The Pi DAC isnt the best, but it seems passable. I've been listening to one all week and have not noticed it being bad. On paper its not first rate. Best I can tell the Pi DSP is generating the DAC via PWM.

I've made some more progress. My living room player is pretty much finished - I cant complain I cant see it ;-) Looks a little better with trim and paint, I just need to add the equipment feet to complete it. It looks better in real life the video camera adds flicker that you cant see in person.
http://www.youtube.com/watch?v=zHSGtInrsDY

Thanks to Patricks kind help the consumer IR works. I could not make the IR decoder reliable, PCA used his wizzy logic analyzer to debug it. He re-wrote the consumer IR code on the front board PIC. After a few problems of my own making I managed to get the IR data returned by the front panel board to match that the EMPEG sends via /dev/IR. Also made the front panel buttons and the LED work. The front board PIC code is now complete.

http://www.youtube.com/watch?v=S2YEHl_ujC0

It still has a few issues but I may be in a position to sell some boards when the final spit and polish is completed.


The player has two protocols really, one for output that is just pixel data and on for input. The input protocol is plain text that Centro interpretets, this can be delivered via a named pipe or via UDP.

If a file exists in the file system, FUSE, NFS, SMB doesnt matter - the player can be told to play it by sending a command and the full file path to the intpreter. Practical result if you have any content than mplayer will decode as audio you can probably play it by reference alone (on the Pi that is).


Yep that works, very simple in fact. Just put the empeg into slave mode, can be done via the command line so it starts in that mode. You could argue though if you can get the player into the car its probably better to just use the player :-)

Jon

We got a prototype box sorted out for the Raspberry Pi based Centro player today. I personally think it looks quite nice!

pca

And the back view.

pca

That looks great! Good work guys.

Beautiful

Is this idea something still in the pipeline or has the emphasis shifted entirely to the hardware side of things?

tanstaafl

To port the application to a tablet running Linux (and just Linux) natively would take a few days. I did some reaseach and could not find a tablet that I could just install a native Linux on, i'm sure its possible though.

To the run the player as an application on a tablet is a months solid work, I am simply not prepared to put that kind of time in for a handful of possible users.

Jon

Not even for ME?

Well, as it turns out, it looks like I may be able to use my empeg again after all.

The Mexican government has made a few changes to the immigration rules. Up until a few months ago, I could remain a Mexican immigrant on "non-inmigrante" status indefinitely. Now they have restricted that to four years, and after that I have to change my status to "permanente", which is advantageous in many areas, such as not having to renew my non-inmigrante status every year, I can legally be employed, a few other things.

The down side is that a permanente resident is not allowed to own a foreign-plated automobile. My particular car cannot be "nationalized" to Mexican license plates, because (1) only cars that are 2007 models or older can be nationalized; and (2) my car was not assembled in a NAFTA country (VIN number starting with a 1, 2, 3, 4, or 5). So, sometime before December of this year I have to drive my car (a 2009 Honda Fit with Navigation) back to the U.S., sell it, fly back to Mexico and purchase a Mexican plated car.

To purchase a comparable car in Mexico will cost me three to five thousand dollars more than I can sell my car for in the U.S., plus the expense of taking my car out of the country and flying back. This is so the Mexican government can protect the Mexican car dealers.

What all this means is this: My "new" Mexican car (it will be about a five year old Honda CRV, I think) will NOT have navigation, but a DIN-sized AM-FM radio which will be replaced by my empeg. I won't bother with a tuner kit. I had one in my last empeg-equipped car and to the best of my memory I tested it to see that it worked, then never used it after that.

So, in another six months or so I will be empeg enabled again. I am really looking forward to that!

tanstaafl.

Over the past week I finally got 5 minutes to myself, to attempt to dissassemble the dash on my 2010 Hyundai. I REALLY want room to fit my Mark2a. But it will require either ditching the stock system (which I actually like) or a lot of hacking structure.

I hopped on here today on a whim, first time in months, and what do I see...

First, great work! Im so glad to see some new life for what is STILL arguably the BEST player (software, and hardware) for car use ever.

Second, I have a few questions, even after reading this whole thread:

- Is the new player software a feature complete empeg re-do? I.e. in all the youtube vids, I see only visuals. Is the other part of the player SW there? I.e. playlist? "down down down" shuffle all? Browse music, enqueue songs, etc?

- The new display board: Heck yea! Finally a solution to my "dimmed out" VFD problem. My only request - can this PLEASE be made so that the empeg can be remote mounted? I.e. max display cable length measured in meters? At least say 50cm? That would let many of us breathe new life into our old empegs... I know for sure myself and my brother, and cousin (who bought one of the first Mk2's) would all consider this.

Where can I read more? I have to share this. I think, despite the "slow" response from the dozen or two people on this board, that once word filters out, and this becomres more documented and "turn key" there will be many more adopters of this.

This is what I have dreampt of (mostly) for ages. Being able to BUILD a modular "empeg" with the SAME GREAT interface/options! One that will fit with a modern car, has different display/control options, etc.

Wow. Great start. Im going to be visiting this thread a lot, and will be ordering a RP today...

J

In regards to the tablet route (not sure I even care, but this would open it up to a lot more potential users) I too found almost NO tablets that you can "just install linux on". I guess the way tablets are manufactured, its a huge task to get the sources for the various drivers/tools/etc necessary.

This looks promising though:

http://www.omgubuntu.co.uk/2013/03/world...but-is-it-legit

https://www.intermatrix.com.au/linuxtablet/preorder.php

So, perhaps soon(ish) we can have an ubuntu unity tablet with a NATIVE empeg/player.

just downloaded the sources.

Just as an FYI that stuff is based on something called libhybris which was born at a slightly inebriated session at Devaamo (a Finnish OSS event) when I said "how hard can it be to use the android bionic drivers with a glibc userspace? We could do ...."

From there Carsten Munk developed the idea in Mer and Ubuntu quite happily stole it, declined to credit the original authors and took it in-house to develop it further. Nice eh?

Still - they eventually saw the light and are now contributing to the upstream code again.

What makes this stuff really neat is that you can now use Mer on top of any Android BSP.

BTW - if you want a truly open tablet then see:
http://makeplaylive.com/

And if you want something to develop a player on then you should seriously consider Mer ... http://www.merproject.org/

.... but then I would say that wouldn't I?

Maybe I am getting old or something ? I read the entire page referenced here, twice, and I still have no idea what this project actually is or what it does ?

The Wiki "About Mer"

Well that clears that up then ........

Nope !! never really used my empeg on anything other than Random play. I did not use Emplode in the past 8 years as it crashes on my Windows boxes. I never really wanted mp3 management as such so I have put zero effort into that part of it. My hope was if people where interested they would take it and develop it, I suspect its a bit of a dead end project but it keeps me entertained :-)

Centro builds what I called (for want a better term) playlists. They are nothing more than a list of filenames from a directory containing MP3 files. Its very crude, the best that can be said is that it works (mostly).

You can enter track by number from the remote control, skip forwards and backwards and that is about it.

I will improve it one day, I concentrated on the hard bits - getting the real time audio/visuals to work on 3 platforms (Linux/X11, empeg hardware, Raspberry Pi hardware), other than those parts it offers little but a few subtle bugs and lots of missing features ! My honesty is why I am not in marketing.....

Cheers,

Jon

OK - it's my baby so I have to post this

First - you really do have a point - we could improve the brochure aspect quite a lot and thanks for taking the time to read the link

But this doesn't feel accurate:


So let me help you with that... the first line is:


So if you're not technical (which I find hard to believe) then following the link on the next line : here's a non-technical page "about Mer" - it says


Does that clear it up?

Sorry I was telling the truth. You have to understand that I started coding on CP/M on Z80 using assembler moving to C a few years later. Had brief few years with Delphi/Windows before landing back in the world of linux/gcc.

I find it difficult to work out what the proposition is with anything that offers a "framework" or "SDK" as these ideas are pretty alien to me. I am relectant to link against a library if I can avoid it :-)

I write code using vi in xterm so nobody could claim I follow fashion !



Yes and no .... I still dont quite follow.

The last two embedded devices I worked on I compiled the kernel and coreutils from scratch - the ones before that I used gerneric Debian. For me personally I tend to code only in plain old C, what I want is linux plus a framebuffer device for whatever display I am using and as little else as possible! This method seems out of favuor, most people seem to want linux to be some kind of base system for a visual studio style front end with a ".NET style bucket of crap all things to all people" runtime ?


Err ok.

My TV (an LG) runs a generic kernel from flash, busybox and custom control software for the tuner/front end. The job of LG would have been to write the device drivers for whatever platform they had built. How does another layer of software speed this process up ? Do you have a generic hardware platform I can use, if not more sofware just slows me down from my point of view.

I am not being difficult, people who code in say Java and want a generic base O/S then this probably ok - but for me personally who just wants generic kernel with support for custom hardware does yet another distro/toolkit help me, how does it differ from say Debian if I want a repository or busybox if I want a small code footprint.


Again, I have no idea what this means in reality.

Dont worry too much that I dont get it - it is not aimied at me I suspect, my phone is a mono Nokia, my TV does not surf the web nor would I want it to :-)
Your pitch sounds like Android without google (not a bad idea) or have I missed the point a bit?

That's fine - you're definitely not the 'target audience' though - we mainly cater for emacs users and konsole

Seriously though Mer is for people who look at the capability of cheap embedded systems and want a much shinier UI or just more in the way of middleware that they don't want to manage themselves.

Most of our devs use vi/emacs and consoles.



Right - we still do that. We actually define an set of CONFIG_* that you need to have in your kernel (and why). You then build your hardware adaptation layer (using Android BSP if you like) and put Mer on top. Your UI layer sits on there. So yes, we're a small and increasingly blurry step up in device class from a kernel/framebuffer.

I see you didn't write a kernel from scratch? Why not? Because if you can answer that then. in a way, Mer takes that rationale and pushes it up the stack to the UI layer (but no higher).

You selected Debian (me too on my desktop) - but if you knew Mer Core has just ~350 packages which are selected and optimised for mobile (RAM and storage space still matter) to form a rational mobile base OS that is pre-built for multiple ARM, MIPS and x86 architectures. If it helped with the "OMG what bits of Debian do I need?" question. If it also used latest stuff like systemd, Wayland and similar (I don't care what people say about them for the desktop/server - they are *superb* for mobile). Then maybe you'd use Mer - after all it's the same source that Debian uses - just less to wade through.

Then yes, there's an 'SDK' which operates at 3 levels - there's a pure command-line SDK which is basically a chroot for us platform-level guys. All vi/emacs and raw gcc, nasm, cmake, $whatever (also we support a really nice opensource cross-compile solution called scratchbox2). Nicely this thing also runs as-is on our build farm stuff - so if you want to scale up to do QA with a team or 2 or more ... that's fine we can go there (and that's another key point - this is not just for 1-man outfits - it scales from hobbyist to Nokia's size)

That *same* SDK is then quietly wrapped in virtualisation tech and shipped with a GUI as an application SDK. Same toolchains, same code releases as your platform or build systems - but with the C++/Qt graphics stuff available. That's nice for QA


So moving on 10 years to 2010

You still need to do that but now does it have ... bluetooth? web browser? wifi? maybe even a remote control with a touchscreen display that needs fancy graphical transitions? Sure you *could* write them - but Mer would let you assemble and prototype that setup in a matter of days. (We know - there's a Chinese consortium making set top boxes using Mer).

I am not kidding - days. Maybe a month if you've never used Mer before and need to brush up on Qt/C++

Sure it'll look like crap until you hire some graphics people to pick nice colours and do fancy layouts - but that won't take long either - Qt/QML is good there.

And this isn't for end-users to hack it onto a phone - it's for a company to 'easily' make a device for their market.

oh, and it just happens that we're making a phone too


Again, I have no idea what this means in reality.

Dont worry too much that I dont get it - it is not aimied at me I suspect, my phone is a mono Nokia, my TV does not surf the web nor would I want it to :-)
[quote]
(I think I covered the Debian/busybox (yes, we'll have a busybox option) points above.)

Fair enough. I'm not radically different. I have smartphones but barely use them. My TV is MythTV but I just watch boring old TV (though Denise does occasionally browse epguides.com to see what order the shows are in)

But I do want my next car stereo to tether to my phone to sync on the move or browse the web or use OSM - but I suspect I'll enjoy building it more than actually using it

The 'services' part is about the systems that you need to run all the infra - automation, QA, build. All opensource of course.



Yes, that's pretty close - but properly open with more linux and no java(*)


(*) ask if you *really* care about running Android apps on it 'cos you can.

Fascinating!

I read that as:

Time time write my own DVR

Thank you

For more discussion: irc is #mer on freenode - I'm lbt of course

Question to PCA: Is there some thought of making the new display board available a a replacement/retrofit for current players? Or (better yet for me) as a "display extender"? I think the bulk of us out there who still use (or want to use) our Empegs would prefer to trunk/glovebox/underseat mount them. I could much more easily find a place to mount a display board (even if DIN size, but only 1cm thick).

My VFD is nearly shot, and as you are all aware replacements are not to be found.

I seem to recall a few display extender projects that never made it all the way to "where can I buy that- even in kit form".

If the cost was anything approaching reasonable, I would pay for a display extender, - regardless if display technology, if it could go at least 1m away from the Empeg. Heck, in my current situation I could probably "hide" the empeg INSIDE my dash, and mount the display extender where my clock is now. That would require only ~30cm of cable.

The existing PCATS display extenders have much, much longer range (ribbon cable) than 30cm!

Still would be cool if the new boards could work with the old player hardware, but I doubt the feasibility of that.

With Colour? the NExt Generation empeg could have been color :-(

It's not impossible at all, but the current version isn't really suitable. It's driven by SPI from the raspberry pi, and the IR and buttons go via a serial port. This isn't the same way the empeg does it.

One the empeg the display is passive and driven by the lcd output port, with a certain amount of logic on the display board to twiddle the LCD signals into the format of the VFD. To make it drive an OLED like this, we would probably have to add another processor, which would read the lcd signals and generate the correct data to drive an intelligent OLED panel.

Something like an STM32F4 series microcontroller would do it. I have been considering making a board like this, but whether there is sufficient demand to make it worthwhile is debatable.

Pca

Could you make a USB display?

That would also support key and maybe rotary feedback.

Could that work from an empeg (with suitable code) and a-n-other device like a raspi or even smartphone?

(edit: thinking to expand the possible user base as this could support a wide range of applications - even PVR and HA)

No USB host (master) in an empeg.

Use a raspi as a middleman ?

usb display -> raspi

Empeg screen-scrape -> raspi over ethernet?

Yeah, ethernet would work. No need for a new screen scrape -- Hijack already has it in /proc/

Cheers

so how viable would that be Patrick?

I'm thinking that it would be nice to have a multi-part board too.

Basically cut the buttons and the rotary sections off the main board and use some extension mechanism (row of header pins) to allow them to be mounted alongside as per
the prototype or in different positions above/below the display or even remotely. That would also allow users to trivially make a custom button-only extension board.

Just thinking of easy ways to broaden the appeal



Hmm - otoh everything is on the internet

Yes, this sounds like an even better idea from many perspectives.

Something more modular, something that would potentially have broader appeal to the general tinkerer community, but conveniently would work well with the default display characteristics needed for the empeg.

I love the idea of two (or three) boards too - more mounting options for those of us so constrained.

How much would be a ballpark figure for seed money to get something like this rolling? I would seriously consider donating $$ to get this project started.

That's if we could come to a "best idea" of how to proceed, and the players PCA, Mark, anyone else needed) would be on board.

I totally agree that having a broader appeal than "just the empeg" would make it potentially more interesting. Having potentially more display size options, and flexible button/control/IR board configurations might also be very appealing. Being able to retrofit into an actual empeg might be nice too, but frankly I think more folks these days would remote mount if they could.

And of course keeping it as simple as possible (without being too simple) would stand a better chance of it actually happening :-)

Being simply a techie/tinkerer/IT geek, I cant really help much other that to offer ideas, input, and possibly seed money. But man, I really would LOVE to see this happen.

Wait, woah, just followed your link to Yoctopuce. That seems really, really intriguing.

The site is a little light on details (I didnt dig far) but are you thinking that one of these little critters, hooked up to a ras pi, with the ras pi reading the screen scrape via ethernet from Hijack, and handling the button presses, etc (from the yoctupuce board)?

Seems like it would work, if I were a programmer I would love to take a crack at it. Hmm... US $110. Plus buttons & Rot Enc somehow. I could live without IR.

Anyone have opinions?