#qi-hardware IRC log for Saturday, 2015-04-18

arossdotmeany advice getting belkin or  serial ups working? not managed to get my small ones computer interface interfacing. 01:13
arossdotmeon debain01:13
wpwrakfun: a tweet as a measure of code size: http://tweetnacl.cr.yp.to/    "TweetNaCl fits into just 100 tweets"01:38
DocScrutinizer05whitequark: ((http://lab.whitequark.org/images/psu-short/short.jpeg)) I completely fail to grok what that BS with exposed tinned traces and copper fill under mask is meant to establish anyway11:23
whitequarkexposed tinned traces for higher current carrying capacity, I'm sure11:33
whitequarkwhy do they do that and ALSO add a copper polygon? I've no idea11:33
DocScrutinizer05either one or the other, both together is BS11:39
whitequarkwell, we've already established they are not the best engineers in the world11:40
whitequarkbtw i've got this thing https://imgur.com/a/vEY4011:40
whitequark8 bucks11:40
DocScrutinizer05awesome and pretty simplistic11:41
DocScrutinizer05all the trickery is in the layout, eh? :-D11:44
DocScrutinizer05now *that's* a layout you can't do with autorouter ;-)11:45
DocScrutinizer05true voodoo11:46
DocScrutinizer05way beyond my expertise11:46
whitequarkafaik no one designs these by hand11:46
whitequarkyou take a FEM simulator and basically brute force until you get the results you want11:46
DocScrutinizer05so actually indeed *sort-of* auorouter11:47
DocScrutinizer05X-band layout is really voodoo11:47
DocScrutinizer05when *everything* is a inductivity and a capacitor and a delay line and a phase shifter and a impedance transducer11:48
DocScrutinizer05look at http://wstaw.org/m/2015/04/18/plasma-desktopsZ1950.png or even http://wstaw.org/m/2015/04/18/plasma-desktopjg1950.png11:54
DocScrutinizer05the latter would be interesting to see the equivalent schematics ;-)11:56
DocScrutinizer05I guess it must consist of at least 2 dozen components11:56
whitequarkthe former is simple11:58
whitequarkit's an LC series circuit tuned at the oscillator frequency11:59
whitequarki.e. an RF block essentially11:59
whitequarkit's one of the standard shapes apparently11:59
whitequarkthe latter is some dark magic11:59
DocScrutinizer05you seem to have found some arcane info sources? :-)12:03
whitequarkknow a dude who's into microwave12:04
Action: DocScrutinizer05 wonders if it would maybe helpful to have a tad if casual learning about that stuff12:04
DocScrutinizer05wow! ta12:05
whitequarkexcept for the microstrip inanity this thing is very simple actually12:05
whitequarkthe very thin conductors complete DC circuits12:05
whitequarkthe "transistor" is not quite just a transistor, it's an MMIC12:06
whitequarkbasically a transistor with an integrated bias circuit12:06
whitequarkspecifically for heterodyne resonators like this one12:06
DocScrutinizer05I guess BAV99 is rather envelope demod12:11
whitequarkBAV99 is just to protect the microwave diode. it has a Vbrkdn of a few V12:11
DocScrutinizer05and what they call "IF" is actually the demodulated interference aka beat product12:11
whitequarkthe "demodulation" (not sure if you can call it so) is the usual result of mixing transmitted with received12:13
whitequarkwhere received would shift by a few Hz due to doppler, and you'll get a signal with those few Hz to the left of microwave diode12:13
whitequark(it's not really... modulating anything, it just transmits baseband and looks for shift)12:14
DocScrutinizer05ring modulator12:14
DocScrutinizer05f1 * f212:15
DocScrutinizer05where f1/2 are signals, not frequencies12:16
DocScrutinizer05resulting in f1+f2 and f1-f2, this time frequencies of the signals12:17
DocScrutinizer05f1-f2 is what we're interested in here. No demod needed12:17
DocScrutinizer05I gather the 1202 upper left is a capacitor?12:20
DocScrutinizer05to DC-decouple the IF out signal12:20
whitequarkit's a pulldown12:21
whitequarkthe IF signal is not DC balanced :D12:21
whitequarkI mean... there's nothing special about these lumped components. the numbered ones are bog standard SMD resistors12:22
DocScrutinizer05hmm, ok12:22
whitequarkthe white ceramic thing is a DC block capacitor12:22
whitequarkagain to protect the microwave diode12:22
DocScrutinizer05so the whole ring modulator works completely passive?12:23
DocScrutinizer05I.E. no power supply?12:23
DocScrutinizer05DC power12:23
whitequarkthe only powered thing here is the oscillator12:23
DocScrutinizer05as simplistic as it gets12:23
whitequarkit works pretty well despite that12:24
DocScrutinizer05mere voodoo ;-D12:24
whitequarkdetects my hand with hundreds of mV on the IF output, if within 50cm12:24
whitequarkdatasheet says up to 20m, I think it will be able to detect large metal like cars in that range12:24
whitequarkbut not people12:24
DocScrutinizer05ooh, 50cm - yes that's sort of easy12:24
DocScrutinizer0520m is absolutely non-trivial12:25
DocScrutinizer05considering the signal levels on RX12:25
DocScrutinizer05I'm instantly thinking of automatic doors where approaching person needs to get detected in ~2..3m distance, which prolly is what those critters are used for in 99% of installations12:26
whitequarkthat is exactly what the description said12:26
DocScrutinizer05did they tell anything about directivity of antenna and about how to mount that thing?12:27
DocScrutinizer05like "15mm above a metal surface" 12:28
DocScrutinizer05or whatever12:28
whitequarkit's fairly directional12:30
DocScrutinizer05I nevertheless guess it would be a good idea to mount it in that way, with the device surface pointing at sensor range12:30
DocScrutinizer05I think it is, yes12:31
DocScrutinizer05emission in a "8" shape along axis through surface of PCB12:31
DocScrutinizer05hey, you got a Rigol?12:32
DocScrutinizer05did you already test my rigol scropts?12:33
DocScrutinizer05rigol-snapshot already works pretty fine12:33
DocScrutinizer05just needs to stop the scope prior to snapshotting, or it will take ages or even fail12:34
whitequarkI don't have ethernet at home12:38
whitequarkso I just use an usb drive12:38
DocScrutinizer05err, you don't have ethernet?12:41
DocScrutinizer05how comes?12:42
DocScrutinizer05you could link the rigol directly to your PC ethernet jack, not even a crossover patch cable needed with contenptorary NICs that do autodetect12:42
DocScrutinizer05honestly, rigol-snapshot is cooool12:43
whitequarkI don't have an ethernet jack in my PC :D12:43
DocScrutinizer05and a 1000 times more convenient than even the USB hookup to PC12:44
whitequarkit's one of the thin ultrabooks, they're too thin for that12:44
DocScrutinizer05USB-attached NIC then?12:44
whitequarkI have the USB thingy but it's just annoying to use12:44
whitequarkfor occasional use I prefer just the USB drive12:44
whitequarkit's permanently plugged into the scope12:45
DocScrutinizer05yeah, I had that too12:45
whitequarkexcept when extracting data12:45
DocScrutinizer05but the unplugging and plugging to PC and copying of files is a PITA12:45
DocScrutinizer05honestly, get a http://www.exp-tech.de/usb-ethernet-ax88772b-usb-ethernet-adapter?gclid=CJ6C-rDx_8QCFSP4wgodXCgAEw12:46
whitequarkI have one12:48
DocScrutinizer05rather sth like http://www.conrad.de/ce/de/product/986780/Netzwerkadapter-100-MBits-EDIMAX-EU-4208-USB-20-LAN-10100-MBits/?ref=category&rt=category&rb=1 - mechanically less dangerous12:48
whitequarkI just don't like the extra wires around12:49
DocScrutinizer05how about a proper AP then?12:49
whitequarkI mean12:49
DocScrutinizer05aka "router"12:49
whitequarkthis is my work laptop12:49
whitequarkI unplug it all the time12:49
whitequarklike every day once or twice12:49
whitequarkthis gets old *fast*12:49
whitequarkI have a router but it is in the hall12:49
whitequarkbecause the ISP uplink is only long enough to reach the hall12:50
whitequarkI *could* lay cat5 around but it's, again, a pain. I have to do something to doorways12:50
DocScrutinizer05you could get a cat5a/6 from ahö to your workbench and to a switch there12:50
whitequarkbuilding obviously has no cat5 in walls because the construction company thinks it's still 199012:51
DocScrutinizer05I have sth nice for you: http://www.conrad.de/ce/de/product/971688/RJ45-Netzwerk-Anschlusskabel-CAT-7-UFTP-1x-RJ45-Stecker-1x-RJ45-Stecker-5-m-Weiss-hochflexibel-mit-Rastnasenschutz?ref=searchDetail12:51
DocScrutinizer05I got one and they are niiiice12:52
whitequarknot bad but 5m is not nearly long enough12:52
whitequarkI looked into it and the path it has to take, it'd be at least 1012:52
whitequarksort of an unfortunate layout12:52
whitequarkmaybe even 1512:53
whitequarkDocScrutinizer05: know what that reminds me of? POTS "noodles"12:53
whitequarkat least here we call these incredibly shitty "two wires in a flat plastic envelope" POTS cables "noodles"12:54
whitequarkhistory goes in circles, etc12:54
DocScrutinizer05only 1.8mm "thick"12:54
whitequarkthe "noodles" had the advantage that you could nail them to the wall via the plastic in center12:54
DocScrutinizer05I wouldn't pierce nails into that cat6 flat cables12:55
whitequarkdoes conrad.de even ship to RU?12:55
whitequarkI don't know of any remailer in DE12:55
DocScrutinizer05I could, if you don't mind12:55
DocScrutinizer05dunno about conrad12:56
whitequarkthat'd work12:56
whitequarkon a second thought I'm going to move in a few months hopefully12:56
whitequarkso probably no point now12:56
whitequarkunless the flat I'm going to rent has an equivalently shitty layout (hopefully not)12:56
DocScrutinizer05take nice things like that cable with you when you move ;-)12:57
DocScrutinizer05you can use it again, for sure12:57
whitequarkwell... could be12:58
DocScrutinizer05anyway, I *strongly* recommend ethernet for Rigol, and when you actually would make up your mind to get decent ethernet and want that cable, just drop me a note with exact instructions how to send12:58
DocScrutinizer05meanwhile: http://maemo.cloud-7.de/maemo5/usr/local/bin/rigol-screenshot and rigol* and u1233 there13:02
whitequarkI wish sigrok supported rigol13:02
whitequarknow *that* would be interesting13:02
whitequarkwell, my rigol. it supports some of them13:03
DocScrutinizer05I think the err whatever-commands are pretty similar, odds are most stuff works for any rigol13:04
whitequarkI looked into adding support13:04
whitequarkthey subtly differ in annoying ways13:04
whitequarklike one rigol accepts an array and the other a scalar13:04
DocScrutinizer05some do, yes13:04
whitequarkwho the fuck writes an API like that13:04
DocScrutinizer05which is pretty much WIP13:07
DocScrutinizer05but seems to work, well, usually. Though I didn't verify the results yet13:14
DocScrutinizer05anyway a downloaded waveform should be sufficient input data for anything like sigrok13:15
DocScrutinizer05you're welcome to "upstream" any improvements you make, e.g. regarding buffer depth configured in Rigol (script assumes 24M)  ;-)13:16
whitequarksigrok is more than just waveform viewer, it can do stuff like infinite capture13:17
whitequarkand configuring the scope13:17
DocScrutinizer05which prolly won't work with DS100013:18
whitequarkwhich is the reason to use it13:18
whitequarkthe SCPI or whatever commands are there13:18
DocScrutinizer05it can't download waveforms realtime13:18
DocScrutinizer05only when stopped13:18
DocScrutinizer05at least afaik13:19
DocScrutinizer05even screen snapshots take 3 times as long when running than when stopped. So in my book it's a hw (bus bandwith) limitation13:21
DocScrutinizer05quite possibly the ASIC that does all the heavy lifting is hooked up to the communications processor with a flimsy I2C or whatever13:22
DocScrutinizer05and the ASIC is not made for doing concurrent communications to processor and same time capturing13:23
whitequarksounds about right13:24
whitequarkis it even an ASIC?13:24
DocScrutinizer05wild guessing, but sort of educated13:24
DocScrutinizer05yep, afaik it is13:24
DocScrutinizer05I once scrutinized a photo "DS1054 internals"13:25
DocScrutinizer05prolly on eevblog?13:25
DocScrutinizer05also somebody in that context said sth like "the bandwidth linit on the 50MHz devices is implemented by a simple switchable RC filter in the ASIC's inputs - this is how you enable DS1104"13:27
DocScrutinizer05which on second thought makes questionable sense since the DS1000 has a well known dedicated discrete ADC chip. But only one, not 4, and it needs prescalers/amps anyway, which might be part of the ASIC as well13:32
DocScrutinizer05anyway once you look at DS1000 screen with that info and from this perspective, you can virtually *see* how messages like "USB connected" etc are sent as bitmap from comm cpu to asic via I2C, for the asic to superimpose them into the display13:36
DocScrutinizer05considering reaction times I wonder how much of stuff like FFT, RS232 decoding etc is done on comm processor as well13:40
DocScrutinizer05the low price of the DS1000 is made possible only by very nifty software exploiting the available hw resources to the limit13:42
DocScrutinizer05you have a relatively simple ASIC doing the samplig and transcoding sample buffer into a display waveform and raster image, and you got a loosely linked auxiliary processor handling all the switches and communication at large, and also doing the more fancy stuff that includes lots of math or for example the help function etc13:45
DocScrutinizer05note how all messages like "function rate limited" or whatever look alike and look awful compared to usual display13:45
DocScrutinizer05err "Paramater limited!"13:46
DocScrutinizer05and how it akes 400ms between you turning the Y-sensitivity knob "beyond the limit" and the "Parameter linited!" showing up on display13:47
DocScrutinizer05sure the asic has some ALU to do for example the left/bottom side value display13:48
DocScrutinizer05"Vpp=..." etc 13:49
DocScrutinizer05also I guess triggering is supported by that ASIC ALU13:52
DocScrutinizer05while the whole menu display, RS232 decoding, FFT etc is done on the aux CPU13:53
DocScrutinizer05it even seems to me that for unknown reasons all display elements coming from aux CPU are mainly B&W while ASIC is more colorful. I have no real good explanation for that yet13:55
DocScrutinizer05possibly a hw limitation in compositor13:56
DocScrutinizer05"compositor" I meant the stuff in ASIC doing overlay of data (from aux CPU, e.g. menus) over the display created locally (the colorful waveform and stuff)13:59
DocScrutinizer05then otoh those menus have a blue highlight, which voids such explanation14:00
--- Sun Apr 19 201500:00

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