« first day (882 days earlier)      last day (4069 days later) » 

1:54 AM
@tyblu where are you?
 
2:27 AM
Has @anindo been missing?
I wanted to talk to him about the new user kit question. I think we need to do a roundup, lots of those exist.
 
 
3 hours later…
5:08 AM
hello all
@Kortuk i think @anindo mentioned some event in this period
 
 
2 hours later…
7:07 AM
good morning @ all
 
@jippie good morning
hows the project going?
 
@Rick_2047 not too bad at all. but there is still a rounding error (resulting in a 256us error)
 
what sort of rounding error?
 
so I have to pull out a function generator today or tomorrow
I have an counter that overflows and I count these overflows
 
okey
?
 
7:14 AM
in the measurement routine I take the overflow counter, shift it left 8 bits and stick the current 8 bit timer value on it.
 
hey, that is clever actually
 
it works unless the counter value at that moment is near 0 or near 255.
 
why?
this should work regardless
 
because you can get an overflow while I'm in the interrupt routine for the zero crossing
 
oh,
 
7:16 AM
and because every instruction in that interrupt routine add at least a 1 count to the counter in parallel ...
 
why don't you stop the counter as soon as you enter the zero crossing ISR?
 
I need to do some smart correctio
 
and restart it when you leave the ISR
 
ultimately I want to average across several cycles
and not stopping the timer will then give me more accurate numbers
maybe I need to include some assembly.
because I loose a lot of cycles PUSH'ing registers on stack
 
can you show me your ISR?
 
7:24 AM
@Rick_2047 I want a good shot at solving it myself first.
But if I can't solve it I'll ping you :)
Also I need to figure out a way to reproduce the issue, without having to wait for mains power to accidentally cross that line.
 
you have any debugger?
debugger + a switch will help a lot
 
I have a disassembler
 
naah, i mean like jtag debugger
which would allow you to pause the controller
 
7:40 AM
nope
@Rick_2047 I'm really thinking about a programmer that supports debugWire, but they cost a lot of money and I doubt if I can reproduce this problem well enough to catch it
But I am convinced that I understand what is happening, I just need to adjust a 'calibration' parameter :)
As I look at it from a 'calibration' point of view and probably being able calculate the calibration values automatically after compile, I can easily store them in EEPROM. Hence my EEPROM questions :)
It's just that when I change the code (or perhaps compiler) the required values may change :-/ Hence my thought of using assembler.
That example program uses the same mechanism I use. It mentiones fuzz = 8, I came up with a pretty similar mechanism, I just have to figure out what exact number I need.
@Rick_2047 Do you know how I would include assembler in the C-source?
 
8:18 AM
@jippie do you want it inline or as a separate file
?
if its just a small piece of code then you can do it efficiently using inline assembly
@jippie this is a good resource for inline assembly
its really easy to insert assembly, but you would have to know how to write assembly
 
8:38 AM
@Rick_2047 inline
@Rick_2047 the timer0 isr
 
yeah, then just write it like this asm volatile ("insert code here");
 
@Rick_2047 compile from C, then copy the resulting disassembly to the source file, optimize it a bit for speed.
 
yeah
that should work
 
@Rick_2047 I want to ensure that the timing stays accurate, even when a new version of the compiler compiles the code
 
heres a question
in the ISR for the zero crossing, when do you read the timer value?
 
8:41 AM
first thing
ISR( INT0_vect ) {
 
i think your isr is too long then
 
uint8_t myTCNT1 = TCNT1;
t = timerOverflows;
then some processing
 
instead of writing a long isr and optimizing assembly, you should try to write a small ISR
 
but reading 5 bytes does five pushes
?
the first thing I do is read the timer and it is already too long?
 
@jippie: I usually set a global flag in my ISR and then process it in my main
 
8:43 AM
no, the "then some processing" part is too long
processing should not happen in the ISR
@StaceyAnne everyone should do this
 
8:56 AM
@Rick_2047 I agree, but i know my ISR will only be called every 20000 clock cycles. That equals some 15000 instruction. So a single if/then should not be an issue.
/me is out to the office
 
@jippie that is not a recommended practice though, consider moving away from it
 
@Rick_2047 i will consider.
/me is out
 
 
2 hours later…
10:50 AM
@jippie, missed you but yep was just wondering about how the eeprom defaults from C went. Sounds like good advice from @Rick_2047 and @StaceyAnne above on your ZCD project, I'd expect to get in the order of 1uS accuracy so trying to calibrate will probably hide a logic problem that'll bite you later.
 
Is there anyone around that has experience with uBlaze, specifically AXI?
 
11:22 AM
@StaceyAnne, can't help there, my most complex FPGA / VHDL code counted from 0-9999 on a 7-segment display and while it worked it consumed about 25K gates so I'm guessing it wasn't good code at that.
 
@PeterJ, It's amazing how far FPGAs have come, It's easy to get away now with coding completely in C for most applications using the uBlaze. All of the VHDL is pre-written by Xilinx. I'm a bit torn about it though, since I prefer VHDL over C.
 
@Kortuk I think i have a solution to Davids answer in this meta question. We could have a blog with articles ranging from simple how-tos, tutorials through to more complex things like choosing a FPGA for a project.
 
11:38 AM
@StaceyAnne, I'm a bit the other way having more of a software background. I bought a Digilent board a few years back for general learning to have an extra skill, but every time so far I have a real project where I think an FPGA would be a great solution I seem to think of a way to use say an ARM and some simple extra logic / clever programming to do it.
That's just me though of course they have their place depending on what you're doing.
 
I saw this done with the DIY SE, as they recently had an article on building sturdy shelves in the garage.
 
@PeterJ, I work mostly in the aerospace/defence/medical industries, where they're mostly used for pre-processing high speed data in preparation for CPU processing.
 
hey @StaceyAnne you are good with fpga right?
can I get your opinion on the papilio fpga platform?
 
@Rick_2047, It depends on the application, but I guess so?
 
@StaceyAnne i want to implement a processor on it
like a general microcontroller
 
11:47 AM
@StaceyAnne, sounds cool, they certainly are good for high-performance / relatively low volume applications, what I work on is normally too mundane to really need them, normally either not enough high speed requirements or something near enough off the shelf.
 
@PeterJ i think that is the thing keeping people away from FPGAs in general
there is not much which an fpga can do and a controller can't
 
@Rick_2047, Papilio looks pretty cool
 
it has a spartan-3E, I would get the 500 one
do you suppose something like an 8051 would fit into that?
 
It is a good starting point, I'm inclined to prefer Xilinx FPGAs, and I like how they have a wrapper around the code, that really helps (initially) with the HDL
 
oh, i am not really a beginner to HDL, have taken classes and have simulated a lot of things in Verilog
but somehow i feel implementing the whole processor v/s a few parts on a physicalfpga will be a lot different
 
11:54 AM
are you looking to implement the processor "manually" as a learning exercise or just for the processor functionality?
 
manually
first I would learn by implementing something like 8051
and then would implement my own design which I have made
 
@Rick_2047, I'm inclined to think you should be ok with implementing a uC on the 500
 
okey then
 
It's been done before, so it looks like it's big enough.
 
where has this been done?
would be a good reference
 
12:02 PM
I'm not exactly sure about the 8051, but here's a discussion about running Linux on a 500 fpgarelated.com/usenet/fpga/show/69376-1.php
The uBlaze is also supported on the 500
try also looking at opencores.org/projects, there's a processor section
 
@Rick_2047, while a 8080 here's something I've been meaning to try on my 3E, space invaders fpgaarcade.com/spc_main.htm but it's probably a good indication a processor of roughly that complexity will fit with some other logic
 
 
1 hour later…
1:15 PM
Would this question be a good migration candidate for EE: diy.stackexchange.com/questions/26394/…
 
1:41 PM
in EE - Ask a Moderator, 4 mins ago, by Kortuk
@Tester101 it is not considered a good fit for EE.SE.
 
2:03 PM
@Kortuk Does that mean it's not populated on the PCB during manufacturing?
 
@NickHalden What?
@BMitch I would bet Super User has a real question somewhere about using lead acid batteries for a computer battery.
 
@Kortuk Haha there was just a question about what "Do not Fit" means in terms of a PCB
 
@NickHalden ha. Smart arse.
 
@Kortuk =P
 
@Kortuk I had asked over there too, they didn't want it either. :)
in Root Access, 25 mins ago, by Journeyman Geek
@BMitch: eh, its one of those super wierd, nichish questions that dosen't really fit well anywhere IMO
 
2:20 PM
@BMitch We have had the exact same question 3 or 4 times on EE.SE, it is not that niche. People want to do it, it just does not fit our site. Moding the battery for your computer seems like Super User but I have been wrong on that more then once.
 
2:31 PM
@BMitch could it be chemistry?
they have an ad on our site about battery composition etc
 
@NickHalden lol, all I know is that it wasn't "Home Improvement" :)
 
2:48 PM
@NickHalden Nope (Chem mod here). This depends on the laptop internals, not the battery
 
3:07 PM
@NickHalden yeah, definitely not.
 
Haha ok I didn't actually read the question =P
 
Howwwdyhoo
lol
Mr. Hankey
 
3:28 PM
@abdullahkahraman morning.
 
Yay!!! Commit to the Arduino SE site! It's about time :-)
3
 
3:46 PM
If enough people feel that it should be, I can pin that link :)
(Better if @Kortuk weighs in)
 
@ManishEarth @Kortuk hates the arduinoSE IIRC
 
@NickHalden Not hates, he just is against the idea when it's basically the result of problems on EE which can be fixed. I have similar feelings
 
@ManishEarth I have no feelings on the topic
 
thumbs up
 
I think I can kill it on ArduinoSE though, I've done a loooooooooot of work with them =P
 
4:01 PM
@NickHalden "hates" is not a corect word.
wait, @ManishEarth already said it, thanks @ManishEarth
 
@Kortuk np :)
 
@Kortuk I seriously don't know why both of you are in the same room - so much redundancy ;)
 
@W5VO hey hey, I just say whatever @ManishEarth says.
 
@Kortuk apparently Steve Wozniak is coming to our campus
 
@W5VO sounds like fun!
you gonna make that meetup this weekend?
 
4:09 PM
@W5VO cool :)
 
@Kortuk hopefully
I may need an invite so I can do it on my phone
it's weird that you can't join them in the same way as on the desktop
 
@W5VO Awesome, I ordered something I could not find at my house for my project, so i am looking forward to spending that hour goofing with things also.
 
@W5VO I ran into him once at a hotdog stand. He dropped by on his Segway. And by "ran into", I mean I was eating my hotdog and he showed up and ordered a hotdog, and then he went on his way wherever he was going.
 
@ThePhoton haha. His modded segway?
 
@Kortuk I should have asked. I don't even know what kind of hotdog he ordered.
 
4:13 PM
@ThePhoton haha. My wife will be unhappy with your story.
 
He had a couple of minions with him, on their own segways.
 
@ThePhoton if I am ever in his place, I will have hordes of minions.
 
@Kortuk It was a deep connection, because it was at Top Dog. I went there because Top Dog was the local hot dog stand at Berkeley, and I was having a nostalgic hot dog. And the guy at the counter told me Woz drops by there every week or so for a nostalgic Berkeley-style hot dog as well.
 
@ThePhoton haha
 
Unfortunately San Jose Top Dog is closed now (or renamed to Supreme Dog so its not the same at all) and so Woz and I can only get our nostalgic hot dogs when we are actually in Berkeley.
 
4:17 PM
@ManishEarth ohh, you can pin it. I can do it for oyu.
 
@Kortuk It's Sunday, April 7th at 7PM. Open to the public
 
@ManishEarth can you tell me how to pin our upcoming event for people to register?
 
@Kortuk the little side arrow on a message> pin
 
@Kortuk It's in the same drop-down menu as the delete function
 
@W5VO Where the hell is the event to access that button, I only see it by going to room>schedule events.
 
^^ Make a message like that
then pin
 
@W5VO where the hell is that.
 
@Kortuk Go to the room info link (top right), and then select the schedule tab
 
@W5VO I am on that tab, I just dont see the button. I am feeling like an idiot right now.
 
@Kortuk you may want to edit the pinned message above to be a bit more explanatory about the event. (Also, you may want to unpin your other "I'm leaving the google hangout on Saturday"... message)
 
4:24 PM
@ManishEarth I was going to have two short and sweet messages, since it clips the message, do you feel different?
@ManishEarth We might want to discuss such policies during the event on saturday :)
 
@Kortuk Nah, your choice. Just that I wasn't sure if you wanted both -- one had a link and one didn't. Wasn't sure if you knew that you could edit other people's chat messages :)
@Kortuk :P
 
@ManishEarth :) You gonna join us.
 
@Kortuk Not me. Mortuk! :P
Not sure though. Probably will
 
@ManishEarth Please do. You can answer any moderation questions. :)
I look forward to meeting my nemesis, Mortuk.
 
21:30 on Saturday -- can do :)
@Kortuk There's only one answer to moderation questions:
 
4:27 PM
@ManishEarth Suspend all users, protect all questions, take a vacation
6
 
haha
 
@W5VO exactly
 
@W5VO did you unpin my other message :(
 
@Kortuk careful, even mods can't repin stuff that they have pinned (you need to summon another mod for that)
 
@ManishEarth I did not unpin it :(
 
4:32 PM
@Kortuk yes
 
@Kortuk So, @W5VO and I can repin it if we need it
(just gimme a link if you want repinning)
Be sure you want it repinned though
 
I am fine for now.
 
@Kortuk ah :)
 
@ManishEarth I added a link to it in yours, just gonna move on.
 
@Kortuk Makes sense :)
 
4:34 PM
@ManishEarth Mortuk better make it this weekend.
 
This pinning business is confusing
is scared of Mortuk
..and Kortuk
 
@ManishEarth little bit.
 
@Kortuk I was actually saying /me is scared of Mortuk, but that works as well :P
 
@ManishEarth haha
 
M or K tuk
 
4:38 PM
The real question, am I a turk.
 
@Kortuk No, but you like pink meat in a can
 
 
2 hours later…
7:02 PM
@PeterJ 1us at what clock speed?
good morning @all
 
 
1 hour later…
MLM
8:07 PM
@jippie Good afternoon!
 
 
3 hours later…
10:57 PM
@jippie evening.
 
@Kortuk Nearly beer o'clock.
 
11:08 PM
@jippie, I was thinking at 20MHz around 20 cycles should be enough for ISR latency and to capture and reset the timer, are you using something slower?
 

« first day (882 days earlier)      last day (4069 days later) »