Rockbox Player - Project to design and build a Free/Open hardware audio player

[toffe]

For designing the case , you can use this http://www.emachineshop.com/.
Download the software for free, design the case in 3d, and it tell you the price you will pay.
This is ideal for a prototype or small quantities.
For the schematics and routing, , Kicad is a good choice I think.
The only thing is that the library is not big and missing components.
I will like to help but don't have so much free time, but I keep looking what you do.
It would be good to discuss on the irc channel also, it is faster than the forum..

[casainho]

For designing the case , you can use this http://www.emachineshop.com/.

- do you have experience with that company? Could you estimate a price for one plastic case, 1 unit?

I will be in IRC during the day :-) - and thanks for helping - stay tunned :-)

[spark]

Spark, can you please say what is your background, experience? - hardware design including PCB? Software? or both?

i have experience in hardware, firmware and software. personally i like to stay close to board level. i would like to contribute in hardware design.

[casainho]

Spark, can you please say what is your background, experience? - hardware design including PCB? Software? or both?

i have experience in hardware, firmware and software. personally i like to stay close to board level. i would like to contribute in hardware design.

Ok, about hardware, what can we do now? What do you think about the dev board?

Since you made the DSPdap, what would you change on that design? - for example, I would stay with the same DAC :-)

[spark]

Spark, can you please say what is your background, experience? - hardware design including PCB? Software? or both?

i have experience in hardware, firmware and software. personally i like to stay close to board level. i would like to contribute in hardware design.

Ok, about hardware, what can we do now? What do you think about the dev board?

Since you made the DSPdap, what would you change on that design? - for example, I would stay with the same DAC :-)

why not use ARM instead of AVR32?
i am a bit skeptical about the use of an AVR32 processor. we should be using the more universally accepted and supported ARM processor. Almost all the ipods, zune, sansa, ... use an ARM processor. i seriously suggest that we should take an opinion from the rockbox software guys as to which processor is the easiest to port to. There should be minimum efforts for them to port rockbox into the rockbox hardware player.
also ARM has better support for linux and other apps.

what was the basis for selecting AVR32 instead of ARM?

[Llorean]

Just a note: Don't think "for them to port to", think "for us to port to."

If you make your own player, you're probably going to be doing your own work for the porting, too. Ports happen by those who have the player, as we say with any hardware port.

[casainho]

why not use ARM instead of AVR32?
what was the basis for selecting AVR32 instead of ARM?

After talking in IRC a few times with developers of RB, almost everyone think that ARM is the best choice. Because of what you said, because there is already hand optimizations in assembly for ARM, etc...

AVR32 because of that dev board that is "dirty cheap", $73. An equivalent dev board with equivalent ARM should be more than $1000 . I think this price is very important to attract or not, developers!

The true is that this AVR32 have also Linux, GCC port, all full documentation, free/open dev tools and an old(transition from AVR 8bits) strong community! However this is new processor.

AVR32 also have DSP instructions, in spite of some old developers of RB said that is not important for RB, others said that is important because can lower cpu energy eat! And thats an important thing at RB! Some developers said that DSP is hard to program... and I would like to have your opinion, since you used a DSP for DSPdap
AVR32 ALSO is a DSP but ALSO is a MCU, DSP instructions can or not be used. DSP instructions are used "directly" from GCC, C code.

And as Llorean said, people that own the hardware will made their port!!! So, I think a cheap dev board and free tools are very important!! If cheap, more people can buy and than more developers.
Also scharkalvin, one that wants to develop, is favorable to AVR32, to this dev board.

To resume: IMO, AVR32 have advantages over ARM, because of very low price dev board, is a MCU/DSP and low power. Please read others old messages from this post!

Spark, If you think that an ARM will be better, please find a dev board and put here a comparison, specially the price and the highlights, the advantages over this AVR32 dev board.

Define the dev board, IMO, is very important! After that we can start developing

[eumesmo]

After development is complete, does a garden variety user need a DEV board in order to build player? As you noticed I'm a total newbie, have you considered scavenging something like the Motorola Motofone F3, for some 30 dollars you get ultra low power screen, keypad, joystick(?), battery, battery controller(?), sturdy frame(slick design), connectors(power, battery, display, keypad), mic, charger, perhaps even other useful items.

[Bagder]

AVR32 also have DSP instructions, in spite of some old developers of RB said that is not important for RB, others said that is important because can lower cpu energy eat! And thats an important thing at RB! Some developers said that DSP is hard to program...

First, "old developers" are against DSP in general for all the downsides a DSP provides. It is marginally better if the DSP is easily accessible using gcc, but only a little. If you really MUST use the DSP in order to get decent battery life that's even more arguments against the AVR I think since it'll then require your AVR port of rockbox to have DSP-adjustments literally all over the Rockbox internals and not just in the parts where you will optimize the codecs. All those modifications must be done by owners of such AVR-based targets and so far that's just this effort...

The current Rockbox code is fairly highly optimized for coldfire and ARM, and re-making all those for AVR will probably be a significant amount of work.

Also, as a side-note, I'm sure you can find cheaper ARM-devboards than 1000 USD, And you'll buy yourself a foundation that is a LOT less work to get Rockbox to run good on...

This said, I'll enjoy getting another architecture that Rockbox will run on!

[scharkalvin]

Also scharkalvin, one that wants to develop, is favorable to AVR32, to this dev board.

I have NOTHING against the Arm processors.  The Avr32 development board that is made by Atmel seemed to have most of everything needed for a GP development platform and it is dirt cheap (as these things go).  If it wasn't being used to develop a DMP device the board would make a dandy Linux network server, router, etc.

We are now using lots of small Arm cpu's in new products where I work, so I expect to get some first hand experience using them.  Most of these parts (2132, etc) are a bit on the slow side for good use with decoders (probably want at least a 50mhz part and what we are using tops out at about 25 I think).  

I've seen some cheap (< $150) Arm development boards on Fleabay but they seemed lacking in what was needed, though they did come with nice i/o (small color lcd, etc).  

Please keep beating the bushes for better ideas folks!  Just keep in mind we want something with a good life cycle (just hate it when someone discontinues the part you are using!).  At least Atmel has a good history in this regard.  They have discontinued very few good parts, and have had close sub's for those that they did.

[spark]

Just a note: Don't think "for them to port to", think "for us to port to."

If you make your own player, you're probably going to be doing your own work for the porting, too. Ports happen by those who have the player, as we say with any hardware port.

Sorry people. it should be "for us to port to". Anyways the rule still applies. if we select an already supported architecture, it will save us a lot of porting work.  

AVR32 because of that dev board that is "dirty cheap", $73. An equivalent dev board with equivalent ARM should be more than $1000

good ARM boards including all the things and even more than the avr32 board are definately available for within $300. I would prefer paying $200 more rather than spending weeks breaking my head porting the RB to a new architecture. Not that i have heaps of money, but in fact time is money.
i will see if i can find a suitable board and post it soon.

Some developers said that DSP is hard to program... and I would like to have your opinion, since you used a DSP for DSPdap

yes they are right. i would prefer if my codec code is written entirely in C and all optimizations are done by the gcc compiler itself. from my practical experience a 200MIPS ARM processor has better performance than a 400MIPS 16-bit DSP considering i am using fixed point C only code and no hand/assembly optimizations.  To effectively utilize the raw power of DSP you need to do a lot of hand optimizations unless the compiler is very powerful to do it on its own.

The current Rockbox code is fairly highly optimized for coldfire and ARM, and re-making all those for AVR will probably be a significant amount of work.

i strongly suggest we stick to ARM since "WE" have to do the porting work.  

[scharkalvin]

Just a thought here.  What about using an old iPac hand held as a building block.
IIRC they have a CF socket (in an adapter "jacket"), a very nice touch screen LCD, and since there IS a version of Linux that runs on this thing (with X support) there should be example code available to document some of the HW.  I don't know what (if any) D/A hw these things have, but I think they DID play mp3's.  And the cpu was an Intel Strong-Arm type.

[yapper]

iPac?  Did you mean iPaq? ... http://en.wikipedia.org/wiki/IPAQ

There is a bunch of info here http://www.handhelds.org/Compaq/iPAQH3600/iPAQ_H3600.html but note it is version 0.2f .... might be a bit "subject to change"  

[spark]

i've made a survey of possible ARM processor candidates.

Samsung   S3C2410A
----------------
http://www.samsung.com/global/business/semiconductor/productInfo.do?fmly_id=229&partnum=S3C2410

ARM920T core (266MHz)

Development board $299   http://www.armkits.com/Product/sbc2410II.asp

Cirrus EP9302
------------
http://www.cirrus.com/en/products/pro/detail/P1066.html

ARM920T core (200MHz) 208LQFP $22

Dev. board $312.5   http://www.cirrus.com/en/pubs/manual/EDB9302A_Tech_Ref_Manual.pdf

Dev. board $180      http://www.olimex.com/dev/cs-e930x.html   
(no USB device)

NXP LH79524/LH79525
-------------------
http://www.nxp.com/#/pip/pip=[pip=LH79524_LH79525_N_1]|pp=[v=d,t=pip,i=LH79524_LH79525_N_1,fi=45994,ps=0]|[5]

ARM720T core (76MHz) 176LQFP $13.66

Dev. board $389      http://www.logicpd.com/products/devkit/nxp/nxp_sdk

Atmel AT91RM9200
----------------
http://www.atmel.com/dyn/products/product_card.asp?part_id=2983

ARM920T core (180MHz) 208PQFP   $22

Dev. board $370      http://www.armkits.com/Product/ATEB9200.asp

please see the attached pdf file for detailed info.

[casainho]

about ARM and AVR32
The AVR32 family will be lower power than an ARM solution and has multimedia-oriented instruction set which is nice. You'll get faster execution for the same clock rate on an avr32, and much faster execution for the same power consumption.

If you're people have experience with ARM, you can think of UC3 as ARM7++ and AP7 as ARM9++/ARM11. If you reckon RockBox can run on an ARM7 then it will run better and with lower power consumption on a UC3. Otherwise you'll need to jack up to the AP7 series (The AP7001 is in a TQFP for easy soldering).

Also, the AP7 has an LCD controller so you can save money on the LCD, not having to buy one with a controller in. I read somewhere upon the launch of the AP7000 processor, it could decode 192kbps mp3 with a clock rate of just 3MHz. Or maybe it was 20MHz. I think it was 3 for mp3, 20 for MP4 video. I think.

Indeed the EVK110{0,1} is more expensive than the NGW100. but is indeed far less power hungry. Also, a UC3 uC isn't any harder to design a board for than an 8-bit uC so you should be able to whack up your own design in no time at all.

If you use google to search this site (google for "foobar site:avrfreaks.net") for threads like "avr32 vs arm" and the suchlike, you'll find some good, long, in-depth discussions which might help
http://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&p=393392

I will look carefully to ARM info dev boards that Spark did post.

About seel it assembled in kits
I am posting here an private email which I think there is no problem! I am contacting various people and I would like to have advices, I am not experienced ''' I think there are good opportunities for this FreeOpen hardware player and RB software

Hello Jorge,

that sounds nice. I think I can help you.

>player for use with RockBox". Do you know what is Rockbox?  --- Rockbox is an
>open source firmware replacement for a growing number of digital audio

Yes I know them. You developed also a USB Stack. I found it when I worked
on my diploma thesis
(http://www.embedded-projects.net/index.php?page_id=186).

>We are in the begining of developing our Free/Open hardware and we will need
>someone that make the pcb, assembly and sell It to us. I belive that companie
>that will do that will be able to make some profit, because Rockbox is very
>good and already famous, think at It as Linux for DAPs :-) and maybe It will
>grow more If this idea will suceed.

Actually I worked on a AVR32 based cheap starter board (about 50 EUR is
my target),
perhaps we can merge your idea and my idea.

>If you can help in make pcb, assembly and/or sell it online, It would be very
>nice to have your opinion and help.

No problem.

>If not, but If you are interested on this subject and can help in design
>hardware, software, etc... or just to know more, please visit us at:
>http://www.rockbox.org/twiki/bin/view/Main/RockBoxPlayer ;
>http://forums.rockbox.org/index.php?topic=6751.0
>
>About USBprog: maybe we will use AVR32 and then we wil need JTAGICE MK2, wich
>is very expensive the original... do you have any news about It clone on
>USBprog? :-)

We build an own AVR32 programmer firmware. Its not based on JTAGICE mk2
because there are already again private instructions from atmels. But we
have first results with the own solution which are used JTAG Boundary
Scan.

How can look our further work?

Whe we have got a working hardware, I can promote RockBox in Germany
(sometimes I write articles in Linux and Electronic Magazines)


I wish a nice good year for us