#qi-hardware IRC log for Sunday, 2014-08-24

DocScrutinizer05according:   Uin: [0..Umax[;  output:b0,b1,b2 [000, 001, 010, 011, 100, 101, 110, 111]; 001:="Uin > Umax/8 * 1", ...  , 111:="Uin > Umax/8 * 7;   comparator0-out=b0:= " Uin > ( (Umax/8 * 1) + {b1 * ( Umax/8 * 2) }  + {b2 * ( Umax/8 * 4) } ) "; comparator1-out=b1:=  " Uin > ( (Umax/8 * 2) + {b2 * ( Umax/8 * 4) } ) "; comparator2-out=b2:= "" Uin >  (Umax/8 * 4)"00:00
DocScrutinizer05speed of conversion aka settle time for LSB: (number of bits) * (comparator propagation delay)00:03
DocScrutinizer05so for an 8bit converter at 25MHz I need opamps with a Ft>200Mhz00:04
DocScrutinizer05hmm, http://de.wikipedia.org/wiki/Analog-Digital-Umsetzer#Pipeline-Umsetzer still isn't it00:08
DocScrutinizer05I only ever seen parallel flash A/D with 2^N comparators for N bits: http://de.wikipedia.org/wiki/Analog-Digital-Umsetzer#Echter_Parallelumsetzer00:09
DocScrutinizer05wonder why. I hardly could have invented a new type of A/D converter, could I?00:09
wpwrakso this is what's really behind the algorithmic captchas of qi-hw: http://www.smbc-comics.com/index.php?id=2259#comic01:53
wpwrakwould you have guessed that the first four panels of this are actually true, at least according to wikipedia ? http://www.smbc-comics.com/index.php?id=2260#comic01:56
whitequarkdid you eventually settle on a DIY scope or what?03:56
whitequarkDocScrutinizer05: re satellite: no, just a regular fuckup, it happens every so often03:57
DocScrutinizer05wpwrak: hmmmm, that looks fine: http://www.eevblog.com/forum/testgear/first-impressions-and-review-of-the-rigol-ds2072-ds2000-series-dso/?action=dlattach;attach=3517505:16
DocScrutinizer05(300some k)05:17
wpwrakyes, segmented memory is sexy ;-)07:24
wpwrakbtw, here's the siglent monste thread i mentioned: http://www.eevblog.com/forum/testgear/siglent's-new-product-msosds2000-series/07:46
wpwrak(well, a small monster - only 44 pages)07:46
wpwraknow has spawned a tech support thread: http://www.eevblog.com/forum/testgear/siglent-technical-support-join-in-eevblog/07:47
DocScrutinizer05aah fine! I already made my way through rigol and spotted the MSO2102A08:11
DocScrutinizer05((looks fine)) it's actually a "printout" of an animated gif, sth that the DS2k rigol can do without any help08:12
DocScrutinizer05hmm, the siglent do not get good reports08:37
DocScrutinizer05also the mech UI looks somewhat less well thought to me08:37
wpwrakthe level knob seems to be the biggest mechanical issue08:38
DocScrutinizer05less buttons, so more submenu levels and more button pressing to do the same setting08:38
wpwrakthere are also some firmware features that clearly need more work. e.g., navigation through 50k segments could take you minutes :)08:39
whitequarklike "let's only record data while things happen"?08:41
wpwrakyes, basically that. take a snapshot on each trigger.08:42
whitequarkoh cool08:42
wpwrakhere's a SDS2000 vs. DS2000 comparison: http://www.eevblog.com/forum/testgear/review-siglent-sds2304-a-comparison-of-features-with-rigol-ds2000-series/08:42
wpwrakthe DS1000Z also has segments, but weaker and optional (so its price either jumps in a higher category or you have to hack it)08:43
DocScrutinizer05wpwrak: thanks for this useful link to the comparison video. It's absolutely clear now that siglent is unusable13:17
DocScrutinizer05slightly better hw (except for the UI aka knobs) but terrible firmware13:47
wpwrak;-) seems that rigol had similar problems when the ds2000 came out. the big problem with the ds2000 is the 2 channels limit. else it would be perfect.13:50
DocScrutinizer05as already elaborated upon, I'm way more happy with 2ch-A+16ch-D13:56
DocScrutinizer05you hardly ever need *analog* recording/watch of more than 2 signals, when you can watch a 16 signals on logical level in parallel and even trigger on them13:57
DocScrutinizer05the waveforms/s capture rate of siglent is sometimes abysmal. The 30min dial-wanking to move form frame 1 to frame 80000 is plain ridiculous. The analysis + playback of history is useless on just one speed and incredibly slow analysis14:00
DocScrutinizer05and the two turn knobs in close vicinity are an absolute killer on top of the dial-wanking14:01
wpwraki suppose our experiences diverge there. i had (have) 2A+16D and find myself longing for 4A. i could imagine adding some D to the A, though, if they're nicely made.14:01
wpwraki expect the dial-wanking to go away as firmware improves. in a way, neither gets it quite right. the analysis mode of rigol is nice, but why not just merge segments and greyscale them ? then one could see a large bundle with each screen.14:03
DocScrutinizer05umm, maybe I don't get it yet, but aiui each segment/frame has its own trigger and they are not supposed to be consecutive/gapless in time domain14:05
wpwrakwaveforms of siglent are okay in the ranges that most matter. in ones where it drops badly you're basically just magnifying the interpolation anyway14:05
DocScrutinizer05nah, you got me wrong I guess14:06
DocScrutinizer05lemme try to find the right timecode14:06
wpwrakso i'd overlay frame #0 with frame #1, etc., through frame #n. each with trigger point at the same position. then divide brightness by n14:06
wpwrakso "zoom" would not be on the time axis (though rigol seem to do that, which looks odd even if it may be convenient) but on the number of waveforms that make up one screen14:08
wpwrakit's the same as intensity grading in normal operation: there you also have multiple waveforms visible at the same time14:08
wpwrakyes. 1 ns/div. most of what you see there is interpolation. just do the math :)14:15
DocScrutinizer05https://www.youtube.com/watch?feature=player_detailpage&v=zOjLNyzxkTQ#t=2123  aka 35:2514:17
wpwrakyes, navigation is very poor there14:18
DocScrutinizer05I'm not buying anything ever based on "hoping that they will fix this issue eventually"14:18
wpwrakan open source scope would be great for that. there, you'd have certainty, not hope :)14:19
wpwrak"it will be fixed. and if i have to do it myself." :)14:19
DocScrutinizer05*maybe* when *I* could fix it14:19
DocScrutinizer05that's why I insist on comprehensive remote control option14:20
DocScrutinizer05it *might* be possible to "fix" the siglent on that level14:20
wpwrakthat's not the same. yes, you can work around some limitations, but there's a huge convenience penalty14:21
DocScrutinizer05also a killer, and start of analysis mode I mentioned (iirc)14:21
wpwrakwell, let's see how siglent proceed with those things. i'm not buying a new scope these days anyway, so i have time :)14:22
DocScrutinizer05waterfall mode in siglent is what you asked for above, no?14:23
DocScrutinizer0541:50 !! No analysis mode in siglent14:25
wpwrakwaterfall mode seems to have an offset, so it shows wave #0 at horizontal 0, wave #1 at -1 div, wave #2 at -2 div, etc.14:26
DocScrutinizer0542:55 s/"minutes and minutes"/"hours and days"/14:27
DocScrutinizer05yes, overlay is the version without offset14:27
DocScrutinizer05all the way up to 44:00 has a clear statement that siglent is useless for a dso, since the most important feature is flawed to oblivion14:28
DocScrutinizer05no use in recording a 20mio points when you can't analyze/search them for that glitch you're looking for14:29
wpwrak(overlay mode) yes, but the step size of 20 should be variable14:30
DocScrutinizer0544:40 is about where it shows the difference between rigol and siglent regarding this14:31
wpwrak(recording) you could proabably download them to a PC :)14:31
DocScrutinizer05then I don't need all that sophisticated stuff in scope at all14:32
DocScrutinizer05also the mere ergonomics of placing the menu softkeys above the 4 BNC for the probes is already giving my wrist aches from just watching this guy operating those14:36
DocScrutinizer0550:38 :-S14:39
wpwrakdunno. i have my scope on a shelf and the bncs are right under the main controls. not a problem.14:39
DocScrutinizer05"I first have to *stop* the scope, then I can move around and manipulate the timebase..."14:41
DocScrutinizer05no, we don't actually need to discuss siglent any longer14:42
DocScrutinizer05rigol it is, MSO2102A14:43
DocScrutinizer05siglent is a poor man's knockoff of rigol14:44
DocScrutinizer05years behind in maturity, firmware and UX wise14:44
DocScrutinizer05I don't need 4chan that are useless14:44
DocScrutinizer05when for same money I can get 2chan that work the way they are supposed to14:45
whitequarkyeah, siglent is kinda crap14:55
DocScrutinizer05>>But most disappointing to me was the lack of any meaningful ability to navigate the huge sample sets that the unit is capable of capturing.  Whether in Frame mode or History mode, it can collect more data than you could ever do anything useful with.  Which is a real shame, since it basically negates a very powerful feature.  It's bad enough that there is no quick access mechanism, just slow knob twiddling, but that's compounded by 15:08
DocScrutinizer05the inability to even vary the Playback speed.  I wonder if Siglent is ready for the barrage of Carpal-Tunnel lawsuits that will follow.      [And guys, I'm only half kidding!]<<  http://www.eevblog.com/forum/testgear/review-siglent-sds2304-a-comparison-of-features-with-rigol-ds2000-series/msg500794/#msg50079415:08
DocScrutinizer05seems i'm not the only one to already feel pain in his wrists while watching that video15:09
wpwraklet's see how long it'll take them to fix it. that review should given them plenty of motivation ;-)15:15
wpwraki'd actually worry more about the two closely spaced knobs. that's much harder to solve.15:20
DocScrutinizer05HAH!!! http://www.eevblog.com/forum/testgear/review-siglent-sds2304-a-comparison-of-features-with-rigol-ds2000-series/msg501116/#msg50111615:21
DocScrutinizer05yes, as I already said that's the ultimative knock out criterion15:22
DocScrutinizer05(buttons spaced to tight)15:22
wpwrak(post) yeah, planting the idea of open source. you never know when and where it may bear fruits :)15:23
DocScrutinizer05while it could get some stopgap fix by disabling the other (trigger) button during rotating the left one, to avoid "stealing focus", it still is a mech UI inconvenience that I don't see myself to get used to15:24
DocScrutinizer05:-D @ FOSS15:24
wpwrak(knob spacing) it does look like a bad flaw. well, maybe after a while you get used to it. for most usability problems it's hard to predict whether they will still bother you after a few weeks of regular use. sometimes even nasty stuff just "disappears". on the other hand, some annoyances annoy forever ...15:25
DocScrutinizer05and it seems siglent hasn't even disclosed/published the SCPI(?) specs/docs15:26
DocScrutinizer05so my approach for a fix by midi tablet would fail as well15:27
Action: DocScrutinizer05 afk, having a look if Nue changed15:27
DocScrutinizer05also enjoying own shower, nothing better in the world really15:28
wpwrakthey seem to be busy writing that document. they've already "pre-announced" to their distributors (who then of course post about it in eevblog :)15:28
wpwrakwhat puzzles me is that i can't find the programming document for the sds1000 series. that one ought be be around and would provide clues of what the sds2000 will be able to do.15:29
wpwrakof course, you have to be careful with such things. there are often weird little bugs and crippling misfeatures. also ran into such things in high-end scopes from tek and agilent.15:30
wpwrak(high-end) the USD 10k+ type15:30
DocScrutinizer05the only thing that can help on that is an agile responsive customer service15:33
DocScrutinizer05alas this type of service is more likely to be found on very small companies, not on the expensive huge ones. Usually15:33
DocScrutinizer05anyway thanks to today's discussion my idea about Rigol changed, till yesterday it had a certain russian army "takes sledgehammer to make it work" appeal to me15:35
DocScrutinizer05that changed completely15:35
wpwraknaw, rigol is a solid company. i think they started as makers of lower-end agilent, then began using their own brand15:35
DocScrutinizer05now somehow siglent inherited that stigma15:36
wpwrakyes, they're a few years behind rigol in profile development. they do look like the next chinese scope company to make it out of the "china crap" swamp, though.15:37
DocScrutinizer05I wonder how reasonable it would be to speculate on hacks/cracks to enable features like enhanced triggering and stuff in a MSO2102A15:38
wpwrakthere's a thread on that, too ... ;-)15:38
DocScrutinizer05yeah, but...15:38
DocScrutinizer05a huuuge field to grok it all15:39
wpwraka bit ugly for the ds2000a series, though. like retrieve old fw, patch it, send it back. in the ds1000z it's just a code you have to enter.15:39
wpwrakand i don't think they cracked the mso2000a yet. it's just a few weeks on the market.15:39
DocScrutinizer05e.g. would a DS2102A be capable of MSO aka LA, just with a key to unlock?15:39
DocScrutinizer05prolly not15:40
wpwrakif it has the connectors, yes :)15:40
DocScrutinizer05they even might be mechanically different15:40
wpwrakbut it seems that rigol are keeping the two apart15:40
wpwrakunlike siglent, which seem to include MSO also in their DSOs. though without connectivity kit. so it's not entirely a software-only change.15:41
wpwrakit also shows in the time: took rigol about 2 years to go from DS2000 (via DS2000A) to MSO2000A15:41
DocScrutinizer05prolly it's also not really "fair" to buy a cheap model, hack it, then complain about stuff not working as expected15:41
wpwraksiglent are now beginning to market the MSO capability, merely half a year after introducing the scope15:42
DocScrutinizer05I wonder if a AWG is worth it15:42
DocScrutinizer05prolly not, rather you'd want a LA-reversed15:43
wpwrakone of my more stupid purchases was a function generator (with AWG and such). i hardly ever use it.15:43
DocScrutinizer05so the thng could actually do tricks like "speak I2C"15:43
wpwrakit can actually do LA-reversed, too. but then, so can, say, a Ben ...15:43
wpwrakand the Ben can respond to what's coming back ...15:44
DocScrutinizer05it would for sure need some *logic* for such trickery15:44
DocScrutinizer05not only braindead playback of a pattern15:44
DocScrutinizer05meh! will ponder getting me a MSO2102A, while thought float during taking a shower15:45
wpwrakenjoy ! :)15:46
DocScrutinizer05sure will15:46
DocScrutinizer05the standards in Spain are... not comparable15:46
DocScrutinizer05e.g. I couldn't even figure how a 24kW water heater would get installed down there15:47
DocScrutinizer05would prolly cause Barbate go blackout15:48
DocScrutinizer05(j/k, not really. But actually electricity installations are kinda scary _still_)15:49
wpwrakmaybe they use gas ?15:50
DocScrutinizer05yup, they do15:51
wpwrakbut yes, in places with a warmer climate, heating equipment is often at a surprisingly poor level. e.g., you'll be much warmer in winter in switzerland than in argentina, even if the outside temperature is 15 deg lower in .ch15:52
wpwrakheating water with gas works fine. i have one such critter. no complaints. and, unlike a boiler, you have an infinite reserve :)15:53
DocScrutinizer05in a heater sized ~50*30*20 mounted tightly into a wooden kitchen cabinet of exactly same size :-o15:53
DocScrutinizer05with door15:54
DocScrutinizer05at least they left out the bottom to give the thing some air to breathe, from below15:54
wpwrak25 x 40 x 75 cm here :)15:55
DocScrutinizer05fed by a 20L propane bottle that sitts inder the kitchen sink15:55
wpwrakthese are dangerous15:55
DocScrutinizer05(50*30*20) the ones with the almost egg-shaped rounded corners15:55
DocScrutinizer05the ignition flame always is "open", not bimetal safety device opening only while heated or pushbutton pressed -  WAAAAH! Duck and cover!15:56
Action: DocScrutinizer05 wonders if such safety mechanisms are mandatory in Germany only15:57
DocScrutinizer05(egg shaped) kinda like this one http://commons.wikimedia.org/wiki/File:Merker_Durchlauferhitzer.jpg15:58
wpwrakthe "pilot" flame is a standard feature in this type of device15:58
wpwrakthere are better ones with electric ignition but the gas-only type needs a pilot15:59
DocScrutinizer05err wut? I'm talking about switching it off when it's already off. By sensing heat and stopping gas flow when no heat15:59
wpwrakyes, then you'd need some sort of bypass valve16:00
DocScrutinizer05we have that on virtually every gas flame no matter where, here16:00
DocScrutinizer05bypass valve?16:00
wpwrakto turn it on. it has to be able to automatically turn on. so if you cut the gas when it's cold, you need to bypass the cutoff to turn on16:01
wpwrakbtw, ovens are different. there, you just hold the button until the bimetal is hot enough16:02
DocScrutinizer05usually those things work with a thermocouple that actually provides enough current to magnetize an e-magnet consisting of some 20 windings of 1.5mm dia copper wire, and you push the steel plate to the magnet maually by pushing a button and thus opening the valve. When thermocouple heats up, the steel plate sticks to magnet16:02
wpwrakso that's an electric model. they can be more complex, yes. but we're talking about the basic gas-only ones, aren't we ?16:03
DocScrutinizer05on water heaters it works exactly same way like oven, just for the pilot flame16:03
wpwrake.g., if electric, you don't need a pilot flame at all16:04
DocScrutinizer05there's no electricity fed to that type of security device16:04
DocScrutinizer05it consists of a thermocouple, a coaxial (dunno why) wire from couple to magnet, and the magnet coil. That's it16:05
DocScrutinizer05the thermocouple seems capable to provide some maybe 0.2V@3A or whatever16:05
whitequarkooooh so THAT is how they work16:06
wpwrakah, i see16:06
DocScrutinizer05>>Abgebende Spannung 30 mV<<16:09
DocScrutinizer05>>erzeut der Thermof├╝hler eine 40 mV Spannung und einen Strom von ca. 10 mA<< o.O16:10
DocScrutinizer05(coaxial) maybe for safety reasons, so nothing could induce a current into it16:12
DocScrutinizer05usually you press a pushbutton which moves the steel plate against the magnet and thus opens the pilot flame's (actually the whole device's) gas valve, and same time the pushed button closes a valve for main flame, defeating huge amounts of gas coming out the thing when pushing button while water flows and pilot flame not burning16:18
DocScrutinizer05then you ignite the pilot flame and wait until it heated the thermocouple, before you release the pushbutton and thus the device enters "normal operation mode"16:19
DocScrutinizer05no such pushbutton on the spanish water heater16:19
DocScrutinizer05which I guess is highly illegal even for Spain16:20
DocScrutinizer05I mean you don't want the gas even for pilot flame coming out of device for hours or days, right?16:34
apeleteHi larsc, you there ?17:52
pcercueilarsc, how many MSPS can the FM-Comms 4/5 push?18:00
larsc54MHz is the analog bandwidth limit18:18
larscthe digital interface can do more18:19
larscbut there is really no point18:19
larscFMCOMMS4 is one I/Q channel, FMCOMMS5 is 4 I/Q channels18:20
pcercueiso 54 MSPS for 4*2 channels in parallel?18:36
larscyes, on each channel18:37
larscso 54*4*2*2 bytes/second18:37
pcercueithat's not much :p18:39
larscI think it is 56MHz actually18:39
pcercueiso that would be around 864 MiB/s max18:42
pcercueiok, thanks18:50
apeletelarsc: I'm modifying dma-jz4740.c to make it work with jz477018:51
apelete 18:51
apeletesince there is no device tree support for jz4740, what's the best way to know which of jz4740 or jz4770 is using the dma-jz4740.c driver ?18:51
pcercueiuse an ID; see our watchdog driver18:51
apeleteI thought about using a platform_devic_id struct indeed, and came up with something like this: http://paste.debian.net/117277/18:52
apeletepcercuei: ah, same trick is used in watchdog driver, would have save me some trouble if I knew :-)18:54
larscyes, that's the way to go18:54
apeletepcercuei larsc: ok, thanks18:57
--- Mon Aug 25 201400:00

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