Back
[00:33:09] <Casper> cap in series won't work
[00:33:13] <Casper> will give too much ESR
[00:34:28] <Sevalecan> then throw them in parallel with two more 10v caps...
[01:01:27] <MarkX> hi
[01:48:24] <MarkX> so i have a weird but very specific problem, hoping someone can help
[01:52:39] <MarkX> basically i have 3 dc hobby motors that need to run, using a lipo power source. i've found a few lipo batteries with 2C rating. the issue is their voltage is 3.4V rather than 5 or 7.4v so I need to use a boost converter... but the ones i've found limit the current so i can never get that 2C rating
[01:52:53] <MarkX> err i can never put that 2C rating to use*
[01:53:25] <xrlk> -)|- vs -|(- on a this pcb that i removed the caps from
[01:53:42] <xrlk> does this help me determine what the polarity of caps should be
[01:55:06] <MarkX> xrlk:
http://2.bp.blogspot.com/-e4Y8eklya-M/T5tsyCnhwxI/AAAAAAAAAHI/CzLyc6APv88/s1600/capacitor1.png
[01:55:22] <xrlk> ok tnx ma
[01:55:24] <xrlk> MarkX,
[01:55:36] <MarkX> no woreie
[01:55:40] <MarkX> worries*
[05:47:40] <Daco> "ENCOURAGE PEOPLE OR WE WILL SQUASH YOU, REALLY!" lol'd
[09:06:37] <hypermagic> http://www.artist-embedded.org/docs/Events/2009/EmbeddedControl/SLIDES/FixPoint.pdf :) float vs fixed in avr
[09:35:29] <Lambda-Aurigae> hypermagic, yeah..if you aren't doing time critical stuff you CAN do floating point on an 8bit microcontroller with no mathcoprocessor.
[09:36:48] <hypermagic> i dont wanna :/
[09:36:56] <hypermagic> 13k vs 3k ?
[09:37:02] <hypermagic> and 50x slower
[09:37:17] <Lambda-Aurigae> I usually just do it as integer math by shifting everything several decimal places over.
[09:38:15] <hypermagic> do you think 23 bit precision is good for a robot joint precision ?
[09:38:29] <Lambda-Aurigae> all depends on how precise and how big the robot is.
[09:38:40] <hypermagic> only like 30cm-1m arm
[09:38:47] <Lambda-Aurigae> I suspect that 23 bit precision is going to be about 13 bits more precise than your hardware.
[09:38:54] <hypermagic> :)
[09:39:22] <hypermagic> 100um is not impossible to achieve
[09:39:56] <hypermagic> it just needs a stable hardware
[09:40:26] <Lambda-Aurigae> 100um repeatability from home-grown hardware will be...you will need commercial grade hardware to build the commercial grade hardware to make the parts for your robot arm.
[09:42:30] <hypermagic> thanks for discouragement :)
[09:49:47] <Lambda-Aurigae> I'm on my third iteration of 3D milling hardware...make the tool to make the parts for the next tool and such.
[09:50:12] <Lambda-Aurigae> I figure I have 3 or 4 more generations before I can get a good 0.01mm repeatability.
[09:50:37] <Lambda-Aurigae> perhaps your machining capabilities are much better than mine...they probably are.
[09:50:46] <Lambda-Aurigae> I'm learning everything from scratch as I build and rebuild.
[10:16:40] <hypermagic> haha
[10:17:00] <hypermagic> you are planning to increase precision by making tools with your tools?
[10:18:10] <hypermagic> well if you have a perfect system and only force decreases precision then if you go slow it will get more accurate
[10:18:42] <hypermagic> Lambda-Aurigae, well i am being lazy and i just use precision things already available :/
[10:19:16] <Lambda-Aurigae> lazy, cheap, good...pick two.
[10:19:32] <hypermagic> all
[10:19:32] <Lambda-Aurigae> most of my gear is made from stuff I scrounged or acquired as free samples.
[10:20:36] <hypermagic> but there will always remain a compromise, speed vs precision, also precision vs weight
[10:21:20] <Lambda-Aurigae> I'm using a combination of wood, scrounged metal plating from copiers, homemade plastic blocks to carve on, and stepper motors/gears/belts from copiers.
[10:21:32] <hypermagic> depends on if you want to mill 1x1cm boards at micrometer precision or something larger preferably faster
[10:22:02] <Lambda-Aurigae> I would like to be able to mill PCBs with 5mil accuracy.
[10:22:21] <Lambda-Aurigae> I can already make most of the parts for a model airplane out of basswood and balsa wood.
[10:22:25] <hypermagic> sounds good, but you will always have trouble with backlash, and things flexing, how you solve it?
[10:22:48] <Lambda-Aurigae> backlash is a problem I'm working on..flex is just a matter of reinforcement.
[10:23:26] <Lambda-Aurigae> I have a little 3-nut-2-spring on a shaft thingie that works pretty well for the backlash.
[10:23:37] <hypermagic> cheap leadscrews are available too for moving
[10:23:54] <Lambda-Aurigae> center nut is attached to the platform...nuts on either side are held in place by clamps to the main center nut and spring loaded.
[10:23:55] <hypermagic> and you could use a teflon block for screw
[10:24:06] <Lambda-Aurigae> tried that..it wears out real fast.
[10:24:13] <hypermagic> did you use ptfe ?
[10:24:21] <Lambda-Aurigae> tried that.
[10:24:25] <hypermagic> :/
[10:24:35] <Lambda-Aurigae> if I used good lead screws it would be better
[10:24:55] <Lambda-Aurigae> but I just use regular all-thread that I run through a die several times to smooth out.
[10:25:48] <Lambda-Aurigae> again, doing my best not to spend any money on it.
[10:26:00] <hypermagic> it would be nice if we could homebrew some magical position feedback tat can do submicron precision
[10:26:08] <Lambda-Aurigae> that's doable.
[10:26:30] <Lambda-Aurigae> with precision enough homebrew tools.
[10:26:35] <Lambda-Aurigae> or a big enough wheel.
[10:26:40] <hypermagic> hmm
[10:26:58] <hypermagic> you mean a kevlar rope and a large wheel and a stepper for example?
[10:27:04] <Lambda-Aurigae> that works too.
[10:27:19] <Lambda-Aurigae> I was thinking a wheel with lots of little copper strips along the outside edge.
[10:27:19] <hypermagic> but that will have errors due to flexing and elongation too
[10:27:40] <Lambda-Aurigae> look at how they do it with high speed inkjet printers.
[10:27:41] <hypermagic> you dont need thta
[10:27:43] <Lambda-Aurigae> they have a strip.
[10:27:59] <hypermagic> you can take a gear and hook up an inductive sensor
[10:28:02] <Lambda-Aurigae> and a sensor that runs along the strip.
[10:28:04] <hypermagic> or even an optocoupler
[10:28:18] <Lambda-Aurigae> the strip has hundreds of lines per inch.
[10:28:28] <Lambda-Aurigae> the sensor is an optocoupler basically..
[10:28:29] <hypermagic> yeah seen that :)
[10:28:38] <hypermagic> but it gets dusty and useless
[10:28:39] <Lambda-Aurigae> very good for positioning.
[10:28:44] <Lambda-Aurigae> that's what a wiper is for.
[10:28:50] <hypermagic> and it is plastic thingy
[10:29:08] <Lambda-Aurigae> on some of the not-quite-so-precision ones they have a 1 to 2 inch diameter wheel mounted to the motor with slots in the wheel.
[10:29:09] <hypermagic> it has large temperature expansion ratio
[10:29:24] <Lambda-Aurigae> not quite so precision but workable.
[10:30:00] <hypermagic> so if we have a precision feedback then we can use a maglev slider and drive it linear
[10:30:08] <hypermagic> or whatever
[10:30:22] <Lambda-Aurigae> yeah...maglev slider gets pricy.
[10:30:28] <Lambda-Aurigae> I like high grade drawer sliders.
[10:30:33] <Lambda-Aurigae> with lots of ball bearings.
[10:30:33] <hypermagic> i obviously meant building one ;>
[10:30:40] <Lambda-Aurigae> which I get out of copiers.
[10:31:32] <hypermagic> what you think about the simple steppr pilling UHMWPE rope method for axes? :)
[10:31:37] <hypermagic> pulling
[10:31:49] <Lambda-Aurigae> no clue what UHMWPE rope is.
[10:32:09] <hypermagic> that is stranded Ultra High Molecular Weight Polyethylene
[10:32:28] <Lambda-Aurigae> I have experimented with using the rail positioning system out of copier scanners....those work well and are pretty precision...
[10:32:38] <hypermagic> stronger than steel, about 7x i think
[10:32:43] <Lambda-Aurigae> go for it.
[10:32:46] <Lambda-Aurigae> probably more expensive too.
[10:32:53] <hypermagic> well i have a few meters
[10:32:55] <Lambda-Aurigae> ok..gotta run and make sawdust while the rain has stopped.
[10:33:01] <hypermagic> guys use it for fishing :P
[10:37:03] <hypermagic> Lambda-Aurigae,
http://netwinder.osuosl.org/pub/netwinder/docs/
[10:37:13] <hypermagic> http://www.amazon.com/Dyneema-Fishing-Strong-Braided-Strands/dp/B00EUWBQ70
