Back
[06:29:04] <Lambda_Aurigae> http://www.survivingtheworld.net/Lesson2201.html
[15:50:49] <hypermagic> hello my friends
[15:55:50] <N1njaneer> Hiyas
[15:56:02] * LeoNerd presumes countrymen and romans are excluded
[15:56:04] <hypermagic> what are you up to? :)
[15:56:55] <twnqx> trying to figure out if "printing" pcbs by cutting them out from 0.5mm thick copper with a lasercutter might work
[15:57:08] <hypermagic> haha
[15:57:26] <twnqx> obviously, single layer only :)
[15:57:32] <hypermagic> you have a 500W exclimer laser ?
[15:57:43] <LeoNerd> Single layer, large traces.. probably good enough if you want to make high-current power-supply boards
[15:57:52] <hypermagic> you need a high power UV laser to burn copper
[15:58:09] <twnqx> nah, i am asking a company who advertises cutting copper sheets...
[15:58:15] <hypermagic> oh
[15:58:16] <twnqx> probably for different purposes.
[15:58:23] <hypermagic> well i tihnk they will cut the fiberglass through
[15:58:43] <twnqx> no, they cut copper. they do not make PCBs.
[15:58:46] <N1njaneer> CO2 will cut FR4 reasonably well if you have enough power, but you will get a very sooty edge. It's not very clean.
[15:59:01] <twnqx> no FR4 or similar involved.
[15:59:09] <N1njaneer> (fiberglass)
[15:59:11] <twnqx> massive copper. and i want to go down to 0.4mm width.
[15:59:18] <hypermagic> i thought you want to "etch" pcb tracks ^^
[15:59:30] <twnqx> no. i want to purely use copper.
[15:59:39] <hypermagic> sounds fun
[15:59:46] <twnqx> :D
[15:59:54] <twnqx> i am waiting for their response
[15:59:58] <hypermagic> but you will place each tiney trace onto a glass one-by-one ?
[16:00:11] <N1njaneer> Not cheap for direct laser structuring. LPKF is one of the current leaders in that, so I'd suggest looking there if you have the need and some deep pockets.
[16:00:30] <twnqx> yeah, wanted to drive over to involve them into the project :3
[16:00:35] <twnqx> they are like 15km from here
[16:00:53] <hypermagic> lazy bastard, that is not a distance even with a bicycle.
[16:01:02] <N1njaneer> Are you in Oregon, or Europe? :)
[16:01:08] <twnqx> europe :P
[16:01:22] <N1njaneer> What is your requirement in machining the copper?
[16:01:36] <twnqx> 0.4mm track/0.2mm space
[16:01:43] <twnqx> roughly.
[16:01:48] <N1njaneer> On what kind of substrate?
[16:01:53] <twnqx> ... none
[16:02:02] <N1njaneer> Just cutting the copper out directly?
[16:02:05] <twnqx> copper sheet in, lots of tiny pieces out
[16:02:42] <N1njaneer> Yeah, you may need direct laser structuring for that.
[16:02:50] <twnqx> ofc
[16:03:07] <twnqx> even milling won't work
[16:03:19] <twnqx> it could with a very specific multi-step workflow
[16:03:38] <twnqx> but the fact that i didn't find <0.6mm heads make it a problem
[16:03:49] <N1njaneer> You'd probably need a carrier substrate for the copper that is removable, else you would need some significant fixturing.
[16:04:03] <N1njaneer> <0.6mm endmills?
[16:04:16] <twnqx> yeah
[16:04:21] <twnqx> like i said, i'd need 0.2 :P
[16:04:24] <hypermagic> milling will work.
[16:04:50] <N1njaneer> It really depends on the geometry of the parts with regard to if milling will work or not.
[16:04:51] <hypermagic> you just need to make extra distance between tracks and you can cut them :P
[16:04:59] <N1njaneer> Another option would be to wire EDM them.
[16:05:07] <twnqx> basically with milling, i'd do a) drill and put in holes for bottom-side SMTs, b) put parts in, 3) remove excess coppper
[16:05:14] <hypermagic> the whole thing will fall apart anyway since you are milling the tracks around
[16:05:36] <twnqx> so two times milling
[16:05:38] <hypermagic> hmm
[16:05:38] <N1njaneer> What is the application that needs such an exotic approach?
[16:05:47] <twnqx> :/
[16:05:50] <twnqx> do i have to say?
[16:06:10] <N1njaneer> I'd hope there's a really good reason for going through so much hassle :)
[16:06:15] <hypermagic> in theory you could lasercut copper on a quartz glass or diamond substrate using an uv laser
[16:06:25] <twnqx> (in fact that's not even the end. the complete result should end up with first silber, then rhodium coating)
[16:06:52] <twnqx> and it's... a LED cube.
[16:07:29] <hypermagic> or you culd vaporize the copper off of the FR4 base with a very tightly focused laser to a tiny spot.
[16:07:40] <twnqx> nah, that would get the copper too thin
[16:07:44] <hypermagic> vaporizing copper with lazer is not easy
[16:07:45] <twnqx> there's 0.75µ at most
[16:08:00] <twnqx> the company i asked claims
[16:08:02] <N1njaneer> Sounds like you'd have to sell them for a couple thousand dollars apiece to recoup your manufacturing costs :)
[16:08:17] <twnqx> 9mm deep reinforced steel, 15mm normal steel, 5mm aluminium cutting
[16:08:36] <twnqx> i am not sure 0.5mm copper should be a problem for their gear
[16:08:49] <twnqx> except possibly for the color
[16:08:56] <hypermagic> the problem with copper is it is a good heat conductor and good lght reflector
[16:09:12] <twnqx> and aluminium isn't?
[16:09:26] <N1njaneer> Aluminum is harder to laser-cut than steel is.
[16:09:43] <N1njaneer> At least with CO2
[16:09:45] <twnqx> obviously
[16:09:51] <hypermagic> you could use aluminium btw no?
[16:09:54] <twnqx> given they can only do 5mm :P
[16:09:57] <hypermagic> you just can't solder it :)
[16:09:58] <LeoNerd> Laser-cutting metal scares me, conceptually. It involves vaporising the metal
[16:10:13] <twnqx> i could try straight silver >_>
[16:10:23] <N1njaneer> Laser-cutting anything involves vaporising the material :)
[16:10:35] <hypermagic> not necessarily.
[16:10:42] <twnqx> yeah, paper just burns!
[16:10:55] <N1njaneer> Paper vaporizes quite readily.
[16:10:59] <hypermagic> using a co2 laser, high pressure air, you only need to get it melt and blow away the melted metal
[16:11:17] <twnqx> do that with my 23x0.4mm pieces
[16:11:22] <twnqx> and the total piece will blow away
[16:11:24] <twnqx> >_>
[16:11:31] <hypermagic> if you don't vaporize it, it will become that what you descripbed as uncertain edges
[16:11:44] <N1njaneer> I have an 80W CO2 here and large air-handler system here to deal with all of the fumes produced. I can assure you most of it vaporizes in to smoke or air particulate.
[16:12:05] <twnqx> can you cut copper? :3
[16:12:10] <hypermagic> :)
[16:12:38] <N1njaneer> No, copper is reflective at CO2's wavelength
[16:12:53] <twnqx> and other metals aren't?
[16:13:07] <N1njaneer> Not as much at particular wavelengths, no.
[16:13:13] <twnqx> i see
[16:13:18] <twnqx> i was contemplating etching, too
[16:13:30] <twnqx> but the depth is scaring me a bit
[16:13:31] <N1njaneer> Which is why you can use a fiber laser to mark stainless without a problem and without any sintering compound added :)
[16:13:34] <twnqx> since i want 0.5-1mm
[16:14:18] <hypermagic> they cut a 60cm diameter steel block with co2 laser...
[16:14:54] <N1njaneer> Yes, once you get in to several kW of power you can start to cut steel and aluminum, depending on the thickness of the material and quality of the cut.
[16:15:03] <hypermagic> laser goes straight you know
[16:15:22] <N1njaneer> Laser cuts with a kerf like anything else due to focus - the trick is to minimize it.
[16:17:14] <N1njaneer> For larger stuff, though, waterjet is often a superior method because you get a much better edge without sputtering that often needs significant deburring operations, and it isn't work-hardened from the heat. Especially useful if it needs welding afterwards. But again, really depends on your needs which is why one single process isn't the ultimate answer.
[16:17:51] <N1njaneer> But that's generally for large stuff. Small stuff is a different game entirely :)
[16:18:14] <N1njaneer> twnqx: You might honestly be better off just etching.
[16:18:28] <twnqx> hm
[16:18:38] <twnqx> pulsed Nd:YAG laser are said to cut copper!
[16:19:09] <N1njaneer> They can, because Nd:YAG can be frequency-tripled in to UV at 266nm
[16:19:16] <twnqx> http://www.lasag.com/en/produkte/festkoerperlaser/kls/ e.g. this one!
[16:19:53] <N1njaneer> Lots of options if money is no object.
[16:20:14] <twnqx> uh. 2.5mm beam diameter? :S
[16:20:50] <N1njaneer> You will generally put it through a focusing lens first.
[16:21:24] <N1njaneer> Larger starting beam diameter will result in a smaller possible spot size through the correct lenses. If the beam is too small you won't get the spot size down without pre-expanding it.
[16:22:41] <N1njaneer> Too small a beam means needing a much higher focal-length lens to focus it down smaller, since there is less area of curvature to work with. But too MUCH curvature to compensate the focal length will introduce spherical abberation, which will cancel the focusing effect. There's fine balance between the two.
[16:23:03] <twnqx> but how do i etch 0.5mm thick copper if i mus not etch away more than 0.2mm in width? :S
[16:23:25] <N1njaneer> You focus the spot-size down.
[16:23:35] <twnqx> oh you meant laser etching
[16:23:40] <twnqx> not acid etching
[16:23:43] <N1njaneer> Yes.
[16:24:06] <N1njaneer> Acid-etching you'd have to compensate the undercutting part to an acceptable level.
[16:24:20] <twnqx> which is bascially impossible, given the sizes
[16:24:52] <N1njaneer> If you can etch from both sides at once you could help compensate this better, but you'd need extremely good registration of the mask on both sides.
[16:25:25] <N1njaneer> You're also going to need a way to plate or seal the copper afterwards to keep it from corroding away.
[16:25:59] <N1njaneer> As well as to allow better wetting for SMT attachments.
[16:26:37] <N1njaneer> Stuff to think about, anyways :)
[16:31:31] <twnqx> like i said, galvanic silver (over the places components - LEDs with embedded controller + SMT capacitors)
[16:31:52] <twnqx> and to stop the silver from corrodium, chemical or galvanic rhodium
[16:32:00] <twnqx> corroding*
[16:32:07] <N1njaneer> Mmm, using WS2811's?
[16:32:08] <twnqx> a little bit of overkill is fun!
[16:32:24] <twnqx> nope, PL.... whatever, 5mm
[16:32:33] <twnqx> also, ws2812 - 11 is just the controller chip :P
[16:32:38] <N1njaneer> Yep.
[16:32:48] <N1njaneer> We have a single-wire solution for LED control like that :)
[16:34:08] <twnqx> yeah, me too :P
[16:34:23] <N1njaneer> No, single wire. As in 1 conductor in, 1 conductor out :)
[16:34:25] <twnqx> the PLs have a slightly different timing, but nothing a bit of assembler can't handle :P
[16:34:36] <twnqx> no ground?
[16:34:47] <N1njaneer> Nope.
[16:35:15] <twnqx> now is that electrically even possible?
[16:35:18] <twnqx> how*
[16:35:28] <N1njaneer> Custom silicon :)
[16:35:34] <twnqx> even then
[16:35:41] <N1njaneer> It's a pretty neat architecture. Power and data over a single loop.
[16:35:53] <twnqx> ah ok
[16:36:00] <N1njaneer> Video refresh rates, too.
[16:36:02] <twnqx> i thought ONLY data
[16:36:12] <twnqx> and no power or ground, and no return
[16:36:29] <N1njaneer> Just two pins - all LEDs wired in series.
[16:36:46] <twnqx> sure
[16:36:53] <twnqx> nothing rf modulation can't handle
[16:37:07] <N1njaneer> No RF modulation. :)
[16:37:13] <twnqx> you'd need some level of RF
[16:37:19] <twnqx> otherwise you get power issues
[16:37:24] <N1njaneer> Nope. :)
[16:37:31] <twnqx> also, how do you get 30A through those cute wires
[16:37:50] <N1njaneer> Don't need to - they're in series. :)
[16:37:55] <twnqx> so?
[16:37:59] <twnqx> i have 0.3W per LED
[16:38:05] <N1njaneer> Higher voltage across the string.
[16:38:10] <twnqx> hm
[16:38:18] <twnqx> but the drop per LED will be widely different, no?
[16:38:27] <twnqx> depending on on or off
[16:38:36] <N1njaneer> Nope, each LED is constant-current controlled by each local driver.
[16:38:52] <N1njaneer> Drivers also generate no heat, only the LEDs :)
[16:39:11] <twnqx> so you put in like 2kV?
[16:39:16] <N1njaneer> Nope.
[16:39:27] <twnqx> but 1000 LEDs with a drop of 2V each...
[16:39:30] <twnqx> in series...
[16:39:30] <N1njaneer> Time-division multiplexing on the charges.
[16:39:34] <twnqx> yuck
[16:39:37] <hypermagic> acid etching is impossible at a steep edge
[16:39:42] <twnqx> multiplexing is the evil
[16:39:43] <hypermagic> it will under-etch
[16:39:52] <N1njaneer> Generally strings have only around 100 or so nodes per string.
[16:39:54] <twnqx> it means you get get the LEDs to 100% on.
[16:39:59] <twnqx> can't*
[16:40:11] <N1njaneer> twnqx: No, the CHARGING is TDM - the actual LEDs are not TDM'd
[16:40:15] <N1njaneer> 100% duty cycles.
[16:40:24] <N1njaneer> Can even do this with 3-5W LEDs, too
[16:40:31] <twnqx> so you have a small capacitor in there, i see
[16:40:42] <hypermagic> you could do it however with like 10x0.1mm repetitive photoresist stage.
[16:40:44] <N1njaneer> Yes, generally an 0603 at each LED
[16:40:59] <twnqx> and all embodied into the case?
[16:41:08] <N1njaneer> Currently no, but working toward that.
[16:41:23] <twnqx> mh
[16:41:30] <N1njaneer> It's a neat design we're working on bringing to market.
[16:41:33] <twnqx> did you take a look at world semi's new design?
[16:41:42] <twnqx> ws2822 i think
[16:41:46] <hypermagic> twnqx, so you apply photoresist, put the mask in place, develop, etch, reapply, mask again, develop again, etch again...
[16:42:21] <hypermagic> and you could just epoxy your 0.5mm copper plate to a blank pcb
[16:42:21] <twnqx> N1njaneer: you wouldn't believe how large my failure rate is :/
[16:42:32] <N1njaneer> I haven't, no.
[16:42:45] <twnqx> they now use parallel data feeds
[16:42:48] <N1njaneer> We do massive WS2811 controllers for a client, though. Ethernet in, thousands of LEDs out :)
[16:42:57] <twnqx> simple enough
[16:43:12] <N1njaneer> Everything else we do with TI drivers - far better control.
[16:43:26] <N1njaneer> Provided you can cluster the LEDs on the same substrate.
[16:43:34] <twnqx> heh
[16:43:42] <N1njaneer> WS wins out when you have wider-pitch topologies :)
[16:43:55] <twnqx> worldsemi now combines an eeprom, a controller and the three leds into one package
[16:44:03] <N1njaneer> Cost per?
[16:44:08] <twnqx> the eeprom stores the led's address, serially assigned
[16:44:21] <twnqx> and then you parallel program them, all data ins in parallel
[16:44:32] <twnqx> so a dead led will be just a dead pxiel :)
[16:44:33] <N1njaneer> Unfortunatly that removes your ability to use the LEDs of choice :(
[16:44:44] <twnqx> they also have the controller
[16:44:49] <twnqx> ws2821 i think
[16:45:16] <twnqx> let me check the price list i have :P
[16:45:37] <twnqx> sop 14, 22 us-ct each
[16:45:50] <twnqx> so i guess i could buy them at 12 or so :S
[16:45:57] <N1njaneer> How much at 10K+ quantities?
[16:46:04] <twnqx> no idea
[16:46:23] <twnqx> last time i just walked into their offices and negotiated
[16:46:45] <twnqx> and got 60% on the pricelist for low quantities
[16:46:55] <twnqx> like 2000 ws2812b etc
[16:47:35] <N1njaneer> Cool :)
[16:47:38] <N1njaneer> Good stuffs
[16:48:09] <twnqx> got them for a friend's online shop
[16:48:27] <twnqx> he laughed his ass off when le learnt that a competitor paid 3x that...
[16:48:43] <twnqx> though he only saved on shipping as i just brought them back in lugagge :>
[16:49:00] <twnqx> s/only/also/
[16:49:57] <twnqx> ah, ws 2821a is 8pin, and 20 US ct
[16:50:06] <twnqx> the 2821b is DMX compatble.
[16:50:36] <N1njaneer> Depends on what you need it to do I suppose.
[16:51:17] <twnqx> i want one of those cree 9.5W RGBW LEDs
[16:51:29] <N1njaneer> We only use Cree here for visible stuff :)
[16:52:10] <twnqx> i was at a booth of a led panel producer in shenzhen
[16:52:19] <twnqx> 32x32 pixel RGB
[16:52:19] <hypermagic> twnqx, so you could etch your 0.5mm copper board if you want.
[16:52:27] <twnqx> 190RMB...
[16:52:28] <N1njaneer> Only manfacturer that fabs LEDs in the US, and highest quality from anyone we've evaluated.
[16:52:40] <twnqx> hypermagic: possibly, but FAR too lazy for that
[16:52:44] <N1njaneer> Plus they give us good pricing since we're buying them 25K at a time :D
[16:53:08] <N1njaneer> Need to in order to guarantee matched binning as tight as possible.
[16:53:16] <twnqx> i get them on taobao for 50% of digikey discounted prices...
[16:53:29] <twnqx> but 25k doesn't seem THAT much
[16:53:38] <twnqx> if i look at some of the video walls in shenzhen
[16:53:55] <N1njaneer> It works for what we need :)
[16:54:11] <hypermagic> twnqx, i don't like that series, i have a white in sape package hanging from my table near me, and it is 80% dead, the chip looks burned, black.
[16:54:30] <hypermagic> same
[16:54:35] <twnqx> hypermagic: not enough cooling/too much current=
[16:54:36] <twnqx> ?
[16:54:53] <hypermagic> i used a current generator for it
[16:55:00] <hypermagic> and it was cooled
[16:55:11] <twnqx> Cree MC-E 4CT RGBW <- that one?
[16:55:17] <hypermagic> a problem with it is if you reflow the chip it will be damaged
[16:55:36] <hypermagic> yes
[16:56:09] <hypermagic> so it is fragile, and for some reason it does not live long
[16:56:12] <twnqx> if you can't keep the profile exactly, or ignore the moisture sensitivity warnings, it will
[16:56:14] <N1njaneer> SMT LEDs are also exceedingly succeptable to moisture absorption.
[16:56:35] <N1njaneer> If they are not handled correctly in manufacturing they WILL fail randomly in the field, especially the reds.
[16:56:44] <twnqx> cree's datasheet says to bake them 24h at 80°C if you excced the moisture levels
[16:56:56] <hypermagic> why you assume im dummy?
[16:57:00] <twnqx> i don't
[16:57:10] <twnqx> it's just one of the most common failure sources
[16:57:13] <hypermagic> i did exactly that.
[16:57:27] <twnqx> and cree's reputaton is not of the kind of failure you described
[16:57:30] <hypermagic> i have +-1C hotplate
[16:57:54] <twnqx> hotplate or oven?
[16:57:58] <N1njaneer> There's a lot of things that factor in to long-term reliability.
[16:58:28] <hypermagic> for a single led chip? :) just wrap it in a little tin foil and crack it up to 80C
[16:58:44] <twnqx> hm
[16:58:45] <hypermagic> and you get a micro oven
[16:58:54] <twnqx> how could the moisture escape fromthe tin foil?
[16:59:02] <twnqx> well, it won't be airtight
[16:59:04] <hypermagic> well you don't close it perfectly.
[16:59:18] <hypermagic> so it will be an oven
[16:59:21] <N1njaneer> Yes, you need a batch oven that will do bake-out. And even that isn't always reliable.
[16:59:23] * twnqx never dries his leds
[16:59:36] <twnqx> but i also hand-solder them :D
[16:59:40] <N1njaneer> Just mount the LEDs as soon as they're out of the package.
[16:59:51] <twnqx> that only works if you buy reels
[17:00:05] <twnqx> i often buy 10 or 20 when i just use 1
[17:00:17] <hypermagic> twnqx, i soldered onto MCPCB. try this by hand.
[17:00:40] <twnqx> heh
[17:00:41] <N1njaneer> If you are getting them from DigiKey or any other legitimate vendor they will be properly sealed in hermetic packaging that has dessicant added.
[17:00:55] <twnqx> probably would put it onto the IR preheater for quite some time before even attempting that :D
[17:00:59] <N1njaneer> We run 3000 part reels of Cree straight through as soon as each reel is opened.
[17:01:02] <hypermagic> so the MC-E package has a heat conducting terminal below it that i soldered too.
[17:01:10] <hypermagic> thus i did a perfect job.
[17:01:26] <twnqx> i admit i wanted to buy the presoldered one :>
[17:01:40] <twnqx> N1njaneer: i am a HOBBYIST
[17:01:44] <hypermagic> yes it may be simpler :)
[17:01:48] <twnqx> i do IT security for a living
[17:02:19] <N1njaneer> I know. And it sounds like you're doing things really well!
[17:02:37] <hypermagic> the seller company who sold these discontinued too, i think because nobody could solder them :)
[17:02:41] <twnqx> well enough my chinese customers pay me enough
[17:03:01] <N1njaneer> I just share what we do professionally since I have experience with reliability and process manufacturing. It's not applicable to everything, but I can at least share what works for us with extreme reliability.
[17:03:08] <twnqx> hypermagic: i would want one unsoldered for my collectibles
[17:03:18] <twnqx> and one soldered for use :P
[17:03:43] <twnqx> N1njaneer: that's how i would do it as well, for commercial use
[17:04:12] <twnqx> i killed a PL LED yesterday
[17:04:28] <twnqx> i think i grilled it while attempting to put the 0402 between its pins, over and over again :(
[17:04:49] <hypermagic> :P
[17:04:51] <twnqx> it was working without the cap...
[17:04:52] <hypermagic> not russian led
[17:05:29] <twnqx> hm. leds are still 70% heat, 30% light?
[17:07:34] <hypermagic> you are looking at it wrong.
[17:07:44] <hypermagic> it gives light from current
[17:07:55] <hypermagic> that is what you want
[17:08:27] <hypermagic> what you can do is improve your super switching power supply that drive it to be 99.999% efficient
[17:09:39] <twnqx> from what read just today, i doubt that is possible
[17:10:40] <hypermagic> i tested a switching power supply of mine 5 years ago and got 98.5% efficiency @ 60W (measured)
[17:11:01] <hypermagic> and technology advances
[17:11:19] <twnqx> primary side, or just dc-dc?
[17:11:29] <hypermagic> dc dc
[17:11:35] <hypermagic> but isolated
[17:11:36] <twnqx> yeah ok, that's possible
[17:12:07] <twnqx> hm, 400V - 5V DC-DC, over the full load range from say 10-90%?
[17:13:07] <hypermagic> 92.5% is not that high tech @ 1kW.
[17:13:27] <hypermagic> you need synchronous rectification and soft switching
[17:15:06] <twnqx> obviously you'll need the matching switching method as well (it seems that DCM seems to do better with sync rect)
[17:15:37] <twnqx> but don't forget that at 1kW you'll also need a PFC that gets to 97% at best
[17:15:51] <twnqx> so you'll end up with 92.5 of 97% :P
[17:15:52] <hypermagic> idc
[17:16:13] <twnqx> and thus you end at 89.7& so far
[17:16:19] <hypermagic> pf only matters to the power company
[17:16:47] <hypermagic> it will not improve your conversion efficiency
[17:17:01] <twnqx> personally i disagree with the first part
[17:17:25] <twnqx> but surely the second is correct
[17:17:44] <hypermagic> raise hands who want 2% lower efficiency in their home computer with pfc for +$50 price
[17:17:46] <twnqx> (unless you factor it transmission network efficiency, in which case no-pfc will lose to pfc :P)
[17:17:55] <twnqx> in*
[17:18:00] <twnqx> 50$
[17:18:02] <twnqx> lol
[17:18:16] <twnqx> more like 5
[17:18:25] <hypermagic> aa... at least 25
[17:18:43] <twnqx> passive pfc will be 2-3
[17:18:44] <hypermagic> more components, assembly
[17:18:45] <twnqx> hell
[17:18:54] <twnqx> i bought a 100W PSU in china with PFC
[17:18:57] <twnqx> for 3€
[17:19:04] <hypermagic> haha ok passive pfc
[17:19:53] <twnqx> i think it has active...
[17:20:08] * twnqx sees two transformers in
[17:20:21] <hypermagic> capacitors are soo active in my opinion too
[17:21:02] <twnqx> there are options that combine them both
[17:21:16] <hypermagic> so did you measure its pf ? 0.99 or better ?
[17:21:17] <twnqx> i mean, both stages
[17:21:20] <twnqx> lol
[17:21:29] <twnqx> at the price i expect more like 0.7
[17:21:36] <hypermagic> lolpfc
[17:21:39] <hypermagic> :)
[17:22:22] <hypermagic> im not complaining, they can"t speak english anyway
[17:22:44] <twnqx> even a meanwell supply will for 40€ will have a PFC
[17:23:00] <twnqx> and really, assembly costs are almost constant for mass volume
[17:23:06] <twnqx> 20 pieces more or less
[17:23:07] <hypermagic> you know what? idc
[17:23:24] <hypermagic> output: 20MW input:500MW
[17:24:10] <twnqx> electricity must be much cheaper in hungary :/
[17:24:30] <hypermagic> on my chinese laser pointer was printed 3MW laser output, and i was so disappointed when it didn't cut through a lamp post :(
[17:24:42] <twnqx> lol
[17:24:57] <twnqx> i have seen worse fakes
[17:25:06] <twnqx> that's just a typo :D
[17:25:15] <hypermagic> no it is not
[17:25:25] <twnqx> they clearly meant mW
[17:25:28] <hypermagic> they dont know what they are doing
[17:25:37] <twnqx> yeah ok
[17:25:38] <twnqx> that too :D
[17:26:36] <N1njaneer> The irony of China knocking off a copying machine...
[17:26:43] <hypermagic> leave me out of theit super high tech science race for cheaper
[17:26:56] <twnqx> lol
[17:27:03] <twnqx> their race for cheaper... is cheaper
[17:27:05] <twnqx> nothign else
[17:27:28] <twnqx> you want to headdesk all day at their engineering capabilities
[17:27:42] <hypermagic> it is cheaper, but the parameters are lies.
[17:27:52] <hypermagic> what is not possible at that price
[17:28:19] <N1njaneer> I've caused some overseas board houses to restate their capabilities after failing to deliver on what they thought they could :)
[17:28:55] <hypermagic> oh cool you tested them for free
[17:29:23] <N1njaneer> It was just a super exotic set of requirements. They do "normal" boards all day long for us without a problem.
[17:29:37] <N1njaneer> I took it to Advanced and they knocked it out of the park on a single-day turn.
[17:30:53] <N1njaneer> That was expensive, but got to the customer on time: )
[17:31:07] <twnqx> http://gd3.alicdn.com/bao/uploaded/i3/T1mRFnFQtbXXXXXXXX_!!0-item_pic.jpg_400x400.jpg
[17:31:15] <twnqx> this is what happens if they try to copy cree...
[17:31:38] <N1njaneer> Oh god lol
[17:31:55] <twnqx> but hey, 50ct.
[17:31:59] <twnqx> claims 4x 3W
[17:39:57] <twnqx> hm. an LED cube with 10W LEDs... now that could be multiplexable without worrying about it becoming to dark :D
[17:40:27] <malinus> twnqx: not if you multiplex at 0.1Hz :D
[17:53:07] <hypermagic> :) all is fun ?
[17:53:23] <twnqx> http://item.taobao.com/item.htm?spm=a230r.1.14.244.z8MmTI&id=37661622712&ns=1&abbucket=3#detail
[17:53:26] <twnqx> yes
[17:53:28] <hypermagic> though i hate that package
[17:53:40] <hypermagic> can you solder it nicely to be heat sink ?
[17:53:43] <hypermagic> i guess no
[17:53:52] <twnqx> no, probably only in commercial environment
[17:54:01] <twnqx> e.g. paste printer, controlled reflow oven
[17:54:09] <hypermagic> tinkerers used to put grease under it and it will fail after it purs out and dries
[17:54:49] <hypermagic> i do reflow all the time but i think that package is a failed concept
[17:55:02] <twnqx> nah
[17:55:08] <twnqx> it works very well i bet
[17:55:21] <twnqx> but i think you still need a real heatsink
[17:55:35] <twnqx> 8W heat at full power is a lto
[17:55:35] <hypermagic> i like the mcpcb
[17:55:41] <twnqx> insufficient imho
[17:55:54] <twnqx> if you look at what they put on 8W lamps
[17:56:02] <hypermagic> you can screw it on a heatsink if you want
[17:56:08] <hypermagic> :P
[17:56:12] <twnqx> i'd say you need that for sure
[17:56:24] <hypermagic> i doubt it
[17:56:29] <twnqx> sure, mcpcb is the step to use between the led and heatsink
[17:56:34] <twnqx> my 7W LED lightbulbs get hot
[17:56:39] <hypermagic> 20W can be dissipated using a metal plate
[17:56:40] <twnqx> at the heatsinks
[17:56:53] <hypermagic> screwed onto the ceiling
[17:57:35] <hypermagic> you overcomplicate it dude
[17:58:01] <twnqx> just saying it will die of heat when not put to *something more than the mcpcb*
[17:58:19] <hypermagic> sure it only conducts heat
[17:58:33] <hypermagic> 2cmx2cm will glow red hot from 20W
[17:58:48] <twnqx> whether a small metal sheet is sufficient is another sorry, you don't need a 10kg copper block with fans
[17:58:55] <hypermagic> you probably thinking of this
[17:59:19] <hypermagic> do you think i have not tried it yet ?:)
[17:59:22] <hypermagic> ;)
[17:59:32] <twnqx> i have no idea what you did
[18:00:00] <twnqx> but i am pretyt sure cree doesn't put the data sheet reliability info with tinkerers in mind
[18:00:52] <hypermagic> i usually like to use 2-5x more at least than needed
[18:00:55] <hypermagic> passively
[18:01:03] <hypermagic> a lamp with fan is lol.
[18:01:37] <hypermagic> maybe you can get away with a fan in your 3MW laser
[18:01:39] <hetii> Hi :)
[18:01:49] <hypermagic> hey hetii
[18:01:58] <twnqx> plasma TVs have fans, too
[18:02:04] <hypermagic> lol lol
[18:02:13] <hypermagic> do not want
[18:02:19] <twnqx> ofc not
[18:02:21] <twnqx> it's retarded
[18:03:53] <hypermagic> it heats your room in winteri bet (and not only in winter)
[18:04:13] <twnqx> around 1998 i had a 20m² apartment
[18:04:22] <twnqx> i could heat it perfectly with PC and 19" CRT
[18:04:36] <hypermagic> :)
[18:04:44] <malinus> those linear particle accelerators had so many usages
[18:04:51] <hetii> I plan to build some library/framework in C where will collect all possible drivers for different devices and want to ask you about the implementation way. I have two idea, first is some of the macro collection that will act as a glue between each layer
[18:04:53] <N1njaneer> I used to use an air-cooled argon laser for heating my dorm room :)
[18:04:56] <hypermagic> yes, great efficiency heating, you get free pc use
[18:05:20] <hypermagic> using an electric heater is dumb when you can do anything useful with that electricity
[18:05:39] <hetii> the second idea is to use proper functions that will register callbacks to end use functions
[18:06:18] <hypermagic> malinus, have you ever had a 290W tdp 3d vga card ?:)
[18:06:28] <hypermagic> or 2 in sli? :)
[18:07:01] <hetii> I want to make is open-source and mcu independed
[18:07:12] <malinus> yes, I had several nvidia. For some reason they always started bugging, and I had to re-flow their shitty card each 3months or so
[18:07:24] <hypermagic> that is bga problem yes
[18:07:35] <malinus> not havin a reflow oven, resulted in me almost litreally baking the vga in a normal one
[18:07:41] <hypermagic> my point was you can rey your hair with the hot air outflow :)
[18:07:45] <hypermagic> *dry
[18:08:43] <malinus> I remember my old gf asked "what are you making? it smells funny", "go look". She was suprised to see what she saw.
[18:09:17] <hypermagic> oh, what? a vga card? ;>>>
[18:09:40] <hypermagic> you know you will die painfully from the fumes ?
[18:10:48] <malinus> I always keep some butterflys with me, when they die, I know it's time to... oh well
[18:11:06] <malinus> I even baked one recently, a 460 ti. Worked for a whole year after that
[18:12:11] <hypermagic> time to grow a new lung? :)
[18:13:30] <hypermagic> just a tip there is a thing called activated carbon filter and fan
[18:13:59] <hypermagic> if you are not to open window but you would still want to live longer
[18:14:34] <hypermagic> http://hackaday.com/2008/08/05/how-to-the-hackers-soldering-station/
[18:19:38] <hypermagic> this is the basic concept
http://3.imimg.com/data3/OT/SV/MY-6015002/smoke-observer-250x250.jpg
[19:04:48] <hetii> Hmm for 32 bit mcu is it worth to use uint8_t type ?
[19:04:56] <hypermagic> yes
[19:05:04] <hypermagic> it is smaller
[19:05:38] <Tom_itx> are they byte or word aligned in memory?
[19:06:24] <hetii> ok but then what is the default range for unsigned int ?
[19:06:40] <Lambda_Aurigae> depends on the platform.
[19:06:47] <hetii> exectly
[19:07:05] <hetii> hmm
[19:07:23] <hypermagic> use fix type :)
[19:07:48] <hypermagic> uint8_t is 8 bits uint16-t is 16 bit ... int16_t is signed 16 bit
[19:08:31] <Lambda_Aurigae> http://en.wikipedia.org/wiki/Integer_%28computer_science%29
[19:08:37] <Lambda_Aurigae> nice little table in there.
[19:08:50] <hetii> The point is that I will build collections of drivers that will share between 8 and 32 bit of mcus
[19:09:32] <Lambda_Aurigae> always use the smallest size variable you absolutely need.
[19:09:36] <hetii> and wonder if its be safe if I use uint8_t and then use that on 32 platform
[19:09:51] <Lambda_Aurigae> unless you are writing for microsoft or java, in which case, use the largest variable you can fit in ram.
[19:10:14] <Lambda_Aurigae> uint8_t should be the same across multiple platforms.
[19:10:28] <hypermagic> Lambda_Aurigae, i'd rather
http://www.nongnu.org/avr-libc/user-manual/group__avr__stdint.html
[19:10:32] <Lambda_Aurigae> it is defined as an 8 bit unsigned integer.
[19:10:46] <Lambda_Aurigae> hypermagic, that's fine if you are only looking at avr.
[19:10:55] <Lambda_Aurigae> but hetii is looking at multiple platforms.
[19:10:55] <hetii> yes but imagine that i define a function like uint8_t getSomething() and will return some value of some mcu register
[19:11:50] <hetii> and wonder if I could have a case in 32bit mcu that this ragister will require bigger variable
[19:11:57] <hypermagic> i use this for pc too and i recommend writing code that knows what its doing
[19:12:16] <Lambda_Aurigae> if you are having to ask this kind of question, I personally think you aren't ready to write this kind of multiplatform library.
[19:13:04] <hypermagic> hetii, you are writing C no ?
[19:13:12] <hetii> yes
[19:13:15] <Lambda_Aurigae> specially as your situation in question depends on what processors, what registers, how you are passing the variables, etc.
[19:13:18] <hypermagic> c deals with movzx and movsx conversions
[19:13:48] <hypermagic> you just write C code
[19:17:23] <hetii> Lambda_Aurigae: ok, but thas why I ask, I have no clue about final mcu that could be used with my library and thats why I wonder if its ok when use fix types
[19:17:50] <Lambda_Aurigae> and I have no way of telling you what you want to hear without a lot more information.
[19:18:31] <hetii> I just thinking loud :)
[19:18:54] <Lambda_Aurigae> if you are writing libraries that will be used across platforms and rely on access to registers then you will have to decide the lowest common denominator for the variables in question.
[19:19:13] <Lambda_Aurigae> or make your library so it compiles different setups for different processors.
[19:21:57] <hetii> ok i`m tired and go sleep now. Thank you for your time.
[19:29:21] <hetii> http://embeddedgurus.com/stack-overflow/2011/02/efficient-c-tip-13-use-the-modulus-operator-with-caution/
[19:31:07] <Lambda_Aurigae> I have to say, in nearly 30 years of programming, I couldn't point to more than a dozen times I've used that particular operator.
[19:31:23] <Lambda_Aurigae> if even that many
[19:31:32] <Lambda_Aurigae> in the last 10 years,,,not at all.
[19:36:31] <hetii> It also show that using smaller data type not always reduce cycles.
[20:20:26] <nn7> Can anyone tell me why my copy of atmel studio would not have _delay_ms listed in util/delay.h ??
[20:22:01] <N1njaneer> nn7: Are you using ASF?
[20:22:17] <nn7> uhh.. I don't know
[20:22:21] <nn7> I don't know what that is
[20:22:36] <nn7> except a windows media streaming format that didn't catch on.
[20:22:49] <N1njaneer> Atmel Software Framework
[20:23:14] <N1njaneer> It's built in to Studio and helps assist with peripheral support by way of libraries.
[20:23:34] <nn7> Never used it
[20:23:59] <N1njaneer> Okay
[20:24:07] <nn7> I tried to start the wizard and it says that my project has no defined board and I need to start a new project.
[20:24:19] <N1njaneer> Do you have a #define F_CPU added, and a target specified?
[20:24:46] <nn7> Yes. I have delay.h linked in, it's actually missing _delay_ms
[20:24:53] <nn7> There's _delay_us, but no _delay_ms
[20:25:06] <N1njaneer> What version of Atmel Studio?
[20:25:19] <nn7> 6.0.1996 service pack 2
[20:25:35] <N1njaneer> I would update to 6.2 -- 6.0 is quite old
[20:26:45] <nn7> hmm... asf wizard did something, gave me the option of downloading drivers, I did that, and my program compiled fine.
[20:26:52] <N1njaneer> My delay.h is showing _delay_ms() in it.
[20:27:14] <nn7> Yeah, it should. My other computer does. I opened up the file and it's missing _delay_ms on this computer
[20:27:40] <N1njaneer> I would upgrade to Atmel Studio 6.2 regardless. 6.0 is very out of date.
[20:28:20] <N1njaneer> delay.h is in C:\Program Files (x86)\Atmel\Atmel Toolchain\AVR8 GCC\Native\3.4.1056\avr8-gnu-toolchain\avr\include\util if you want to check it directly
[20:31:20] <nn7> I actually just let Atmel Studio open the version that popped up when I click on the "delay.h" in my code.
[20:31:33] <nn7> That way I was certain it was opening the version that it knew of
[20:32:02] <nn7> It upgraded to 3.4.1.95 and it's happy now
[20:39:04] <N1njaneer> There you go :)
[21:19:24] <nn7> N1njaneer, so here's the same file:
https://android.googlesource.com/toolchain/avr-libc/+/edcf5bc1c8da8cc4c8b560865d2a54b73c1b51d3/avr-libc-1.7.1/include/util/delay.h.in
[21:19:39] <nn7> but where that one has _delay_ms, mine has just delay
[21:19:59] <nn7> I'm wondering if I did some mass find and replace and atmel studio happily changed that file, too
[21:20:28] <nn7> that's the only thing that makes sense to me
[23:54:47] <rue_more> i2c, is the MSB or LSB sent first?
[23:55:19] <Casper> why didn't they standardise all that? :(
