#milkymist IRC log for Thursday, 2011-10-06

kristianpaul!seeen antgreen01:23
kristianpauloops02:10
wolfspraulaw: you there?06:10
awwolfspraul, hi yes06:11
wolfspraulwe may have found another rc3 bug06:11
wolfspraulthe idea is to add a 4.7k pullup on CE0 and WE# of the nor flash06:11
wolfspraulWerner is currently investigating whether that fixes anything, but from the datasheet it definitely looks like the right thing to do06:12
wolfspraulso the question for you right now is:06:12
wolfspraulcan you find out how difficult it is to rework existing rc3 boards by adding a 0402 resistor in those 2 places?06:13
wolfspraulwe don't know yet whether we want to do it, just collecting data points06:13
awthose two pins are hard to solder since there's no alternative pads ONLY nor chip's pins on WE#(pin 55), and CE0(pin 54)06:15
awwe're accumulating a total 70pcs "Avail-fix2b-l3" h/w patches with short L3 and lockflash version of flash version.06:17
aw70pcs here includes 25pcs "Packed - For Sale" and those ones sent out. With Werner's auto testing idea to verify if those two pull high resistors can suffer from exponential distribution and temperature-oriented later. :-)06:20
aweven if the results are good for fixing, the required fixes about two 0402 resistors I currently don't have idea to easily rework them in one batch time. ;-)06:23
wolfspraulwait06:23
awbut it's good that one pointed. :-)06:23
wolfspraulwe don't know much about CE0 and WE# yet06:23
wolfsprauljust collecting data06:24
awsure06:24
wolfspraulI am not worried about the units we are testing, packing and shipping now06:24
wolfspraulwe can definitely add CE0/WE# to the rc4 list06:24
awyup...but I checked already that about the reworks if needs later. like i said, it's quite a hard though. :-)06:24
awsure06:24
wolfspraulthen before we finalize rc4, we need to make a design verification on at least one rc3 board that those 2 pullups will not cause problems06:25
wolfspraulso you can definitely already write down a note in the rc3 errata about ce0/we#, so we don't forget06:25
awChristopher Adams will come here soon. ;-)06:25
wolfspraulthe other question is whether we should fix it on rc3 boards06:25
awsure okay...06:25
wolfspraulfor that, let's see what werner finds out about them first06:26
wolfspraulso you have to solder the resistor directly onto the feet of the nor chip?06:26
wolfsprauland then a wire running somewhere, or no wire?06:26
wpwrakyeah. and add a little wire to 3.3 V06:26
wpwraki'm actually more worried about the wire than about soldering to the NOR06:27
wolfspraulcan the resistor be on the other side of the wire?06:27
wpwrakyes. that would be an option06:27
awyup...wait couples days after his results are if related to remaining considerations.06:27
wolfspraulhow long is the wire?06:27
wpwrakcould be just a few mm, to the next point that has 3.3 V06:28
wolfspraulaw: ok, so right now only this for you: write down ce0/we# into the rc3 errata06:28
awmust have two small wires to VCC and connects 0402's one terminal pad then solder to nor's two pins06:28
wolfspraulaw: maybe the resistor can be on the other side of the wire, not directly at the nor pins06:28
awwpwrak, this fix for me or you may not have problems on soldering, but i really have no gurantee on the quality though.06:29
awsoldering one board to have design validation is good to synchronize together06:30
awbut even if your results are great enough about two resistors, i still no confidence on such two. :-) I am serious.06:31
wolfspraulsure understand06:33
wolfspraulaw: first step: add to rc3 errata06:33
awokay06:33
wolfspraulwerner will verify whether they improve anything, or add regressions06:33
wolfsprauleventually sebastien will have a chance to quickly look at and double-check this idea06:34
wolfspraulfor rc4, it's already clear we want this unless we find regressions06:34
wpwrakaw: (quality of rework) yes, that would be my concern, too. it would be fragile in the best case. one light touch, and it would break06:34
wolfspraulfor rc3, right now it looks like we will not add them06:34
awyup...i quitely prefer not to add those two resistors added to nor's two pins. it's not about i don't want to do. it's most related to that I am not sure those two wires connected to nor pins will break easily or not.06:36
awso i suggest to apply this to rc4 once Werner's result about this is good, and surely I can add this patch to mine and later confirm with that.06:37
wpwraksounds good to me06:40
wolfspraulI think we should skip the standby state when booting, i.e. when powered arrives over the DC cable, M1 should immediately boot all the way to rendering.23:59
wolfspraulwhat do you think?23:59
--- Fri Oct 7 201100:00

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