#milkymist IRC log for Wednesday, 2013-01-09

azonenbergwpwrak: i have a board here http://uploads.oshpark.com/uploads/project/top_image/xcxzJIC7/i.png that is along those lines but a little more complex04:49
azonenberg4 layers since i wanted to make it tiny :P04:49
azonenbergFT232H for JTAG plus an oscillator and an XC2C32A/64A CPLD04:50
azonenbergshould be a self-contained dev board04:50
wpwrakhah, so many excess components :)04:51
wpwrakmy idea is rock bottom simplicity. but yes, it'll need a ben to be useful.04:52
azonenbergI wanted enough LEDs and buttons to be able to develop my toolchain on it04:52
azonenbergi wanted access to a lot of pins04:52
azonenbergotherwise you could be simpler04:52
wpwrakit's basically an extension of wolfgang's idea to have a disposable fpga module, in case one gets damaged by configuration experiments.04:52
azonenbergmakes sense04:52
wpwrakbtw, do you have a rule-of-thumb estimate for the current on the 1.2 V rail of an lx9 ?04:54
azonenbergI haven't stressed my lx9's much04:54
azonenbergBut, low04:54
wpwraklike 100 mA ?04:55
azonenbergThat's probably high04:55
wpwraknice :)04:55
azonenbergi pulled ~100 from an LX25 at 160 MHz with a NoC prototype and a softcore CPU04:55
azonenbergbut my lx9 boards are not instrumented04:55
azonenbergso i can't give you numbers04:55
wpwrakyeah, a ballpark number is all i need. just to know what sort of LDO it would need.04:56
azonenbergI've been using 1117-series on my little boards and it's fine04:56
azonenbergThose numbers are for 3v3, 2v5, 1v2 rails on the entire board in the upper left04:56
azonenbergmost of the 3v3 is consumed by the 10/100 ethernet04:57
wpwraklooks pretty massive ;-)04:57
azonenbergwhen the DDR RAM is active it pulls like 500mA from the 2.5V rail between the core, I/O, and FPGA04:57
azonenbergI don't think i've ever gotten an lx25 above 150 mA on the 1v2 rail04:57
azonenbergso 100 as a ceiling for an lx9 sounds entirely realistic04:58
azonenbergyou can see all of the little boards are lx9 based04:58
azonenbergHave you looked at my lx9 mini board btw?04:58
wpwrakthe cheapest 1.2 V LDO at digi-key does 250 mA. perfect :)04:58
azonenbergi have the kicad files for the little lx9 boards in that picture public04:59
wpwrakyes, yours looks like a good starting point for massive simplification :)04:59
azonenbergDepends on how ismple you want05:00
azonenbergi wanted enough peripherals that it was still useful05:00
azonenbergbut its far from disposable in that state05:00
wpwrak(starting point) e.g., i trust that you get all the details like caps right. having a reference reduces the amount of chasing :)05:00
azonenbergwell it works05:01
wpwrak(far from disposable) yeah. my aim would be just a bunch of headers and the minimum set of on-board peripherals. i.e., maybe 4 LEDs. maybe a button if it should be really fancy.05:02
wpwrakanyone doing complex stuff will probably a more standalone board anyway. i.e., something more like yours.05:03
wpwraks/probably/probably want/05:03
azonenbergoh and the cpld board is self-contained05:08
azonenberghas onboard jtag05:08
wpwrakyeah, the evil ftdi05:08
azonenbergannoying but i havent yet made anything of my own that is better05:08
azonenberghave you guys had bad experiences wit hthem?05:08
azonenbergi spent a lot of time hunting down bugs from their driver, libftdi was just as buggy05:08
wpwrakwith a slightly smaller one (232, not 2232), yes. basically doesn't let you properly toggle bits. so you must get the 2232. besides, documentation is (was ?) woefully incomplete.05:09
azonenberg232h allows mpsse mode05:10
azonenberg232r does not05:10
wpwraki think it was the r. lemme see if i have the design still around ...05:10
azonenbergyeah, the r is uart only + basic bitbang05:11
azonenbergthe h has mpsse05:11
wpwrakyeah, FT232RL05:11
azonenbergthe original 2232 did not do mpsse either05:11
wpwrakthe basic bitbang would have been enough if it worked05:11
azonenbergit didnt? o_O05:11
wpwrakloses data05:12
wpwrakmaybe it's something i did wrong, but without proper documentation i couldn't find out what05:12
azonenbergi had issues with data being lost on reads too05:12
azonenbergon the 2232h05:12
azonenbergtook me like a week to figure out a hack of repeated reads and polls that seems to work05:12
azonenbergi do not like it, but it's the best i've found so far05:13
azonenbergi will be making a custom jtag dongle soon thouhg05:13
wpwraki eventually switched the project to using a ben. felt good to be in control :-)05:13
wpwrakthe ben is great for all that in-circuit programming. just take an UBB, make whatever adapter weirdness your board requires, write a few lines for the ISP (if none doesn't already exist), done05:14
wpwrakeven pretty nasty protocols are no big deal05:15
azonenbergwell the goal here was on-board programming05:15
azonenbergi wanted something i could integrate int oa board and make it not need external hardwar05:15
wpwrakyeah, ftdi like projects like yours ;-)05:16
wpwrakbut seriously, ben+UBB is a great choice. UBB is as simple as it gets and programming all that stuff in linux but with direct access to gpios is incredibly satisfying after having experienced some of the alternatives05:18
azonenbergwell i already have libjtaghal working for most of this stuff05:20
azonenbergi just need to build myself a smart in-circuit debug module05:20
azonenbergso i dont have four usb packet latencies being added to every operation05:20
wpwrakyeah, round-trips, the dark side of USB :)05:21
azonenbergmy new module will have an fpga and gig-e05:21
azonenbergand do as much smarts as possible in the adapter05:21
wpwrakoh, and the ben also provides power and a clock. all for free :) the only case where this arrangement failed me was for RF. there, the clock proved to jittery. i mean, the device "worked". but you didn't want to see what it put into the ether ...05:22
azonenbergmost of my boards need more current anyway05:23
azonenbergand some have unusual clock requirements05:23
azonenbergthere are some things i'm sure its good for but this isnt one :p05:23
wpwrak(nasty spectrum) here are various tries "before" (using ben-provided clock): http://downloads.qi-hardware.com/people/werner/wpan/rework-20110113/05:24
wpwrakand this was with a local crystal: http://downloads.qi-hardware.com/people/werner/wpan/redesign-xtal/05:24
wpwrakwell, "before" ... there was a "before" even before that: http://downloads.qi-hardware.com/people/werner/wpan/rework-20110112/05:25
wpwrakthe art of making the signal disappear in the noise ;-)05:25
wpwrakshould make a good question for an EE exam. explain all the things that are wrong with this picture ;-)05:27
azonenbergCan the MM1 do 1080p btw?05:29
azonenbergor only 72005:29
wpwrakthe highest i've seen is 1024x768 "XGA"05:30
wpwraknot sure if lekernel has a faster variant, though. he might.05:31
azonenbergwell spartan6 cannot drive 1080p with the SERDES onboard but it could drive an external HDMI PHY05:33
azonenbergwhich would allow it05:33
wpwrakah, i see. well, M1r3 has just VGA :) and M1r4 (with DVI) never made it beyond a paper exercise05:49
wpwrakabsence of demand vs. cost of making a prototype run05:53
Fallenouwpwrak: I thought the remaining R3 units had been reworked as R4 units09:01
wpwrakFallenou: only some fixes were made. e.g., i think they now all have a "good" reset circuit.11:01
mwalleFallenou: what tests are failing?23:45
--- Thu Jan 10 201300:00

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