#milkymist IRC log for Monday, 2011-10-17

lekernelI just got myself one of those composite USB recorders and a VGA-to-composite converter10:56
lekernelthe nice thing is that the VGA-to-composite converter also has a VGA pass-through, so I can simply insert it between the M1 and projector to record future performances ...10:57
lekernelwith this setup, the final picture quality is a bit lousy, but acceptable11:01
wpwraksounds like a nice concept. "final picture" is the one projected, the one grabbed, or both ?11:18
lekernelthe one grabbed11:35
Action: lekernel goes looking for suitable gigs in Berlin ...11:36
lekernelhmm there's something on nov 11.. quite late11:37
wpwrakif you had a little SATA block somewhere, you could grab to a harddisk ;-) that would solve all these pesky quality issues for good11:42
lekernelcould the disk handle uncompressed video?11:46
lekernelthat's 18MB/s sustained11:47
wpwraki think the faster ones do something like 50 MB/s these days11:48
lekernelsustained write?11:48
rohyes. especially that11:48
wpwraklemme see .. c't always has those benchmarks ...11:48
rohlinear write/read is the easiest to handle11:48
wpwrakc't got for the latest seagate 4 TB a sustained write minimum speed of 50 MB/s11:51
wpwraknot sure how much buffering you need, though11:51
wpwrak(linear) yeah. start at block 0. write until reset ;-)11:52
wpwrakthen plug into PC and process the date11:52
lekernelbuffering shouldn't be a big issue, we can easily add more framebuffers than the current 3 into the M1 SDRAM, and add them to the HDD write queue instead of freeing them :-)11:52
lekernelSATA uses the same signaling as HDMI, right?11:53
wpwrakhuh ? really ?11:55
wpwraki would expect it to be wildly different11:55
lekernelboth are LVDS... with some small difference it seems, but nothing that can't be taken care of by reconfiguring the FPGA I/Os11:56
wpwrakthingking of it .. i wonder if uSD or at least SD couldn't handle 18 MB/s, too11:56
lekernelmaybe it can, but with an horrible overengineered protocol11:56
lekernelnot sure SATA is better though11:56
lekernelif we have DVI/HDMI on board, we can also hack a simple passive DVI->SATA adapter to connect the drive :)11:57
wpwrak(sata) i'm sure they had their committee meetings, too :)11:58
wpwrakah yes, that sounds nice11:58
lekernelI think we should try to switch to the DVI connector (analog + digital) soonish... then we can have passive VGA and HDMI adapters (and composite), and also hacks like this one :)12:00
lekernelit 'just' becomes a matter of FPGA design ...12:00
wolfspraulyes we can try to switch to dvi-i in rc412:01
lekernelon the PCB side, it should just be about replacing the connector and wiring the extra digital signals to appropriate FPGA pins12:01
lekernelwolfspraul, what do you think of "digital kaleidoscope"?12:12
wolfspraultell me more12:13
lekernelwell, instead of "video synthesizer" ...12:14
wolfspraulah ok, you mean to market m1 like that?12:17
wolfspraulI shall start reading about kaleidoscopes a bit more, will let you know in a bit...12:17
kristianpaul(dvi-i) sound like lekernel found a new exp header :-D12:23
wolfspraulthat's also why I want dual-link if possible (enough space)12:25
wolfspraulgeting some more wires out may be helpful later in hacking12:26
lekernelthere are restrictions on FPGA pin placement for TMDS signals, we need to check that first12:26
Action: lekernel is hacking together a variable high voltage supply to trace the curves of the PWL triode12:28
wolfspraullekernel: ok can you check that? maybe some recommendation for the routing guys?12:34
lekerneland yes, I would expect some special routing requirements too...12:37
wpwraka "kaleidoscope" is a cheap and simple device ("mirrors and colored glass"). that could produce the wrong associations. even though many effects are kinda kaleidoscopic12:39
wpwrak"digital kaleidoscope" sounds to me like the same product category as "lava lamp", i.e., more the gadget kind than a professional tool12:46
wpwrakof course, if M1 manages to define a new product category, that perception would change ;-)12:46
Action: lekernel is abusing a nice variac from a 70's x-ray scanner13:02
kristianpaullekernel: the fact that CSR dont have a select input may lead to troubles when writing data at high speeds or continously right?13:37
lekernelhuh? no14:04
lekernelyay! volts bonanza. now let's hook up that triode :)14:05
kristianpaulbut is okay ignore the wb_sel_i anyway?14:05
lekernelif you're always using full-word write access, yes14:06
kristianpaulah ok14:06
lekerneland btw that's a wishbone feature, not csr14:06
kristianpauli know :)14:06
kristianpaulBtw if i'm syncing clock domains between a conbus master and an another csr bus with a separate clock, is okay to do just sync ack, cyc and stb signals as if they we're flags?14:09
kristianpaulah yes also we :)14:09
kristianpauli'm currently just doing the cdc using your psync from the minimac2 core, but a memory test again this another csr busgave m1 0.38% error when validing data writes for about 4 seconds :(14:11
lekernelyay, i'm measuring some 1mA of cathodic emission at 95V :)14:45
Action: kristianpaul just sent a mail to the wb developer15:13
lekernel...and now controlling it with the grid. the triode is good! :-)15:48
lekernelhmm, the thing isn't quite linear, especially for high |Vg|. the grid fails to block all electrons...16:04
wpwraksetting up some death ray guns for the coming zombie attack ? :)16:08
lekerneltesting the triode I got from http://lekernel.net/blog/2011/09/prywatna-wytwornia-lamp-where-diy-meets-vacuum-electron-devices/ :)16:09
lekernelI made nice emission and amplification plots ...16:10
wpwrakretro electronics ;-)16:10
lekernelit's not the best triode in the world, but it definitely is a working and usable triode16:10
lekernelwhich is impressive :)16:10
wpwrakM2 will be a little larger than M1, i guess ? with all the vaccum tubes ...16:11
lekernelI stopped playing with vacuum tubes ca. 2005 :) but this one was too cool for me to resist16:11
wpwrakif you stick to the general concept, that would probably also be the first tube-based FPGA16:12
--- Tue Oct 18 201100:00

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