Would this work for a handheld DVR?

THe RPi is completely unsuitable for this kind of thing:

You are talking about "small batch production runs with it for a business start-up" - the RPi project is currently not a source of anything but even if they sort out the current problems it would be a poor basis for a serious product because the RPi project has no commitment to being a source of OEM boards.

But it's a bad choice for other reasons -  the IO is very limited and you will have dififculty interfacing CAN or high performance ADC to it.

I don't know if it will do the video stuff you want.

AFIK The Beaglebone design can be cloned (check this before making too much comitment) so if it works for your project you can make your own custom boards if you get to production.

But  I'm puzzled - what will be special about your device - if you Google for handheld dvr you find zillions of products form about £30 upwards - it looks like a dificult market to penetrate with a solution made of plugged together retail devices. You'll be up against people who are happy to develop custom chips but you're worried about the complexity of reading buttons via GPIO.

Michael Kellett

Well, I agree with Michael Kellett, the RPI as is is unsuitable for what you want at the moment, but maybe its not because it lacks the hardware to do what you want, but more because there is no support at the moment for what you want to do with that hardware; let me explain.

The RPi does have hardware (interfaces) for Video Input and LCD output, but there is no software support for them (yet), and the interfaces are very modern, and therefore may not be easy to interface to the kind of devices you want to use. It has these interfaces because the Broadcom SoC (chip) that is used is actually a chip meant for smartphones and tablets and such, and therefore naturally has these interfaces, (for smartphones and such) the RPF decided to bring them out to a connector for possible future use.

For video input the RPI has a 15-pins connector on board for a CSI camera, this interface is meant for the type of camera's typically found in smartphones. It connects to such a camera module with a paper thin flexible foil (with metal traces on them). This interface thus provides a video input option, but perhaps not in a form that is directly useful to you. The foundation has announced they are working on a camera module for this interface.

Its to be noted that this DSI camera "speaks" directly to the GPU, which can only be controlled through a software interface, so its hard to say what useful things you can do with the captured video, the (ARM) CPU is probably too slow to encode the video signal into any kind of video format, but the GPU might be able to, IF such a thing is supported by it, using the available software interfaces.

Regarding an LCD, obviously you can hang a PAL (NTSC) or HDMI LCD screen on it using the RCA and HDMI connectors, but the SoC also has a DSI output connector (also a 15 pins connector of the same type as the DSI connector), which can be used to drive raw LCD panels. But for this interface too, hardly any documentation exists at the moment, nor does a driver. DSI is very powerful and can not only transfer video data to the panel, it also can control the LCD panel and read out a touch interface.

So maybe sometime in the future something like you want will be possible, but not at the moment, and not unless you are very knowledgeable about these interfaces, and the software that is needed to make them work, and only then if you can coax the GPU to do what you want, which may be impossible without support from Broadcom.

Hi Martin

I certainly agree with much of what you have said. I am an optimist at heart and when I first came across the RPi I thought that we might just have a chance of making things a little bit more positive with respect to opening-up some quite clever electronics applications via open-source strategies.

I’m a retired engineer with an interest in aerial, digital still and video photography using actively stabilised, RC, fixed wing platforms. I wish to use the RPi (when I get mine?) with Aptina MT9D111 2MP and MT9P031I12STMH ESMT9P031I12STMH ES 5MP SOC camera modules (among others) to store the captured images onto SD card. These modules where designed for and are used in mobile phones and similar systems hence my particular interest in the RPi and its Broadcom SoC processor. The question I should like to ask of anybody who has an informed opinion on the subject is: Is there any support for such devices at present or in the pipeline for using digital camera sensors, board modules, SOC’s etc. especially for full HD (1080p@30fps H-264) video and, say, 5MP or 8MP still images? If so, where, when and how, can we get details of such work? The software for these tasks has been written many times, for many devices: it only needs someone with a little generosity (it doesn't have to be the latest, all-singing, all-dancing version) to release it into the public domain for the use of all hobbyists, students, educators, tinkerers etc. Maybe we need to pressure RPF into negotiating with Broadcom to give more support for this AOI especially in terms of legacy drivers and their hardware.

An ever hopeful tinkerer, 

John Barton Wood

John,

searching the forum on the original raspberry PI site might provide some answers, as well as a place to ask question, or ask for features.

the site (if you do not know) is here: http://www.raspberrypi.org/

@ Michael, Thanks for your comments.

I did think that the RPI wouldn't be well supported, as it's primarily designed for teaching kids about computing, so your (and others) comments about lack of production support in the future confirm my suspicions.

ADC or CAN would be connected via I2C/SPI, so that would be quite easy.  I had assumed that there would be some kind of Board Support Package available via the Linux community, but it seems that this is also some way off (if at all), so it does look like the RPI would be unsuitable for anything other than a one-off homer project.

As for why I would want to try to compete with the major ASIC manufacturers - I don't!  The myriad cheap devices that you can get anywhere aren't suitable for my target market - sports use. None of the devices you can buy cheaply allow an external CVBS video source to be input (which is important for siting a small bullet camera in a small space on the inside of a race fairing, or on a roll-cage - and imagine trying to ride a mountain bike downhill one-handed, while taking video shots on a mobile-phone with the other!).  Plus, any device with a hard-disk can't withstand the vibration levels you would get in a rally car (most affordable SD-Card video cameras are quite poor quality).

In respect of interfacing buttons to GPIO, you're right that it's a very simple thing to do to just wire up some switches to GPIO, but when you try to fit those switches into a waterproof case, and still make them usable with gloves on, that becomes a whole new problem!  That's why a touch-screen would be the way to go for a waterproof user interface.

Doug.

Martin,

Looking at the spec, it does sound like all of the elements I am looking for are already 80-90% there, in terms of hardware alone, but it sounds like the support I would need to turn it into a product isn't quite there yet, at least on the software side.

I have years of experience in embedded hardware and software development, so hardware mods wouldn't be an issue with this board, but I am brand-new to LINUX, so if there isn't already a fair bit of support for this board and the interfaces it already carries, or at least a BSP, then I should probably look elsewhere.

I've already found suitable boards elsewhere, but these are mostly sourced from China, or Taiwan.  Not that there's a problem with Chinese or Taiwanese kit these days, it just would have been good to use some home-grown kit from the UK!

I'll go back to my Taiwan boards for now, but I'll keep an eye on RaspberryPi developments and see if I can switch to it at a later date.

Thanks for your comments - very helpful.

Doug.

@John,

I'd be interested to hear how you get on.

Doug.

It sounds like an interesting project, Douglas, and even if you decide to use some other ARM board, your experiences could be quite useful to the Pi community.  Please keep us in the loop periodically.

Douglas,

     I will also be listening intently for any progress you can make.

     I would also add that the RPi has had many optimizations made during the development that allows the driving of specific displays that the chipset supports as well as the ability to decode certain types of audio and video.  Not all formats are supported by hardware, which means that some would have to be done by software.  This is where the previously mentionned optimizations make the software efforts rather inefficient.

     I've been using TV tuners with Linux for years and even now with systems with multi-core processors running at 2-3GHz and 4-8GB of memory, I can still have issues encoding and decoding HD video.  You will find that DVR-type systems like Apple TV & WD TV are extremely optimized for audio and video encoding and decoding but would not be very useful for other types of work. 

     I believe that we will have to wait until many people have their RPis in hand putting them through their paces to get a real read on how it operates under the large variety of potential uses.

Are you sure the functionality you want isn't already available? For instance, there are 720P H.264 "keychain cameras" for US$30 available now. These are small enough to put just about anywhere.

http://www.ebay.com/itm/808HD-11-HD-DV-KeyChain-Lens-Camera-Cam-H-264-720P-New-/160595839562?pt=LH_DefaultDomain_0&hash=item2564420a4a