#qi-hardware IRC log for Wednesday, 2016-04-20

arossdotmeMistahDarcy, bobthebuilder@maker:/home/bobthebuilder/projects$ xdg-stanard eoma-cf | mkdevice handheld00:11
arossdotmeheh :)00:11
arossdotmebobthebuilder@maker:/home/bobthebuilder/projects$ xdg-standard --url eoma-cf00:12
arossdotmehttp://rhombus-tech.net00:13
arossdotmeheh bit of fun00:13
arossdotmebut with a point00:13
arossdotmelolz....00:14
arossdotmefor device in "laptop-15inch portablegamesconsole box mox freedombox handheld tablet-7inch, tv, microwave " mkdevice $device00:15
arossdotme;done00:15
arossdotmefor device in "laptop-15inch portablegamesconsole box mox freedombox handheld tablet-7inch, tv, microwave " ;do mkdevice $device ; done00:16
DocScrutinizer05wpwrak: \o/01:43
DocScrutinizer05arossdotme: ???01:45
DocScrutinizer05ohmy rhombustech, phoneblokes01:47
DocScrutinizer05the idiocy of a consumer society that's *creating* demand to keep production and economy *growing* can _not_ get cured by smarter design exploiting sorts of modularity. Rather that modularity (if it would even work) furthers the compulsive hw upgrade idiocy01:49
DocScrutinizer05what is needed is a new approach to product longevity01:51
DocScrutinizer05why do we need new computers today to browse webpages that didn't work on PCs 10 years ago? what was wrong with the internet as it been 10 years ago? why it had to get transformed into this crap we have today?01:52
DocScrutinizer05when you're concerned about this avalanche of upgrades in every aspect of our technical life, don't search for better modularity but just say NO01:54
arossdotmeDocScrutinizer05, you have valid points, i ponder about them too...01:55
arossdotmei do wonder wont the humans just buy tons of cpu cards... but then i think thats a lesser of the evil then hole new devices. I also know and acknowledge that social/human brain/society/psychology issues cant be fixed by better tech. only by better people01:58
arossdotmebut then wont pepole buy lots of new device variants but then i think that they can also invest in just one 15inch laptop thats extra well made instead of lots of cheep ones that ware out after 2 years of abuse.01:59
arossdotmeso if the psychology it can help the plant=us but if its not then it can continue the unsustainable  status quo. unlike the current status quo it also enables a more sustainable way of having devices that are fast and can be upgrade (or shall i call it swap-up) easily02:02
arossdotme*  so if the psychology along the right sort of lines02:03
arossdotmei also see how all the fancy stuff makes what used to be a simple article a several mb of ram hog. is you need a 8gb+ ram to have more then 30 tabs open02:05
arossdotmebut i can also see the side of it making, opening possibliltys of experiments in design of information.... then on a small side it also means thanks to webrtc i might be able to video chat with people once in my life :)02:07
arossdotmehmm aggh02:07
arossdotme:/02:07
arossdotmeDocScrutinizer05, for me and if i may, i think for luke too. modularity is not the god of anwser but instead is a side ways answer to getting free software computer hardware, buy expanding our needs we can enable possibilities that favour us and help others, help innovations (new cpus,socs,archs,small scale manufacture/designed devices like a gitar effects box, gnu+linux microwave i think would befit a lot from eoma-68), oh and also 02:12
arossdotmean hw platform for reducing waste with the right mental attitude02:12
arossdotmesounds like the marketing for rhombus-tech needs err innovation.... 02:14
arossdotmecus i dont remember "mental attitude" being mentioned02:14
arossdotmeso with what i just said, whats your thoughts back? id like to know :)02:15
arossdotmethanks btw02:15
DocScrutinizer05my take on 'modular' stays unaltered and in line with many others here: it's basically not really feasible and multiplies resource demand for same class of product, while you hardly ever can save enough resources during lifetime of the product to compensate for that02:17
arossdotmek02:17
wpwrakunless you're building PC mainboards. then it still works, barely ;-)02:18
arossdotmeso what do you think is feasible? if its not a silly question. neo900 but on a mass production scale with a few $mill to fund it?02:18
DocScrutinizer05when you see that mudular design needs maybe 2 to 3 times the resources, you need to extend the product lifespan by same factor, plus the resources needed for module updates during that period on top02:19
DocScrutinizer05and then you just achieved break-even02:20
DocScrutinizer05use your $device just a few months longer and you do way more for environment and resource savings than with any modular design02:21
DocScrutinizer05Neo900 is not really a modular design. If it does anything for "green" (except trying to use politically and environmentally correct resources) then it's the strict planned-anti-obsolescence in the design02:23
DocScrutinizer05we hope the device will be useful for at least 5 years02:23
arossdotmeyea :)02:24
arossdotmeim afraid i dont see what you mean or how it "needs maybe 2 to 3 times the resources"? with socs its basically a connector between the soc/ram and the device. the ardunio-like chip that does various bits (like gpio to basic audio even) is no worse (if not better heh) than whats in x86 laptops02:24
DocScrutinizer05and for sure will work that long, and longer02:24
arossdotmeneo900 well made by germans :D02:24
arossdotme(well not just germans i think)02:24
arossdotme(?)02:24
DocScrutinizer05(2 to 3 times resources) your biggest savings nowadays in electronics are from integration02:25
DocScrutinizer05when you want modularity you can't use integration02:25
arossdotmeyea, the arduino-like chip, doesnt that take care of most of the integration?02:25
arossdotmeand the brains are integrated into the cpu card. cpu,gpu,video,ram.02:26
arossdotmeoh and usb02:26
DocScrutinizer05nowadays you got system-on-a-chip almost everywhere02:27
arossdotmeso is it not the integration cut into 2 halfs that go together?02:27
DocScrutinizer05zilch modularity02:27
DocScrutinizer05no, yiou can't cut a chip into two halves02:27
DocScrutinizer05that makes it two chips and both have cost for packaging etc, which otherwise are only one time and now you pay for them 2 times02:28
wpwrakarossdotme: arduino is a great example for the enormous cost of modular hardware :) most arduino circuits could be implemented for a tiny fraction of the cost of arduino + shield by just putting the MCU and whatever peripheral in their own circuit02:28
DocScrutinizer05this is exactly where my factor 2 to 3 (to even 5) comes from02:28
arossdotmenot the cpu or gpu or ram etc but why not the pwr, audio,gpio separately? like a brain-on-chip and a gpio/io-on-chip ?02:30
wpwrakarossdotme: the modular design forces you to over-design the base module, because it has to fit a large number of use cases, then adds interconnection elements (headers), and if you're unlucky, you need components that work around constraints of the interface. e.g., a voltage regulator or some mux/demux circuit02:30
DocScrutinizer05ohmy, this is all one chip nowadays02:31
DocScrutinizer05maybe 202:31
DocScrutinizer05wpwrak: level shipters, phy chips, signal shapers at large...02:32
arossdotmesocs with every thing in one is also from a market of throw away devices so what does it matter. these days are there not moves to reduce that all-in-one nature of socs? i remmber that was it allwinner had less in there mutl core soc....02:32
DocScrutinizer05*sigh*02:32
arossdotmesorry02:33
arossdotme:/02:33
DocScrutinizer05adding "GPIO" to a chip costs 5 ct, adding a separate GPIO chip costs 2 bucks02:33
DocScrutinizer05now consider the whole SoC only costs 5 bucks02:33
wpwrakarossdotme: the main drive is to offer as many of the features you may want. if i get Soc A that has feature Y I need for my design, and soC B that needs an extra chip for it, it's usually a design win for A.02:34
wpwrakDocScrutinizer05: you sure buy your GPIOs expensively ;-)02:34
DocScrutinizer05I talked about TCO02:34
arossdotmewpwrak, yea isnt though when there are optional things in the standard? theres no optinal thing in eoma-68. just the basics, and sane voltage levels02:35
DocScrutinizer05throw in 3 GPIO extender chips instead of a SoC that has 48 more GPIO pins02:35
arossdotmeok i guess i see where you comming from....02:35
wpwrakarossdotme: there isn't much of a "standard" in the SoC world. chip manufacturers make some bundle, with things they hope will attract customers02:36
wpwrakalso, most chips have lots of unused features. that's not a problem, since it actually makes everything cheaper02:36
arossdotmethe old days of separate chip for $thing was very expensive and lead to bulky products. hench to have a mp3 player arduino is $$ cus it you  needs this and that chip for the basics and it all gets messy02:37
wpwrakso instead of producing N x chip A with core + X and M x chip B with core + Y, you make (N + M) x chip C with core + X + Y02:37
wpwrakmuch more efficient02:37
arossdotmetaking of making everything cheeper. all these small production quantities products stuggle to get there hands of socs cus they are so small and if they do get soc it costs extra. with the eoma-68 approach. all projects benefit from of original bulk bug of the socs to make the brain cards. so its lot cheeper for those projects to get socs into there products02:39
arossdotmebulk bug = bulk buy02:40
DocScrutinizer05chip C with core + X + Y is cheaper than either chip A with core + X and chip B with core + Y02:40
arossdotmeyou know i never said it was cheaper.  02:41
arossdotmeits only cheaper if it means you didnt have to buy a hole new device to get something faster.02:41
arossdotmethen its a lot cheaper than a few $ 02:42
DocScrutinizer05it's not02:42
wpwrakchances are that by then technology has advanced also in other areas. so you'll want a more modern display, too02:42
DocScrutinizer05a modular §device will be 2 to 5 times more expensive02:42
wpwrakand perhaps some new interface has come along as well, e.g., USB C. then you need an adapter chip that translates whatever old you have to USB C. and so on.02:43
DocScrutinizer05so any savings you might be able to do are only relative to that way higher price02:43
arossdotmethan a few or some $ diff between soc in device vs soc not fixed into device02:43
arossdotmewhat if the aim so not cutting edge but good enough?02:43
arossdotmeusb 3.1 is going to be around for the next decade, no?02:44
DocScrutinizer05prolly not02:44
wpwraki mean, even with PC mainboards, you often build the whole core system in one step. you don't build it incrementally. because after a relatively short while, some of your modular interfaces will become obsolete02:44
DocScrutinizer05I bet they already work on USB4 standardization, and a draft of USB502:44
arossdotmeyea but so what? usb4+ is for the next decade and will properly but by light and not wire?02:45
wpwrakso there it's not about longevity but about flexibility when combining parts. also, individual items tend to be relatively expensive. so the relative cost of modularizing is smaller.02:45
wpwraki think the USB generations are actually accelerating ;-)02:45
DocScrutinizer05indeed02:46
arossdotmescreens, for a portable device $screensize at full hd is good enough for this decade right?02:46
arossdotmepcs where about latest cutting edge within budget ranges02:47
DocScrutinizer05and claiming "it's just good enough for the next NN years" is a point on my side, since you're better off with a decent long-lifecycle monolithic device then, rather than with a modular stuff for 5 times the cost02:47
arossdotmeso there was no long term then a year upgrade component path, just the original flexibility of what you config your new system too.02:48
arossdotmeidk cus well look at the bom of the eoma-68 cards and laptop02:49
arossdotmeeven at small quantitys its pretty dam cheep. 02:49
DocScrutinizer05look, even server manufs go for SoC instead of the archetypical modular server design02:50
DocScrutinizer05I dunno and actually don't care much about eoma02:51
arossdotmei forget the numbers but i think end prices for laptop with brain card it was around £300-400 thats what i would pay in the uk for $brand laptop02:51
arossdotmeok i respect that02:51
DocScrutinizer05the typical laptop once had CPU and RAM socketed02:52
arossdotmeserver are also specialist, cpu, sata and ethernet.02:52
DocScrutinizer05so you could upgrade the CPU a little, and you can extend the RAM02:52
wpwrakpreciousss sssocketsss02:52
arossdotmeand theres no cpu or ram socket on a eoma-* one cus there as one, fixed, inside the card02:52
DocScrutinizer05now they move to soldered CPU and RAM, because of the cost for the mere sockets02:53
arossdotmeand cus there good enough02:53
DocScrutinizer05you seem to switch direction of your arguments02:53
arossdotme2x core 1.*ghz+ 2gbram+ min specs good enough..02:54
DocScrutinizer05so??? where's the modularity?02:54
arossdotmearrgh i was fearing i might come across like that02:54
arossdotmecus i want 4cores with 4gb+ ram in my netbook but i cant cus its not in a module :(02:55
DocScrutinizer05o.O02:55
arossdotme*faster 4cores, already got 4cores :)02:55
DocScrutinizer05sorry I can't follow anymore. What you're after, what's your point?02:55
arossdotmemainy i want more ram but im stuck to intels limit :(02:55
wpwrakmaybe we should just assure arossdotme that we also think eoma-68 is a great idea. then everyone is happy and we can go back to making our efficient monolithic designs ;-)02:57
arossdotmeive used my netbook for 4years plus now but i really need more ram at lest. if it was a eoma-68 netbook then i could have swapped the soc up to something newer 02:57
arossdotmewpwrak, :D ok ok i try to push my luck. 02:57
DocScrutinizer05ERRRR???  >>and theres no cpu or ram socket on a eoma-* one cus there as one, fixed, inside the card<<02:58
DocScrutinizer05vs >>if it was a eoma-68 netbook then i could have swapped the soc up to something newer <<02:58
DocScrutinizer05you comletely lost me now02:59
DocScrutinizer05are you sure we even use the same terminology?02:59
arossdotmewondering that my self....03:00
DocScrutinizer05quite usual marketing strategy to use fuzzy weird incorrect terminology to make a selling point out of nothing03:01
arossdotmeaww thats not my intention :( 03:01
DocScrutinizer05not yours03:01
DocScrutinizer05maybe eoma's03:01
arossdotmeno not eomas also03:02
arossdotmebut i wonder about the modular word now03:02
DocScrutinizer05~wiki module03:02
infobotAt https://en.wikipedia.org/wiki/Module (URL), Wikipedia explains: "{{Redirect|Module}} 'Modularity' is the degree to which a system's components may be separated and recombined. The meaning of the word, however, can vary somewhat by context: *In biology, modularity is the concept that organisms or metabolic pathways are composed of modules. *In construction, modules are a bundle of redundant project components that are produced en masse prior to ...03:02
arossdotmelolz i knew that was asking for problems that sentence. what i was getting at is that i feel your in the mind set of modularity used to be about cutting edge chips that you upgraded/swamped each chip to kind of "perfection". wheres for me its not to that level, modularity is hmmmmm03:03
arossdotme...03:03
DocScrutinizer05sorry, chip complexity is not directly related to modularity03:04
arossdotmeawww03:04
DocScrutinizer05it's just it doesn'tmake sense to ahve a modular function $X if same function $X is already to be found in the SoC and the SoC doesn't get more expensive because of that03:05
DocScrutinizer05functions in SoC come almost for free03:05
arossdotmefor a not the best example, sry: more like a err fire place. you can swap out the burner for a diff design but its still a fireplace, a brick wall with an chimney03:05
DocScrutinizer05so it makes sense to cram in as many as you could imagine and find pins for03:06
arossdotmein the monolith design yes03:06
arossdotmetotal sense on that case03:06
arossdotmeon=in03:06
DocScrutinizer05a chip is *always* monolithic03:06
arossdotmeaggh03:06
DocScrutinizer05(inless it's a very special multi-die chip)03:07
arossdotme:| this is hard 03:07
DocScrutinizer05it's basically the packaging that makes a lot of the chip cost. Split the chip in two halves and more than double the cost03:07
arossdotmei guess i need to be a full hw tech to talk with an full depth hw tech03:08
DocScrutinizer05(well, simplified picture, but still...)03:08
arossdotmeyes03:08
DocScrutinizer05think of a swiss army knife. It doesn't make sense to build that thing 'modular'03:09
DocScrutinizer05actually for chips it's worse: next year's twice as large knife with double the tools costs same as this years predecessor model03:10
DocScrutinizer05and both are by magnitudes cheaper than a "modular" approach where each tool is independent03:11
arossdotmeyes comparing monlithic to monlithic. what about totial cost of ownership, having $device for the rest of you life. your never going to have one laptop. what if your new laptops were staged. each say decade you got a new $device instead of several hole devices03:11
arossdotme*hole new devices03:12
DocScrutinizer05sorry, I'm busy. BBL03:12
arossdotmeok03:12
arossdotmethanks for your time03:12
arossdotmei know its a pain and i imaine it feels like im not listening. i am reading and aknowaging to myself what you write and think, im.... trying to well i guess it feels like im trying to breakthough and get to another argument an higher level but then you see holes at this lower level and so are trying to fill them in :)03:15
arossdotme*imagine03:15
arossdotmei know its a pain and i imagine it feels like im not listening. i am reading and acknowledging to myself what you write and think, im.... trying to well i guess it feels like im trying to breakthrough and get to another argument and a higher level but then you see holes at this lower level and so are trying to fill them in :)03:16
arossdotmethanks so much for keeping on trying to discuss with me. :)03:18
arossdotmesee ya03:18
arossdotmegot this amp board but with a damaged ferrite. while it seams to work. this newbie/user wonders what cons there are from it being damaged? http://bayimg.com/FAIkiaagH03:32
DocScrutinizer05hard to tell. Prolly will work within parameters (possibly getting a little hotter than usual) until you drive it at high load where it might distort or shutdown or overheat and burnout. Also for sure it creates more electrical noise (EMI)03:41
DocScrutinizer05if it's a class-D with the ferrite beads just in the speaker lines to eliminate noise, then you don't need to worry much about anything03:44
DocScrutinizer05actually the damage to the ferrite looks more severe than it is. The properties of that inductivity are prolly almost unaltered03:49
arossdotmeok thx03:57
arossdotmeyes its class-d03:57
arossdotmetpa3118 blt mono04:04
arossdotmepblt04:05
arossdotmethe board was less than £4 inc ship for mono 60w04:07
arossdotmeamazing and it sound fairly good. im concerned about THD at high levels, need to test it again but with 24v dc instead of 19v dc i guess to get the best THD. noise floor was fairly good to04:09
DocScrutinizer05hmmhmmmhmmmmm http://www.diyaudio.com/forums/group-buys/269855-wiener-tpa3118-amplifier-card.html05:14
DocScrutinizer05oh, that's another board05:18
DocScrutinizer05so yes, the ferrite coils are just output filters, shouldn't do any harm, except marginally worse EMI radiation on speaker cable05:48
DocScrutinizer05of course when you start to worry if your speaker is 7 or 8 Ohms and tune the filtering to match the speaker impedance, then you can't accept such defect ;-)06:05
arossdotmeoh hmm, i might be using 8ohm speakers, tested it on one last night.... hmm12:06
DocScrutinizer05nevermind, this was tongue in cheek19:49
DocScrutinizer05as a matter of fact, you don't even know for sure what "Ohms" your speaker is19:50
DocScrutinizer05since it's no resistor after all, but a complex load19:50
DocScrutinizer05the Ohms are a roughballpark value to distinguish N (4) from 2N (8) or 4N (16), any more precise values don't make any sense at all, and the filters in such D-class amp don't need tuning to the speaker either19:52
DocScrutinizer05just like such amp doesn't need a 500 bucks 16mm^2 oxigen-free copper power cable19:53
DocScrutinizer05that's all voodoo19:54
DocScrutinizer05the poor sort of voodoo19:54
DocScrutinizer05lemme put it this way: cheap designs prolly simply omit the ferrite chokes and accept acting non-compliant to FCC rules regarding RF noise emission19:57
DocScrutinizer05you'll need a shortwave radio (or spectrum analyzer) to detect the difference from your broken ferrite19:58
DocScrutinizer05and even then I doubt it's easy to tell19:59
DocScrutinizer05type and length of speaker cable will have way more impact on RF emission than that little broken part of the choke20:00
DocScrutinizer05and absolutely no impact on the intended purpose of this circuit: playing audio20:01
DocScrutinizer05I couldn't tell that from beginning since I didn't know the circuit / chip used20:02
DocScrutinizer05now I can tell for sure20:02
--- Thu Apr 21 201600:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!