[asterisk-users] building a phone

[Gordon Henderson]

On Fri, 27 Feb 2009, Tzafrir Cohen wrote:

> Hi folks
>
> A common wisdom here is that one should use a proper hardware phone
> rather that an extra software on the user's PC. Why is that such a big
> issue?

Both my laptop and desktop PC have poor quality microphone inputs. I use a 
USB "phone" on my laptop when out and about.

> One thing that bothers me with the current crop of hardware SIP phones
> is that they are hopelessly properitary.

SIP is an open standard...

> So what would it take to build a fully-adaptable phone?

Persuade an existing manufacturer to provide an SDK for their phone...

And thinking about it, don't Snoms run Linux? anyone asked if an SDK is 
avalable?

> Here are some of my thoughts. This is not anything I plan to do soon (if
> at all), but I really find it strange that there aren't such phones
> already.
>
>
> == Small Quantities:
> When you look at such systems it becomes aparant that you can get much
> nicer prices if you buy large quanities. But this is something that will
> be a problem. Not only for prototying. The fact that you're limited to a
> strict hardware setting is very limiting. No mixing and matching like in
> a standard PC. I'm not exactly sure how to overcome that.
>
> == Platforms:
> There are many embedded platforms nowadays. I assume that the relevant
> application requires some non-trivial CPU power. I would exclude e.g. a
> 486-based systems. My target phone should be able to handle at least two
> concurrent Speex calls. Preferrebly 6 speex calls and above.
>
> OTOH, I can't afford a monster CoreDuo. I need a quiet system with no
> fan. Thus the target CPU may be higher end VIA or Atom. Not sure about
> Geode.
>
> There are also some interesting ARM-based boards around. I'm completely
> unfamiliar with them but I suspect that they may prove to be cheaper.

Custom DSP or ARM.

Don't forget power requirements too. PoE may well be able to supply 15W, 
but imagine a building with 100 x 15W phones... Everything I've plugged 
into my meter idles at about 2W (Grandstream, Snom, Siemens, ATL)

> == SIP Software:
> Not really sure here. There must be something close to usable already, I
> guess.
>
> == Micro Browser:
> Hell no!

"Executives" want it...

> The device should have an LCD display, and the content of that display
> should be programmable. Programming it using a HTML renderred is a bad
> design decision.
>
> The device should be a good phone. It should not attempt to be a web
> browser, as it will be a lousy one.

I've actually used the RSS reader in my Grandstream Video phones.. It's 
usable, but I take your point about the full web thing!

> == Handset:
> I suppose that an obvious starting point for a handset is "skype phones"
> such as USB handsets from yealink. Far from an optimal design, but a
> driver already exists.

Seen these:

   http://www.fit-pc.co.uk/meet-fit-pc.html#tiny

And of-course the power line thing mentioned here a few days earlier - 
eg.

   http://www.theinquirer.net/inquirer/news/143/1051143/pc-plug-coming-soon-wall-near

if you're going to use a USB phone, these platforms are ideal, if a little 
pricey..

I did play with a yealink phone and Linux some time back - get the display 
and keyboard working and it'll be very functional... (I got it to almost 
work with Zoiper, so a custom app. would be easy) Alas, I gave it to a 
friend, then go anothe from the same source (tesco), thinking it would be 
identical - and it was - in packaging, but was a totally different phone 
)-:


> == Ease of Use:
> A phone must be usable. The target device must be something my mom can
> use. However that does not mean it must be easy to program. It must be
> programmable and hackable. But I can live with a complicated user
> interface for that. If such phones become successful and useful, better
> interfaces will eventually be written.

Dial the number, push the green button, off you go...

Gordon