| Author |
Message |
|
|
 Posted: Apr 30, 2012 - 02:31 PM |
|
Joined: Dec 29, 2010
Posts: 466
|
|
Your project should make an excellent xmega board.
I hope you're able to build it around those beefier xmegas. |
|
|
| |
|
|
|
|
|
|
Posted: Apr 30, 2012 - 06:21 PM |
|
Joined: Feb 13, 2007
Posts: 1026
Location: Gillies, Ontario
|
|
Thanks, I look forward to giving back to this community! I guess the big question now is... do I wait for the XMega384 or just proceed with the XMega256 I am using now?
Brad |
_________________
I Like to Build Bikes : http://www.AtomicZombie.com
I Like to Hack Stuff : http://www.LucidScience.com
|
| |
|
|
|
|
|
|
Posted: Apr 30, 2012 - 10:19 PM |
|
Joined: Dec 29, 2010
Posts: 466
|
|
Hi Brad,
With the limited resources (SRAM and flash) on the XMega256, why would you want to release a "light" product?
Your libraries should consume a great portion of that flash and exhaust the SRAM. What about resources for user programs? The XMega384 clearly has more than enough resources for this purpose.
Lets face it, if your intentions are to have us perform development/research, then we will need another board, one with more resources (the Xmega384). Otherwise, your board will not be much more than one great tech demo.
Also, you may want to re-think general user need when designing your board. People will need a development board for the Xmega regardless(not just the Xmega384!), so you may want to design a board with general-purpose development in mind.
I think most of us are tired of buggy crapware our Atmel has released over the years and want something decent.
Just look at Gabe with his boards, he designed a more than competent board that defined the Xmega32A4U, and many users now use his boards for purposes other than stock scope functions.
Just don't go about hording any of those Xmega384s! Otherwise, you'll end up pissing off a lot of freaksters.
Just my two cents. |
|
|
| |
|
|
|
|
|
|
Posted: May 01, 2012 - 02:57 PM |
|
Joined: Feb 13, 2007
Posts: 1026
Location: Gillies, Ontario
|
|
All good points, thanks for the input.
Currently my proto uses 15000 bytes of the 16384 bytes of the SRAM, but only about 10% of the program memory (it's an extremely optimized assembly system).
I agree that having only about 1K of SRAM for system vars is a bit tight! The 384 would offer a bit more breathing room there (2K free) and offer a second video buffer for auto back buffer flipping.
Currently, the system just draws as much as it can during the vblank and then clears the screen on the next frame transition. This still allows easy programming without needing the user to care about much timing and any interrupts, but constrains the system to the XMega to keep up to the 60 frames per second rate.
In my tests, drawing 16 sprites of 32x32 pixels using 256 colors was possible on the X256. Since sound routines are embedded into the sync period, they do not consume any main resources.
I do have a fully functional FPGA accelerated AVR game system ready to go as well, but want to start with a basic open source version first, using only a single XMega and a few passive components. I guess I am just testing the waters as getting 100 or more boards ready is not an inexpensive process!
Built this one 2 years ago, but my DIY bike website has been taking up all of my time lately...
My idea of the perfect inexpensive XMega proto board is one with all pins available and nothing else but a DC jack for a wall adapter. Basically, a breadboard ready Xmega.
I appreciate your input, as I hope to launch AVRCade.com either this year or next and hope it becomes a sister site to this one, dealing mainly with Freaks who want to learn and have fin with video, sound, and music based AVR projects.
Wow, I sure drove this thread off topic!
Brad |
_________________
I Like to Build Bikes : http://www.AtomicZombie.com
I Like to Hack Stuff : http://www.LucidScience.com
|
| |
|
|
|
|
|
|
Posted: May 01, 2012 - 03:39 PM |
|
Joined: Dec 29, 2010
Posts: 466
|
|
I thought Xmega384s will have 32k SRAM.
Making available all of the xmega pins will be of great benefit to users.
Another feature that most of us would appreciate is 3-5 headers for 3.3V connections (grnd and pos+) and a 5V regulator with similar 3-5, 5v connections. It gets frustrating finding extra connections to power our 3v/5v accessory devices without resorting to soldering!
Asking for ADC hardware support for external voltage references will be asking too much, but please do consider a 5v regulator in addition to providing availability for more power connections.
To save on costs you can make it mandatory for users to provide the 5V reference, then use a 3.3V regulator to power the Xmega and the several voltage reference pins. |
|
|
| |
|
|
|
|
|
|
Posted: May 01, 2012 - 04:10 PM |
|
Joined: Feb 13, 2007
Posts: 1026
Location: Gillies, Ontario
|
|
Yes, 32768 bytes of SRAM, which will use as follows...
15000 bytes : Video buffer #1
15000 bytes : Video buffer #2 & flood fill temp
00050 bytes : Used by A/V engine
00256 bytes : Reserve for stack
02462 bytes : Free for main vars
My AVRCade boards (including Lazarus-64) are video game systems, so they will have only the built in joystick switches, A/V jacks and a PDI port, nothing else.
If I can make it worth the cost of selling boards, then I intend to release a pile of simple XMega board with features much like the ones you listed.
I think a DIP breakout board with a DC jack on the top would be perfect for those who just want to dig into XMega without spending huge $$$ on peripherals they don't want. This is what I would have wanted...
- XMega 256 or 384 board with DIP pins
- 48 IO pins, VCC33, and GND
- DC jack on top for 9-12 V
- PDI port on top
I would also do similar boards for 512K SRAMs, DACs, ADCs, and all of the other wonderful chips out there that don't come in DIP format and often find their way bolted to my AVRs.
Brad |
_________________
I Like to Build Bikes : http://www.AtomicZombie.com
I Like to Hack Stuff : http://www.LucidScience.com
|
| |
|
|
|
|
|
|
Posted: May 04, 2012 - 02:06 AM |
|
Joined: Feb 13, 2007
Posts: 1026
Location: Gillies, Ontario
|
|
A present just arrived in my country mailbox...
Yeah baby, Xmega384! Now I am going to pop them in my video project and see what they can do at 50MHz.
Stay tuned...
Brad |
_________________
I Like to Build Bikes : http://www.AtomicZombie.com
I Like to Hack Stuff : http://www.LucidScience.com
|
| |
|
|
|
|
|
|
Posted: May 11, 2012 - 09:59 PM |
|
Joined: Feb 13, 2007
Posts: 1026
Location: Gillies, Ontario
|
|
Good news to report!
It seems that the XMega384 is working stable at 57.272MHz whereas I could only push the Xmega256 up to 46.534MHz. These odd clock speeds are a result of my base 3.5795 clock module and the PLL multiplication factor.
PLLx16 has been my goal as this allows my system to deliver 256 NTSC colors, all synthesized directly in software. Having the color burst base frequency of 3.579MHz multiplied by 16 allows 16 phase shifts (colors) and another 16 shades of each color. I managed a lot of colors from the XMega256 as well, but only as 13 colors by 16 shades.
I still have to run a battery of tests, but at this point, there is no problem when reading the full 384k program space using ELPM or anything else.
My program occupies 99.5% of the 384K, is pure optimized assembly, and uses 100% of the available 32k of SRAM.
Will report back in a few days with video screencaps as well.
Brad |
_________________
I Like to Build Bikes : http://www.AtomicZombie.com
I Like to Hack Stuff : http://www.LucidScience.com
|
| |
|
|
|
|
|
|
Posted: May 14, 2012 - 06:53 AM |
|
Joined: Nov 28, 2004
Posts: 3628
Location: San Diego, Ca
|
|
| Brad, did Atmel sample you those Xmegas ?! I can't see where to buy them ... |
_________________
1) Studio 4.18 build 716 (SP3)
2) WinAvr 20100110
3) PN, all on Doze XP... For Now
A) Avr Dragon ver. 1
B) Avr MKII ISP, 2009 model
C) MKII JTAGICE ver. 1
|
| |
|
|
|
|
|
|
Posted: May 14, 2012 - 01:50 PM |
|
Joined: Feb 13, 2007
Posts: 1026
Location: Gillies, Ontario
|
|
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
