#robotics Logs

Jan 11 2020

12:01 AM rue_mohr: all doing good, no reallocated sectors
12:06 AM rue_mohr: WD drives
12:06 AM rue_mohr: black on my workstation
12:06 AM armyofevilrobots: I've had good luck with those too.
12:06 AM rue_mohr: Model Family: Maxtor DiamondMax 10 on the shop fileserver
12:06 AM rue_mohr: the shop fileserver uses an ultrascsi drive for the OS
12:06 AM rue_mohr: I'v had to replace the controller :S
12:06 AM rue_mohr: I now have no spares
12:06 AM armyofevilrobots: I might have a 15k scsi kicking around...
12:07 AM rue_mohr: not quite the drive I thought it was
12:08 AM rue_mohr: anyhow
03:09 AM fling_ is now known as fling
03:36 AM logicmoo is now known as dmiles
04:02 AM polprog: rue_mohr: ive got buspirate working as a jtag, i dont have any xilinx specific programmers
04:03 AM polprog: also, i need to download the software first, but sure, id appeciate that if you targz'd some example code :)
05:29 AM Jak_o_Shadows: I don't have the porches right
06:01 AM Jak_o_Shadows: hang on, it's fine
09:07 AM veverak: Any tips about how to make small stackable boards?
09:07 AM veverak: pcb
09:07 AM veverak: have to use 1.27 headers, but can't find their stackable version :/
10:04 AM rue_bed: sockets with loooong pins
10:45 AM polprog: veverak: maybe try SMD male on one side and smd female on the other side?
10:55 AM veverak: polprog: hmm
10:56 AM veverak: they actually take a lot of space
10:56 AM polprog: i dont see any other option really, unless you find long ones like PC/104 uses
10:57 AM polprog: the pci connector on these is a male-female with a shround, goes thru the board
10:57 AM polprog: but it's 2.00mm
11:05 AM veverak: :/
11:05 AM veverak: maybe I should scan 1mm connectors
11:32 AM polprog: how many pins?
12:40 PM veverak: 4x 2x4
12:42 PM veverak: https://scontent-vie1-1.xx.fbcdn.net/v/t1.15752-9/fr/cp0/e15/q65/82307156_617613558995831_9141872837684887552_n.jpg?_nc_cat=105&efg=eyJpIjoidCJ9&_nc_ohc=DElgh-yOlmMAQmZDJYYnyWgT8ZcTNGnUZSKNe96WrThBbkra2DDvIZ_FA&_nc_ht=scontent-vie1-1.xx&_nc_tp=12&oh=b430c4181a570ffb5f4525b3df9625eb&oe=5E97C881
12:42 PM veverak: polprog: it is 24x24mm thing
12:43 PM veverak: 16 signal pins + gnd,3v3,vcc
12:57 PM polprog: veverak: hmm
12:58 PM veverak: https://eu.mouser.com/ProductDetail/Hirose-Connector/BM46B-12DS-035V53?qs=sGAEpiMZZMvffgRu4KC1RzIOOrpfsZzR%2FkitmecZP6khDn%252BZgk%2FFQA%3D%3D
01:09 PM veverak: but these are too low :/
01:49 PM BitEvil is now known as SpeedEvil
01:50 PM rue_mohr: oh I'll do those examples
01:50 PM rue_mohr: start installing webpack now and you might be done by next week or so...
01:57 PM Steffanx: stop using a pc from < 2000
01:58 PM Tom_L: why?
02:04 PM Steffanx: then it wont take a week to install
02:04 PM Tom_L: but then he'll run out of things to wait on
02:05 PM Steffanx: Ah, ok.
02:05 PM Steffanx: Cant beat that
02:05 PM Tom_L: what you been up to?
02:11 PM Steffanx: Not too much. Work. Not using AVR. Some random other projects like a CNC. There?
02:12 PM Tom_L: you have a cnc?
02:12 PM Tom_L: running what software?
02:13 PM Steffanx: For now its just grbl, because i just needed something easy.
02:13 PM Steffanx: needed = wanted
02:13 PM Steffanx: and cheap.
02:13 PM Tom_L: you build the cnc?
02:13 PM Steffanx: Yes
02:13 PM Steffanx: Just took some ideas from here and there
02:14 PM Tom_L: nice
02:14 PM Tom_L: i built one here a while back
02:14 PM Tom_L: what's it made out of?
02:18 PM Steffanx: Aluminium mainly. Here some random pics http://imgur.com/a/Qnicnpa
02:19 PM Steffanx: Some are a bit old and not up to date.
02:19 PM Steffanx: i designed it so i could semi-easy do everything by hand.
02:20 PM Steffanx: Some parts i still have to replace, because they are 3d printed. Although it works surprisingly well
02:21 PM rue_mohr: Steffanx, nice
02:22 PM Tom_L: yeah looks good
02:24 PM Steffanx: And some parts are just kinda hacked in. Like the end switch in the 2nd pic. Have to find a better mount for that.
02:24 PM Steffanx: find = design
02:25 PM Steffanx: and then copy Tom_L's touch probe
02:26 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Column_Mill_VMC.jpg
02:26 PM Tom_L: i designed and built that
02:27 PM Steffanx: Did you use the real hiwin stuff? ;)
02:28 PM Tom_L: absolutely
02:28 PM Steffanx: hah
02:28 PM Steffanx: That was kinda outside my hobby budget. And it works fine as it is now.
02:28 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/spindle_cooler/Mill1.jpg
02:29 PM Tom_L: used my old sherline spindle
02:29 PM Steffanx: Same with the Ball screw spindles
02:29 PM Tom_L: there's lots of build pics there
02:31 PM Steffanx: I just use a makita spindle for now. The noise it makes is the most annoying part of it.
02:31 PM Steffanx: *spindle = router
02:31 PM Tom_L: router?
02:31 PM Tom_L: yeah, you'll burn cutters up with that if you don't feed them enough
02:32 PM Steffanx: hehe yeah, it can only spin fast or super fast.
02:33 PM Tom_L: i've cut steel on mine but i gotta go slow with it
02:37 PM Steffanx: Cool :)
02:46 PM Steffanx: but yeah. Then you have this done and then.. you need proper dust collection. Maybe a new controller, new spindle.
02:46 PM Tom_L: i'm running linuxcnc on mine
02:47 PM Tom_L: spindle is really the only downfall on it
02:47 PM Tom_L: but it's fine for me
02:47 PM Steffanx: Yeah, i was talking about mine.
07:21 PM derek_aoer is now known as armyofevilrobots
09:43 PM rue_shop2: wow, working out how to do phase measurement from timestamps is evil
11:24 PM Tom_L: don't see how that could be accurate
11:27 PM rue_shop2: I have two rising and two falling edge timestamps for two channels
11:27 PM rue_shop2: I have to work out wtf to know which value to measure between
11:28 PM Tom_L: rising and falling edge will give you the period
11:28 PM rue_shop2: I want the phase difference between the two channels
11:28 PM Tom_L: then measure the difference between rising edges of the other channel and the first one
11:28 PM Tom_L: you need the period to determine the phase angle
11:29 PM rue_shop2: it depends on when the readings came in
11:29 PM Tom_L: not at the same time?
11:29 PM Tom_L: if they're un'related you can't do it
11:29 PM rue_shop2: cause you dont know if you started your readings when which wave was high of low, and if the channels were leading or lagging
11:29 PM rue_shop2: there is one master counter
11:29 PM rue_shop2: the hardware snap shots it when the edge happens
11:30 PM rue_shop2: I have 4 of those
11:30 PM Tom_L: 2 rising 2 falling edge stamps should still give you the phase angle
11:30 PM rue_shop2: 1 for rising edges, channel 0
11:30 PM rue_shop2: 1 for falling edges, channel 0
11:30 PM rue_shop2: 1 for rising edges, channel 1
11:30 PM rue_shop2: 1 for falling edges, channel 1
11:30 PM Tom_L: so rising to falling of 0 gives the period
11:31 PM rue_shop2: there is a cache of two timestamps each
11:31 PM Tom_L: rising 0 to rising 1 gives the phase
11:31 PM rue_shop2: and unless you kinda sort them, you dont know who came first
11:31 PM rue_shop2: unless there was a falling inbetween
11:31 PM Tom_L: or you'll be 180 out of phase...
11:32 PM rue_shop2: hmm
11:32 PM Tom_L: are 0 & 1 the same period?
11:32 PM Tom_L: ie a quad encoder...
11:32 PM rue_shop2: if I take the earliest of the first and the latest of the second, and modulus the answer by 360
11:32 PM Tom_L: or resolver
11:33 PM rue_shop2: hmmmm
11:33 PM rue_shop2: I, hmmm
11:33 PM Tom_L: dunno what it's for just that it's evil.
11:33 PM Tom_L: so you say
11:34 PM rue_shop2: well, you bring up a good point
11:34 PM rue_shop2: the error when its all done is a multiple of 360 degrees
11:35 PM rue_shop2: I'v been focusing on adjacent edges
11:35 PM Tom_L: find the period
11:35 PM Tom_L: then take the percent of the period offset
11:35 PM rue_shop2: my problem is knowing which edges to use for the calc
11:36 PM Tom_L: how do you know what direction a quadrature is turning?
11:36 PM Tom_L: same concept
11:37 PM rue_shop2: I dont, but its all time-stamped so it can be 'searched'
11:39 PM Tom_L: the period of both signals is the same right?
11:40 PM rue_shop2: it should be
11:41 PM rue_shop2: if I'm testing a ringing filter, .... I'm beyond the design of the analyzer
11:41 PM Tom_L: then you can determine which one is leading
11:41 PM rue_shop2: but I have to search
11:41 PM rue_shop2: and thats whats bugging me
11:45 PM rue_shop2: here is the idea you lead me too
11:45 PM rue_shop2: diff ALL the combinations of rising edge times,
11:45 PM rue_shop2: the smallest absolute value of them all is the winner
11:46 PM rue_shop2: then compare that against the period
11:46 PM Tom_L: possibly
11:46 PM Tom_L: and that would determine the direction
11:46 PM rue_shop2: yup, as +-
11:46 PM Tom_L: the winner is 0, clockwise, the winner is 1, couterclockwise per se
11:55 PM rue_shop2: I wonder if there is any point in resolving the phase of a filter past 1 degree
11:58 PM rue_shop2: oh huh
11:59 PM rue_shop2: I guess I only need to do this for one of the inputs, as the other one should just be the - value of the first