#linuxcnc | Logs for 2012-02-01

Back
[04:33:19] <Loetmichel> mornin'
[04:42:59] <Loetmichel> anyone knows a Zen Tooolworks DIY CNC? some ppl i have met have one and i want to make them a new mount for the dremel/proxxon
[04:43:16] <Loetmichel> can someone gice me the distance of the mointing screws?
[04:43:19] <Loetmichel> give
[04:44:32] <Loetmichel> 'cause THAT is a PITA but no moount
[04:44:36] <Loetmichel> http://www.hackerspace-ffm.de/wiki/index.php?title=Datei:Foto1519.jpg
[07:22:57] <mrsun> think a quick change holder for the lathe would work if it was made in alu ? :)
[07:24:57] <jdhNC> why AL?
[07:25:55] <mrsun> cause thats what i know my mill can do :P
[07:26:02] <mrsun> havent tried steel tbh =)
[07:30:36] <archivist> I dont think ally would be strong enough
[07:34:10] <mrsun> problem is that the toolpost on the lathe is to high for my tools :(
[07:34:17] <mrsun> cant get them to centerline with the work
[07:34:21] <mrsun> stupid lathe
[07:46:51] <jdhNC> get smaller tools
[07:47:02] <jdhNC> shave the toolpost?
[08:00:39] <shade88> Hello. How i can to increase steplenght and stephold above 500 000 ns?
[08:01:17] <shade88> in stepconf wizard
[08:03:13] <cradek> why do you want that?
[08:07:42] <shade88> because we plan to use our own driver stepper motor. on which we will send signals more than a 1 ms
[08:10:08] <cradek> do you realize that even if you're only half-stepping, you'll only be able to make 1 about rotation per second of the motor with that step length?
[08:14:09] <jdhNC> he's young, has a lot of time left.
[08:15:14] <cradek> aren't we all?
[08:16:22] <shade88> 1 mm per step. acceptable to us
[08:17:06] <cradek> must be a very unusual setup!
[08:17:19] <cradek> if it's just a limit in stepconf, you can post-edit the hal it generates
[08:18:28] <cradek> the stepgen manpage does not mention any limits, so I bet it's just stepconf that's limiting you to reasonable values
[08:18:50] <shade88> well. Thank you. Try do it
[08:18:57] <cradek> welcome
[08:19:01] <skunkworks> are the steppers run with relays?
[08:27:31] <JT-Shop> Jymmm: pong
[08:32:03] <jdhNC> opto-muscular couplers
[08:32:42] <Gromits> Has anyone else followed the thread on isolcpus on the mailing list?
[08:33:39] <JT-Shop> got too confusing for me to follow it
[08:34:06] <Gromits> The crux of it is pretty simple.
[08:34:57] <Gromits> I was just running linuxcnc on my second core and Gene Haskett is right that it really speeds up the machine
[08:35:26] <Gromits> The question is, are you doing something bad but running a process on the core that you have dedicated to the rtai kernel...?
[08:35:34] <Gromits> but==by
[08:39:14] <skunkworks> The few times I have tried it - I didn't see that much of an improvement.
[08:41:22] <pcw_home> Or does isolcpus simply disable one CPU requiring taskset so regain its use?
[08:42:03] <Gromits> Well, that is a good question, but the info on it claims it pegs the rtai kernel to the cpu designated in isolcpus.
[08:42:49] <Gromits> skunkworks: you have tried isolcpus? or taskset?
[08:43:11] <skunkworks> isolcpus
[08:43:17] <Gromits> isolcpus makes my machine much slower, but the latency test is consistently lower
[08:43:53] <skunkworks> it may have only seen a few percent maybe improvement in latency...
[08:44:04] <pcw_home> So something in LinuxCNC automatically uses the isolated CPU? isolcpus by it self just removes a core from the scheduler
[08:44:05] <skunkworks> it seemed to settle faster.
[08:45:12] <Gromits> by the way, Gene claims that latency is low when using taskset to put it on the idle core. That is true for me too, but if I run another process like glxgears (using taskset) the latency is terrible. Worse than just running on the stock machine.
[08:46:31] <pcw_home> glxgears on the LinuxCNC core> you expect that to be terrible
[08:46:50] <Gromits> why?
[08:47:02] <Gromits> nothing else is running there but the latency test...
[08:47:26] <Gromits> and the latency test only using 2% of the core...
[08:47:35] <pcw_home> latency test with taskset glxgears without?
[08:48:19] <Gromits> no, both with - putting them both on the "idle" core
[08:48:31] <pcw_home> sure that would be a mistake
[08:48:39] <Gromits> why?
[08:49:04] <Gromits> When running the latency test you are suppose to tax the machine...
[08:49:08] <pcw_home> Because you want one core free for LinuxCNC
[08:49:24] <pcw_home> and only LinuxCNC
[08:50:00] <Gromits> Yes, in operation mode... but I am talking about running the latency test...
[08:50:36] <Gromits> Not WHILE linuxcnc is running - but before you run it, to understand the latency of the hardware
[08:51:36] <pcw_home> So if you run the latency test with taskset (to use the isolated CPU core) and everything else on the other core whats the latency?
[08:53:29] <Gromits> I don't know. I would have to use taskset for ALL processes. That would be extremely difficult to do...
[08:54:11] <Gromits> I have used it to put glxgears on the core with the latency test which was telling enough...
[08:55:20] <pcw_home> But thats exactly what you are trying to avoid
[08:56:21] <pcw_home> and what gene avoids with his taskset script AFAIK
[08:56:27] <Gromits> in operation, yes. I would never do that.
[08:56:51] <Gromits> In operation, I would just use taskset to run linuxcnc on the idle core, and that is all.
[08:57:42] <Gromits> The issue I am raising, is that typically with a new system you run the latency-test to understand how fast the base thread can be set to.
[08:58:05] <Gromits> If I do the same thing, but using taskset to put the latency-test on the idle core, then the numbers look great.
[08:58:50] <pcw_home> right but unless you run the latency test in the same CPU environment as LinuxCNC its results dont apply
[08:58:51] <Gromits> BUT - if I run another process on that core as well (which is something you should do while running the latency-test normally), the latency numbers are bad
[08:59:18] <Gromits> But that is exactly what I am doing by using taskset to put it on that core.
[09:00:28] <pcw_home> which seems silly since that's not the final running environment so not representative of LinuxCNCs performance
[09:01:07] <Gromits> You mean running glxgears (or anything) as well? I agree.
[09:02:30] <Gromits> But if the latency-test is ever valid, then why wouldn't it be valid using taskset?
[09:02:40] <pcw_home> running glxgears is fine, deliberately running it on your carefully isolated CPU makes no sense to me since that would never happen in a running system
[09:03:44] <Gromits> I agree - since we have an isolated core - there should be nothing else there beside linux cnc...
[09:04:35] <Gromits> I was just asking the question as to why latency on the second core is worse than the latency on the first - or both?
[09:07:15] <pcw_home> Dont know. maybe if LinuxCNC "owns" one CPU, its core threads/processes never get flushed from the CPU cache
[09:08:23] <Gromits> hmm, might be possible...
[09:08:54] <Gromits> I wish someone who understood the rtai kernel and the scheduler would weigh in. There is just a bunch of speculation and random trials.
[09:09:00] <pcw_home> skunkworks did you try Genes taskset trick with isolcpus?
[09:09:14] <skunkworks> no - not yet
[09:10:13] <pcw_home> I would but our test machine is just single core
[09:10:39] <pcw_home> I should upgrade to a better dumpster machine
[09:11:31] <cradek> I think our code asks rtai to put all the *realtime* threads on the highest cpu
[09:11:43] <cradek> my understanding is that if you isolate that one, you'll get better latency numbers
[09:12:24] <cradek> I am not sure if this is the setup gene is trying to get. There's confusion/disagreement floating around about whether the nonrealtime userland parts should also be on that cpu
[09:13:28] <Gromits> that makes sense. Isolate it so nothing runs on it by the default scheduler, then tell the kernel to puts specific threads there...
[09:13:51] <cradek> yes I think that is exactly what happens
[09:14:08] <pcw_home> I wonder if that has been tested, seems like Gene got a huge improvement
[09:14:10] <pcw_home> (but maybe its because having the non RTAI parts of LinuxCNC running on the same core helps somehow)
[09:14:14] <cradek> someone even posted the chunk of code
[09:14:46] <cradek> pcw_home: I've done it on my dual P3-1000, and got much-improved latency numbers, but that's been ages ago
[09:15:12] <Gromits> I just tried it as well, and I can concur with Gene - there is a huge improvement of the responsiveness of my machine
[09:15:12] <cradek> I'm not even sure it was the current lucid kernel
[09:15:19] <pcw_home> isolcpus or isolcpus+taskset?
[09:15:26] <cradek> Gromits: define "it"
[09:15:27] <Gromits> isolcpus and taskset
[09:15:42] <Gromits> I have had isolcpus on for a long time.
[09:16:14] <Gromits> When I did it, the latency-test numbers improved quite a bit
[09:16:15] <Gromits> However, my machine has been quite slow
[09:16:46] <Gromits> I have just lived with it. Several people in the shop have commented on how sluggish the machine is.
[09:17:13] <pcw_home> what Motherboard?
[09:17:33] <Gromits> Now, if I use taskset to move linuxcnc to the idle core, everything - both linux cnc AND all other random things I am doing are much much snappier.
[09:17:34] <cradek> what's your base thread? for any speed of machine there's a base thread setting that'll make it suck.
[09:17:40] <Gromits> D510MO
[09:17:59] <Gromits> I have no base thread - only a servo thread as I am using 7i43
[09:18:17] <cradek> huh that shouldn't be too sucky then (although it's a very low power machine)
[09:19:00] <Gromits> Before doing isolcpus the machine was fine - not the fastest - but not noticably poky either
[09:19:15] <Gromits> taking away a core had an obvious impact
[09:20:20] <Gromits> we were doing step gen on the MB (at the very beginning) which is why we cared about latency
[09:21:04] <Gromits> Obviously I care less (not in the snide way :-) now that the Mesa card is doing step gen...
[09:23:47] <cradek> what's your servo thread rate then? do you have a huge ladder or complex hal or anything like that?
[09:24:42] <Gromits> servo thread is default, I have tried to set it down below 750,000 in the past and always gotten realtime errors. i will try it again the on the other core and see if that changes
[09:25:28] <Gromits> We have a little bit of hal and a bunch of python, but not an outrageously large amount. no ladder.
[09:26:45] <skunkworks> I have also not been able to set the base thread much less than 1ms. I think we figured it was the ladder - but I have not investigated it too much.
[09:28:49] <pcw_home> I current have a 6 KHz servo thread for our testing. pretty sure CL is a time hog
[09:28:50] <skunkworks> the latency test on the k&t I think is between 10000 and 15000
[09:30:30] <Gromits> well, perhaps a small improvement in the servo thread. Looks like I can set it to 600000, but not lower without getting the realtime error message...
[09:31:48] <skunkworks> I wonder if you could put the ladder in a 1ms thread and made the 'base' thread faster...
[09:34:01] <cradek> sure, you could put ladder in a slower thread.
[09:34:11] <pcw_home> doesn't that mean the the CL thread would be interrupted by the servo thread?
[09:34:12] <pcw_home> (not sure if that a problem or not)
[09:34:24] <cradek> yes I think it means that
[09:34:33] <cradek> I agree with not being sure if that's a problem
[09:35:41] <cradek> if you could separate reading/writing encoders/dacs (for servo thread) and for reading/writing gpio (for slower io thread) you'd surely be fine
[09:35:48] <cradek> not sure if that's possible, though
[09:38:50] <shade88> i want to change file hal_parport.c. and i added #include<stdio.h> into then. but when i execute make command, i have exeption "no such file or directory". what should i do, to make it work?
[09:38:58] <skunkworks> what is odd - the 'scan time' within the cl editor - show to be pretty fast.. like <50us if I recall correctly
[09:50:07] <cradek> shade88: realtime modules run in kernel space, where you can't use stdio or other library calls.
[09:53:43] <shade88> <cradek> is there a way to write to a file the values are sent to the parallel port?
[09:54:20] <cradek> file io is also not possible
[09:54:32] <cradek> use halscope instead
[09:56:27] <shade88> i go read about halscope. thank you
[10:02:33] <skunkworks> what is odd - the 'scan time' within the cl editor - show to be pretty fast.. like <50us if I recall correctly
[10:02:37] <skunkworks> heh
[10:02:39] <skunkworks> oops
[10:13:48] <Gromits> psha: hi
[10:17:40] <psha> hi
[10:17:51] <Gromits> i have a glade question: should the "Layout" Container in glade work?
[10:18:09] <psha> what's layout container? :)
[10:18:45] <psha> don't know )
[10:18:46] <Gromits> you know how there are vbox and hbox?
[10:18:49] <psha> i've not used it
[10:18:53] <Gromits> really?
[10:19:04] <psha> i've used *box, fixed and table
[10:19:07] <psha> probably others too
[10:19:29] <Gromits> hmm, one of the example I came across used it...
[10:20:04] <psha> it looks like 'fixed'
[10:20:17] <psha> probably it's mhaberler's example
[10:20:25] <Gromits> yes, may be.
[10:20:39] <Gromits> it looks like fixed but has vert and horiz arrows next to it
[10:20:48] <psha> scrollable fixed?
[10:21:20] <Gromits> perhaps
[10:21:31] <Gromits> whenever I use it, no panel appears in axis
[10:22:41] <Gromits> when I hover mouse over it, it claims to be "Layout"...
[10:31:20] <psha> probably it shrinks to 1x1 size
[10:31:32] <Gromits> ah
[10:40:43] <Gromits> yep, that was it. thanks!
[10:45:01] <Gromits> i think it was shrinking to 0x0, btw :-)
[12:02:36] <Jymmm> JT-Shop: Nitrogen regulators, are there two kinds; pressure and flow? Are they interchangable at all, like from flow to pressure?
[12:04:21] <jdhNC> regulators regulate pressure. flow is determined by pressure and orifice size
[12:06:21] <Jymmm> Well, this is CF/m not PSI... http://www.harborfreight.com/regulator-gauge-94841.html
[12:08:10] <Jymmm> err CF/hr
[12:08:33] <jdhNC> there is a restricted orifice of known size inside the output barb. The low side gauge is just labeled in flow units for that orifice size
[12:08:56] <jdhNC> obviously, the gauge is only sensing pressure, not flow.
[12:09:27] <Jymmm> So, if I remove the orifice and replace the gauge with a PSI, it'll work?
[12:09:44] <jdhNC> yes
[12:10:11] <Jymmm> will it be accurate?
[12:10:28] <jdhNC> how accurate do you need it to be.
[12:10:47] <archivist> and accurate at what pressure or flow
[12:10:52] <jdhNC> a cheap gauge is a cheap gauge. It will be 'accurate' up util the flow is restricted at the barb
[12:11:09] <Jymmm> +- 2psi
[12:11:16] <Jymmm> gove or take
[12:11:20] <Jymmm> give or take
[12:11:32] <jdhNC> at what pressure?
[12:11:44] <Jymmm> 10 - 60 psi
[12:12:52] <jdhNC> still depends on flow rate, but probably
[12:14:07] <jdhNC> that's just a single stage reg though, it will change based on input pressure
[12:14:19] <fragalot> how long ago did the #emc channel die again? lol
[12:14:31] <fragalot> only just noticed >.>
[12:14:32] <Jymmm> and a dual does what exactly?
[12:14:46] <Jymmm> fragalot: It's still alive, sorta.
[12:15:01] <jdhNC> first one regulates HP down to something reasonable, the second only has to deal with mostly regulated
[12:15:22] <fragalot> Jymmm: meh :P
[12:15:23] <Jymmm> Horse Power?
[12:15:31] <jdhNC> high pressure
[12:15:35] <Jymmm> high press
[12:15:41] <Jymmm> took me a while =)
[12:16:18] <Jymmm> jdhNC: Are there dual reg that will still allow 0-250 psi ?
[12:16:59] <jdhNC> should be, don't recall seeing one that went that high though
[12:17:34] <jdhNC> you can use two single stage regs in series. Second one doesn't need to be high pressure
[12:17:38] <Jymmm> You might have to reload this page twice, but is $80 a good price? http://www.grainger.com/Grainger/HARRIS-Medium-Duty-Nitrogen-Regulator-5KZ42
[12:18:01] <jdhNC> or if you don't care about pressure increase over (longish) time, it wont' matter
[12:18:56] <Jymmm> jdhNC: Well, I kinda do, but if it's cheaper just to refill the tank than pay $500 for a dual reg, I'll refill the tank =)
[12:19:18] <jdhNC> I have quite a few regs, never paid more than $30ish
[12:19:49] <Jymmm> Got a spare you wanna sell =)
[12:20:20] <Jymmm> I can only find one used sadly.
[12:20:43] <jdhNC> there should be hundreds on ebay
[12:21:09] <jdhNC> that grainger picture isn't the real gauge, it has the wrong connector
[12:21:37] <Jymmm> Yeah, I'm going by the desc CGA-580
[12:22:04] <Jymmm> Though, I'm not sure what "Purging" means in this context.
[12:22:06] <jdhNC> that looks like a cga540/o2 reg
[12:22:45] <jdhNC> it self-purges when the pressure increases on the downstream side
[12:23:07] <Jymmm> Oh, that sucks
[12:23:18] <jdhNC> it's generally a good thing
[12:23:36] <Jymmm> Well if it's a safety thing, ok. If it's a SOP thing, then no.
[12:23:53] <jdhNC> it just purges until the pressure reaches the pressure the reg is set to
[12:24:29] <jdhNC> otherwise you couldn't set the pressure with the downstream side closed off
[12:24:38] <Jymmm> ah
[12:24:43] <jdhNC> without turning off the supply side on pressure decreases
[12:26:27] <jdhNC> that same reg is on ebay for $90 BIN, even the same pic
[12:26:35] <Jymmm> heh
[12:28:11] <Jymmm> Hmmm, I just realized I also need 12CF/s, fudge.
[12:28:20] <jdhNC> damn
[12:28:41] <Jymmm> 42K CF/hr, fsck me
[12:29:07] <jdhNC> you can't afford that :)
[12:29:36] <Jymmm> I can't afford the TANK, much less the regulator =)
[12:30:14] <jdhNC> I lease 4/each Helium & O2, that's painful enough.
[12:30:54] <Jymmm> Well, if I can get 5s @ 12CF/s, that's a 80lb tank.
[12:31:13] <jdhNC> lb?
[12:31:25] <Jymmm> Nitrogen
[12:31:35] <Jymmm> I have a 40lb tank
[12:32:02] <jdhNC> gas by the lb?
[12:32:03] <Jymmm> err cf
[12:32:15] <Jymmm> I have a 40cf tank
[12:32:18] <Jymmm> sorry
[12:32:38] <jdhNC> I get gas in 300ish cf tanks, normal sized gas bottles
[12:33:15] <Jymmm> I just didn't have the room to storage that size tank
[12:33:36] <jdhNC> it's only a 8.5" tank?
[12:34:35] <Jymmm> Heh, hell who needs a regulator, just crack open the valve @ 2000PSI, I should be able to get the 42K Cf/hr no problem =)
[12:35:49] <Jymmm> jdhNC: And it doesn't fit under my bench.
[12:37:32] <jdhNC> wtf do you need with 700cfm?
[12:38:02] <Jymmm> jdhNC: Fire suppression in my laser engraver
[12:38:23] <jdhNC> I would dispute that number.
[12:38:41] <Jymmm> the laser cabinet is 12CF
[12:39:03] <Jymmm> I doubled that for fudge factor
[12:40:24] <Jymmm> I'm guessing 5 seconds if flooding the compartment would extinguish a flame, though there still may be heat/ambers.
[12:41:04] <Jymmm> That should give enough time to shut down the laser, and stop the blower
[12:41:31] <Jymmm> or at least turn off the blower and close a gate valve
[12:42:24] <Jymmm> jdhNC: make sense?
[12:42:41] <jdhNC> what is going to trigger all this?
[12:43:12] <Jymmm> heat sensor and/or BRB (Big Red Button)
[12:43:52] <jdhNC> get a fire extinguiser, plumb it in to the enclosure
[12:44:01] <Jymmm> Why?
[12:44:08] <jdhNC> it's cheap and easy
[12:44:32] <jdhNC> http://www.fireboy-xintex.com/portable-fire-extinguisher.htm#M
[12:45:06] <Jymmm> Dry extinguishers are caustic, CO2 will promote condensation, and Nitrogen is cheap and I have an article on Nitrogen as a replacement for Halon for fire supression from NIST.
[12:45:47] <jdhNC> those aren't co2
[12:46:07] <Jymmm> No, those are dry, which is caustic
[12:46:22] <jdhNC> Unlike dry chemical systems, these portable extinguishers leave no residue whatsoever.
[12:46:23] <Jymmm> And EXPENSIVE
[12:46:33] <Jymmm> I looked at them from te marine supply
[12:47:27] <Jymmm> http://www.westmarine.com/webapp/wcs/stores/servlet/ProductDisplay?productId=96210&catalogId=10001&langId=-1&storeId=11151&storeNum=50469&subdeptNum=50502&classNum=50504
[12:47:43] <Jymmm> That has a auto / manual release
[12:47:56] <Jymmm> so you can plumb into engine compartment.
[12:48:09] <Jymmm> my bad on the dry / clean part.
[12:49:12] <Jymmm> jdhNC: This is what happens when you go pee while your laser is running.. http://i37.tinypic.com/sqh6r7.jpg
[12:49:29] <jdhNC> cool
[12:49:59] <Jymmm> Yeah, $30K pile of ashes and not covered by her insurance company.
[12:50:48] <Jymmm> She had ran 12 jobs that morning so far, no issues. On the 13th one, POOF!
[12:52:04] <Jymmm> I'm not sure if she used the bathroom or signed for a package, but if was just within a couple of minutes
[12:55:52] <Jymmm> jdhNC: http://www.bfrl.nist.gov/866/HOTWC/HOTWC2006/pubs/R0000296.pdf
[13:04:25] <jdhNC> Extinguishment times vary from 16 to 80
[13:04:31] <jdhNC> seconds
[13:04:46] <Jymmm> for a full room, right?
[13:05:39] <jdhNC> yep, huge room (relatively)
[13:06:17] <Jymmm> He also says he accunts for 32% O2, instead of flooding 100% Nitrogen
[13:06:57] <Jymmm> So I'm hoping I'd get a little better performance even if overkill
[13:07:38] <Jymmm> at least in theory =)
[13:30:18] <A2Sheds> http://reprap.org/wiki/ConductiveMaterials I updated some of the info. Maybe this will inspire someone.
[13:39:38] <mrsun> hmm, drilling int he lathe, the hole becomes conical over 20mm depth ... wth =)
[13:40:04] <mrsun> either the chuck is pointing far far away from the gah whats its called, the part where you put the centers and the drills :P
[13:40:12] <mrsun> or something else is very wrong :P
[13:40:22] <mrsun> (first time using the lathe since i got it :P )
[13:40:25] <jdhNC> tailstock
[13:41:39] <jdhNC> or the bit is wandering
[13:42:42] <mrsun> jdhNC, i center drilled first then drilled ... so dont know how much it can wander =)
[13:45:42] <jdhNC> are you using a regular drill chuck in the tailstock?
[13:45:50] <mrsun> jdhNC, yeah
[13:46:56] <jdhNC> so, fix it :)
[13:47:20] <mrsun> jdhNC, the drill chuck or what? :P
[13:47:25] <Loetmichel> *eyes watering* *SNEEEZE* ... THAT was too much CA... burning eyes and nose :-(
[13:48:02] <jdhNC> the drill chuck, or the headstock to ways alignment or the tailstock to ways alignment
[13:48:48] <mrsun> jdhNC, yeah ... i guess ill have to do some measuring =)
[13:49:20] <Loetmichel> <- is glueing the 12th cable spool ... a few more and my wire stock is sorted and detangled ;-)
[14:43:54] <joe9> i like this irc, has a pretty congenial attitude. glad I stumbled on it.
[14:45:26] <cradek> don't tell anyone that, jerk
[14:53:42] <JT-Shop> lOL
[14:53:45] <JT-Shop> L
[14:54:13] <fragalot> Anyone here have any idea abt the strongest metal glue you can think of?
[14:54:20] <fragalot> currently considering loctite 435
[14:55:09] <fragalot> it's a very small contact surface (4x10mm) & welding isn't an option (it's the tang of a straight razor)
[14:55:41] <jdhNC> I don't think any glue will work well.
[14:56:23] <fragalot> least I can do is try
[14:56:32] <jdhNC> sure
[14:56:42] <mikegg> JB weld!
[14:56:52] <fragalot> mikegg: loctite 435 is stronger than jb weld
[14:56:56] <fragalot> and faster
[14:56:58] <fragalot> and cleaner :P
[14:57:24] <mikegg> blasphemy!
[14:57:30] <fragalot> fact.
[14:57:38] <jdhNC> jb weld isn't so hot for rejoining small pieces, if you could fillet it, it would be good
[14:58:10] <fragalot> i'm trying to preserve the stamp as much as possible
[14:58:15] <jdhNC> in my (completely worthless) opinion, the 435 would be as good as anything else, which is to say, not very good.
[14:58:24] <fragalot> yeah
[14:58:29] <fragalot> well i'll give it a go
[14:58:34] <jdhNC> it will hold
[14:58:41] <jdhNC> right up until it doesn't
[14:58:56] <fragalot> if it fails, i'll spot weld the top & bottom of it with my TIG & then spend ages with a file getting it to look good again
[14:59:01] <jdhNC> heh
[14:59:19] <jdhNC> resistance weld
[14:59:36] <fragalot> trying not to do anything invasive on the text that's stamped into it
[14:59:59] <jdhNC> don't you think it would be good to reduce the need for abortion as much as possible?
[15:00:06] <jdhNC> <wrong channel>
[15:00:10] <bill20r3> so very wrong.
[15:01:12] <mrsun> fragalot, braze it? :)
[15:01:21] <mrsun> ahh no heat? :PO
[15:01:28] <fragalot> mrsun: yeah tht'd kill the temper
[15:01:42] <fragalot> spot welding with the TIG would be as far as i'm willing to go in terms of heat
[15:02:11] <fragalot> I've considered drilling holes in the 2 pieces lengthwise, but... that's not gonna happen with this temper, lol
[15:04:46] <mrsun> from what i can tell the alignment is off by 0.1mm over like 400mm
[15:05:05] <mrsun> have to get a better bar to measure with i think =)
[15:05:55] <fragalot> what are you measuring?
[15:11:47] <mrsun> headstock to ways
[15:12:07] <mrsun> using the average method =)
[15:12:16] <mrsun> tho, i havent checked if the bed is twisted
[15:13:00] <jdhNC> did you check tailstock at various extensions?
[15:13:12] <mrsun> nop
[15:13:32] <mrsun> jdhNC, just did this test fast while i was throwing some wood into the "furnace?" ...
[15:14:21] <jdhNC> furnace? it's 22C outside
[15:15:55] <mrsun> ...
[15:15:57] <mrsun> -10C here
[15:16:26] <fragalot> all of the machines in the shop at work today were frozen solid
[15:16:42] <fragalot> I found out when I fired up a 7HP electric motor that did nothing but hum at me
[15:16:46] <fragalot> :/
[15:17:23] <mrsun> fragalot, haha nice =)
[15:17:27] <mrsun> no heat? :P
[15:17:55] <fragalot> heater didn't work either lol
[15:18:38] <fragalot> it's a portable diesel burner, someone borrowed it a few years ago & apparently it worked fine then
[15:18:47] <fragalot> tried to fire it up today,... nothing
[15:19:05] <fragalot> opened the filter only to find a very thick slurry... so I think they put parafin or something nasty like that in it
[15:19:37] <fragalot> so that thing isn't going to work until it either warms up again or I clean the entire unit out
[15:19:40] <mrsun> fragalot, home heating oil ? :)
[15:19:52] <mrsun> it goes to a slurry when cold
[15:19:57] <fragalot> possibly
[15:27:49] <Jymmm> fragalot: what about one of those propane tank mounted heaters?
[15:29:36] <Jymmm> fragalot: Like this, but without the defective part... http://www.cpsc.gov/cpscpub/prerel/prhtml05/05148.html
[15:43:34] <fragalot> Jymmm: http://www.labeltec.be/files/library/GA_100.jpg we've got a few of these too but none that could heat the shop up properly
[15:43:47] <fragalot> Jymmm: the portable diesel one we have now is basically a vivox on wheels :P
[15:44:06] <Jymmm> fragalot: Too bad I dont know what a "vivox" is =)
[15:44:16] <Jymmm> fragalot: How big is this warehouse?
[15:44:22] <Jymmm> err shop
[15:44:44] <fragalot> 400 square meters with plenty of , er.. "vent holes"
[15:45:03] <Jymmm> fragalot: It sounds like you guys need to leave a heater running 24/7 just to prevent freezing.
[15:45:09] <fragalot> pretty much
[15:45:23] <fragalot> it's really a losing battle anyway
[15:45:33] <fragalot> it's basically a glorified shed
[15:45:54] <Jymmm> Yeah, when a 100,000 BTU heater can't heat the place, you're screwed for the most part =)
[15:46:11] <fragalot> :P
[15:46:28] <Jymmm> fragalot: Have you thought of just moving OUTDOORS, where it's warmer
[15:46:35] <fragalot> I actually did
[15:46:42] <fragalot> did some extreme outdoor welding
[15:46:42] <fragalot> lol
[15:46:51] <fragalot> it was nice & sunny
[15:46:58] <Jymmm> I know my garage is colder right now, than it is outside.
[15:47:10] <Jymmm> weird that way.
[15:47:21] <fragalot> it's -5°C outside right now & my room still has the pilot light on
[15:47:26] <fragalot> good insulation++
[15:47:32] <Jymmm> heh
[15:50:02] <fragalot> Jymmm: in either case - I don't like the type of heater you linked
[15:50:23] <fragalot> it's just infrared, doesn't actually HEAT anything, it just FEELS warm, lol
[15:50:34] <fragalot> switch it off & it's ice cold again
[15:50:35] <Jymmm> Well, isn't that the point?
[15:50:41] <fragalot> it is for people
[15:50:48] <fragalot> not so much for people handling steel all day
[15:51:02] <Jymmm> ah
[15:51:57] <fragalot> still pissed that i broke that razor :'(
[15:52:14] <fragalot> it was worth a good 200 euro before I broke it
[15:52:25] <Jymmm> face razor?
[15:52:29] <fragalot> yes
[15:52:32] <fragalot> straight razor
[15:52:39] <fragalot> ex-mint condition
[15:52:45] <Jymmm> That's one hell of a beard you have there
[15:53:13] <fragalot> broke it while attempting to straighten the tang the same way i've done on the others i've restored
[15:53:18] <fragalot> first one to break
[15:53:40] <fragalot> also the most valueable, go figure, heh
[15:54:06] <fragalot> Well the loctite 435 *SEEMS* to hold up to what it needs
[15:54:12] <fragalot> sofar, atleast
[15:57:10] <Jymmm> Well, sorry to hear that.
[15:59:20] <fragalot> shit happens. :)
[16:00:06] <Jymmm> Yeah, but still.
[16:12:18] <Valen> fragalot: weld it up, grind, heat treat polish QED ;->
[16:13:09] <fragalot> Valen: not going to risk botching the heat treat on a blade that's got it done to perfection
[16:13:34] <Valen> well if its broken, its not that perfect anymore ;-P
[16:13:42] <fragalot> esp. with a 6/8" full hollow grind down to a .12mm thickness
[16:13:50] <fragalot> Valen: the BLADE is still fine :P it's the tang that isn't
[16:14:13] <Jymmm> fragalot: duct tape =)
[16:14:19] <fragalot> I also don't want to ruin the engravings
[16:14:35] <fragalot> so if I do weld, it's going to be on the top & bottom of the tang & just spot welds with a TIG
[16:14:51] <Jymmm> fragalot: and heat sink the blade itself?
[16:14:54] <fragalot> all I have to do then is match the pattern of the .. ribbly bits
[16:14:55] <fragalot> Jymmm: yeah
[16:15:02] <fragalot> have it sit in oil or water while i tack it
[16:15:08] <Nick001-Shop> Using apt-get update at the buildbot- will it just get updates and bugfixes or will it change the ini files and such?
[16:15:24] <fragalot> but that's if I break the glue bond when stropping it tomorrow
[16:15:31] <fragalot> if it survives that, i'm golden
[16:15:40] <Jymmm> Nick001-Shop: It is best to BACKUP everything before doing an upgrade.
[16:16:11] <fragalot> anyway off to bed for me
[16:16:12] <fragalot> gnite :)
[16:16:18] <Jymmm> G'Night fragalot
[16:16:37] <Jymmm> fragalot: Pleasant sharp ass razor dreams!
[16:16:53] <Nick001-Shop> I thought it was an update, not upgrade
[16:16:56] <fragalot> 's fine i've got others to keep my cleanly shaven :P
[16:17:08] <fragalot> none as shiny, but they're funtional
[16:17:31] <Jymmm> Nick001-Shop: My bad, you're right. Still good to backup when doing an update too.
[16:18:17] <Nick001-Shop> just the configs or is there other stuff to backup?
[16:18:53] <Jymmm> Nick001-Shop: *I* am not sure.
[16:19:23] <cradek> our apt packages do not edit configs that are in your home directory
[16:19:41] <cradek> you should have backups of course, but not because of the upgrade
[16:21:48] <Jymmm> If only he had a tape robot.... NOT!
[16:23:22] <Nick001-Shop> what else should I backup besides the config section and how do I store that backup on the same drive as EMC2? I guess I'm asking how to have multiple instances of emc2b on the same drive or do I get into trouble with the boot record
[16:23:47] <cradek> make a copy of your configs somewhere other than on that drive
[16:24:02] <cradek> a second copy on the same drive is the poorest type of backup imaginable
[16:24:30] <cradek> disk drives are ephemeral
[16:24:52] <Jymmm> QUESTION.... Do you think there is a legal reason why device come with a power adapter (wal-wort) as a "plug in" instead of being hard wired into the device directly? OR, do you think it's just cheaper for mfg's to buy them in bulk with the connectors already?
[16:25:08] <Nick001-Shop> I know - just trying not to have to screw more stuff into the box
[16:25:30] <cradek> wall warts make the part you can drop in the sink or tub low-voltage
[16:25:30] <jdhNC> my kids think that having the only copy of their stuff on a thumb drive is reasonable.
[16:26:07] <cradek> also they make the part you hold or sit on your lap lighter and cooler
[16:26:38] <jdhNC> they also make teh device region independent
[16:26:39] <Jymmm> cradek: Right, but why not HARD WIRED (soldered to the PCB) instead of a jack? I'm sure the jack adds to the cost when you get into the 100K's quantity.
[16:26:50] <cradek> oh I get your question
[16:27:07] <cradek> well on a laptop that'd suck
[16:27:08] <Nick001-Shop> can the thumb drive be bootable?
[16:27:09] <Jymmm> jdhNC: Most come 110/220 universal, at least voltage wise.
[16:27:27] <cradek> but for a clock radio, sure it could be hardwired, and I think some are
[16:27:35] <jdhNC> imagine a desktop device with a hardwired wall wart
[16:27:49] <cradek> I bet the real reason is they all come from the same factory
[16:27:50] <jdhNC> you could never route the wire with the wart on the end
[16:27:54] <Jymmm> cradek: Sure, but think of things like cordless phone, clock, or even a wireless router.
[16:28:08] <cradek> oh true - you can't put the wart through the hole in the desk
[16:28:34] <Jymmm> Yeah, but that's few and far in between
[16:28:38] <cradek> but is there a legal reason - who knows
[16:28:56] <jdhNC> Third Annual CNC Workshop
[16:29:04] <cradek> you might have an easier time getting something approved/certified if the HV parts are "separate"
[16:29:06] <jdhNC> I don't think you can get to Ann Arbor, MI from here.
[16:29:18] <Jymmm> cradek: Ah, that's true.
[16:29:27] <cradek> like "china already worked all that out for us"
[16:29:56] <cradek> so they can get it approved once, then sell the same products to everyone
[16:30:18] <cradek> I'm talking out my ass, though, I have no idea
[16:30:31] <Jymmm> cradek: Makes sense, I just wasn't sure if it was a liability thing. As onse it's soldered in place it becomes a "whole unit"
[16:30:39] <Jymmm> once
[16:30:53] <cradek> yeah that sure seems like a likely guess to me
[16:31:29] <Jymmm> cradek: I like your idea better though, the certifications just make sense that way.
[16:32:14] <jdhNC> I'd be unhappy if my wall-warted devices were hardwired
[16:32:15] <Jymmm> If one batch of "CE approved" PS are defective, they can just swap them out.
[16:32:36] <jdhNC> I'd be happier if they were all the same size coax with + in inside
[16:36:54] <Valen> modularity would be the reason
[16:37:12] <Valen> manufacturer of psu's makes them by the bazillion
[16:37:48] <Valen> you have 10000 parts your not going to tool up a factory to save you a few cents by not having the jack
[17:21:04] <Jymmm> Valen: soldering two bare wires is far cheaper than a jack I'd think.
[17:22:54] <Jymmm> But, they could use the same device, then include a wall-wort with the correct plug for the region, US, EU, etc.
[17:23:16] <Jymmm> Standards, you gotta love em!
[17:24:03] <Jymmm> How the computer industry got IEC connectors as standard is beyond me.
[17:32:18] <Valen> Jymmm: your presuming the jack costs what you pay in a store for it
[17:32:28] <Valen> it probably costs 5 cents or less
[17:32:44] <Valen> and means that the whole PCB can be mechanically stuffed
[17:33:07] <Valen> IEC connectors are pretty handy really, I see them on more and more electronics these days
[17:44:43] <FinboySlick> A2Sheds: You got your hands on a raspberry Pi yet?
[17:49:44] <A2Sheds> FinboySlick: no plans, I'm actually designing these right now http://elinux.org/Embedded_Open_Modular_Architecture/EOMA-68
[17:51:16] <A2Sheds> using the A10 and AMD APU's http://rhombus-tech.net./allwinner_a10/ http://rhombus-tech.net./amd_g_series/
[17:59:43] <Jymmm> Valen: Even at $0.05/each when you have to buy 200K, it adds up.
[18:00:20] <Valen> yeah, now factor that against getting custom psu's made, and paying somebody to do the soldering
[18:00:45] <Jymmm> Valen: they solder in a jack, or two wires, same thing.
[18:00:57] <Valen> jack is done by the PnP machine
[18:01:07] <Valen> goes through the wave solder bath
[18:01:11] <Jymmm> you sure?
[18:01:49] <Valen> lots of jacks are SMT these days too
[18:01:49] <Valen> (thats why they break lol)
[18:02:07] <Jymmm> oh, ok.
[18:03:34] <Jymmm> Well, with localization of the wall plugs, I can see having a jack in the device. I just wanted to make sue it wasn't a legal liability thing/practice.
[18:03:49] <Jymmm> s/sue/sure/
[18:04:53] <Valen> its probably because they can get the wallplug "ticked" as a seperate item then
[18:05:02] <Valen> regulatory approvals arent cheap
[18:05:32] <FinboySlick> A2Sheds: That would actually fit my needs even better than the Pi.
[18:05:34] <Jymmm> Yeah, it's the plug VS hard wired I was questioning.
[18:06:53] <Jymmm> Valen: But if its' wave soldered and then can swap out wall-wort with localized wall plug, then it makes sense. one device + couple localized version of wall-worts.
[18:10:01] <FinboySlick> A2Sheds: When can I buy? ;)
[18:11:22] <Valen> the problem is because the device is then connected to mains it has to presumably go through mains rated safety
[18:11:29] <Valen> double insulated everything etc etc
[18:12:35] <A2Sheds> FinboySlick: boards start shipping this month, the factory in China is just getting back up after the spring festival
[18:12:50] <FinboySlick> A2Sheds: No onboard storage at all, right?
[18:16:01] <A2Sheds> FinboySlick: 4GB flash
[18:16:21] <FinboySlick> A2Sheds: Wow, then it's absolutely perfect.
[18:16:35] <A2Sheds> the 68 pin connector supports SATA
[18:19:22] <FinboySlick> A2Sheds: Got that. What's to be expected cost-wise?
[18:25:38] <A2Sheds> FinboySlick: once it ramps up $25-35/card with the A10, the AMD APU version will be equivalent to a similar mainboard price
[18:26:42] <FinboySlick> Is there an FPU on the A10?
[18:27:50] <FinboySlick> Nevermind, just figured out you meant Cortex A10 ;)
[18:27:52] <A2Sheds> yes, it's an ARM cortex a8 1.2GHz
[18:28:19] <A2Sheds> http://rhombus-tech.net./allwinner_a10/
[18:28:49] <A2Sheds> Luc Verhagen is releasing open drivers for mali as well
[18:29:19] <A2Sheds> http://www.phoronix.com/scan.php?page=article&item=arm_mali_reverse&num=1
[18:29:52] <FinboySlick> That would be important to my *other* project :)
[18:29:53] <A2Sheds> I might have a version of the card with an FPGA to interface to mesa IO cards
[18:30:44] <FinboySlick> Actually, all I need is something capable of booting linux with 8-10 GPIO and an USB connector.
[18:31:23] <FinboySlick> And an FPU.
[18:31:29] <A2Sheds> the FPGA version would also have a DDR3 NAND drive (ddr3 behind the NAND interface to be used as fast swap)
[18:32:03] <A2Sheds> yes, it covers those requirements
[18:32:25] <A2Sheds> #arm-netbook is that channel
[18:32:33] <FinboySlick> It's simple enough that I'm begining to wonder if I could run it all off an FPGA.
[18:33:00] <FinboySlick> Hmm...
[18:33:14] <Valen> FinboySlick: rasberry pi?
[18:33:34] <A2Sheds> uclinux has been running on xilinx and other FPGA soft cpu's for a while
[18:33:34] <FinboySlick> Valen: Well, it seemed like a cheap platform for what I'm trying to do, but even that has lots of extras that I don't need.
[18:47:07] -card.freenode.net:#linuxcnc- [freenode-info] help freenode weed out clonebots -- please register your IRC nick and auto-identify: http://freenode.net/faq.shtml#nicksetup
[19:04:02] <sliptonic-shop> I'm configuring a new computer to drive my cnc router and laser. I get max jitter numbers around 9700 on the base thread but unless my base period is set at 34000 or greater I get the RTAPI unexpected delay error.
[19:04:36] <sliptonic-shop> This is way worse than the old P4 machine with similar max jitter numbers. Any idea where to start looking?
[19:15:37] <Jymmm> Dont use on-board video, disable things in BIOS you done need (FDD, serial ports, etc)
[19:15:45] <Jymmm> s/done/dont/
[19:15:57] <FinboySlick> And disable hyperthreading if you have it.
[19:31:40] <Valen> sliptonic-shop: its not uncommon to have a single latency spike when starting openGL
[19:31:57] <Valen> run your latency test, then start glx-gears
[19:32:06] <Jymmm> What it is after 24H is what matters.
[19:32:08] <Valen> if you get just the one spike its probably ok
[19:32:24] <Valen> just don't start any openGL apps ;->
[20:22:59] <sliptonic-shop> Valen Jymmm I've left the machine for 24 hours at a stretch and I see pretty consistent numbers around 9700 - 10000. But I get the RTAPI message every time I launch emc and home it. It's just that the latency numbers are so different from what I have to put in for base thread that has me confused.
[20:23:28] <Valen> sliptonic-shop: did you start glx-gears *after* latency test?
[20:23:48] <sliptonic-shop> Yes. 6 or 8 copies.
[20:24:55] <Valen> you got no spike when you started glx?
[20:25:23] <sliptonic-shop> No. Not really.
[20:25:39] <Valen> you can look at latency within axis as well
[20:25:53] <Valen> theres some thing you can perhaps plot
[20:26:05] <Valen> forgot what its called
[20:28:26] <sliptonic-shop> Something in Halscope?
[20:29:12] <Valen> yeah
[20:29:25] <Valen> one of the things you can scope is latency or something like that
[20:29:29] <Valen> i forget its actual name
[20:36:29] <cstop> I have the Lat test going now on an IBM Thinkcenter that I use as my desk web crawler. Sevo thread Max jitter is 290130, Base thread is 304700. No hope for this one ;-)
[20:55:17] <Jymmm> cstop: you shouldn't / I wouldn't rely upon a laptop
[21:01:24] <cstop> The Thinkcenter is a small form factor Box Thinkpads are lap tops
[21:02:53] <Jymmm> All I saw was "IBM Think..." Which technically is Lenovo, not IBM, unless this is like 10 years old =)
[21:03:15] <Jymmm> or guberment surplus
[21:04:26] <cstop> Everything I have is at least ten years old ;-) I like old stuff!
[21:05:01] <cstop> ya know ya can't buy 'em like this any more
[21:10:16] <skunkworks> Is there no i/o on 7I78
[21:15:31] <skunkworks> no
[21:29:28] <cstop> it takes an additional serial card to add I/O
[22:00:19] <pcw_home> or just the other port if 17 TTL I/O is enough
[22:04:02] <mikegg> http://www.reddit.com/r/LinuxCNC/
[22:04:09] <mikegg> anyone else a reddit junkie ?
[22:06:43] <jdhNC> I thought reddit was porn
[22:09:35] <mikegg> eh, there's other stuff too
[22:12:51] <mikegg> pictures of cats, and atheists
[22:17:17] <skunkworks> pcw_home: oh - didn't think of that...
[22:20:33] <pcw_home> or any standard parallel port type breakout if you want isolation/ buffering or whatever
[22:20:59] <skunkworks> pcw_home: that is a very cool product..
[22:23:11] <pcw_home> I should have done it sooner, I was afraid the PCI stuff would be too hard but it was pretty easy
[22:24:26] <Valen> pcw_home: pci stuff?
[22:24:37] <pcw_home> PCI core
[22:24:55] <Valen> inside a fpga?
[22:25:53] <pcw_home> Yes part of hm2 for 5I25 like cards that have no bridge chip
[22:26:27] <Valen> nice
[22:26:41] <Valen> pci has always scared me electrically, and now pci-E just makes it insane
[22:27:10] <pcw_home> Tradeoff is you cannot bootstrap a un-programmed FPGA card except via JTAG
[22:27:40] <Valen> bit of a pain
[22:27:51] <Valen> parallel port jtag wiggler? ;->
[22:29:08] <pcw_home> But the Spartan 6 has multi-boot capability so even if you pull the plug while writing the EEPROM you are OK
[22:29:10] <pcw_home> (yes a parallel port wiggler will do)
[22:29:28] <Valen> multiboot is handy
[22:29:41] <Valen> I have always wanted to play with fpgas
[22:29:53] <Valen> i always seem to want bunches of PWM's
[22:30:05] <pcw_home> should only need JTAG if you make a pretty bad mistake or 2 mistakes in a row
[22:30:06] <Valen> and doing that with micros gets irritating
[22:30:44] <pcw_home> PWM gen is small enough Im sure you can have 1 per pin if you need it
[22:31:00] <Valen> i also want small and cheap
[22:31:17] <Valen> for one app I want 7x PWM's with deadtime
[22:31:21] <Valen> ideally ;->
[22:31:37] <Valen> including the fets and a radio , in the size of a matchbox
[22:32:17] <pcw_home> FPGA are not really cheap unless you have high volumes (and need funny power supplies)
[22:33:11] <Valen> thats always been the barrier to entry for me
[22:33:14] <pcw_home> probably some couple of $ DSP is the best bet
[22:33:30] <pcw_home> some have 8 or more PWMs
[22:33:34] <Valen> they don't seem to have the deadtime generators though
[22:33:48] <Valen> that only seems to come from motor controllers
[22:34:02] <Valen> and they don't often have more than 3 or 4 PWMs
[22:34:19] <pcw_home> 7 PWMs with highside/lowside/deadzone?
[22:34:35] <Valen> that'd probably do i spose lol
[22:34:45] <Valen> I like the micros with FPGA around them, never used one though
[22:35:18] <pcw_home> Some of the PICs we use have 8 PWMs but they may only have 4 if you need h/l/d
[22:35:33] <Valen> yeah, thats what i've been seeing
[22:36:02] <pcw_home> also they have the high res PWM (960 MHz)
[22:36:15] <Valen> mhz? or khz?
[22:36:35] <Valen> I'm only after ~10khz max
[22:36:40] <pcw_home> but its really only good to 480 MHz (thats the basic clock)
[22:37:00] <Valen> you sure your not adding a 0 or an M to those numbers?
[22:37:21] <Valen> 96mhz in a pic i can just about see ;-.
[22:37:24] <Valen> 960mhz?
[22:38:05] <pcw_home> no thats the PWM clock so ~9 bit PWM at 1 MHz 12 bit at 125 KHz
[22:38:44] <Valen> yeah, i just still wouldn't believe any component in a pic running at ~1ghz
[22:38:56] <pcw_home> PLL
[22:39:09] <Valen> i have avr pwm running at 60+ mhz for pwm stuff
[22:39:37] <Valen> got a part number on one of those handy?
[22:39:42] <Valen> got me all curious now
[22:39:58] <pcw_home> deliberately done for Hi res PWM ( older chips ran PWM at CPU clock so 40 MHz or so)
[22:40:00] <pcw_home> TI also does this
[22:40:13] <Valen> oh i've seen it done
[22:40:16] <Valen> i've even used it
[22:40:23] <Valen> that avr had a core clock of 16mhz
[22:40:29] <Valen> pll for pwm clock
[22:41:45] <pcw_home> DSPIC33FJ16GS502
[22:41:57] <pcw_home> for example
[22:45:44] <pcw_home> TI has a 5.5 GHz equivalent PWM frequency (180 ps steps) (and maybe more on some chips)
[22:46:12] <Valen> I'm guessing they cheat somehow
[22:47:13] <Valen> some kind of skewed clock doohickey
[22:47:17] <pcw_home> They do, its some kind of analog delay imterpolator or some such
[22:48:39] <pcw_home> kind of like Xilinx's DLLs which are just a tapped ring oscillator
[22:48:43] <Valen> how many macrocells would a pwm use ?
[22:49:09] <Valen> roughly
[22:49:41] <pcw_home> You can make a really tiny one with SRL16s so you could have 1000s in a decent sizedFPGA
[22:50:18] <Valen> just browsing through the cheap section of them in my suppliers catalog
[22:50:25] <Valen> anyway i must get some work donw
[22:50:33] <Valen> thanks for the heads up
[22:50:38] <pcw_home> the one in HM2 uses a common reference counter and comparators
[22:51:22] <pcw_home> its not too big either so the smaller FPGA will easily fit more than you are likely to need
[22:51:39] <Valen> I wonder about doing the rest of the code in it
[22:51:44] <pcw_home> s/smaller/smallest/
[22:51:47] <Valen> i spose i could make it pretty simple
[22:52:03] <Valen> so just radio > fpga
[22:52:45] <Valen> theres usually some slightly complex stuff involved, and a brushless motor controller, but that shouldn't be too hard
[22:53:17] <pcw_home> you can put maybe 8-10 8 bit 133/150 MHz processors in a very small FPGA
[22:53:20] <pcw_home> (XC6SLX4)
[22:53:49] <Valen> those might be physically small
[22:54:00] <Valen> but pretty spendy and bga?
[22:54:05] <pcw_home> We would normally use a DSP for that cost wise
[22:54:29] <pcw_home> FPGA will be maybe $7 vs 3 and needs 1.2V
[22:54:57] <Valen> do they have higer voltage IOs?
[22:55:27] <pcw_home> But if you need custom/flexible I/O the FPGA wins
[22:55:46] <Valen> I'll probably just use 2x AVRs
[22:55:55] <Valen> one for 2x brushed motors and 1x for the brushless
[22:56:00] <pcw_home> No newer FPGAs are 3.3V only or even 2.5V
[22:56:17] <Valen> not going to direct drive fets off that
[22:57:28] <pcw_home> FPGAs are also nice in that you can tailor the peripherals to your needs (if we had to do what SSLBP does with standard UARTS we would have given up)