it should be ok then
1mm ~ 18 AWG, 2mm ~ 12 AWG
if you say ;)
the thicker the better ;)
that's what the pretty chart tells me
but > 2mm is probably overkill for these
ask jmkasunich , he should know ;)
* alex_joni heads to bed
whenever you wonder about wire and current, just google for AWG ampacity
then use the worst-case numbers from their charts, or heavier gauge
jmkasunich_ is now known as jmkasunich
did CIA post a success message?
I must have been disconnected when that happened
.it was about 5 minutes after you said "we'll know in a few minutes"
has anyone looked at analog widgets for halvcp? i just built gtkdial from http://www.gtk.org/tutorial/x2917.html,
it needed gtk+-2.0
analog has been discussed, but not implemented AFAIK
tomp: I've wanted to do that, just haven't had the time
its a bit buggy, but 'works'
I should see if I still have a copy of the source to the scope/meter/logic analyzer/function generator software I wrote in college
it looks like a meter but is an input widget... i had to bury the widget file inside the example .c becuz i couldnt figger the compiler directive
though it's probably less work to just use the widget sets that already exist (I had to write my own, and they're in Turbo Pascal)
anyone had any experience with dhttpd?
time to go home, printer is done running in the shop
thanks for answering my random n00b questions
i\d like to brag a bit.. my last retrofit is at http://tag-edm.com,
and i just made some movies of big cuts, clunky interface & shakey cameraman, but lotsa smoke
maybe someone can tell me if the vids render ok on different browsers
it renders fine (the first one so far)
it is loud ;)
oh yeah TURN DOWN YOUR VOLUME (sorry)
and you can't read the meters - in fact, I had to look to see that they were meters
dunno how to edit that yet
look up "LiVES", I think
1st is at 225 amps, meters clear-er on other vids
nice editor with lots of features
what's the control software?
LiVES is on the live cd from rastasoft,, still learning it
ok - in the second vide, I can see the needle (near the end), but can't read the text on the meter
the control is from Taiwan , FineTech and is a pc running dos and using HCTL1000 chips fro motion
the meter is marked 0 100 200 300 amps. midscale is 150 average ampers
if you can, you should make it so the video doesn't loop automatically
yeah, gotta spend some $ to get real editing... I used flash output becuz i think more people have that than .. others
I'm using Mozilla 1.7.12 on Windows, and I have too many video players installed to count, just so you know one place where it works ;)
dynebolic is the linux distro with LiVES
but did it play >in< the page or thru a helper window?
flash isn't great, since there's a lack of documentation (and no 64-bit player)
in the page
thanks, and which flash, 7 or 9?
I suspect I have - err - I don't know ;)
cool videos :)
I knew people who had these generators and swore they couldnt get 150 amps out , I had it up to 250 but the tool was big enuf to handle it
thier problem was thier machines, not the generator
how big was the actual tool in the videos?
1st vid had 4" dia with 1/4" flush hole, last 2 had 12x18" die for a universal ( big like in train motor )
graphite to steel positive polarity
ok. the Z contraption looked pretty big, but I couldn't see under the fluid ;)
it's actaully way bigger now, they sent me an orbiting head to attach, at least an extra 300 lbs, good thing i put 3kW on the head
03jmkasunich 07HEAD * 10infrastructure/farm-scripts/index.shtml: update
jmkasunich_ is now known as jmkasunich
tomp: running flash 7 in Firefox - the videos play in the page
Are you using a camcorder? For the screen shots, you might try a still camera set to macro mode - use the video output.
is imagebin.org/6692 the latest picture?
oops, jtr: i used fujix camera, it does avi, and i do use macro for closer than 12"
thanks about flash 7 ( 9 seemed to be a bit weird on ie6 )
be back later...
tomp: was thinking macro for the meters - may be an artifact of the flash medium. never did any web video.
seems like ubuntu is having network problems...
Could not connect to us.archive.ubuntu.com:80 (22.214.171.124), connection timed out
either the encoders on the motors are cleaner, or quadrature is immune to that noise, because the lathe (other than the spindle encoder) works great
Rugludallur_ is now known as Rugludallur
jtr: i cant usemacro for vid, it only works for still, & some artifacting is 2meg swf from 46meg avi. orig avi is not real clear either :-(
03cradek 07HEAD * 10emc2/src/emc/motion/control.c: ignore FO when homing
tomp: Yeah, you lose a lot with compression - and cutting to a good sharp closeup of the meter will look like faking.
03cradek 07v2_1_branch * 10emc2/src/emc/motion/control.c: ignore FO when homing
tomp: my still camera (olympus d-490z from 2001) has a live composite video output. Used it as an input to a VCR before.
jtr: i got svideo out and never tried it... but the prob seems to be once the camera decides focus at beginning of record, it keeps that focus
hmm - I was recording with the camera on a tripod - focus length never changed. Have to play with that.
the manual for my centent drives mentions using a zener diode and fuse to protect the drive/power supply from back emf
they suggest a 1 watt diode ... but don't mention a voltage
I'm guessing the diode allows a surge of back emf to go directly to ground without shorting the psu during normal operation?
f, all the smart people are asleep already
the diode probably shouldn't be a zener
it goes pointing "toward" the supply, so that back EMF can get dumped to the supply, but no current will go to power the drive
[06:28:03] <ejholmgren> http://www.centent.com/_private/cn0162.pdf
you probably want one that's fairly beefy, not just 1A, and the reverse breakdown voltage should be higher than your supply voltage
page 7 and 8
are you talking about D5?
ah - I had misread your question as a diode "across the fuse"
I have a 24V ps
I don't think the diode is needed then - it's pretty unlikely that you'll get >80V when the supply is 24V
but for that circuit, the zener would need to be a "slightly higher than 24V" value :)
the ps is ~18A regulated 24V
do you think C1 is necessary?
regulated isn't ideal, but that should work OK
if C1 is similar to the 100uF cap that Mariss recommends "within 1" of hte power terminals", then yes, I'd use it
if it's just for supply filtering, then it shuoldn't be needed
C2 is the first one
C1 would be filtering
I suppose I could look at the schematic again :)
nope - C1 isn't needed for a (good) regulated supply
it's a nice Lambda
it's better to use a bulk supply in these applications - a regulated supply will probably not have the headroom of a transformer/bridge/cap
I have a 60V, 18A bench supply (a nice one - Sorensen), but I wouldn't use it for motors, other than for testing
well ... time to put on the carharts and take the wife's dog out for a bathroom break
hav efun. it's time for bed here :)
see you later
it's 5 degrees outside and our yard isn't fenced
woo freakin hoo
later, thanks for the help
wow - it's still above freezing here
lots of it piled up :)
same with me
I just got home from a big lan party :/
good morning all!
what were you playing at your lan party?
nothing, I participated in some wargame
that's about the gaming I did :)
you mean, tabletop?
no, I mean as unix box with code challenges ;)
jepler: did you have to do some funnky stuff to serve from your dsl? the 2 service providers here block port 80 outgoing.
skunkworks: the local ISP's "small business" plan doesn't block anything
jepler: really funny, though we got lost on level 8 of 10
skunkworks: I think I pay about $10 more than the "home" plan, each month, and I get 5 static IPs (a "/28", I think)
its funny. here at work we pay an extra 10 for 1 static and still no port 80 ;)
this plan is just under $50 a month
jepler: pretty happy with the pluto so far?
skunkworks: yes, I am.
Good. I hope to be able to play with this week.
I have not gotten the other interface cards yet. I will give them this week.
like I said on my last blog entry, I want to get a new version uploaded in the next few weeks -- HDL source code, emc2 .comp, and precompiled firmware
if you want to get a copy of the current source or firmware before that, I'll e-mail them to you
do you know how you'll go about connecting this to your drives & encoders?
The plan is a: buy or desolder some headder pins off of a old mother board to solder on the pluto and find a ribbon cable to cut up. b:same as a except make a 'break out' board for the other end of the ribbon cable to hook into.
b:same = b: same
oh, you don't have any .1" headers laying around?
depends on how lazy I am :)
suprisingly - no :)
Yes, I did as a matter of fact
one header came with the package, didn't it? (the 26-pin one)
No - not with mine.
I thought you had mentioned that. All I got was the 10 conductor ribon cable
maybe I should email them
it might be because I bought the flashy too
it came with a 12 (?)-pin female connector, and they matched
that could be.
it says nothing on the site that that is what you get.
we have a nice soldering station at work here that I was goint to sit down and see how much luck I had with desoldering some headers.
and go from there
they're not expensive until you pay the shipping -- http://www.jameco.com/webapp/wcs/stores/servlet/ProductDisplay?langId=-1&storeId=10001&catalogId=10001&productId=514571
:) those are only 3 amps max.
* skunkworks is kidding
anyone else like the smell of new circuit boards?
I'm sure it's cancer-causing :-P
I am sure it is ;)
anyway, I want to decide on a "standard pinout" for pluto_servo, because you can only change the pinout by recompiling the firmware
I should look to see if I have any old isa cards that are single sided with headers. would be a lot easier to desolder
I'm not thrilled about it, but I may end up adding a wire from one parport pin to an fpga pin, so that I can reliably reset the device from emc
No problem with that.
jepler: when do you need to reset it?
skunkworks: I want to reset it whenever you exit emc2
ooh it's coffee time here
jepler: would you want to hook a 1k or so ohm resister in series with it? in case the fpga pin gets turned into an ouput?
n/m - the reset is prabably the 'reset' pin
skunkworks: yes, I don't think it can be configured as an output
I can fit 4xPWM 3xQuadrature, 1I, 1O on the main connector; the rest of the inputs on the JTAG connector; and the 4th quadrature and the rest of the outputs on the right-hand connector
some people may be able to use just the main connector
I wonder if I have enough gates left to do some kind of watchdog timer: if the board doesn't get a write in 5ms or so it stops all the PWMs
or if those few gates left are better spent on digital filtering of the index pulse
jepler: watchdog would be VERY cool
jepler: could it just reset itself?
I'll have to think more about that
the FPGA "reset" pin already has a user-controlled IO pin connected to it. That's how the existing "self-reset by command" is supposed to work
that's the big remaining problem for my system
if I go with the "hook reset to parport signal" route I think I'll have to program that pin as an input too
maybe you should ask the manufacturer about the reset problem?
I'm not convinced it's not my fault
I still don't know why we have to do that bizarre dance to get it fully reset
yeah it's peculiar how it gets "some" power from my parport but not yours
but here's my "alternate theory": the timing of your EPP is different than mine, and the FPGA sees an edge of the parallel port "strobe" pin after the reset
that has the effect of starting the programming cycle, and forever after it's 1 bit off and it'll never work
could noise do that?
we should plug it directly into the machine and try that
I've also worked with a length of parport cabld
that cable I brought in my bag the first night
john bolton resigns
hmmm - 1G Lexar JumpDrives, $9.99
SWPadnos: where was that? office max?
reading the finer print, it's $30 minus a $20 rebate
(+/- $0.01 ;) )
well, both MicroCenter and Lexar are reputable companies, but it's still a PITA
the scam is they have MIR because most people forget or don't do them
they also have generic 1G drives that are actually $9.99 each, no rebates
I think MIRs are dishonest/scams no matter what company
they're usually a bit shady
if you are anal enough they're a great deal - -you're getting a better price at the expense of the people who don't send in MIRs or screw it up
hmmm - that reminds me. I should be getting $100 from IBM from the 75GXP class action suit - I wouldn't have expected it to take 18 months
funny that 1/3 of the cost of my 60G drive will now buy a 300G+ drive and leave spare change :)
funny that 1/3 the cost of my 850 meg drive will now buy a 300+G drive and leave spare change
I think I still have one sub-1G hard drive in my house but that machine is no longer in use
ooooohhh - you want to play, huh?
nah I'm sure you'll win
there's a commodore 64 with datasette in a basement closet
I was just going to say, funny that 1/8 the cost of the 10 meg hard drive we bought for our high school BBS would get you a 300G+ drive, and leave spare change
your BBS had a hard drive? Luxury!
oh yeah - even the 1541 floppy drive was more than a HD these days
a real BBS was run from 2 floppies
not 2 floppy drives -- 2 floppies. you swapped them after loading the BBS software
we just got a core duo (not a core 2 duo) and it seems to perform quit well :)
well, we finally raised the money to buy it because RBBS (or TBBS) needed 4 floppies, so if the computer crashed over the weekend, we couldn't bring the BBS up until Monday
you boot with 2, then swap to the "run" pair
jepler: you mean one floppy that you flipped over after booting?
cradek: oh yeah I forgot about that
heh - back when floppy discs actually had some quality
I'm sure I don't have my floppy drive punch anymore
I bet mine's in the box with the 1541 and everything else
I wonder if I can find my Atari stuff still
I even had a converter that would run a real parallel port printer
were there enough I/Os on that left-hand port to make it simple? Or was it something sophisticated?
yeah that's how I ran my huge daisy-wheel
it plugged into the normal printer port (the round thing)
(I don't know how it worked)
huh, it must have been fairly sophisticated then
that port was a clocked serial protocol if I remember correctly
serial -> parallel converter
we had a serial port box for your commadore. I could never make it work.
but it was easier back then - you only had to worry about the error lines, there was no bidirectional stuff to worry about
it was multi-device too -- the first printer was device 4 and the first drive was device 8 (by convention)
so that converter must have had a fair amount of brain in it
see, that's one place where the Atari was better: you actually got to use names (D: or D1:, D2: for disk drives, C: for cassettes, etc)
programming video on the C64 was aPITA as well - the graphics memory was arranged like a bunch of character cells, so the first 8 bytes are the first 8x8 pixel group (in the top left), the next 8 are the second character cell, etc
sounds sort of like how the TI was
I think the graphics memory arrangement was "normal" on the TI
I don't remember though - I only used BASIC on that machine
it only required a modest amount of arithmetic to find the right pixel
same here. has been a while. I remeber having to program the graphics for each character sized spot.
anything was better than the apple II which had some weird permutation of the lines
sure, Y/8*(8 rows) + X/8 *8 + Y mod 8 (all ints, so the division loses the fraction)
jepler: have you thought about soldering wires directly to the fpga to use the extra pins yet? ;)
hmmkm - better to write it as (Y>>3)*320+(X>>3)*8+(Y mod 8)
err - 640, not 320
will this work? http://emergent.unpy.net/files/sandbox/two-pins-driving-reset.png
SWPadnos: commodore graphics were 320 wide
no, 320 - 40 chars wide
sorry - only had half a cup of coffee so far today ;)
the office coffee wasn't so bad today. I remembered to bring milk to dilute it
is that weak pull-up internal oto the reset line?
what's the range of values (out of curiosity)
well I think there will be one on nReset and one on the OC pin (the fpga pin)
let me see if the datasheet says
I probably should have saved that datasheet locally
hmmm - maybe I did
oh good, I did
.. hm this says something about it being pre- and during-configuration, but I thought you got an internal pull-up when you made a pin OC
hmmm - does the datasheet go into the use of an I/O to self-reset the chip?
there's an inherent race condition there
I didn't see anything about it in this datasheet, but the fpga4fun board is set up to work that way
you can possibly configure a pin to output a high when reset, and that'll act as a handshake
looking at page 24
you set the output to 0, and when the clear occurs, the pin gets loaded with a 1, which de-asserts /RESET
is the best power supply for running steppers a simple unregulated design, ie. transformer, bridge rect, and a big cap?
it is, from what I've gathered (but I'm no expert in this field)
i have made a component HAL using serial line
i need the rtai_serial.ko module
where i must insert it ?
Didier: you can use sudo insmod rtai_serial.ko
you can even put that in the .hal file as
loadusr "sudo insmod rtai_serial.ko"
but you need to configure sudo to load things without asking for a password
by itself emc2 can't load the rtai_serial.ko without root privileges, because it might be a security risk (not a module supplied by emc2)
after the line loadrt hal_atmel
i have sudo like this
to make the rtai_serial.ko module loadable by emc's module_helper, insert it in module_whitelist
where is the module_whitelist ?
then use emc_module_helper insert /path/to/rtai_serial.ko to load it
it's in src/module_helper/module_helper.c (hard-coded in the source)
i have found the whitelist, then i try
then if you want it to always be loaded and unloaded, you can change scripts/rtapi.conf.in to liist it
currently we don't have a way to automatically load and unload additional rtai_ modules depending on HAL modules
if [ "@RTAI@" = "3" ] ; then \
MODULES="adeos rtai_hal rtai_ksched rtai_fifos rtai_shm rtai_sem rtai_math"
i have seen it, yes
it's the best way to load this module
the best way is to make something more flexible than what currently is availab.e
this may be the best of the simple ways to do it
ok i will add the module in rtapi.conf.in
that's in addition to listing it in module_helper.c
i need to do the 2 things ?
yes, do both
SWPadnos: yes, the datasheet does say that using a FPGA I/O to perform reset should work:
SRAM configuration elements allow ACEX 1K devices to be reconfigured in-circuit by loading new configuration data into the device. Real-time reconfiguration is performed by forcing the device into command mode with a device pin, loading different configuration data, re-initializing the device, and resuming user-mode operation.
I saw that paragraph, but didn't fully analyze it
it doesn't say that a direct connection between an I/O pin and the reset pin is sufficient to "force the device into command mode", or that it's done from an internal drive - it could also refer to a supervisory microcontroller that resets the device and loads a new config
it seemed ambiguous to me
you think that if RESET is not low long enough, it will leave the device in an inconsistent state
that's one possibility that my critical (pessimistic/paranoid) engineering mind came up with :)
the reset has to propagate across the device, so you should connect the reset input to the furthest output (all the way across the die), but the easiest thing to do is to connect the closest pin ...
euh, i have the message "emc_module_helper invalid usage args...
Didier: does the error message list the module name you added?
... insert @MODPATH_rtai_serial@
I guess you'll also have to amend configure.in to find the module path for rtai_serial
that is about line 884 in my copy of configure.in -- search for MODPATH_adeos and look nearby
i look for...
found the line
why the command loadusr "sudo insmod..." is bad ?
because of the remove ?
because a proper sudo setup will ask you for a password
yes, and the remove is a problem too then
There might be some debate on what is "proper", but the default on ubuntu is to require a password to be entered.
i have changed : rtapi.conf.in, module_helper.c, and configure.in
i have configured, maked...
you have to run "autoconf" after changing configure.in
jepler: these cards come with a burned cd labeled avr porgrammer, eprom emulator, motorola programmer, and avr jtag
skunkworks: the pci-8255?
these are quite the cards by appearence. and have weight to them :)
still have my mailing address?
yep - I will send one out today.
hey is it time for free christmas hardware?
oh - nothing is free ;)
cradek: I promised him a HAL driver for it
I figure - one more i/o option for emc2
is it several 8255s?
rtai_serial isn't mounted !
i will try manually
start emc, insmod ,loadrt hal_
on my ubuntu5.10 machine with realtime kernel, it doens't look like rtai_serial is available
manually, it's ok
i will see that tomorrow
thanks for all
ignoring this issue of whether the FPGA can reset itself with an output pin tied to the reset input, will this scheme with one push-pull and one OC work to pull the pin low from either output?
I would think so - as long as the resister from the printer port is low enough to stear the pull up and not to large to cause current issues with the current reset circuit.
not to small to cause current issues with the existing reset circuit
jepler: this cd just seems to be a generic programmer cd. There seems to be no examples of this card on it. No manual also. I could try to email the company - to see if they have anything else
skunkworks: huh, I wonder if it's the wrong CD or something
I was thinking that also. I will send off a email and see.
not much. trying to decide if I can remove .1 headers from a mother board
or for you 2.54mm headers ;)
digikey is your friend. S2011, I think
gold plated, 2 row (S1011 is the single-row version), cheap, low cost shipping, and it should be overnight since they're ~100 miles from you
but I want it now!! :)
S2011E-36-ND, $2.74 - that's the 72-pin (2x36) version
tomorrow is now ;)
they used to be a little less expensive, but these are the RoHS compliant version
then I have to figure out what else I need... You might be right again.
those break appart - right?
yes, but you need to be a little careful to not break the plastic around the end contacts
bah, crappy rohs
I have a ton of razor blades laying around just for that purpose :)
I usually hold the header in one hand, and break off the pins I want with needle-nose pliers
or use a small wire cutter
O.o have you seen the post about needing a password on the EMC users list?
wtf is that?
probably should give him the activation code also
mostlikely wants to get into the cvs
the password is 0xDEADBEEF
or 0xDEADC0DE, for the master password
cvs doesn't need pass
deadc0de you mean
I like DEADBEEF
in fact, I think I'm hungry
have you ever gotten that?
I would think you would if you where a developer.
as in, a hex error message that *is* DEADC0DE
no, I haven't
just ate my left-overs from adwardos pizza wagan. (big cheese burger)
looks like the owner of my work is going to get wildblue. That should be interesting.
high speed, but very high latency
right - I had directway for a while. Worked ok - just don't expect to play fps games.
my pings would average 1 second
yep. a friend (~15 miles from me) has wildblue, and he gets pings in the 800-1000 ms range
but it was a lot better than dialup for downloading (to a point - you where capped)
he's managed to get hundreds of megs of downloads, like SolidWorks updates and the like
... at reasonable speeds. though he may have mentioned that it was "ultra-fast" at first, then settled down to a slower rate after 10-20 M
ok. time to run errands. bbl
do you know what the limit is? directway had 168mb every 4 hours (little more complicated than that but..)
ok - looks like it is a lot better than directway.
my local ISP must be one of the few good ones. No capping bullshit, they tell you up front your BW quota and how much it will cost you after that
we are talking sattelite - limited bandwidth - they have to do some sort of limiting.
in practice I never hit the transfer quota even when I torrent heaviliy (see an ubuntu CD for a week or something)
I don't know if I have ever had issues here at work. We transfer a lot of art files
that's really bad
jepler: do you want me to send the board anyways today - and hope that I can email/ftp the rest of it to you?
oh I missed that these were sattelite systems
skunkworks: yeah go ahead with the board
then again, sattelite
skunkworks: I am pretty sure I already have the programming info I need (the thai pdf for the board plus a generic 8255 datasheet)
or satellite, even
I hope I get it this week -- if I do I bet the driver'll be in decent shape by monday
what board is this?
I saw that
[19:22:00] <jepler> http://futurlec.com/PCI8255.shtml
3 8255s on the PCI bus, memory-mapped I/O
in case any of you read thai (I don't): http://www.etteam.com/download/10PC_INTERFACE/1008/Man_ET_PCI8255_V3.pdf
but page 7 is pretty self-explanatory
jepler: are you working on that board and the plutu-p simultaneously?
Lerneaen_Hydra: It seems like I'm always working on more than one project
haha, just as usual
in various stages of completion and a few on the backburner
actually, I think I am close to finishing off the pluto_servo project, and i don't think I'm going to start right away on pluto_stepper
yeah very different from ASCII
"Internal registers and the PIB address bus of the Tiger320 can be accessed either by I/O or
I suppose I'll benchmark and see if one or the other has higher performance
I suspect memory might, because most modern processors have a write buffer that can contain many transactions
emc can do dma?
nothing in emc uses dma yet
jepler: how will the end user flash the pluto-p in linux?
is it even possible?
Lerneaen_Hydra: the HAL driver uploads the firmware every time
is it flash based or something else?
Lerneaen_Hydra: this device doesn't have flash or EEPROM, the configuration is stored in SRAM
didn't the pluto-p need a windows based environment?
so uploading it every time is the simplest way to do it
oh, I see
skunkworks: are the 8255 really socketed like in this photo?
cradek: yes - socketed
the free development tools only run on windows, but the firmware upload procedure was very simple and easy to implement on linux
skunkworks: that's great
cradek: great because you can replace them after you've fried them?
how many did you fry?
jepler: oh, so the binary is uploaded on runtime, and you need windows to compile it?
serial cards always used to have their line drivers/receivers socketed too - not anymore
* anonimasu nods
jepler: is there any way to devel in a linux environment?
Lerneaen_Hydra: you can buy their development environment for Linux, but it is too expensive for a hobbyist
and windows is free?
Lerneaen_Hydra: I had an available license for windows XP, so that wasn't an issue.
but I think you can buy an XP professional license for much less than their linux development environment
no, sorry, that's not what I meant, is the windows dev environment free?
well 50% of everyone has one..
yes, you can download it for free
oh, that's kind of shitty
xp license that is.. probably
windows devkit for free, linux devkit for lots of $
should be the opposite :p
no, they should both be free
they should make their money selling the hardware
* anonimasu nods
altera's main competitor, xilinx, has a free linux dev kit. but I didn't want to spend the time developing my own board, so I bought this altera-based board
is this from the same company that made the pluto-p?
so it's the fpga more than the board
altera makes the FPGA chip, some other guy designed the pluto-p board around one of altera's chips
when I go into full scale production of my FPGA servo and stepper boards I'll make a fresh design with a xilinx chip :-P
jepler: nice. :)
the parts (minus motors) for my entire lathe setup are < $150
the L298 is a nice and cheap dual H bridge
for desktop mills and lathes, it looks like 2A/46V is plenty for an appropriate stepper
cradek: that's probably excluding the PS too?
er, yes, appropriate servo
jepler: yes excluding PS too
jepler: a transformer/cap is plenty good enough if the motors are 24v and the driver handles 40+v
I'm a bit worried about what happens to the L298 if you stall the servos -- that datasheet I was looking at showed no-load current of <<1A, stall current of >10A
10A is way above the maximum rating of the L298, and there's no current limiting in the circuit
did you test the stall current of your motors?
I don't remember
maybe it's even more important than you thought to avoid running into the limits
you'd sure have to check that for a production kit
of course you'll get a following error so you won't run at 10A for long
well it will sure stop right away thanks to the tight FE limits
but certainly you might for a few MS
(megaseconds? did I just say megaseconds?)
yes you did
all your teraseconds are belong tu us
I think it would be easy to build a current limit with an op-amp and maybe a 555 timer
when it hits the current limit force the associated enable pin low for a fixed time
how about just current sense -> comparator -> enable?
I think it would "chop" then
I meant comparator when I said op-amp
but without an off time, won't the L298 just start turning off and on as fast as the comparitor's output can change? after all, the current across the resistor will immediately drop to zero.
yes but doesn't the motor inductance give you a time constant like a stepper's chopper drive?
when ENABLE goes low, all the transistors in the bridge turn off, so there's nothing connected to the sense resistor
it's just a resistor to GND
and to the comparator input
that's why the L297 uses a fixed off time chopper
you've obviously thought it through more
it doesn't mean I'm not mistaken
why not use some RC thing, with a schmitt trigger (slowly charge/discharge the cap through the resistor and trigger when X volts is reached)
instead of the lm555
yes, something like that would work too -- another way to get a longer off
less stuff that can go wrong too ;)
less complexity = good
then again, this is going to connect to a consumer computer ;)
I think you can do something like this: http://emergent.unpy.net/files/sandbox/current-limit.png
(I may have the comparator inputs backwards, and you want to have hysteresis)
yeah looks about right
how much current does the comparator source?
once C1 charges enough, enable goes low but it doesn't go high again until c1 has dischaged through r1+r2
oh, wait, I misread the direction
yeah that will work
1 ohm is a bit little
maybe you meant 10k?
1V/amp is easy to do calculations with though
R2 is the current-sense resistor
you want a very low value so that it doesn't dissipate much power
wait, what does the h-bridge do?
it switches the power for the servo motor
oh, that's what's usually connceted to ground?
like usual then
uh, IIRC an h-bridge only has +Vcc and ground, right?
no +,-,gnd, right?
the "bottom" of the h-bridge is connected to an output called "sense"
if you don't do current sensing, you just connect it to GND
then I think I've got it
if you do current sensing, it's by putting a power resistor between SENSE and GND so you can measure current as voltage
so current will only travel from up and down
I was thinking that when current travels from ground to the negative rail the comp would go all wonky, but that's a non-issue with only a V+
if R2=1 ohm and R1 is a much larger value, then you get 1V/amp. C1 charges at a rate defined by R1/C1 and eventually trips the comparator. The h-bridge turns off, and C1 discharges at the same rate
the hysteresis of the comparator helps define the time the op-amp is turned off for
that's all just details though
you should have less than 1 ohm though :p
10m ohm or so
Lerneaen_Hydra: sense resistor depends on the current usually
but I'm thinking >30 amps here
seen .01 ohms for this purpose
and I don't want 900 watts over the sense resistor ;)
I am looking for .01 ohm resister.
that would be good for about 30A if my math is still ok
skunkworks: I scrapped a few off old DC servos
same here. some .03 iirc
Lerneaen_Hydra: the l298 has a limit of 2A per bridge, so a 1 ohm current sense resistor is not too bad
oh, right :p
4 watts is managable
skunkworks: two in parallel would be .015
2 amps, thats nothing though
which voltage? high?
up to 40 something
cradek's using 24V for hus lathe
hmm, that sounds like a small-scale servo
it's a small lathe
here I was thinking tons of metal and it's only a few kilos
it still kicks the ass of his stepper mill in terms of rapid speeds though
we need some more videos ;)
this is sherline/micromill territory
I like a slow g0 feed though, becuase I'm a wuss even since someone had a semi-large accident in the big, really really fast lathe at school
700mm/min is perfect ;)
it's so small that it doesn't really matter
I think the rapids are 1250 mm/min -- and about 150mm of travel :-P
that wasn't that fast though
I would have expected something like 7m/min
what's the resolution?
1/6000 mm now
what's the repeatability?
If anyone cares, it's portable (can run from a USB Stick), Has AES encryption, and syntax highlighting: http://www.snapfiles.com/get/codebank.html
thought so ;)
[20:37:13] <skunkworks> http://www.linear.com/currentsense/17-fault_sensing.pdf
I would like to do something like page3 'Conventional H-Bridge Current Monitor'
that way the resister outside the h-bridge.
(and I can use jmk's great low impedance traces)
skunkworks: what's that?
jmk re-did the h-bridge that I was working on so that it has the shortest traces possible. really cool.
this green crayon in the middle. http://www.petegoldlust.com/carvedcrayons.html
how would one create that on the mill? A axis, moving a long... ball nose endmill, past the centre axis line?
to get that double helix?
oops, bottom thumbnail, where it shows an array of crayons
g0x0 g0a180 g1x3a3780
very tempting to try it
or just run the mill all the way through, and do it in one pass :)
I've done one similar to that
not below the centerline though
has everyone seen the pencils too?
know why I'm asking about the centre line? because it's a double helix.
the one full of circles is cool too.
the real problem would be cutting it without melting or breaking it
I saw one at some place with four enterances, looked really nice
yeah, saw the pencils ones a while back.
I wonder how they chucked the items.
A-L-P-H-A: I'm sure those were cut by hand
outsource them to children in india
cradek, you seriously think so? you're probably right, because of the spacing isn't perfect...
A-L-P-H-A: sure, it's art, not machining
A-L-P-H-A: those are manual
A-L-P-H-A: we're the only ones who would think of it in terms of CNC
I want machinable art.
yeah, it's easy and you can make many :D
fun to watch too
[21:11:01] <eholmgren> http://www.infofreako.com/jad/enpitsu-e.html
not so much fun to clean pu.
that's all hand carved
it's WAX... it'll stick to EVREYTHING.
use a 1/8th ballnose to carve out the crayon.
maybe a 1/16, and do the yellow one with what looks almost like a honey comb stucture.
[21:12:17] <alex_joni> http://www.infofreako.com/jad/pencil/1rasen6ju-e.html
[21:13:19] <alex_joni> http://www.infofreako.com/jad/pencil/7nisou-e.html
[21:13:25] <SWPadnos> http://www.infofreako.com/jad/pencil/15bjoint-e.html
I wish those pictures were better
yeah - they do look a bit like cell phone photos
[21:14:52] <alex_joni> http://www.infofreako.com/jad/pencil/10gyakuset-e.html
ack, twin spirals would be a PITA to machine
cradek: all pre-1995
SWPadnos: they are almost 10 yrs old
this one says 97
looks to be best made by splitting a real pencil, and gluing it back together after machining.
95 was a guess :)
say a 1991 one
a real pita
some are 15 years old!
that's older than you!
er - oops ;)
SWPadnos. I'm getting old. I'm 27.
I think not ;)
I may be wet behind the ears, but not that wet :p
I'm getting old, you guys should be dead, and mumifying.
[21:17:04] <Lerneaen_Hydra> http://www.infofreako.com/jad/pencil/2kikko-e.html
Lerneaen_Hydra: try the twin double spiral
03jepler 07HEAD * 10emc2/src/hal/components/ (freqgen.c stepgen.c): when the number of steps per second is not attainable, log a message (once per step generator)
jepler: is that once per stepgen and initialisation of stepgen, and not per attempted step?
once per stepgen .. for instance, if you say 'loadrt stepgen step_type=0,0,0' you'll get at most 3 such messages in that run of emc2
I understand it as: if it ever encounters a problem it logs it once, and shuts up from then on
no 100mb error logs
unfortunately, this is dependent on the order in which setp's are done (adn I'm not sure how to fix that)
oh - maybe not, if all the setp's are done before "start"
SWPadnos: jepler says the tests are run when the stepgens are enabled
for emc that usually means "machine on"
the setp's are even before realtime start iirc
ah - ok
but I've been wrong before :)
it would probably be OK to print once per enable attempt, actually
try to turn machine on, do a few manual setp/sets, try to turn machine on ...
SWPadnos: I suspect few will try it that way
probably true, since it seems most people want a pre-made config for their machine ;)
I got my order out to digikey.
good - you should have it tomorrow or Wednesday :)
I was looking for a decent diff amp for current sensing - but could not find any that I could buy ;)
I used to get things overnight (when orderedgground), but these days, it doesn't seem to be that way
how big is a so package?
about this big: ||
:) i mean - hand solderable?
or is it wide SOIC?
skunkworks: everything is hand solderable
the pin spacing should be 0.050, so they're pretty easy if you have a relatively fine-tipped soldering iron
given enough time and nerve
even bga.. but you need a hotair gun for that
you don't even need the 1/64 " tip for those ;)
was looking at this for s and g's.. http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1009,C1126,P7521,D5087
but I would have to buy 700 from digikey
any reason you can't just use an op-amp?
but I would be a little fuzzy on the rest of the circuit. technically I would need a single ended output.
"s and g's" ??
I would like to put the sense resister external to the h-bridge.... In series with the motor. if that makes sense
shits and giggles
was thinking it might be some arcane motor drive term
whatcha trying to build skunk?
like the last page here http://www.linear.com/currentsense/14-motors_and_inductive_loads.pdf
mdynac: rather large h-bridge... but no current limit yet (and I am not sure if I need it yet). http://www.electronicsam.com/images/KandT/servostart/almost.JPG
you should have it - there are things like short-circuits in the real world
SWPLinux: right... I don't know if I live in the real world though
SWPadnos: is it making sense what I am trying to do?
I don't know either - you could be a delusion created by the evil genius
I will definatly do my best to blow this one up ;)
how many amps wiil the bridge supply?
tops right now is 20 amps continuous. the mosfets are rated at 44a but that is not possible unless they are dipped in liqid nitrogen.
I will probably try to find some better ones once I proof this out.
what schemo are you using?
mdynac: do you have eagle?
[21:59:41] <skunkworks> http://www.electronicsam.com/images/KandT/servostart/schem/
placing the sense resistor inside the bridge is nice, too bad about it requiring a special op-amp
it is more for layout than actual componants. the drivers are ir2111
jepler: you mean outside?
skunkworks: no, inside like in the last PDF you posted
ah - I see what you mean
you can measure the current even while the whole bridge is turned off
measuring it at the low side, outside the bridge, sure is easy
there are small PCB-mount isolated DC:DC converters for not too much money (like $7-12)
you can use one of those, plus an optoisolator or similar (on the output of the current sense circuit) to get overcurrent back to the low side
It just makes it so I don't have to fool with the power traces. and I think it is pretty neat.
you could do something like have a sense resistor on both of the bottom side transistors, and use the higher of the two (or the sum, perhaps)
i have an emc2-pre2.2-cvs-head question......
uh-oh - that sounds specific ;)
what's up with the index pulses?
for homing or threading?
I don't think that code has changed in a long time, and I don't know if it's been tested.
it works fine on 2.0.4
but not in HEAD?
it is commented out on the pre2.2
it crashes until i comment them out
do you mean you had to comment out the lines in your .hal files, or that something has been commented out in the souce code?
can you post the ini'/hal files you're using, or are the stock config?
if there's an error, can you put it in pastebin or something?
when the motenc config was done, canonical encoders weren't defined
well it came commented out....so uncommentted then and crash.....no such signal or something.....
mdynac: it might be that some names have changed
#linksp Xindex <= motenc.0.enc-00-index
#linksp Xindex => axis.0.index-pulse-in
I presume you are talking about those
I think it is this change to the motenc hal component: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/hal/drivers/hal_motenc.c#rev1.11
yeah in the ini fie?
in the .hal file
those need to be axis.0.index-enable
er, no, that's not it -- that's before 2.0.x was released
jepler: I'm sure the motion controller got new pin names
the motenc config is probably older than that
whatever the actual problem is, this should be on the UpdatingConfigs page
once we figure it out
hmmm the 2.0.4 emc2 index pulse works jut fine....
* alex_joni agrees
mdynac: can you be our eyes and hands for a while?
we don't have a motenc lying around
at least I don't
i must get this machine running with the M52 command,
backup your config and other usefull motenc related stuff
yes it was the motion controller
I'll update the wiki page
mdynac: next, open the file motenc_motion.hal
and change axis.0.index-pulse-in to axis.0.index-enable
same for axis.1.index-pulse-in to axis.1.index-enable
and for axis.2.index-pulse-in to axis.2.index-enable
jepler: this is a backport for 2.1
alex_joni: I will fix the sample configs in HEAD and the 2.1 branch
jepler: good, I'd do it myself.. but can only do it tomorrow
but lets wait and see if there's nothing else needed
well er my machine is at work....not here, nor can i xchat at work yet.......a big problem i am trying to remedy, but the sysadmin guy has not shown up to install xchat for me....
mdynac: ok, don't sweat it
it's probably closer than it is now .. please report back when you know more :)
03jepler 07HEAD * 10emc2/configs/motenc/motenc_motion.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07HEAD * 10emc2/configs/ppmc/ppmc_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07HEAD * 10emc2/configs/stg/stg_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
but i will change the file tomorrow morning and check it out
(if you want to print it out, this may work better: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/motenc/motenc_io.hal.diff?r1=1.4;r2=1.5)
03jepler 07HEAD * 10emc2/configs/vti/vti_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07HEAD * 10emc2/configs/univpwm/univpwm_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07HEAD * 10emc2/configs/m5i20/m5i20_motion.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
er, that's not the right change -- hold on
03jepler 07HEAD * 10emc2/configs/univstep/univstep_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
that's a different diff
why is that now index-enable (vs something like index-in)?
[22:18:27] <jepler> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/motenc/motenc_motion.hal.diff?r1=1.5;r2=1.6
here's the right change: ^^
SWPadnos: that's the defined name in the canonical encoder interface
it's a two way pin
and the name is more descriptive
ah - ok
motion sets it high when index can reset the counter, driver sets it low when it has reset the counter
ok - forgot about the two-way thing
I was thinking it was the actual index output pin, but I guess that isn't there any more
so i just edit the hal file and i get my index pulses back?
oops - bbiab, my turn to cook dinner :)
mdynac: yes, hopefully so
mdynac: if nothing else is still wrong :)
this is great news.....i was so close to running the machine under adaptive voltage control today.........
hmmm.. might be that the motenc driver needs to be fixed
* alex_joni is reading through it..
alex , just how bad is it?
mdynac: I think you also need to change motenc.0.enc-00-index to motenc.0.enc-00-latch-index
"hopefully" that will make it work properly
but it's hard to say by just reading code and comments :(
it would be best if you would be by the machine
I "think" that you also need to connect an aditional pin to that signal
true, hoopefully tomorrow i can get xchat installed at work.....
anyone else looking at the hal_motenc.c?
alex_joni: no, I was just blindly fixing the axis.N pin name, not worrying about the other ...
jepler: maybe if you feel like it .. 2 sets of eyes might see better
03jepler 07v2_1_branch * 10emc2/configs/univstep/univstep_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07v2_1_branch * 10emc2/configs/m5i20/m5i20_motion.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07v2_1_branch * 10emc2/configs/stg/stg_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07v2_1_branch * 10emc2/configs/univpwm/univpwm_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07v2_1_branch * 10emc2/configs/ppmc/ppmc_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07v2_1_branch * 10emc2/configs/vti/vti_io.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this
03jepler 07v2_1_branch * 10emc2/configs/motenc/motenc_motion.hal: some time ago, the pin index-pulse-in was renamed to index-enable. Update the sample configuration files to reflect this