#qi-hardware IRC log for Friday, 2012-03-02

wolfspraulArtyom (in absence) thanks a lot for detailing all your thoughts and decision making process!01:55
wolfspraulthat's very helpful to understand...01:55
wolfspraulI didn't understand what judgment of mine you referred to though - did I say anything?01:56
wolfspraulI planned to email the Qi list pointing to your great blog post, but haven't done that yet...01:56
wolfspraulanyway, all good. thanks for the background info!01:56
wpwrakwhitequark: no idea. try and find out ;-) i wouldn't expect very visible damage. and it may be at a hidden layer. but you never know ...02:02
kristianpaulindeed, Reliaility !03:15
Artyomwolfsraul: I meant this statement: "wolfsprauldismissing arm is not smart and good imho". May be I just didn't understand the meaning correctly. But anyway I decided to explain why I had switched to MM SoC and put ARM away for some time.03:58
wolfspraulthat was in NO WAY at all a reference to you04:08
wolfspraulin fact you are very balanced and objective in comparing all this, which helps me (and others I think)04:09
wolfspraulit's more a reminder to self to appreciate what ARM has done (and keeps doing), and take Milkymist to the next level, instead of trying to catch up04:11
wolfsprauland of course, your findings may take you back to ARM, and why not?04:11
wolfspraulwe are passionate about making things work, I think04:11
wolfspraulI would do that too, I want to get stuff to work...04:11
wolfspraulmeanwhile we improve Milkymist SoC, which will eventually be chosen because IT IS the best, not because it's a dogmatic choice04:12
Artyomwolfspraul: I think we undertand each other fine now :) As I said I could misunderstand the true meaning of your statement ;)04:20
whitequarkcan anyone answer a relatively simple question?10:14
whitequarkDocScrutinizer: maybe you know?10:15
DocScrutinizer51whitequark: maybe I can answer, but only qhwn you pose the question here14:59
DocScrutinizer51no way I'll open a webpage to rea it now, on mobile15:00
DocScrutinizer51anyway the URL suggests you want to plug mic-in to errr line-out?15:01
DocScrutinizer51PLRASE KILL E!!16:10
DocScrutinizer51or rather, kill billyboy gates16:11
Action: DocScrutinizer51 can't stop headdesking16:12
DocScrutinizer51c:\foo\bar\boo\CPU-A\conf\brd.cfg  gets displayed in sourceinsight as brd.cfg[c:\foo\bar\...\conf\]16:14
DocScrutinizer51BUT THERE'S ALSO CPU-B!!!16:15
DocScrutinizer51guess what's display for cpu-b brs.cfg file in source-inight16:16
Action: DocScrutinizer51 continues headdesking, while pondering suicide or 500kg TNT16:17
DocScrutinizerin explorer you got a decent full pathname in top addr textbox. Not so in "Open" and "Save as" dialogs :-(((19:19
DocScrutinizerFSK wincrap!19:19
DocScrutinizeris there any tool to resize non-resizable dialogs, on windoze?19:21
DocScrutinizerprolly not, and even if you could resize the "window" of that dialog, I'd bet the widgets inside won't resize accordingly - unless the unlikely luccky case the requester is coded in Qt19:24
whitequarkoh, was afk19:25
whitequarkDocScrutinizer: yeah, I have a GSM module with line-out and mic-in (electret)19:25
whitequarkactually, two modules19:25
whitequarkand I want to plug the line-out of one to mic-in of another19:26
whitequarkshould be pretty trivial, but I was unable to google it19:26
whitequarkbonus points for not letting TDMA noise to the line19:26
whitequarkboth the line-out and mic-in are differential19:27
DocScrutinizeruse a R-divider to attenuate, with a 2kR across the mic lines and a 50..1000kR series to line, in series with a 100nF decoupling C19:27
DocScrutinizeruse a 1k load R across line-out19:28
DocScrutinizerfor the TDMA noise, you might or might not run into problems. Worst case you need a NF-transformer19:31
DocScrutinizerbut then, you as well can use some active circuitry, like e.g an optocoupler19:32
whitequarkah. I thought that I could just use coaxial cables (I have a lot of these) for connecting the audio i/o to a shielded board19:34
DocScrutinizerdepending on particular line-out differential design, you might already have sufficient bias on one of the lines to GND, to drive the opto's LED with just a series R as external components19:34
whitequarknot sure if that would help actually19:35
DocScrutinizerwhitequark: regarding TDMA noise your main problem is common GND of both modules19:35
DocScrutinizerthe TX draws massive current bursts and thus introduces TDMA noise floor to its own GND level and in turn also to all inputs and outputs19:36
whitequarkmy attempt at schematics: http://higgs.rghost.ru/36815621.image19:37
whitequarkhave I got it right?19:37
DocScrutinizerof course the differential input on the other module is meant to suppress those common mode noise components in signal19:37
DocScrutinizerbut that usually doesn't pan out19:38
DocScrutinizer(right?) err, yup, just common GND missing at bottom19:38
DocScrutinizerone line +19:38
whitequarkyes, about common gnd19:39
whitequarkwith this scheme, I just connect LINE- to MIC-, right?19:39
whitequarkand they're not connected to the GND of the modules19:39
whitequarkthus they're not affected by noise floor19:39
DocScrutinizerbut they are still affected by noise on TX module's GND as it will "lift" whole circuit's GND reference level19:40
DocScrutinizerso both Line+ and Line- will have common mode noise19:41
whitequarkhm, yes, but how differential input won't reduce that to zero?19:41
DocScrutinizeran ideal differential input will actually do that19:43
DocScrutinizerproblem is most of those "differential" inputs are not perfectly balanced19:44
whitequarkoh, the non-ideality of electronics ruining our clevermost plans...19:45
whitequarkif that would be the worst case, is an NF-transformer something like I could just go, buy and solder in?19:46
DocScrutinizerooh, and you want a 100nF in your GND connection as well19:46
whitequarkso, http://rghost.ru/36815801.view19:47
whitequarkcool, thanks. I'll buy the components tomorrow and assemble the device19:48
whitequarkah, one more thing19:48
whitequarkthe module has SMT antenna pads, so they recommend to lay out a PCB, then calculate a 50 Ohm trace and so on19:48
whitequarkI'm going to wire-wrap the entire device, so I want to buy a 50 Ohm antenna with a cable, cut its SMA connector off and just solder it directly to pads19:49
whitequarkwould that work?19:49
whitequarkok thanks again. I need to run now19:51
DocScrutinizerif it's an unbalanced 50Ohm output, I.E. one of the two pads is GND19:51
DocScrutinizerwhitequark: you should calculate R2 so it attenuates your line-out level to sth like 100mV for 2k mic load R19:54
DocScrutinizererr R119:54
DocScrutinizerfor R1 you might reconsider 1k according to properties of your line-out. If that's e.g. no line-out but a dedicated speaker output, you may need a 100R rather than 1k, maybe even a 33R though I doubt the line-out amp is that picky19:56
DocScrutinizer1k is proper load for standard line-out as found on any arbitrary home audio equipment19:56
DocScrutinizerstandard line-out maybe 2Vss, so R1=50k19:57
DocScrutinizerpoti should be log type19:59
DocScrutinizerwith the spead end (the origin) towards mic-20:02
DocScrutinizerwhitequark: (trafo) http://www.conrad.de/ce/de/product/516201/20:33
whitequark(antenna output) yes, it's unbalanced20:39
whitequark(line-out) it can drive a speaker directly if it's not very powerful (a headphone), or they also have an example with additional amp20:40
whitequarkLoad Resistance (typ) is 32 Ohm20:40
whitequarkand the diff. swing is 2.2V20:40
whitequarkas per the mic in, its working voltage is 1.5V, and diff. input voltage should be 740mV RMS to keep THD <5%@1kHz20:42
whitequarkat 0dB 20:42
whitequarkand 15.9 mV rms at 20dB20:42
whitequarkI guess I should drive it with higher voltage, as the amp would add more noise at greater attenuation levels, right?20:43
whitequark(http://www.8051projects.net/e107_files/public/1300865705_32137_FT45814_sim900d_hd_v1.02.pdf, page 43)20:46
--- Sat Mar 3 201200:00

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