Hello everyone,

Have you ever wanted to tell the cable or satellite company what you really wish their set top box could do? Well here is your chance!

Please pardon the slightly lengthy post as I want to make sure I share the appropriate background. I would really appreciate the feedback of the KnoppMyth community. Likewise if you're interestd in the current pricepoint on a fanless/diskless front end, there is some good info in here for you too.

WHAT THE CABLE INDUSTRY IS DOING

I work for a cable company and I've been doing set top box research for the past few months. As you may or may not be aware there is a concerted effort in the cable industry to introduce the next generation of set top boxes, running cross platform stack called OCAP (open cable applications platform). The goal being to make boxes from different vendors interoperable and provide a common platform that developers can write to. The real goal, of course is to drive down price and move the box from a cable company cost item to a consumer electronics item (picture the box embedded in your next TV set or DVD player).

You can read more about the OpenCable specification here: http://www.opencable.com

The gist of an opencable STB is that it has a MIPS processor (in the 250Mhz-400Mhz RISC range), often runs linux (although not always) and runs a java stack upon which the applications reside (the apps being downloaded from the cablco either in the digital TV stream or via a cable modem). The box has a certain minimum RAM and flash footprint etc. Also, the new boxes almost always contain a DOCSIS Set Top Gateway (DSG), which is just a fancy name for an embedded cable modem, so the box itself will have broadband access internally via the CATV F-connector.

I've been looking at the boxes and you can get in to a nice digital (standard def) box for around US$125 an up. So lets set that as a baseline. This is a basic box with no PVR support or anything fancy, but able to run OCAP apps, EPG etc etc.

WHY I DON'T WANT TO FOLLOW THE CROWD

The problem that I'm having is this... Our company is planning to build a brand new fibre deep (close to the home but not to the home) cable system, passing about 32,500 homes (we already own a smaller system in a different location). I want to make sure that we make a good decision about the STB platform that we choose because we'll be living with it for 3-5 years and I think the OpenCable based boxes (at least at the target price point) are under powered, and don't take in to account how people really would use a nice interactive set top. I, like Issac, think the promises of the boxes have never lived up to the reality of what we all want to do with them and I think we can do a heck of a lot more.

By way of background, I've been using Myth since .08 (so thats about, what, 4 years now?) and KnoppMyth almost from its inception. Hats off to everyone who has contributed to both projects.

HOW YOU CAN HELP

Now, on to where I am seeking feedback:

I think the right box for us is essentially an x86 based set top running linux and X windows, with an interface based on or similar to Myth. The goal being to keep component cost to a minium by using off the shelf parts, and keeping the application layer as open as possible and allowing a world of legacy software (think great games even, from 1-2 years ago) to run on the box. I also want to be able to tap the widest possible open source development/content community, and not require that stuff be ported or portable to a RISC architecture, at a seriously reduced capability level. We want to easily and rapidly be able to deploy applications to meet the needs of our subscriber base and also have a future hardware roadmap that tracks nicely with the lowest end of the PC market. I just think it makes a lot more sense than introducing yet another semi-proprietary platform, and I think it can be done at or near the same $125 per box starting price point.

Some examples of applications that we're looking to run include:

-Weather tracking (hurricane tracker specifically, pulling XML feeds from the National Huricane Center)
-News headlines and photos (XML)
-Sports info and discussion
-News/Stock/Weather/XML ticker
-IM On net (our cable network) and off net (gateway to MSN/AIM/Yahoo)
-Box to box video conferencing via simple USB web cam w/potential gateway to offnet IM
-Photo sharing tied to a web portal (web and on box - show mom your photos)
-Video clip sharing via data storage at the cableco
-Streaming Internet audio and video
-Interactive competitive games across the user base like Friday Night Trivia
-Legacy PC games (MAME style and more complex like Doom/Quake/Sims/Civ)
-Town council related content
-Community buy/sell/trade content linked to web site
-On screen bill presentment
-Caller ID on TV (We'll be doing telephony as well)
-Visual voicemail and call log review
-Anything else you can dream of!

Those are some of the ideas, and as you can see many overlap (at least somewhat) with what Myth can do. Myth itself may or may not figure directly in the solution. The body of hardware/software and user experience knowledge here is ideal to help better understand peoples views so I wanted to toss it out for discussion.

In phase two the box will also receive/decode/play digital TV streams but for various reasons thats outside the scope fo this discussion and there are well developed chipsets and hardware to support this.

The more I look at the opencable boxes the more I see a lot lacking. Which got me to thinking, has the intersection of tech progress and price finally reached a point where we can do a fanless/diskless set top box for a similar price point to the proprietary RISC based boxes... I realize this question has been bandied about the forums before but I wanted to raise it again and share some of my research and perspective. Also, we'd be bringing a reasonable unit volume to the table.

ROUGH SPECIFICATION

A rough specification of what I'm looking for:

Processor: AMD (or other) x86 native low power/low heat in the 1Ghz-1.8Ghz range
Memory: 256MB
Flash: 512MB or 1GB via IDE to CF adapter
Net: 10/100 ethernet (to be connected to cable modem or alternatively CM embedded)
Graphics: NVidia GeForce family (via either AGP, PCI or PCIe slot)
I/O Ports: >=2 USB2, IR receiver, Coax pass through, Coax out (for people w/o SVid), SVid, RCA Out (stereo)
Input: IR Remote (Wireless Keyboard?)

Optional but nice:

VFD Display
CF/SD/MMC reader for adding photos/videos to your collection
Other suggestions?

Form factor: Mini-ITX is a possibiliy, maybe even Mini-ATX, could also be custom but would prefer to stick with industry standards

The box would run a standard 2.6 x86 kernel and X.

The box would be connected to a cable modem and any box related software and/or content would be accessible at speeds approaching 10 Mbps, regardless of the users normal internet "tier", as we can create discrete service flows in the modem based on where the traffic is destined. Local net traffic has virtually no cost to us, so always bear in mind that you have a huge amount of bandwidth back to the cable company server farm to play with, which is why we can keep the storage footprint so small on the client end and avoid the HD requirement (until we get in to a PVR box, but thats another discussion for later).

FIRST ATTEMPT AT PRICING

For my first cut of this I decided to see how far I could get just buying off the shelf components in a unit of 1 each from NewEgg. I made out impressively well I think.

Here is what I selected for a total of $187.05:

AMD Processor Duron 1.8 Ghz $48.00
http://www.newegg.com/Product/Product.a ... 6819104160

ECS Mobo $38.50
http://www.newegg.com/Product/Product.a ... 6813135153

Case/PS $28.99
http://www.newegg.com/Product/Product.a ... 6811164011

GF MX4000 AGP 64MB $23.25
http://www.newegg.com/Product/Product.a ... 6822998003

CF Card 512MB $16.99
http://www.newegg.com/Product/Product.a ... 6820183152

Memory 256MB $15.33
http://www.newegg.com/Product/Product.a ... 6820161635

CF to IDE Adapter $11.99
http://www.newegg.com/Product/Product.a ... 6822998003

CPU Fan $4.00
http://www.newegg.com/Product/Product.a ... 6835124001

I still don't have my IR remote or a keyboard in there (very small cost), nor is the form factor right. Otherwise, that is essentially what I'm thinking. A final case will likely have to be custom as all the low profile cases carry a hefty premium on initial looking and we'll probably want to customize the appearance.

The fact that the components are available on the market today, at the consumer end for only ~$62 above my target budget of $125 tells me this should be workable. That's a pretty poweful system, far in excess of the original XBox which was a Celeron 733, 64MB of memory and an 8GB HD as well as the DVD drive.

THE BIG TICKET ITEMS

As the processor, motherboard and case represent almost 62% of the total, they are the obvious targets for lowering the cost.

AMD GEODE EMBEDDED OPTION

Doing some further research, I'm found the AMD has introduced it's Athlon Mobile processor in a fanless form factor under a new name, targeted at embedded applications like this. I do not have any pricing information, but you can read about the Geode NX1700 here http://www.amd.com/us-en/ConnectivitySo ... 37,00.html

The NX1700 runs at 1.4Ghz but lets me lose the fan and I expect its price will be comparable or lower than the Duron I have priced here. It is also a Socket A chip, so if my research is correct, it can be dropped in any Socket A mobo.

AMD shows a reference board on their web site that has everything I need (at least to be comparable with the above) onboard with the exception of graphics (and the custom stuff like the RF output and RF passthrough etc). You can read about it here: http://www.amd.com/us-en/ConnectivitySo ... 9863_10837^11090,00.html

It doesn't have an AGP slot but I'm sure somewhere there is a comparable board that does, in the mean time it does have a PCI slot that could be used for the NVidia card in a pinch.

So, this is what my current research has turned up on the hardware front. It is looking increasingly likely that off the shelf components are available in the right price range, and some volume ordering as well as commitments to use older stuff (like the older NVidia chipset) can likely bring the price down.

BRING IT ON

I'd love to hear other suggestions for applications, hardware, considerations and anything else you want to throw in to the pot!

Thanks!

-Jeff

Hmm 43 reads and no replies... not sure what to make of that one...

Thanks for the reminder... This is Off-Topic.

_________________
cesman

When the source is open, the possibilities are endless!

I think it stopped being interesting at this point.
No HD? No PVR? Why bother?
I think that many started reading your post thinking "oh boy, here's a major STB player, thinking of investing millions into Myth Boxes, and they're going to contribute 30 developers to the project, etc., etc., etc., going to give as much as they take, and it's going to be great for everybody."
Or, "an evil STB company is going to try to hijack Myth development, let's see what they're up to!"

You don't need to seek my input, you already have it... MythTV.

Yeah, I'd like my cable company (NTL, Uk) to transmit something that I can pipe into the back of my Myth box. But since I assume you're representing an american firm, I don't think you can help me.

But as others have pointed out, by not doing HD, it'll alienate most of your starting market, the early adopters. They've already adopted HD tellys, they've probably got HD OTA, and they're waiting for Blueray/HD-DVD.

If you're not aiming it at early adopters, but instead the mass market, it's got to be cheap, real cheap. And a PC based solution probably wont be.

_________________
Mephi
-----------
R5.5
Opteron 2.Ghz
1Gb Ram, 200Gb Maxtor PATA
Avermedia 771 DVB-T
Hauppauge Nova-S2 (waiting to be installed)
MS MCE2 Remote

Well it's disappointing how this has turned, but I'm not surprised. Thanks to those who did read the post.

As an aside, it's disheartening that when a cableco brings forward an idea and trys to engage the community they get rebuffed, yet if I brought forward an obscure solution based on a propritary chipset, everyone would line up to try and hack it and get a mythfront end to half work on it. Look at the Hauppage MVP.

Open hardware makes more sense for open software.

-Jeff

I think you might be confusing "rebuff" with "not really interested".

Don't get me wrong, but when I look at the list of applications, they aren't enough to interest me in buying such a box. For most people, the "killer app" is video. For me, it was a jukebox and video was a "bonus app", but I'm not exactly typical.

To sell such a box, the core app or apps need to be defined. I.e. The ones someone walks into an electronics store to purchase. Everything else are just market differentiators to help separate the box from the competition.

Now if you really wanted to interest the people here, you might talk about a PC cable card that provides access to the encrypted HDTV stream allowing storage of the encrypted video on an HD combined with a video card with embedded encrypt/recrypt functionality including PIP, OSD overlays and playback speed controls. All that, while allowing open source SW to access the drivers.

It is my belief that we will get that. OpenCable is heading in that direction. The only question is whether it will have to be hacked for open source or if the industry will admit that this is a good compromise.

Cheers,

Allen

_________________
ASUS AT3N7A-I (Atom 330)
TBS 8922 PCI (DVB-S2)

I think that depends on the market. Here in Sweden there is (as far as I know) only ONE channel broadcasting HD. And that is a satellite channel. For DVB they have not yet even decided on what HD solution to use.

DVB was not introduced here to get better quality, it was introduced to lower the costs for distribution.

One thing to keep in mind if you are going for the bottom of the market hardware is that you probably wont be able to easily get the same stuff in quantity in a year's time. This means you will end up having to support a variety of hardware platforms. This can make things much more difficult (although not impossible of course).

I think I wasn't actually clear about the place the box takes in the business model.

The box is not intended for sale at retail, it would be bundled in with the internet service. The model would be something like $15/month for the service, including the box rental, in order to enable the kinds of apps that I outlined, viewed on your TV. Many of them are handled, to one extent or another, in the overall MythTV package today so thats a potential starting point and the front/back model myth is based on could lend itself well to a larger scale service delivered by the broadband provider.

(As an aside, and with a tip of the hat to the KnoppMyth and Myth development groups, I plugged my QuickCam4000 in last night to try out MythPhone and everything just worked. The only thing I had to do was set the source to /dev/video2 in the config. Very nice)

There is also some confusion about the video (TV) component, let me clairfy.

We are looking at delaying the launch of our (traditional) video service and focusing intially on Internet and telephone service. My reference to HD above as not actually to high definition signals, it was to the box having a hard drive. As the box won't be performing typical cable TV stream related functions, it won't be a PVR in this iteration or even handle cable TV content - thus the ability to go with a CF Flash based solution for the OS and any locally saved content. When we do launch video content we intend to offer a box, which may or may not be PC based, that does have proper DVR functionality.

For further background, a note about pricing. Today, in lower volumes (ie not Comcast/Cox/Charter etc), you are looking at a starting point for a digital box that can run OCAP at ~$125. The prices rise rapidly from there based on processing power and features. Many of the PVR boxes, esp with 5.1 and high def run mid $200's to mid $300's. The boxes are far from cheap and as you can see, when you start pushing mid $300's on a box there is a lot you can do with PC hardware. I agree, at the skinny end of the price scale its a lot tougher, but I did want to provide the relative scale so people have a sense of what the box cost is for the traditional solution. You may not pay that to your cable operator, but those are the kinds of figures the operator is paying for the boxes, they are just often subsidized to the consumer.

The video content the box we're discussing would access would be internet bsaed, and likely in shorts format. Be it YouTube clips, movie trailers, or any other content available in a stream from the internet at large or that the user might choose to store on their storage space at our head end.

With regards to the hardware - I agree, the development cycle continues so ther are two options. Either purchase a sufficient stock for our needs for a given period (say 18 months), or - and this is much more appealing - since we're looking at entirely standard PC parts - simply select a new motherboard, chip and NVidia graphics card at the appropriate price point when the next order of boxes is required. I have been discussing volume purchasing chips with AMD. Again, because it's a closed project and the boxes aren't for sale, having two or three hardware "revs" in the field isn't a huge problem unless we're getting in to apps that require more firepower than the older boxes have.

-Jeff

Agreed. That's why Dragon is an open specification.

As others have suggested, I think the main reason for the lack of replies is the fact that the hardware doesn't look like it'll do much. It's kind of hard to rally behind a hardware platform that is designed with price in mind, not performance. It's just not exciting


Now I'm confused. If the hardware will be open, people could build their own, right? And what happens if people design their own boxes to use with your infrastructure?

In any case, if you'd like to talk to me about how I went about designing Dragon and testing the parts, just send me a private message. I can be of general help with the technique and process.

_________________
KnoppMyth R5.5
MythiC Dragon v2.0
Join the KnoppMyth Frappr!

Sorry, semantics bit me. When I say closed project I mean that we're not aiming for a retail product, we're aiming for a platform that is connected to our network. Closed in the sense you won't take the box to Oregon and use it with your cableco. It's network specific. If an end user wanted to build their own, that would be great.

A cable company actually thinking of allowing it's user to build thier own box?! Are you kidding?! Perhaps now you know why I stated MythTV above....

Actually, not having encryption to begin with would make things a lot easier

_________________
KnoppMyth R5.5
MythiC Dragon v2.0
Join the KnoppMyth Frappr!