#qi-hardware IRC log for Friday, 2015-04-17

DocScrutinizer05testing hot: http://wstaw.org/m/2015/04/17/plasma-desktopdH1954.png  http://wstaw.org/m/2015/04/17/plasma-desktopls1954.png  35min already, PC and moni, not bad18:13
DocScrutinizer051h http://wstaw.org/m/2015/04/17/plasma-desktopTf1954.png18:37
Action: DocScrutinizer05 should attach a hair dryer :-)18:38
DocScrutinizer05alas the mein purpose of this little experiment failed: heat UPS up a bit to boil out that nasty PVC smell - but thanks to built-in fan starting as soon as PSU operates in battery-powered mode, the whole device got *cooler* instead warmer18:41
DocScrutinizer05another funny sidenote: first 19% of shown battery capacity vanished during first 60s of operation, after 120s it been down to 77% and then eventually even recovered to 78% after a few minutes18:43
Action: whitequark just got around fixing the lab PSU18:45
whitequarkthe damn chinese have screwed a heatsink mounting screw to a PCB so tight it went through the mask18:46
whitequarkand of course, the screw has touched a conductor at +15V, rectified right from the transformer18:46
whitequarkthus putting the PSU ground at... -15V from PSU -18:47
whitequarkthen, I decided I should for once do all correctly and plugged in both PSU and scope into a socket with actual ground prongs.18:47
whitequarkand connected scope ground to PSU -.18:47
whitequarkthe wire went white hot18:47
whitequarkthe PSU amp meter just sat there at 99918:48
whitequarkwho the *fuck* lays out their PCB like that18:48
DocScrutinizer05sounds almost as nasty as my universal laptop et al PSU with the PCB throughhole solder points sitting straight on a 0.2mm plastic film on top of the alu heatsink18:48
whitequark... that is in fact how I fixed it, I put mylar film between that screw and PCB :D18:48
whitequarkno pointy things near though, so it should be fine18:49
DocScrutinizer05you seen the photos of mine, do you?18:49
whitequarkyeah you told me18:49
whitequarkI also don't understand the love of these chinese vendors for heavy transformers18:50
DocScrutinizer05ugh, this UPS makes funny noises18:50
whitequarkthe PSU doesn't have a single FET, non-through-hole component, two-layer PCB, or even FR-4 inside18:50
DocScrutinizer05transformers are easy to understand and build18:50
whitequarkseriously, that thing would cost half as much to ship18:51
whitequarkhad it been built as SMPS18:51
whitequarkI wonder if they heard that power FETs exist18:51
whitequarkIt sure doens't look like so, glancing at the PCBs18:52
whitequarkthe LEDs have a metal mesh put at them, to emulate that VFD display feel18:52
whitequarkI wonder if they revised the circuit since 1982 or not18:52
whitequarkProbably not18:52
DocScrutinizer05it works ;-)18:53
DocScrutinizer05and 34 monkeys got killed during R&D ;-P18:53
DocScrutinizer05I told you about that russian mini bench vice I had in 1975 or so? which had spackle filler all over it and when that came off due to an occasional hammer hit that missed the target, the steel under it had a shape like Hephaestus had hangover when smithing it18:59
whitequarkno you didn't19:01
whitequarkbut that sounds about right19:01
whitequarkI wouldn't be surprised if the smith literally had a hangover19:01
DocScrutinizer05http://wstaw.org/m/2015/04/17/plasma-desktopqY1954.png after 90min. note the shown load percentage which now jitters between 9% and 11%, with load unchanged. at start is was jittering between 6 and 819:06
DocScrutinizer05voltage is also quite a bit jittery, even more on device display than on web interface19:06
DocScrutinizer05but hey, PC and monitor are happy since 90 min already19:07
whitequarkeh, doubt the SMPS would care19:07
DocScrutinizer05obviously ;-)19:07
whitequarkthey'd work like down to 90V from 220V, a dozen or so jitter is nothing19:08
whitequark^^^ another reason everything should use SMPS, damnit19:08
Action: DocScrutinizer05 picks his decent Fluke DMM19:08
DocScrutinizer05err Agilent19:08
DocScrutinizer05what I thought, a rock solid 233.[7-9]Vrms19:10
DocScrutinizer05ooh nope, 233.4 now19:10
DocScrutinizer05PSU display jerks between 229 and 23119:11
DocScrutinizer05err now it goes quite jerky even on DMM19:12
whitequarkidle thought: many (but I think not all) SMPS PSUs rectify everything right away19:13
whitequarkso you could basically power them with 300 VDC or so19:13
DocScrutinizer05yes, most do19:13
whitequarkI've seen some truly bizarre schematics for PC ATX PSUs19:14
whitequarklemme find it19:14
DocScrutinizer05ok, one thing's for sure: I don't need the 10s periodic beep to remind me the UPS works on battery mode. The snare and hum it produces is quite enough so I won't forget about it19:16
DocScrutinizer05ouch, zilch inrush current limiting before the 470u19:18
DocScrutinizer05oops nope there's a NTC, almost missted it19:18
whitequarklook above D1419:19
whitequarkthe fuck that trafo is doing?19:19
DocScrutinizer05maybe sort of primary feedback19:22
whitequark8-9-10-11 of TL494 is C1 E1 E2 C2 of its output transistors19:23
DocScrutinizer05actually I guess the winding next to R4 is doing a positive feedback that keept whichever transistor conducting until C7 chharged and current through C7-T3-T2 begins to decay19:24
whitequarkI'm told this transformer does three things at once, it provides feedback for the overcurrent protection19:24
whitequarkit allows control of the transistors on the primary side19:24
DocScrutinizer05overcurrent limiting might be another possibility, though it wouldn't work well for that19:25
whitequarkand it does a sort of autogeneration with a controlled transition19:25
whitequarkall at once19:25
DocScrutinizer05actually with C7 there can't be much of any overcurrent19:26
whitequarkmore like secondary side overcurrent19:26
whitequarkTL494 will increase duty cycle based on feedback, and at some point it'll exceed primary side specifications19:27
DocScrutinizer05nah it's a positive feedback19:27
DocScrutinizer05nasty: no overvoltage protection on secondary side, at all19:27
whitequarkunno, maybe. my brain hurts trying to decipher this circuit, with the primary coming through the control trafo19:28
DocScrutinizer05the whole voltage stabilization is a pathetic joke19:30
DocScrutinizer05*my* brain hurts trying to decode ZIC1 TL431C lower right19:33
DocScrutinizer05they're doing something pretty weird for 3V3 "stabilization" there19:33
whitequarkhaha wow19:33
DocScrutinizer05OOOH THERE is the stabilization, all to the 19:38
DocScrutinizer05wow, UPS just cut me out without prior notice19:39
DocScrutinizer05OOOH THERE is the stabilization, all to the way to the left19:39
DocScrutinizer05the PC went down to suspend-to-disk, but when I plugged in UPS to mains and then started up PC again, a ~90s(?) later the UPS went into shutdown mode nevertheless (some hickup from before in either the control sw on PC or the UPS itself, thinking it's still in shutdown procedure) and that of course killed off the PC cold, without suspend to disk19:56
DocScrutinizer05[Fri 17 April 2015] [21:39:38] <DocScrutinizer05> OOOH THERE is the stabilization, all to the way to the left19:57
DocScrutinizer05is last in IRC log. I thought I sent more, but that got lost in disk cache I guess19:58
DocScrutinizer05UPS, always good for an unsolicited shutdown ;-)20:05
whitequarkas I have said20:10
whitequarkwho the fuck lays PCBs out like that20:10
--- Sat Apr 18 201500:00

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