Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2020-08-26

00:08:07 Join ac_laptop [0] (~ac_laptop@
00:36:45***Saving seen data "./dancer.seen"
01:13:29 Join __Bilgus [0] (41ba23be@
01:16:36__Bilgusspeachy ( logs) Xduoo X3 battery bench 12 hours
01:44:55 Quit ac_laptop (Ping timeout: 240 seconds)
02:36:46***Saving seen data "./dancer.seen"
02:40:57 Join advcomp2019_ [0] (
02:40:58 Quit advcomp2019_ (Changing host)
02:40:58 Join advcomp2019_ [0] (~advcomp20@unaffiliated/advcomp2019)
02:43:55 Quit advcomp2019 (Ping timeout: 240 seconds)
02:57:31 Join advcomp2019__ [0] (
02:57:31 Quit advcomp2019__ (Changing host)
02:57:31 Join advcomp2019__ [0] (~advcomp20@unaffiliated/advcomp2019)
03:00:25 Quit advcomp2019_ (Ping timeout: 240 seconds)
03:03:54 Join lebellium [0] (
03:39:32 Quit __Bilgus (Remote host closed the connection)
04:36:50***Saving seen data "./dancer.seen"
05:07:08 Join johnb2 [0] (
06:36:53***No seen item changed, no save performed.
06:37:36 Join sakax [0] (~r0b0t@unaffiliated/r0b0t)
06:54:13 Quit lebellium (Quit: Leaving)
07:06:23 Quit Oksana (Read error: Connection reset by peer)
07:22:52 Join ac_laptop [0] (~ac_laptop@
07:27:16 Quit johnb2 (Ping timeout: 265 seconds)
07:53:24 Join akaWolf [0] (
08:36:55fs-bluebot_Build Server message: New build round started. Revision f791df1, 280 builds, 9 clients.
08:36:56***Saving seen data "./dancer.seen"
08:40:06speachy__Bilgus, a recalibrated curve and runtime current estimates are now committed.
08:40:59 Quit mendelmunkis (Remote host closed the connection)
08:41:01 Join mendel_munkis_ [0] (
08:50:00 Join johnb2 [0] (
08:50:51fs-bluebot_Build Server message: Build round completed after 837 seconds.
08:50:54fs-bluebot_Build Server message: Revision f791df1 result: All green
08:52:54 Quit tchan (Read error: Connection reset by peer)
08:52:58 Join tchan1 [0] (
08:53:08 Quit tchan1 (Client Quit)
08:53:27 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
08:56:44speachythat 12h runtime represents 50% better than the manufacturer's claims.
08:58:33johnb2There exist 2 different batteries: one with 1500 mAh (model has a button ESC, one with 2000 mAh).
09:02:15speachyjohnb2: for the x3 you mean?
09:02:53speachyprobably isn't any way to tell them apart at runtime though..
09:03:12johnb2maybe the website still refers to the first version (1500) or RB is doing better.
09:03:29johnb2No, probably not.
09:04:05johnb2AFAIK, it is just different labeling and different battery for the two models.
09:04:06speachyI don't think the OF utilized dynamic clocking, if the other ingenic/linux-based players are any indication.
09:04:57johnb2I used to have an X3 a few years ago - and just ordered one seeing you working on it ;-)
09:05:01 Quit akaWolf (Ping timeout: 264 seconds)
09:05:53johnb2I found some better photos of the device taken apart in a German forum and asked the guys whether I can upload them to the wiki.
09:06:36johnb2On head-fi some folks replace the dark screen with some transparent plastic.
09:07:09johnb2These photos show the screen from inside:
09:08:50johnb2In this thread they have a link to a replacement battery on Aliexpress, which are hard to find.
09:08:53 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
09:09:09johnb2Is it ok to add a link to it on our Wiki (X3 page)?
09:10:10 Join __bILGUS [0] (41ba23be@
09:10:27__bILGUSJOHNB you never cease to amaze me lol
09:15:02johnb21500 is the older version, BTW.
09:18:14speachyjohnb2: I ordered five of those batteries a couple of months ago. one's in my X3 now. they're not an exact replacement though.
09:18:28speachyand yeah, go ahead and add the link.
09:18:53speachyI assume the second internal photo is the smaller-capacity battery?
09:22:16johnb2That tiny Picture from post 93? He wrote, that it was glued too heavily, so he didn't want to rip it off to check capacity.
09:23:11speachythe aliexpress batteries have a regulator IC and are slightly shorter on the long dimension than the original.
09:23:54speachyalso slightly thicker. If I tighten down the case screws all the way a couple of the buttons don't operate properly.
09:24:24johnb2I read some threads where they said the player/battery got very hot during charging.
09:24:45johnb2Maybe that's without the protection circuitry.
09:26:20speachyyes. two reasons for that −− the first is that the CPU was running flat-out, even when idle. the second is due to the charging/battery circuitry.
09:26:40speachythe CPU is clocked _far_ lower now, and it's probably feasible to drop it further.
09:28:50 Quit kugel (Ping timeout: 256 seconds)
09:30:12 Join kugel [0] (
09:30:12 Quit kugel (Changing host)
09:30:12 Join kugel [0] (~kugel@rockbox/developer/kugel)
09:35:02speachycan't do anything about the fact that the USB power supply circuitry seems to always draw over a watt.
09:35:15speachy(charging is purely HW controlled)
09:35:23__bILGUSI dig that leather case idea
09:36:20__bILGUSwe can turn the charging on and off we could do d duty cycle of 3 on 2 off\
09:36:50__bILGUSgive it a cool down period
09:37:09speachy__bILGUS we are supplied a GPIO that tells us if we're charging. that's it.
09:37:34__bILGUSoh well that expected I suppose but unfortunate
09:39:35__bILGUSI was writing asm in god bolt lastnight no clue where the code for greydata came from in C but for one it doesn't go at 8bytes a pop it goes by a single byte and I think some of the logic is off on the shifts
09:40:27__bILGUSI'm secent at arm asm so i'm back porting the clip+ routine to c
09:40:58__bILGUSdecent, (not typing though apparently)
09:41:22speachyfixed the typo on the charger IC.
09:41:36speachyMCP73_8_63. :)
09:43:04speachyhas no external comms capability. I'd be willing to bet that there's no temperature sensor hooked up to it.
09:44:06__bILGUSthat is a bit scary guess this gets a metal box to charge in along with samsung devices
09:44:45__bILGUSminus the back it kinda in a metal box
09:47:09speachywonder if there's a drop-in replacement for that SSD1306 screen that doesn't suck so much.
09:48:44speachyso.. I guess the original battery curve as committed is the 1500mAh unit. that explains the patch in Poretsky's tree to hack in two battery types
09:50:19__bILGUShow did he do it as far as selection goes?
09:50:42__bILGUShow did he do it as far as selection goes?i think a menu with settable mah values would suffice
09:50:52speachythe discharge curves are pretty different
09:51:03speachythere's already an option in the UI to select battery capacity
09:51:27__bILGUShow did he do it as far as selection goes?i think a menu with settable mah values would sufficei thought so after I said that
09:51:37speachyhe abused the "battery type" configuration that was intended to select between alkaline and NiMH cells
09:52:08__bILGUSoh that is a decent idea
09:53:04__bILGUSyou could also just infer
09:57:00 Quit sakax (Read error: Connection reset by peer)
09:57:13johnb2speachy shows how the tinted screen eats up the brightness of the display.
09:57:22 Join sakax [0] (~r0b0t@unaffiliated/r0b0t)
09:57:46__bILGUSno I guess you couldn't was gonna say if runtime is > 6 hrs switch it but that would be annoying to have to run the first build for >6 hrs to get it to calibrate
09:58:46speachyjohnb2: yeah, I should do that to mine. but I was thinking a higher-res screen in the same physical case/board footprint.
09:58:52__bILGUSjohnb that is significant I wonder if I could get glass waterjet cut like that
09:59:18__bILGUSspeachy that screen reminds me of the Clip+ its not that bad
09:59:44speachyit's the same resolution as the clip+ but even without the tinted window it's a lot dimmer.
10:00:19__bILGUSwe can probably up the drive strength
10:01:43__bILGUSI think for the glass if I could get a rectangle blank of the right size I coul probably radius the corners with a diamond wheel
10:02:33speachy__bILGUS: we're already maxing out the brightness. I think anything else would require hardare mods.
10:03:30__bILGUSI'd have to look at the lcd code a bit closer but we should be able to set olded drive strength independent of brightness
10:04:22__bILGUSor irrespectivelying
10:04:56speachyI think changing that will require a hardware mod. section 8.8 of the SSD1306 datasheet
10:05:15__bILGUSoh and it has a 'dim mode setting' on the lcd controller wonder what that does
10:07:09speachyI don't see any reference to "dim mode", what version of the datasheet are you referencing?
10:07:19johnb2 : they use a Nintendo replacement screen as replacement for the "window"
10:07:27__bILGUSwell poop these are very rigid hardware as far configurability goes
10:08:29__bILGUSjohnb the plastic makes it feel cheap
10:09:18__bILGUSpiece of thick soda glass with some polish would raise it a bit out of case and feel very nice under finger
10:09:20speachyhuh, they cut the DS screen cover down.
10:09:28johnb2That is true :-) but probably rather easy to implement.
10:09:58__bILGUSill be doing it before I start using this in my truck
10:10:27speachy__bILGUS: word of warning −− this thing gets _hot_ when left in the sun.
10:10:58__bILGUSI'm in the habit of bringing it in in the summer
10:11:05speachy__bILGUS: that's the 1305, not the 1306..
10:11:37__bILGUSI found it by looking for the commands I didn't see anything listing the LCD controller
10:12:10speachyno dim mode on the 1306
10:13:34__bILGUShow did you decide it was a 1306?
10:14:03speachy1306 is 128x64 resolution. 1305 is 132x64
10:15:42__bILGUScrowd goes wild, you powers of dadduction have exceded my own
10:16:07speachyno idea about the actual OLED display though.
10:16:28__bILGUSand they probably dont either
10:17:18__bILGUSwell this should make my lcd exploration a bit simpler
10:18:31__bILGUSdamn I was excited when I thought it had that split screen like the clip+
10:19:06speachywell, you could swap out the display with one of those if you felt so inclined. :)
10:20:03__bILGUSnah thats just too much trouble
10:21:03__bILGUSwonder if this one has a dma too
10:21:07speachyjohnb2: were there actual links to the screens they used? I'm having a hard time finding something that's not a "stick-on"
10:22:31johnb2yes, but don't recall them. They said, just search for Nintento DS replacement screen on *bay
10:22:46johnb2which I did. They were around 8 bucks.
10:22:55speachy__bILGUS: I don't think we can take advantage of the 4760's LCD controller.
10:24:09johnb2that should yield two screens
10:24:17speachy__bILGUS: and we have to drive.. 12 pins?
10:26:08speachyhmm. the bitdelay() function isn't sane given the CPU reclocking.
10:27:17__bILGUSthat area I don't believe to be sane in multiple ways
10:27:42__bILGUSI'm still playing with it atm
10:29:20__bILGUSi think the display is in parallel mode
10:29:31speachyyes, it's parallel in the 8080 mode, I believe.
10:29:41speachythat bitdelay needs to be >=40ns
10:29:59speachy(it's really the "data setup time")
10:30:11__bILGUSthat probably explains some of the grey lib wonkyness good catch
10:30:41speachycould probably use one of the many HW timer peripherals to do this, assuming the pin can be remapped.
10:35:49speachyhuh, this thing has a PS/2 keyboard controller
10:37:00***Saving seen data "./dancer.seen"
10:38:15__bILGUSkitchen sink?
10:45:20speachyokay, we can't rely on the LCD controller or any HW peripherals to accelerate this
10:46:44speachythe I/Os used for the display are only muxed to the display controller and GPIO.
10:50:05speachyI'm using the OS Timer to implement udelay, we can piggy back on that (or do something similar with another timer) for the lcd delays.
10:54:18__bILGUSthat is probably sufficient I think thats how I did it for the Clip+ rewrite or maybe that was the SD
10:58:30speachyugh, my udelay is actually mdelay. oops.
11:02:52speachyif I switch it to using PCLK rather than EXCLK, that'll give us a 96MHz base.
11:03:16speachywith a prescaler of 4, that'll give us 24 clocks per us, which is coincidentally 41ns.
11:04:18speachymake that 42ns. so with that we'd have a worst-case 84ns setup time.
11:05:19speachythe next lowest prescaler is 1.. 10.41ns per tick, worst-case setup time of 52ns
11:05:53speachywant me to make that change?
11:07:13__bILGUSsure I pull in the new stuff before I finish the grey routine
11:07:55__bILGUSI should probably do that on master and try it too
11:18:02 Quit johnb2 (Quit: Nettalk6 -
11:19:29speachyokay, it's written. appears to work.
11:48:16 Quit prof_wolfff (Ping timeout: 265 seconds)
11:52:25 Quit fs-bluebot_ (Ping timeout: 264 seconds)
11:53:01 Quit bluebrother (Ping timeout: 264 seconds)
11:54:24 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
11:56:20 Join fs-bluebot [0] (
11:59:04fs-bluebotBuild Server message: New build round started. Revision f4bf27c, 280 builds, 9 clients.
11:59:19speachy...the new udelay() function was completely _broken_.
11:59:22 Quit bluebrother (Disconnected by services)
11:59:26speachyembarrasingly so
11:59:27 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
12:01:16fs-bluebotGerrit review #2697 at : xduoox3: Fix delay funciton in LCD code to be cpufreq-agnostic (WIP) by Solomon Peachy
12:04:45 Quit __bILGUS (Remote host closed the connection)
12:07:11speachyIt's running on my player now.
12:08:58 Join johnb2 [0] (
12:09:14 Join mendelmunkis [0] (
12:10:58 Quit mendel_munkis_ (Ping timeout: 246 seconds)
12:12:55fs-bluebotBuild Server message: Build round completed after 832 seconds.
12:12:57fs-bluebotBuild Server message: Revision f4bf27c result: All green
12:36:08 Join lebellium [0] (
12:37:01***Saving seen data "./dancer.seen"
12:54:44 Join MrZeus [0] (
13:03:10 Quit johnb2 (Quit: Nettalk6 -
13:31:30 Quit beencubed (Quit: Leaving)
13:37:06 Join beencubed [0] (~beencubed@
14:18:27fs-bluebotBuild Server message: New build round started. Revision a52eb1d, 280 builds, 9 clients.
14:30:25fs-bluebotBuild Server message: Build round completed after 717 seconds.
14:30:26fs-bluebotBuild Server message: Revision a52eb1d result: All green
14:37:02***Saving seen data "./dancer.seen"
14:37:32 Join prof_wolfff [0] (
15:11:50 Quit ac_laptop (Ping timeout: 260 seconds)
15:28:10 Nick mendelmunkis is now known as mendel_munkis (
16:37:06***Saving seen data "./dancer.seen"
17:01:31 Join ac_laptop [0] (~ac_laptop@
17:06:44 Join mendelmunkis [0] (
17:08:25 Quit mendel_munkis (Ping timeout: 240 seconds)
18:17:06 Quit pamaury (Ping timeout: 260 seconds)
18:31:15 Join akaWolf [0] (
18:37:09***Saving seen data "./dancer.seen"
18:53:24 Join Oksana [0] (~Wikiwide@Maemo/community/ex-council/Wikiwide)
19:42:05 Quit MrZeus (Ping timeout: 240 seconds)
19:56:01 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
19:56:23 Join fs-bluebot_ [0] (
19:58:51 Quit fs-bluebot (Ping timeout: 265 seconds)
19:59:28 Quit bluebrother^ (Ping timeout: 256 seconds)
20:02:25fs-bluebot_Build Server message: New build round started. Revision 8990c90, 280 builds, 9 clients.
20:16:32fs-bluebot_Build Server message: Build round completed after 848 seconds.
20:16:34fs-bluebot_Build Server message: Revision 8990c90 result: All green
20:35:25 Quit mendelmunkis (Ping timeout: 240 seconds)
20:36:20 Join mendelmunkis [0] (
20:37:10***Saving seen data "./dancer.seen"
20:38:30 Quit lebellium (Quit: Leaving)
20:42:26 Quit ac_laptop (Ping timeout: 258 seconds)
20:52:26 Quit sakax (Quit: Leaving)
21:28:25 Quit akaWolf (Ping timeout: 240 seconds)
22:02:56 Quit mendelmunkis (Ping timeout: 258 seconds)
22:09:50 Quit beencubed (Ping timeout: 258 seconds)
22:22:12 Join beencubed [0] (~beencubed@
22:26:33__builtinspeachy: out of curiosity, how are you generating those screenshots?
22:26:44__builtinmanually or fancier?
22:36:23 Join mendelmunkis [0] (
22:37:13***Saving seen data "./dancer.seen"
22:38:20speachyvia the uisim screenshot hotkey
22:38:45speachyif I wanted to be clever I could probably take the clipplus screenshots and convert them to all the same solor.
22:39:05speachyIf I wasn't going through the manual to fill in keymaps too it would be a lot faster..
23:20:55 Quit TheSeven (Ping timeout: 240 seconds)
23:21:16 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
23:32:08 Nick mendelmunkis is now known as mendel_munkis (

Previous day | Next day