX case... The motherboard isn't standard :P
[14:06:03] <FinboySlick> http://www.supermicro.com/products/motherboard/Xeon/C220/X10SLM_-F.cfm
[14:06:24] <FinboySlick> They released this new line but didn't bother to come up with updated chassis.
[14:06:36] <FinboySlick> Or if they did, they didn't document the updated part numbers.
[14:07:23] <FinboySlick> Even in their list of 'optimized for' on the website, you look at the chassis and the backplate is completely different from the board. And in 1U, those aren't swappable.
[14:07:42] <CaptHindsight> figures
[14:09:00] <FinboySlick> pcw_home: http://pastebin.ca/2447700
[14:09:12] <FinboySlick> The bit that bothers me is line 08
[14:22:54] <FinboySlick> Wow, I'm going to have to do a lot of reading before I get the funky new terminology in that bios :P
[14:35:47] <JT-Shop> Holy Crap Batman, I'm selling stuff at my Spyder Store!!!
[14:37:47] <andypugh> You seem to be miss-matching your comic-book hereos.
[14:38:35] <FinboySlick> Bah, they're having Daredevil play Batman anyway.
[14:42:50] <JT-Shop> Yikes! I need more stock now
[14:44:12] <CaptHindsight> JT-Shop: did you start a trend? Now I'm seeing Spyders everywhere
[14:50:04] <JT-Shop> I've been designing some things for Spyders and have finally managed to get some ready for sale and they are flying off the shelf
[14:50:24] <JT-Shop> http://gnipsel.com/spydercart/
[15:06:34] <andypugh> For your next trick, a dual inflator that you can adjust on the move :-)
[15:11:43] <jdh> real time constant pressure, compensating for uneven tire heating.
[15:13:23] <CaptHindsight> andypugh: what do they use for the seals on the rotary connection at the hub?
[15:14:21] <andypugh> I don't trouble myself with trivial implementation details.
[15:14:47] <CaptHindsight> thats right you're a scientist, those small details are for the engineers :)
[15:15:44] <jdh> http://airgotireinflation.com/products/trailer-inflation/
[15:16:42] <CaptHindsight> Rotary Union Assembly
[15:23:38] <CaptHindsight> http://www.deublin.com/industry/automotive/idev/media_air/ probably would also require a larger hub and different rims
[16:02:40] <fedo> hello
[16:02:57] <kengu> hello
[16:03:07] <kengu> umm.. sort of good night pretty much
[16:03:09] <fedo> i've got a problem with the installation
[16:03:24] <fedo> of my maschine
[16:04:42] <kengu> i didn't have problems so I am not good in that matter
[16:05:32] <fedo> you are lucky with emc
[16:05:33] <fedo> ?
[16:05:59] <fedo> no problems during the installation ?
[16:08:03] <kengu> none. just install and settings and the laser started to move.. as soon as i got right cable and not a twisted parallel cable
[16:23:04] <DJ9DJ> gn8
[16:27:49] <andypugh> I suspect fedo didn't get the help he was hoping for :-(
[16:40:26] <JT-Shop> looked like a bit of language barrier and lack of paitence
[16:41:35] <kengu> true. i was onto something else and then when i was about to ask about the problem ..umm.. lack of patience.
[17:35:42] <andypugh> pcw_home: How many ssi encoders do you support?
[17:37:22] <PCW_> only one as thats all I could afford
[17:37:39] <PCW_> 32 max
[17:38:47] <JT-Shop> running the BP in shorts and flip flops is never a good idea even if you saw Muller Nick casting aluminum in shorts
[17:38:49] <PCW_> might be 64 if I dont use the program start with a data bit per encoder
[17:40:33] <JT-Shop> I like this "This may be less obvious if you don't count in binary."
[17:40:48] <PCW_> Lets see, orange hot molten aluminum and bare-feet / legs Sounds safe
[17:41:17] <JT-Shop> as seen on the internet
[17:41:45] <JT-Shop> he did have a proper apron and face shield and gloves on
[17:42:32] <JT-Shop> PCW_: the MPG hook up use 5v from anywhere to power the MPG?
[17:42:54] <Tom_itx> yeah
[17:43:08] <PCW_> Yes but it needs to have a common GND with field power
[17:43:32] <PCW_> (since you are using field inputs)
[17:44:01] <andypugh> JT-Shop: There are some strange chaps on that Internet
[17:45:16] <JT-Shop> is the 5v input and the 24v input on the same GND?
[17:45:32] <JT-Shop> andypugh: yes there is
[17:49:08] <PCW_> No, not unless you connect them (the 24V IO is completely isolated from the PC 5V)
[17:49:37] <JT-Shop> I have 5v power supply to the small terminal block, I'm not using the PC 5v
[17:50:41] <PCW_> Yeay so the MPG GND must connect to the field I/O GND somewhere
[18:09:23] * JT-Shop wonders if the street sweeper from Rocky and Bullwinkle is looking for part time work
[18:31:21] <EngEC> Hi there fellow mates. Does anyone have an hal comp of an >=12bit ADC?
[18:32:26] <andypugh> What interface?
[18:35:59] <EngEC> the idea is to control a spindle motor that have an analog input
[18:36:40] <andypugh> You want a DAC then?
[18:37:03] <andypugh> PWM is easy and already exists..
[18:37:28] <EngEC> I was wondering if anyone already have it done. The basic idea was to convert dec->bin and mounting an hardware DAC to supply the motor driver with the defined voltage values
[18:37:49] <EngEC> that's a problem with 25us base thread
[18:37:57] <EngEC> yep, but my driver uses frequency of 5kHz
[18:38:02] <andypugh> The issue is what protocol you use for communication out of the PC>
[18:38:40] <andypugh> I am not sure what the driver frequency has to do with anything.
[18:38:59] <andypugh> I am meaning PWM to voltage to drive control.
[18:39:40] <andypugh> What hardware interface are you currenlty using?
[18:40:20] <EngEC> VSD-E from ganite devices (http://granitedevices.com/index.php?q=servo-drive-vsd-e)
[18:41:41] <andypugh> No, I mean how does LinuxCNC communicate with your hardware?
[18:42:00] <EngEC> paraport
[18:42:21] <andypugh> Hmm, then I am thinking that PWM is the way to go, as that only uses up one pin...
[18:43:27] <EngEC> If the frequency of the PWM is 5kHz, I'll have 200us of a period to use, and with 25us base thread (for make-pulses()), I just have 200/50 = 8 velocity modes (just in one direction)
[18:44:34] <andypugh> My spindle is perfectly fine with 500Hz PWM base frequency
[18:45:13] <andypugh> Just how fast do you want the spindle speed to vary?
[18:46:05] <EngEC> exactly, the frequency is low, giving you more pulses per period. (i.e. 1/500 /25u = 80)
[18:46:21] <andypugh> Yes.
[18:46:41] <EngEC> -4000 to 4000 RPM with 1 RPM variation
[18:47:03] <andypugh> And that works fine. In fact 50Hz is probably OK too. After all, a VFD can't _possibly_ react faster than that.
[18:48:04] <EngEC> the manual says that frequency of PWM is 5kHz
[18:48:08] <andypugh> And nobody complains about VFDs not being responsive enouhg.
[18:48:16] <andypugh> Which manual?
[18:48:28] <EngEC> ow, I read it wrong sorry
[18:48:54] <EngEC> VFD?
[18:49:25] <andypugh> VFD = Variable Frequency Drive. Like you would use to drive an AC induction motor spindle.
[18:50:16] <andypugh> The fact that a normal motor runs fine on a 50Hz base frequency suggests to me that there is ample bandwidth at 50Hz
[18:51:00] <andypugh> And I really don't think you need 1rpm resolution on spindle speed. My milling machine was built with 8 gears.
[18:51:15] <EngEC> mine is an DC motor, and the manual of the driver states that frequency of PWM input should be between 5~30kHz
[18:51:26] <andypugh> At 50 Hz you have 800 gears :-)
[18:51:56] <andypugh> I thought you wanted to control it on analogue voltage?
[18:52:50] <andypugh> A DAC won't give you 5kHz either.
[18:55:12] <EngEC> but with the analog input I wont need a frequency control, I'll have a level control-type
[18:56:13] <EngEC> and then the emc2 wont need to "lose time" with the frequency pulse
[18:56:39] <EngEC> instead of using fast-thread it will only require a slow one
[18:59:14] <andypugh> I don't see how you expect to drive a DAC from the parallel port in the slow thread.
[18:59:51] <EngEC> because I wont need a fast refresh rate
[19:00:11] <andypugh> What I am suggesting is using a parallel port pin and PWM to create an analogue voltage level to control the drive.
[19:00:16] <EngEC> with the frequency control (either pwm or pulse) will make me lose resolution for the spindle motor, because It has the base-thread limitation, with the level control type (voltage level) I will not have that kind of problem
[19:00:24] <andypugh> What controls your axis motors?
[19:01:58] <EngEC> its a lathe, so the axis are controlled by 2 stepgen (they are stepper motors)
[19:02:18] <EngEC> there I dont have a problem
[19:04:13] <andypugh> If you want a 12-bit DAQ controlled by the slow-thread that is an update rate of 125Hz. Not really any better than the PWM (which can manage the equivalent of 8-bits, but will dither to give better effective resolution).
[19:04:40] <andypugh> The other alternative is PDM, then the effective resolution is set by the filter components, not the base thread.
[19:05:51] <andypugh> 25kHz PDM into an RC circuit with a 125Hz time constant will probably end up behaving quite a lot like your DAC.
[19:06:29] <EngEC> sounds like a great solution
[19:07:11] <EngEC> but wouldn't I lose some resolution?
[19:07:46] <andypugh> Well, it's quick and easy too, though the basic setup (just a resistor and cap) will only give you a unipolar 5V. Do your drives handle speed/dir or do they need +/- 10V?
[19:08:45] <andypugh> PDM keeps a running total of what proportion of the time it has been "On"
[19:09:04] <andypugh> So it's rather hard to say what it's resolution is.
[19:09:29] <EngEC> they have +/- 10v (which can be offset-ted and scaling)
[19:09:53] <EngEC> i.e. can be between 0~5v with no problem
[19:10:21] <andypugh> If your spindle inertia gives it a TC of 10Hz (and considering that AC motors don't noticably vibrate at 50Hz that seems reasonable) then you should see 1:2500 resolution.
[19:10:58] <andypugh> How are you for IO pins, home switches, step rate, that sort of thing?
[19:11:25] <andypugh> Because the Mesa 5i25/7i76 just solves all these problems at a stroke.
[19:11:37] <andypugh> But does cost $200
[19:11:39] <EngEC> yep
[19:12:16] <EngEC> I've already searched, I have a budget problem mate
[19:12:28] <EngEC> I cannot use it
[19:12:51] <andypugh> In that case, try PDM into a resistor/cap and see how that works out.
[19:13:45] <EngEC> I will, thanks alot ;)
[19:15:07] <andypugh> I suspect that any cap that is sensoble will enforce a resistor that meets this limit, but keep the P-port current down under 3mA
[19:15:35] <andypugh> You set PDM mode by requesting 0 PWM frequency, if you are wondering
[19:16:17] <EngEC> I am using an interface with a transistor, so the power isn't an issue for the port
[19:16:28] <EngEC> I've read and will try
[19:16:37] <EngEC> thx ;)
[19:16:58] <andypugh> Good luck. I guess I will be here the same time tomorrow
[19:31:33] <andypugh> night