#linuxcnc-devel | Logs for 2014-12-05

[08:41:42] <skunkworks> The servo issue was the brushes (2 of the 4 where broken)
[08:42:35] <skunkworks> dad says the tach is brushless
[08:57:52] <skunkworks> 'Brushless dc generator' wonder how that works - some electronics in it?
[09:02:03] <skunkworks> a bit of electronics under the cover..
[09:09:20] <kwallace> Here is my brushless generator, it's AC, but with could be DC with some diodes. http://www.wallacecompany.com/machine_shop/Witte/dcp_6811.jpg
[09:09:32] <kwallace> Yay NASA
[09:10:46] <cradek> skunkworks: it was brushes and they replaced the ballscrew and then sold the machine? ouch...
[09:12:08] <skunkworks> or Yay!
[09:12:16] <cradek> heh yeah
[09:13:44] <skunkworks> he says the comutator looks good. 2 of the brushes where broken and 2 looked like they had been in there forever..
[09:14:41] <skunkworks> kwallace, lathe spindle motor?
[09:18:51] <mozmck> Anyone know if there is an overview of the differences between Gmoccapy and Gsreen somewhere?
[09:23:09] <cradek> I don't know of any
[09:24:07] <mozmck> ok. I'll keep looking at it.
[09:24:22] <kwallace> skunkworks, no, it's a 10 kW AC generator. I posted it because a brushless tachometer might use the same brushless system. A bit of a stretch I suppose.
[09:28:16] <kwallace> Yeah, I can see these look just like it. http://www.marshbellofram.com/servo-tek/products/tachsyn-series-brushless-dc-tachometers/
[09:31:46] <skunkworks> heh - I found that page earlier
[10:02:26] <pcw_home> A selenium rectifier?
[10:02:45] <skunkworks> old school
[10:06:19] <pcw_home> Hmm maybe thats a Varistor (since there are Silicon or Germanium rectifiers )
[10:10:37] <kwallace> It's a diode to turn the rotor AC into DC for the field excitation. I had to take it apart and clean each disk to get it to work. I'll most likely replace it with a modern diode.
[10:11:53] <pcw_home> Ahh, but you cant clean the disks of modern diodes :-)
[10:50:30] <bjmorel_work> Anyone have a good DC Servo Driver for Dummies link ?
[10:54:34] <cradek> what's the real question you're trying to answer?
[10:57:13] <bjmorel_work> I have an X2 that I'm finishing converting to CNC and I want to replace the stock motor with a DC servo, trying to wrap my head around all the diff. VDC ratings on ebay servos.
[10:58:20] <bjmorel_work> * Spindle motor *
[10:59:31] <cradek> ah spindle motor
[10:59:47] <cradek> that generally doesn't need to be a servo
[11:03:09] <bjmorel_work> Yeah, doubt I'll ever get into tapping or tool change with the X2. Basically I would like the SX2 brushless motor, but not willing to pay the $600ish to get the parts from LMS
[11:06:18] <pcw_home> Hmm tapping does not require a servo spindle
[11:06:41] <skunkworks> why not go vfd and 3 phase motor?
[11:07:09] <cradek> that's the ticket
[11:07:23] <pcw_home> and relatively rough orient should work with a VFD
[11:07:50] <cradek> if you want an automatic tool changer on an X2 isn't that the least of your problems?
[11:08:39] <cradek> I think it's morse taper
[11:09:10] <cradek> but yeah you can tap with any spindle motor that reverses cleanly (a vfd is just fine)
[11:10:22] <bjmorel_work> Really I thought the VFD was the way to go ( I work for an industrial parts dist. ) as I could get brand new with a really good price, but the weight of even the smallest 1/2 - 3/4HP motors on the little mill would be a bit much.
[11:19:55] <kwallace2> Hows about: http://www.ebay.com/itm/111533747619 and http://www.ebay.com/itm/171191507878
[11:21:35] <kwallace2> One of my treadmill motors is at the bottom here: http://www.wallacecompany.com/machine_shop/LinuxCNC/pwm_fet/pwm_fet.html
[11:24:28] <bjmorel_work> What kind of weight are those ?
[11:30:26] <kwallace2> Gosh, I don't really know. They have a steel housing, ceramic magnets, copper and steel rotor, so fairly heavy. Guessing 15 to 20 pounds.
[11:31:13] <kwallace2> They have lots of low end torque.
[11:31:52] <bjmorel_work> That puts it in line with a 1/2HP 3600RPM motor, but the price sure is right. I know I've read about a lot of people doing that conversion.
[11:32:08] <bjmorel_work> Of course, than I don't get the fun of really learing servos.
[11:36:07] <kwallace2> http://www.cnczone.com/forums/benchtop-machines/108150-x2-spindle-motor-upgrade.html
[11:38:38] <kwallace2> Here is my servo experiment: http://www.wallacecompany.com/machine_shop/bridgeport/00046-1a.jpg and http://www.wallacecompany.com/machine_shop/bridgeport/
[12:21:38] <archivist> I wonder how many thou the x2 column bends with the weight of the motor
[12:56:58] <bjmorel_work> Considering the number of "Stiffen the Column" articles I've see, probably quite a bit. If I didnt get such a "craigslist steel" on it, I would have went for a bigger machine.
[12:59:26] <archivist> I added some angle to my home brew mill to stiffen the column
[18:32:05] <andypugh> cradek: have you seen my Arduino pulse clock exploits?
[18:33:02] <andypugh> radio time module + arduino nano suddently makes every old and pretty lab counter look like a super-accurate clock.
[18:51:38] <kwallace> I only found the Arduino hotel room exploit.
[21:23:44] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff 622cf1b 06linuxcnc 10configs/sim/axis/moveoff/hookup_moveoff.tcl 10configs/sim/axis/moveoff/moveoff_demo6.ini 10scripts/moveoff_gui 10src/hal/components/moveoff.comp hookup_moveoff: error handling and ini settings * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=622cf1b
[21:23:44] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff 9167e5a 06linuxcnc 10configs/sim/axis/moveoff/hookup_moveoff.tcl hookup_moveoff.tcl: eliminate hardcoded threadname * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=9167e5a
[21:28:29] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff e76171c 06linuxcnc 10configs/sim/axis/moveoff/hookup_moveoff.tcl hookup_moveoff.tcl: eliminate hardcoded threadname * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=e76171c
[22:07:19] <pcw_home> replaced the CPU in my desktop and now it runs hm2_eth at 4 KHz (barely 2 Khz before) Cache seems to be the secret
[22:07:52] <seb_kuzminsky> the new cpu has the same (or similar) frequency & number of cores, but more cache?
[22:08:22] <pcw_home> same cores 3.16 VS 2 GHz but 6M cache vs 1M
[22:08:32] <seb_kuzminsky> nice
[22:08:41] <seb_kuzminsky> yeah cache is hugely important
[22:09:33] <pcw_home> before, a plot of read time looked like grass on halscope, now its flat with occasional bumps
[22:09:47] <seb_kuzminsky> pcw_home: that's really cool
[22:09:59] <pcw_home> (and the "grass" grew when you moved the cursor)
[22:10:30] <pcw_home> replaced the old e2180 with a e8500 ($15.00 on ebay)
[22:11:25] <pcw_home> I did break 2 of the stupid plastic heatsink clips so the heatsink is working in gravity mode
[22:13:18] <skunkworks> been there..
[22:14:45] <pcw_home> This is a Gigabyte H41 motherboard, Ethernet latency seems better then the newer H81+G3258
[22:15:13] <pcw_home> trying 5 KHz...
[22:19:44] <pcw_home> 4 khz plot
[22:19:45] <pcw_home> http://imagebin.ca/v/1jf1ls1mm9Sc
[22:27:58] <skunkworks> nice!
[22:28:42] <pcw_home> Naw wont quiet run at 5 because there just not enough margin for the previous cycle write to sserial read
[22:29:42] <pcw_home> but nice that old MBs can work this well with hm2_eth
[22:30:11] <Tom_L> is it more hardware particular than other interfaces?
[22:30:40] <pcw_home> Less I think
[22:31:29] <pcw_home> USB Ethernet adapters dont work
[22:32:09] <pcw_home> certainly less fussy about hardware than parallel (EPP) interfaces
[22:32:24] <Tom_L> nice
[22:32:47] <pcw_home> but still not as fast as PCI if you want a very high servo thread rate
[22:32:58] <Tom_L> that stands to reason
[22:33:12] <Tom_L> it's right on the buss
[22:33:50] <Tom_L> with that in mind, why are hdd going serial again?
[22:33:59] <skunkworks> we are going to convert the mazak using a 7i80.. should be fun.
[22:34:20] <skunkworks> they have been serial for a long time...
[22:34:24] <Tom_L> skunkworks, i hope you're making a page for your camera experiments
[22:34:51] <skunkworks> when I get time I will put it on the wiki
[22:35:02] <pcw_home> I think with Preemt-RT and the size of the stock linux network stack a bigger cache really helps Ethernet latency
[22:35:41] <pcw_home> Yeah thats really neat (a lot of people want that for pick/place)
[22:37:36] <skunkworks> I would have had a video a week ago - but failed at making a video with my phone. (thought it was recording but it wasn't)
[22:40:33] <skunkworks> I always thought cache was bad for realtime..
[22:42:14] <pcw_home> having to reload the cached is....
[22:42:51] <skunkworks> right
[22:45:49] <KimK> pcw_home: Hi Peter, with all the changes over the last year or more to linuxcnc, real-time kernel, hostmot2, mesa firmware, etc., etc., do I need to get old bitfiles that you made for me remade? Or are they OK even if over a year old? I've been away from a particular project for quite awhile, but I'm hoping to get back to it soon.
[22:46:21] <pcw_home> no need to update unless you need a new feature
[22:46:34] <Tom_L> just new functions
[22:46:39] <Tom_L> if you're not using them...
[22:47:02] <pcw_home> I dont actually think there have been any bug fixes recently
[22:48:07] <KimK> OK, thanks. I'm not sure about new features or if I need them, I'll look into it.
[22:48:40] <Tom_L> sserial would be one probably
[22:48:51] <pcw_home> I do have that Cincinatti I/O board sort of working
[22:51:38] <pcw_home> yeah, sserial or anything with code in the FPGA has many more revs than hardware
[22:51:51] <pcw_home> sserial is up to V44
[22:51:53] <KimK> sserial? Yes, I'm using a pair of 7i44's.
[22:52:48] <pcw_home> the resolver code was patched to make it get along with the Ethernet cards and their 100 MHz ClockLow
[22:53:59] <KimK> Ha, I'm using resolvers too, but not yet with the 7i44's. I'd like to eventually, though.
[22:54:29] <KimK> Glad to hear about the Cinci card, anything I can do to help?
[22:54:43] <Tom_L> pcw_home that would be more a firmware update than bit files wouldn't it?
[22:55:41] <pcw_home> the resolver code is in a ROM in a bitfile so its still a bitfile issue
[22:56:49] <pcw_home> (but this is only need for Ethernet FPGA cards so wont affect PCI/EPP cards)
[22:58:41] <KimK> So then, what do you think? Would it would be a good idea to remake the bit files? Or will I be OK?
[22:59:15] <pcw_home> Should be OK
[23:02:22] <KimK> Now that I think about it, maybe I *was* using resolvers with 7i44's (just testing for now), I'd have to check. I think it was 7i49, 7i49, 7i44, 7i44, I'll check on it and get back to you later.
[23:04:00] <KimK> Anything I can do from this end to help with the Cinci card?