Back
[00:33:26] <Loetmichel> mornin'
[01:55:07] <DJ9DJ> moin
[06:02:31] <Heinz_60> hi all
[06:39:59] <skunkworks> JT-Shop,
https://www.youtube.com/watch?v=i057pzjrML4
[06:45:07] <jthornton> looking good!
[06:45:55] <skunkworks> very happy with the results. linuxcnc is awesome :)
[06:46:27] <skunkworks> plug and play with the enco compact5 pc.
[06:47:48] <syyl_> and something easy like threading for the beginning? :D
[06:47:50] <syyl_> cool
[06:47:54] <jthornton> I've been having fun rigid tapping on the BP knee mill since I added the encoder to the spindle
[06:48:01] <syyl_> i like that little lathe
[06:49:31] * jthornton snickers at mail list over forum discussion
[06:49:42] <jthornton> what a time sink that one is
[06:53:48] <jthornton> no one will convince the other that their way is better LOL
[07:05:24] <Tom_L> http://www.electronicproducts.com/Electromechanical_Components/Motors_and_Controllers/Smart_servo_drive_has_EtherCAT_interface.aspx
[07:23:57] <jdh> wonder if there are any more of those Enco lathes around.
[07:43:26] <skunkworks> jdh, they are often on ebay
[08:01:13] <jdh> I've seen them before, but kinda pricey.
[08:01:32] <skunkworks> they are :)
[09:52:52] <geo01005> anybody know if I can directly drive a solenoid air valve with the 24 VDC outputs on the 7i76 card? Or do I need to add a diode for emf suppression.
[09:53:50] <Jymmm> 25ยข sounds like cheap protection no matter if you can or not.
[09:55:08] <cradek> did you see pages 13 and 14 of the 7i76 manual?
[09:55:48] <Jymmm> Manual, he's the guy that makes my tamales!
[09:56:47] <roh> geo01005: just add the diode. you never know, and it doesnt hurt and costs like 3cent
[09:57:02] <geo01005> I did read the manual, but somehow glossed over that.
[09:57:09] <geo01005> Thanks cradek.
[09:57:22] <Jymmm> Well, what does it say?
[09:57:39] <geo01005> VOLTAGE CLAMPS The output driver chips used on the 7I76 have built in Zener diode clamps to clamp inductive turn-off (fly-back) spikes. This means that flyback diodes are not normally required on inductive loads. If high current inductive loads are switched at high frequencies, they should have flyback diodes to limit power dissipation the in 7I76's driver chips
[09:58:30] <roh> but i guess you need some driver, depending on what the 7i76 can sink/source on current
[09:58:30] <Jymmm> Nice. pcw_home...Yo, define "high current" ?
[09:58:32] <cradek> I only looked because I would have been shocked if it wasn't carefully spelled out in there... they're pretty careful about the manuals.
[10:00:12] <roh> ah. 350mA .. so depending onyour relay/valve it should be ok
[10:00:45] <pcw_home> They are OK without diodes, but may do a thermal shutdown if you cycle very fast (> 100 Hz or so for a big contactor)
[10:00:48] <roh> i'm looking at page19 of the manual.
[10:01:06] <roh> just add the frickin diode on the valve directly (or in the plug)
[10:01:28] <pcw_home> Disadvantage of diode is turn off is slower
[10:01:52] <roh> there is really no sense in being cheap there. inductive back-emf is evil and kills hardware in much more evil ways than just the obvious 'defective port' way...
[10:01:53] <Jymmm> pcw_home: So no modofied 50Hz sinewave ghetto-inverter huh? lol
[10:02:16] <pcw_home> normally though you dont care about this and without a diode you generate more switching noise
[10:03:38] <geo01005> I'll install some diodes in the connector. The coils on these solenoids are 175ohm, so about 138ma load.
[10:03:56] <pcw_home> At 24V the output will swing from +24 to about -18 (40V zener) without a flyback diode
[10:04:14] <geo01005> It might be ok without it, but I'll add them.
[10:04:56] <pcw_home> Just get the polarity correct :-)
[10:05:23] <geo01005> Yeah, no kidding.
[10:07:30] <geo01005> By the way thanks for making these 5i25-7i67 kits.
[10:07:36] * roh just got 2 valves with solenoids for the mill... just need to mount em
[10:08:04] <geo01005> This is much better than the 7i43 plus breakout board mess I had on the last machine.
[10:08:51] <pcw_home> well the FPGA boards are generic, the 7I76/7I77 are more targeted for machine control
[10:09:14] <geo01005> Well I like it.
[10:22:04] <geo01005> pcw_home I see that there is now a 7i88 on the price list for a thermocouple board. Is that in development?
[10:23:30] <pcw_home> yes
[10:25:28] <geo01005> Cool. I would have used that I this machine I'm building right now. Instead I ended up using three standalone PID controllers.
[10:25:32] <pcw_home> 8 channels J,K,T, or E
[10:25:34] <pcw_home> firmware has been rather a pain, but its getting there
[10:26:18] <geo01005> (PID Controllers are for controlling the temperature of some heat sealing heads)
[10:27:46] <geo01005> With the price be similar to the 7i87?
[10:28:01] <pcw_home> Ive heard the PID is probably not the best for temperature control (asymmetry in heat/cool rates, and lags)
[10:29:01] <pcw_home> Yes similar (a little more since it has 2 good temperature sensors on the card for cold junction sensing)
[10:29:02] <geo01005> yeah, the thermal time constant is long. I'm just using the controller in a bang bang relay control.
[10:29:47] <geo01005> I would have used linuxcnc but acquiring the temperature is difficult.
[10:29:57] <geo01005> But the 7i87 will change that.
[10:29:57] <pcw_home> and a "isothermal" bar bolted to the card
[10:30:22] <pcw_home> I really like thermocouples since they are so rugged
[10:31:15] * Jymmm introduces pcw_home to Mr Lava
[10:31:43] <pcw_home> I saw a great demo of pounding on the thermocouple with a hammer and watching the temperture steps
[10:31:54] <Jymmm> heh
[10:32:10] <Jymmm> New meaning to "thermal Shock"?
[10:32:30] <pcw_home> try that with your thermistor...
[10:32:48] <Jymmm> I do all the time... once.
[10:33:44] <Jymmm> pcw_home: does you new thermo board have the ability to connect a 2x16 LCD ?
[10:35:09] <pcw_home> no, its a sserial remote so connects only to our FPGA stuff
[10:35:43] <Jymmm> pcw_home: It be slick if your FPGA boards had a USB/uSD slot and maybe a uC to eliminate the need for a PC for anything other than initial config.
[10:37:28] <Jymmm> pcw_home: How much can you load on a FPGA (in MBytes) typically?
[10:38:01] <pcw_home> A lot have SPI flash for config and embedded uprocs are used for sserial, host interfaces, resolver etc
[10:39:04] <pcw_home> most FPGAs have only a modest amount of internal RAM (ours have maybe 12K to 128K bytes)
[10:39:39] <Jymmm> pcw_home: I get that (ISP, I2C, etc), but the uSD slot idea is MUCH simpler, and even a layman can swap out and update a uSD card.
[10:39:41] <pcw_home> though many support DDR RAM interfaces
[10:40:39] <pcw_home> Yeah but SD cards/connectors are rather unreliable
[10:41:05] <pcw_home> I would not like to depend on one
[10:41:34] <Jymmm> pcw_home: Every smartphone has one and they get tossed around, etc. Not THAT unreliable.
[10:42:09] <Jymmm> pcw_home: The ones that LOCK in, not slide in.
[10:43:22] <pcw_home> not really good for industrial settings
[10:44:14] <Jymmm> Well, a PC card isn't much better in that respect.
[10:46:08] <pcw_home> right but for our uses a SD card gains little but adds additional risk
[10:46:09] <Jymmm> pcw_home: But, I've used uSD cards on industrial application (actually oceanic research vessel) wihtout a hitch.
[10:46:47] <pcw_home> its possible but has 0 advantages for us
[10:48:37] <Jymmm> pcw_home: Maybe not advantage (per se), but I can see the potential for a stand alone system here. And that could open up a LOT of market potential.
[10:49:19] <Jymmm> pcw_home: Maybe just find high quality uSD connectors with proven industrial reliability
[10:50:19] <Jymmm> pcw_home: Whats the est price on your thermo board?
[10:50:29] <pcw_home> I would prefer a stand alone system to be updatable via Ethernet
[10:51:15] <Jymmm> pcw_home: Under $300?
[10:51:54] <pcw_home> I think $119
[10:52:16] <Jymmm> pcw_home: Ok, have you seen the prices on standalone themral controllers?!?!?!?!?!?!?!?
[10:52:48] <Jymmm> They are expensive, very kludgy to use, not very intuitive, and limited functionality.
[10:54:01] <Jymmm> If I could have a standalone fpga thermo board that I could toss into a box, add a 2x16 LCD, and make it do whatever I wanted... that would be AWESOME!!!
[10:54:38] <Jymmm> pcw_home: Hell, if I could shove LinuxCNC into a fpga, that would be awesome (remote GUI of course)
[10:55:39] <Jymmm> pcw_home: Fuck, if they can do it on a RasberryPi, imagine on a FPGA!!!
http://www.youtube.com/watch?v=WLIo7TA82xE
[10:55:57] <pcw_home> You dont need a FPGA, just a ucontroller with decent 12 bit A-D and some temp sensors ans a good low noise op amp
[10:56:23] <Jymmm> pcw_home: your missing the FPGA Glam =)
[10:56:34] <pcw_home> (low power ucontroller as you want minimum heat flux on the card)
[10:57:07] <pcw_home> (to keep input terminal; block as isothermal as possible)
[10:57:39] <Jymmm> pcw_home: Well, will you as least seriously consider it?
[10:58:30] <Jymmm> pcw_home: This aint your Momma's FPGA anymore =)
[10:59:03] <Jymmm> pcw_home: Google "fpga bitcoin" =)
[10:59:57] <pcw_home> that was our main challenge on the 7I88, getting its fast enough with a weak low power processor
[10:59:58] <pcw_home> to get 30Hz update rate while maintaining 60 KHZ or so input oversampling
[11:00:28] <Jymmm> ah
[11:01:21] <pcw_home> I dont know the bitcoin stuff and the ultrafast transaction stuff seem somewhat scammy/silly to me
[11:01:21] <Jymmm> pcw_home: what about the Cortex-A15 and the like?
[11:02:27] <pcw_home> for what?
[11:03:31] <Jymmm> "low power processor"
[11:03:57] <Jymmm> pcw_home:
http://www.tomshardware.com/reviews/bitcoin-mining-make-money,3514-5.html
[11:04:05] <pcw_home> ~100 times to much power and ~10 times too much ciat
[11:04:21] <pcw_home> s/ciat/cost/
[11:04:54] <Jymmm> Well, there are others too. I just picked the fastest one.
[11:06:19] <Jymmm> pcw_home: All I'm saying is just keep an open mind about a stand-alone FPGA as I described.
[11:06:41] <pcw_home> sure
[11:06:45] <Jymmm> =)
[11:08:00] <Jymmm> pcw_home: One example... My laser uses a embedded 386 mobo. It's getting up there in age. It be nice to have an alternative for it.
[11:08:31] <Jymmm> pcw_home: I dont want to pay $1200 to replace it.
[12:21:55] <IchGuckLive> hi all
[12:25:29] <Heinz_60> Hi @ll
[12:28:44] <t12_> with the 8i20 and an associated encoder going back to an 5i25 for example, does servoing happen in the 5i25 or the control software?
[12:49:57] <skunkworks> t12, linuxcnc does the pid.
[14:47:03] <skunkworks> heh
http://www.machsupport.com/forum/index.php/topic,17093.msg176667.html#msg176667
[14:47:52] <archivist> :)
[14:52:17] <andypugh> For those who don't read the forum, somone has just presented some remapped lathe codes (G71.1 turning, and a G76.1 for NPT threads, which might cheer Stev Blackmore up :-)
[14:52:40] <skunkworks> ooh - cool
[14:54:11] <andypugh> http://www.linuxcnc.org/index.php/english/forum/20-g-code/26760-new-gcode-cycle#36582
[15:03:48] <CaptHindsight> I still have to fire up Mach on the Matsuura. Anything in particular I should watch out for? Or in general everything? :)
[15:04:50] <cradek> yeah, watch out for asking questions in the wrong places :-)
[15:06:16] <CaptHindsight> cradek: I want compare it to Linuxcnc. Maybe write an article.
[15:06:28] <CaptHindsight> it came with mach installed
[15:16:26] <andypugh> I suspect that you will find that it is entirely adequate.
[15:34:54] <gabewillen> It definitely works, what it lacks is the extendibility and customization that is were linuxcnc & hal shine.
[15:35:58] <gabewillen> Not to mention the small fact that it runs on Windows... which is only good to me for one thing, and that is solid works.
[16:20:34] <DJ9DJ> gn8
[16:42:09] <andypugh> First steps to getting the Rivett to rotate under power:
https://plus.google.com/photos/108164504656404380542/albums/5889085582650378897
[16:43:26] <Tom_garage> is that the one you're making the special screws for?
[16:47:08] <andypugh> Aye, that is be.
[17:54:01] <t12> these geckodrive beanbag geckos are starting to accumulate
[18:31:41] <Jymmm> Is there something I can "coat" spring steel straps with to prevent rusting (water proof?) and won't "rub off" too easily?
[18:32:00] <Tom_itx> chrome
[18:32:21] <Tom_itx> urethane
[18:32:34] <Tom_itx> plastick dip
[18:33:01] <Jymmm> PU is too "slick", PD would get knicked alot due to usage.
[18:33:10] <Jymmm> and too thick.
[18:33:34] <Jymmm> a good paint? spray coating?
[18:33:56] <Tom_itx> imron
[18:34:02] <Jymmm> imron???
[18:34:11] <Tom_itx> epoxy paint
[18:34:18] <Jymmm> water thin?
[18:34:25] <Tom_itx> i dunno anymore
[18:34:28] <Jymmm> k
[18:34:41] <Jymmm> I was almost think something like "blueing"
[18:34:45] <Tom_itx> was used alot in aircraft finishes
[18:34:54] <Jymmm> like a firearm
[18:35:00] <Tom_itx> so blue it then
[18:35:11] <Tom_itx> cad plate
[18:35:12] <Jymmm> I have nfc what/how that's done
[18:35:17] <Tom_itx> heat
[18:35:26] <Tom_itx> not alot though
[18:35:39] <Jymmm> it's not a coating?
[18:35:45] <Tom_itx> blue?
[18:35:50] <Tom_itx> not if it's done right
[18:35:56] <Jymmm> like a gun barrel
[18:36:08] <Tom_itx> you can fake it but the real stuff is done with heat
[18:36:21] <Jymmm> ah, hmmm
[18:36:41] <Tom_itx> grind a piece of steel and watch it turn blue
[18:36:58] <Tom_itx> that blue part is the temp you want the whole thing
[18:38:33] <Jymmm> ah, here we go...
http://www.amazon.com/Rust-Oleum-7881830-Appliance-Enamel-12-Ounce/dp/B0009XB40I/kkorg-20
[18:49:26] <CaptHindsight> aluminum plate it,
http://www.alumiplate.com/
[18:50:33] <CaptHindsight> the rustoleum is far more economical however
[19:03:44] <Tom_itx> Jymmm, it's epoxy like i suggested only cheaper
[19:08:14] <ve7it> Jymmm,
http://www.cosmolinedirect.com/?gclid=CMHM_6blm6UCFQQ-bAodHgRMIw
[19:40:31] <Jymmm> Tom_itx: thanks
[20:19:19] <andypugh> Jymmm: Any electroplate should work (nickel is pretty). Zinc galvanising is effective. Use a stainless material, or perhaps switch to a beryllium copper spring material if this is a really difficult environment.
[20:23:17] <t12> http://kottke.org/13/07/human-powered-helicopter-wins-250000-sikorsky-prize
[20:23:18] <t12> this is absurd
[20:24:23] <Jymmm> andypugh: It's just a "pocket" adjustable tripod and clamp for my smartphone.
[20:25:08] <andypugh> Ah, in that case you should just use stainless steel.
[20:28:28] <andypugh> t12: I question your choice of adjective
[20:31:17] <andypugh> And I would love to see what it would do with Mark Cavendish at the pedals.
[20:31:50] <andypugh> From an engineering point of view, I wonder why they kept the back wheel? Enery store?
[20:50:19] <t12> i dont think absurd is bad
[20:50:20] <t12> i mean more
[20:50:23] <t12> its visually absurd
[20:50:58] <t12> maybe rear wheel is to behave like gyro for stabilzation?
[20:52:58] <Tom_itx> u sayin the pilot is a wuss?
[20:53:07] <Tom_itx> andypugh^^
[20:53:45] <Tom_itx> i say let him hoover at the celing and cool the building
[20:54:32] <ReadError> http://www.youtube.com/watch?feature=player_embedded&v=VqjlhtKIToo
[20:55:25] <andypugh> I think "absurd" belittles the achievement. I like "Marvellous" in this case. Perhaps "Stupendous". I was tempted by "Incredible", bit I do believe it, so perhaps "Impressive" is good too.
[20:56:12] <Tom_itx> might be like trying to control a blimp outside though
[20:56:15] <Tom_itx> or worse
[20:57:23] <andypugh> This is an achievement where you need to choose your adjectives with care. For me this ranks with watching the first shuttle launch,
[21:02:14] <t12> i dont think absurd impies not (achievement|marvellous|stupendous|icredible|impressive)
[21:08:57] <andypugh> I rather think it does. "Absurd" means jocular and foolish.
[21:17:02] <andypugh> I present, as evidence of my opinion:
http://en.wiktionary.org/wiki/absurd
[21:17:58] <andypugh> Though clearly I would have looked further had that disagreed with my opinion, as the internet is home to a billion mutually incompatible truths.
[21:19:48] <CaptHindsight> http://mentalfloss.com/article/16814/who-flew-wright-brothers
[21:25:31] <andypugh> I think it is probably fair to say that the Wright's were the first to achieve _controlled_ flight. Maxim flew, but didn't even intend to, for example.
[21:26:40] <sam_arg> I've been having trouble getting the latency-test to run. I did have the lapic issue identified in the troubleshooting guide but even after setting it in grub, it won't run. What else can I try?
[21:26:52] <sam_arg> Oh, I verified lapic was set in /proc/cmdline too
[21:27:06] <andypugh> Any clues in dmesg?
[21:27:31] <Tom_itx> the mrs'itx' is directly related to the wright bros
[21:28:01] <sam_arg> Dmesg still reports "ERROR, LOCAL APIC CONFIGURED BUT NOT AVAILABLE/ENABLED."
[21:28:03] <Tom_itx> doesn't mean i can fly though
[21:31:09] <andypugh> sam_arg: That might be an unsurmountable obstacle with your motherboard.
[21:31:20] <andypugh> sam_arg: There may be help here:
http://wiki.linuxcnc.org/cgi-bin/wiki.pl?TroubleShooting#emc2_doesn_t_run_missing_lapic
[21:32:25] <sam_arg> I'm quite familiar with that page :)
[21:32:57] <sam_arg> I don't see anything in the bios associated with it
[21:33:12] <sam_arg> and I can 100% confirm the lapic is set with grub. dang
[21:33:35] <andypugh> Options include a diferent PC (if you have a house full, then this is less of a problem than if you don't) or dropping back to the 8.04 kernel (but that is so old now that it might not support your hardware, but if it does, it will still support the latest version of LinuxCNC, so is not to be ignored)
[21:34:48] <andypugh> For $45 you can buy a beaglebone and struggle at the other end of the curve :-)
[21:35:05] <sam_arg> hmm decisions
[21:35:28] <sam_arg> It's an epia-ml mobo. It was pulled from a washing machine and I was going to set it up for a local hackerspace's cnc router
[21:36:00] <sam_arg> The washing machine had an altera fpga in it too lol
[21:36:11] <andypugh> A washing machine?! That changes things, you _have_ to make that work!
[21:36:21] <sam_arg> Yes
[21:36:29] <sam_arg> I've already connected it to a projector
[21:36:44] <sam_arg> The laundry machine was running centos with selinux
[21:36:45] <Tom_itx> andypugh are you messin in the bbb croud?
[21:36:55] <Tom_itx> what's the advantage of using it?
[21:37:03] <andypugh> Tom_itx: I own one, I booted it...
[21:37:11] <Tom_itx> the software still has a long ways to go doesn't it?
[21:38:04] <andypugh> I think it's nearly there, actually, bit still more trouble than just buying a mini-itc Atom board.
[21:38:28] <sam_arg> I use a d525 for my mill. It's great
[21:38:35] <Tom_itx> can you still get the d525 atom?
[21:38:46] <sam_arg> hmm
[21:38:49] <Tom_itx> yeah i have one too but they came and went pretty quick
[21:38:58] <andypugh> sam_arg: Daft question, but the lapic is actually enabled in the BIOS?
[21:39:31] <sam_arg> That's a good question
[21:39:37] <sam_arg> I've looked, but don't see a setting for it
[21:40:41] <sam_arg> what terms is the setting generally described as?
[21:41:34] <andypugh> Lapic, Apic. other than that, no idea. The acronym is
http://en.wikipedia.org/wiki/Advanced_Programmable_Interrupt_Controller
[21:43:36] <andypugh> Dropping back to the uniprocessor 8.04 LiveCD then upgrading to the latest LinuxCNC (which will work perfectly) looks like your best plan.
[21:44:22] <Tom_itx> i'm using that on one old pc, works fine
[21:45:00] <andypugh> Actually "perfectly" might be an over-statement. "With only the current bugs" is probably closer.
[21:45:31] <Tom_itx> features
[21:45:51] <andypugh> No, they only become features when documented.
[21:46:37] <sam_arg> easter eggs?
[21:46:42] <Tom_itx> i'm concerned about programming this new router i ordered
[21:51:45] <sam_arg> well this is frustrating :
http://www.mite.cz/__NOVE/__Nviaminiitxfiles/ML/UM_EPIA-ML_v10.pdf
[21:52:01] <sam_arg> That manual talks about "ACPI Function"
[21:52:08] <sam_arg> and says there is an enabled/disabled state
[21:52:17] <sam_arg> but the image of the bios screen right above it doesn't show it
[21:52:20] <sam_arg> nor does my bios
[21:53:35] <sam_arg> I guess that is associated with the power savings features. I've messed around with that with no dice
[21:54:46] <sam_arg> ah, I see they copied/pasted from another machine that has the option
[21:55:11] <sam_arg> oh lordy I'm dislexic
[21:56:12] <andypugh> I don't think we have any real howlers, just curiosities like:
http://sourceforge.net/p/emc/bugs/315/
[21:57:37] <andypugh> Hmm. Given a 2004 date i would go with the vintage (8,04) distro and then put the 2.5.3 LinuxCNC on top
[21:58:19] <andypugh> (I might mean 2.5.4, whatever is newest while not being experimental)
[21:58:25] <sam_arg> right
[21:58:39] <sam_arg> Thank you for your thoughts on that. I would not have thought to try 8.04
[22:00:34] <andypugh> We only started to require the Lapic when we built the first multiprocessor RTAI kernel, up to that point we just quietly wasted half (or 3/4, or 7/8) of your CPU power
[22:01:20] <sam_arg> with my d525, I think I have a pig process running to eat one core.. not sure if I've stopped that since then
[22:02:50] <andypugh> That's not sensible, use isolcpus instead to leave a core free for RTAI
[22:03:31] <sam_arg> Yea, I will be doing that. The original reason I did it was based on some linuxcnc documentation. After reading through to figure out this lapic stuff I ran across it
[22:03:59] <sam_arg> I shall miss my space heater ;.;
[22:07:50] <andypugh> I was just discussing the docs in another channel. There is some stuff in the Wiki which is simply wrong now.
[22:11:04] <sam_arg> Documentation seems troublesome with this project regardless of how it's done due to the diversity of platforms this runs on
[22:11:31] <sam_arg> Has there ever been tought to support an "official" platform?
[22:11:52] <sam_arg> beagle D:
[22:12:07] <sam_arg> Ohh, I have had soo many headaches working with that omap
[22:13:07] <sam_arg> I do have the shining achievement of placing the omap/ram bga pop by hand and reflowing it successfully
[22:16:07] <andypugh> sam_arg: For simple, as you already know, D525 or D510 if you can find one. I think there is a spectrum of "harder" as you move away from there. I suspect that the apogee of tricksyness is probably a MacOS build, which ought to not be impossible, but would be a challenge.
[22:24:01] <andypugh> I have a motor, and the cable blanking plugs are clearly marked "PG16". But the thread angle is equally clearly not 80 degrees. I wonder what is going on there?
[22:24:20] <CaptHindsight> AMD APU boards are working, some are <$70 with APU
[22:25:30] <CaptHindsight> E series APU boards have the APU soldered onto the mainboard
[22:26:25] <andypugh> Time to snooze
[22:26:30] <sam_arg> goodnight
[22:26:33] <sam_arg> thank you for your help
[22:26:34] <CaptHindsight> http://www.newegg.com/Product/Product.aspx?Item=N82E16813157228 for example
[22:28:11] <CaptHindsight> the intel Atom boards have some GPU issues
[22:29:10] <sam_arg> hmm
[22:29:37] <CaptHindsight> sam_arg: there should be a whole new list of boards and hoto's for tweaking soon
[22:34:18] <CaptHindsight> pcw_home: what were the GPU issues you were mentioning with the Atom boards?
[22:38:38] <pcw_home> I dont recall other than the 2xxx series dont have decent linux video drivers
[22:39:54] <pcw_home> (also actual jitter is worse than reported with the latency test)
[22:47:04] <CaptHindsight> AMD Phenom-II CPU's and 780 chipset mainboards with integrated graphics are under 10uS latency jitter and are dirt cheap now
[22:53:22] <CaptHindsight> AMD A series APU's are down to $40 ea and socket FM1/2 boards are as low as $45
[22:54:25] <CaptHindsight> we hope to have all the GPU hardware accell working with RTAI and Linuxcnc shortly
[22:58:51] <CaptHindsight> http://www.jetwaycomputer.com/NC85.html Jetway has a few Mini-ITX mainboards with GPIO headers
[23:00:14] <pcw_home> I wonder how these do latency wise:
[23:00:16] <pcw_home> http://www.newegg.com/Product/Product.aspx?Item=N82E16813128598
[23:01:18] <CaptHindsight> that is the question
[23:01:55] <pcw_home> those are basically slow I3's
[23:02:42] <pcw_home> (but about twice the native CPU of the AMD 350D's)
[23:04:21] <CaptHindsight> we just figured out how to get the hardware accell working with the A series with RTAI
[23:05:36] <pcw_home> video accel?
[23:05:42] <CaptHindsight> yeah
[23:06:24] <CaptHindsight> so $45 for the MB and $40 for FM1/2 APU
[23:06:44] <CaptHindsight> with HD7000 series graphics performance
[23:08:14] <CaptHindsight> http://www.newegg.com/Product/Product.aspx?Item=N82E16819106013 HD6xxx on the lowest cost one right now
[23:08:36] <pcw_home> Sounds good, so how close is RTAI +current kernel to being useable?
[23:09:10] <CaptHindsight> "Linux precise-rtai-x86 3.5.7-rtai+ #5 SMP Thu Feb 28 00:21:45 MST 2013 i686 i686 i386 GNU/Linux
[23:10:56] <CaptHindsight> that was the latest working kernel with RTAI
[23:12:13] <CaptHindsight> someone is working on an in kernel debugger to figure out once and for all what the causes of the jitter are
[23:12:44] <CaptHindsight> it doesn't mean it can always be fixed, but at least it may be traced
[23:18:13] <pcw_home> Hmm I wonder how you trace code thats not interruptable...
[23:20:59] <pcw_home> (or when things are blocked by hardware)
[23:21:51] <CaptHindsight> probably won't go this far but i have access to the entire AMD CPU and chipset virtual platform
[23:23:43] <CaptHindsight> it emulates all the mainboard hardware
[23:32:22] <CaptHindsight> http://tinyurl.com/qg9tyzd might be of interest Bridgeport BOSS 8 CNC
[23:34:34] <CaptHindsight> http://tinyurl.com/pe8eo5m Cincinnati Milacron Mill