#qi-hardware IRC log for Tuesday, 2013-10-15

DocScrutinizer05ysionneau: thanks for the heads-up01:34
DocScrutinizer05wpwrak: >>the content jumps vertically about once every 1000 screen updates<< burn-in protection01:48
DocScrutinizer05I'm pretty sure it's a feature, not a bug ;-)01:48
wpwrakyeah. or maybe it's just happy and jumps for joy ;-)01:49
DocScrutinizer05at least I've seen similar behaviour on several screens01:49
DocScrutinizer05early LCD had sth similar, when they still had issues with electrode deterioration01:50
DocScrutinizer05and of course plasma screens01:50
DocScrutinizer05funny enough CRT rarely ever did that01:50
DocScrutinizer05despite they as well had burn-in problems01:51
DocScrutinizer05my current LG TV jumps the "NO SIGNAL" message all across the screen, every 5 seconds01:51
wpwrakit only happens when transferring data. static content is stable forever.01:51
wpwrakcould of course be a botched burn-in protection ;-)01:52
DocScrutinizer05yeah, they for sure don't have a timer on board in the OLED, eh?01:52
wpwrakwell, who knows what magic the controller chip can do :)01:53
DocScrutinizer05it's definitely botched as otherwise they would move slowly up and down, instead of jumping01:53
DocScrutinizer05but that would cost a more complex loop, with maybe 50 bytes more code01:54
DocScrutinizer05plus one bit more storage, for the up/down flag01:54
DocScrutinizer05for the boost: it's prolly normal that there's a low-ESR path from input to output, as long as it's a step-up converter01:57
DocScrutinizer05not though for chargepumps01:57
DocScrutinizer05I think I don't have the schematics open here ATM01:58
DocScrutinizer05so I can't tell for sure what's up01:58
wpwrakthere are FETs in the chip that could cut the path. but they don't. ah well.02:00
wpwrakand yes, it's a step-up converter: http://www.skyworksinc.com/uploads/documents/AAT1217_202050B.pdf02:00
DocScrutinizer05hah, step up. always has a virtual diode forwarding input thru choke to output02:02
DocScrutinizer05those critters also don't really current-limit. You can pull mega-amps from output at input voltage02:03
wpwraki'm sure that, for a few picoseconds, you can :)02:04
DocScrutinizer05hehe, hardly, the inductance of the pin is too high02:05
wpwrakbut then there's the parasitic capacitance. a while 5 pF, all yours to burn ;-)02:06
DocScrutinizer05yeah, the block diagram of the chip *suggests* it has a switch that cuts out output. But that isn't meant to work that way, and you don't know what P/N-gaps are *really* on the die02:06
DocScrutinizer05could even be OVP or clamp diodes02:08
DocScrutinizer05see figure 1 in http://www.skyworksinc.com/uploads/documents/AAT1217_202050B.pdf02:10
DocScrutinizer05> >In conventional synchronous step-up converters, a con -duction path exists from input to output through th ebackgate (body diode) of the P-channel MOSFET durin gshutdown. Special application circuitry can disconnect th eload from the battery during shutdown (see Figure 1).<<02:13
DocScrutinizer05it's all already there ;-)02:14
DocScrutinizer05and i'm once again lucky that my uneducated blbla was in line with the explanation in the datasheet ;-P02:15
wpwrakah yes, that's where they were hiding it. thanks :)02:15
wpwrakhehe ;-)02:15
wpwrakinteresting that they mix FET and BJT, instead of just using two FETs02:17
wpwrakBipolar Junction Transistor02:18
DocScrutinizer05more interesting are the 2%..3% higher efficiency with an external schottky02:18
wpwrakyeah, burning 1.5 uA in the transistor is still about an order of magnitude below that02:21
DocScrutinizer05wtf is the deal with FB?02:22
DocScrutinizer05or rather, with R1, R202:22
DocScrutinizer05are they for decoration?02:23
wpwrakit's for the adjustable version02:29
wpwraki picked the fixed version, which has them inside the chip02:29
DocScrutinizer05weird, they say >>The AAT1217 has two fixed output voltage options: 3.3 Vand 5V. An internal resistor divider is connected to th eFB pin inside the package which eliminates the need fo rexternal feedback resistors.<< Still they mention the external R1/R2 dimensioning for exactly those two output voltages. And use the only values for R1/R2 in the application note that doesn't make any sense at all02:30
DocScrutinizer05this seems pretty important to me: >>In applications where th einput power source lead inductance cannot be reduce dto a level that does not affect the converter perfor -mance, a high ESR tantalum or aluminum electrolyti ccapacitor should be placed in parallel with the low ESR ,ESL bypass input ceramic capacitor. The introduction o fthe high ESR capacitor dampens the high Q network an dstabilizes the AAT1217.<<02:34
DocScrutinizer05though they write it's usually not a problem for PCB with their short traces, I'd rather not rely on it, particularly when using a CR2032 as input, and when shielding and EMI is a concern02:38
wpwrakwell, someone may prefer to buy just the adjustable version and add cheap resistors.02:38
DocScrutinizer05ooh there are different versions?02:38
wpwraksee page 12 :)02:39
DocScrutinizer05AAAH I see02:40
DocScrutinizer05-1.2 -3.302:40
wpwrak1.2, confusingly, isn't 1.2 V but 1.2 A ;-)02:40
DocScrutinizer05duh, 1.2Ampere?02:40
DocScrutinizer05that's.... o.O02:41
wpwrakwell, or maybe MHz. can be either.02:42
wpwrakin any case, it's not V ;-)02:42
wpwrakat least it's not gigaparsecs per pound force :)02:46
DocScrutinizer05ok, finished my "breakfast"02:49
DocScrutinizer05(iow closed the pdf-reader)02:49
wpwraktasty little pixels :)02:51
DocScrutinizer05yeah, I wonder if it was Wolfgang who minted the "eats datasheets for breakfast"02:52
wpwrakanyway, i need to see if the display will actually draw the advertized current in sleep mode. maybe i just got that wrong. if it's happy with 1-2 uA, then i won't bother trying to beat that02:52
DocScrutinizer05beware of dangling floating inputs02:54
DocScrutinizer05they easily cause excess quiescent02:54
wpwraki may have held the chip in reset. could be that it didn't like that.02:55
DocScrutinizer05(sorry for mentioning it)02:55
DocScrutinizer05yep, sure02:55
wpwrakwhile (reset) try_to_get_out();02:55
DocScrutinizer05reset pulls low quite a number of internal reset inputs02:56
wpwraki usually try all the likely combinations, see which ones actually result in the lowest current. sometimes, doing it "wrong" works best.02:57
DocScrutinizer05so it can be _very_ power hungry02:57
DocScrutinizer05wpwrak: in austria we can't finance the Neo900 with down payments. It's forbidden to accept money from non-bank persons/entities and pay interest for it. Only banks may do that :-o I guess austria is quite silly a country11:46
rohDocScrutinizer05: it gets much more complicated than that ;)11:48
DocScrutinizer05even more funny, it's not the private person that's lending the money and accepts the interest on top of payback who's getting sued but the one who takes the money11:48
rohDocScrutinizer05: remember the tax stuff for gta01?11:49
DocScrutinizer05been ante Joerg11:49
rohshipping from california.. means: sales tax depends to which us state or country you sell to11:49
rohbut when you ship from california to california... its gets more complicated.. then there is sales tax depending on the district.11:50
DocScrutinizer05yeah, USA are also weird11:50
rohtotal mindfuck. i typed in dozends of different tax levels into that stuff11:50
rohand it changed multiple times. thats why we went shopify later...11:50
rohbecause there was also the issue of different shipping prices depending on the wherefrom, whereto etc11:51
rohDocScrutinizer05: webshop-application-logic and maintainance service11:51
DocScrutinizer05well, I never looked into that ;-P11:51
rohthey run the shop and the servers, and take care of such nasty details... and stuff like ups shipping price calculation... i nearly went mad till i found that11:52
DocScrutinizer05or actually I might have done, at a time when GTA01 wasn't available anymore11:52
rohand that was before i went checking out how to accept a creditcard internationally11:52
DocScrutinizer05steer clear of credit cards11:53
rohnot possible when you want to sell in the 3rd world11:53
DocScrutinizer05It still strikes me funny that in Austria no money lending as long as not one of the 2 entities is a bank11:56
whitequarkuse bitcoin !12:13
viricdamn bitcoin12:16
wpwrakuse silk road ! ;-)13:46
wpwrakDocScrutinizer05: i didn't spend enough time in austria to know about such details :) but i wonder whether this actually applies to you - if you sell something in .de, are you obliged to find out whether there are local .at laws that affect it ?13:47
wpwrakDocScrutinizer05: besides, is this really an "interest" ? i can't imagine that discounts and such are forbidden in austria.13:48
viricwpwrak: hm what was that silk road?13:49
viricI think I recently saw it, and I forgot all again13:50
viricaaah yes, it was in the news.13:50
viric"FBI seized over 26,000 BTC"13:51
viricI imagine my parents hearing that in the news... fun.13:52
larschow is that worse than s/BTC/USD/ ?13:54
wpwraki think it's 1 BTC = 150 USD or such, so it's about 150 times worse13:56
whitequarkI think the guy got around $60m or something, in total13:56
wpwrakhmm, i guess i'll have to quote from "Loreley" (?) :)19:25
apeletelarsc: Hi, are you there ?20:29
larsc5 minutes20:31
apeleteI'm trying to figure out how to implement dma interrupt for jz474020:33
larschm, we never used dma with the current driver20:34
apeleteI don't know if the jz4740 is able to use Mentor Graphic's Inventra dma engine20:34
apeletelarsc: udc driver does not implement dma ?20:34
larscthere is supposed to be dma support but it is not really documented in the datasheet20:34
apeleteI saw that20:35
larscthe udc driver implements it, but it is disabled by default20:35
apeletedo you think we could use the Inventra dma engine for the musb glue layer or not ?20:36
larscI don't know20:36
apeletehmm, looks like dma is going to be a tricky part :-)20:37
apeleteI tried using the interrupt function from ux500.c, but I'm getting the error messages:20:38
apeletemusb-hdrc musb-hdrc.0.auto: No DMA interrupt line!20:38
apeletemusb-hdrc musb-hdrc.0.auto: musb_init_controller failed with status -2220:38
apeletemusb-hdrc: probe of musb-hdrc.0.auto failed with error -2220:38
apeletelarsc: well, won't hold you longer, maybe we can discuss this another time20:39
--- Wed Oct 16 201300:00

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