Re: [Linux-aus] Hello everyone! and Grant Request

[Michael Bennett <mib@homemail.com.au>]

Mark Tearle wrote:
> On Sun, 26 Jun 2005, Tim Ansell wrote:
> 

<snip>

>>
>> FXS devices are still very expensive however (the cheapest being around
>> $AUS 70 per line). A FXS line is required for every internal phone, so
>> while it Asterisk might be able to get away with a small number of FXO
>> lines it needs numerous FXS lines.
> 

Is $70 per line really that expensive? If so, what price is affordable?
What price would it cost you to manufacture your FXS and in what quantities?

Can we just go to the manufacturer that sells at $70 per line model and
put in a group order for 1000 units and see what discount we get?

> 
> *snip*
> 
>> The clear benefit of this project is that cheap FXS devices become
>> available. This could dramatically increase Asterisk's penetration into
>> areas which would not normally consider VOIP PBX to be cost effective
>> savings, such as small business and general home usage. It will also
>> help further the goal of moving to a completely open system, one which
>> includes open hardware.
> 
> 
> Quick question:  how are you going to deal with compliance and
> certification to Australian Standards for these devices?
> 

It is required to meet electrical safety standards, EMC standards and
get a c-tick before it can be marketed in Australia. This does make it
very hard for homebrew projects to be commercialised.

>> Request:
>>
<snip>
>>
>> Which both charge $US 33 per PCB. I believe it will take me between 4
>> and 8 prototypes before the final design is finished. This means the
>> cost will be around ~$US 200. I did hear that some people at the
>> linux.conf.au might also be able to do this cheaply, but have lost
>> contact with them.

You may want to try wire wrapping for prototypes. It's an old technique
but it works, is reliable, its easy to do and easy to modify a
prototype. It also is cheaper than multiple PCB prototypes.

>> I also would like some help with the cost of parts. I have managed to
>> source most of the chips as "free samples" in sufficient quantity to
>> not need to buy any. However I need to purchase other SMD components
>> such as resistors and capacitors. Most of these however only come in
>> groups of 50 or 100 (at $AU 0.10 cents each). So far I have spent
>> around ~$AU 200 buying parts for usage on this project. I will soon
>> spend another ~$AU 50 for more parts. I was hopping to get a small fund
>> (maybe $AU 200-$AU 500) which I can use to buy (and re-emburse) these
>> parts. Specialised SMD to DIP converters and similar for prototyping
>> could also be purchased from this fund. Any expensive parts brought and
>> are left over would be available for other projects.

I think this request opens up a lot of interesting possibilities (can of
worms) for LA sponsorship and illustrates the difference between open
software and open hardware. If LA were to sponsor this project and it
was successful we would end up with an publically exploitable design for
a FXS. This wouldn't be much use by itself and we would still go out and
buy the $70 per line FXS because a design in hardware still needs to be
manufactured for it to be useful.

If I were to take that design and build for only myself it may cost me
$140 per line in quantities of 10. If I got some of my friends together
and had it manufactured in quantities of 100 the price might be $70 per
line. If it was a small scale manufacture in quantities of 1000 I might
be able to half the price in which case it would be useful at the
'affordable' price point.

Then if I'm manufacturing it and selling it, I need to provide a
warranty under law which I don't need to do if it is software. I need to
comply with all regulatory requirements which I don't under software. It
also costs real money to apply a patch in hardware or fix a fault.

Would LA want to be involved with any part of this after the design has
been completed? If not who will assume the risks associated with
hardware that are not present with software?

>> As I have only recently started participating in hardware side of the
>> Open Source world, I am lacking a few tools for working on it. A
>> quality soldering iron would be a welcome addition (current using a
>> cheap $14 soldering iron). Other welcome tools include an Oscilloscope
>> (a second hand one would be cool - been using the Uni's currently),
>> universal chip programmer and a quality digital multimeter (using
>> another cheap one). All these devices vary quite a lot in price and
>> second hand are just as good.

My suggestion is become a hardware engineer and work for a company which
lets you borrow their equipment :-)

The cheapest you'll find test equipment is to use the PC based solutions
such as bitscope. I'm not sure of their support for linux though.

>> As the Open Source software for producing PCB is still primate I have
>> been using a piece of software called "CadSoft Eagle". While this
>> product is not "free as in freedom" there is a version which is "free
>> as in beer" (which runs on Windows, Linux and MacOSX). This should
>> allow everyone to view and produce PCBs from the design.
>> While I have been using the free version, I am running up against it's
>> restrictions. I would like to purchase a full copy of the Standard
>> version for Students which costs $US 298.50, I understand if people
>> think this is inappropriate usage of the money.
> 

This is one of the problems with hardware. The tools are so expensive.
The cheap commercial grade PCB CAD package costs US$10,000 and the base
model good commmercial packages start at about US$30,000. There's an
open source package called Gnu EDA (GEDA). I think it is now at a
useable stage. However with hardware design tools you get what you pay
for. The more expensive the tool, the better your design can be verified
in software so that when you eventually get around to building it there
is a higher probability that it will work the way you expect it first time.

I would encourage anyone who wants to work on free software but doesn't
know what to do to have a look at developing electronic design
automation (EDA) tools. We desparately need good tools in this area.

This project sounds like a cool project and we don't have enough people
doing cool open hardware projects (or even boring open hardware
projects). I wish you all the best in it however I don't see its
benefits for the LA community justifying the sponsorship.

Best Regards
Michael.


Michael Bennett
mib@homemail.com.au