[hpsdr] [OZY] QFP or BGA?

[Eric Ellison]

Lyle et.al.

I figgred I'd jump in here before I read further in the follow on messages.
In Phil_C's last message replying to Phil_H's requests he asked "Where is
the Excedrin?" 

Where is Alex when you need a cold blanket? He's in the frey! 

We have an explosion of Talent, Projects, Designs, Parts, Questions,
Manufacture options, FPGA's or CPLD's on every board BGA's yea/nay, and sort
of an atmosphere of pleasurable Chaos! I for one am really having fun
gorging on datasheets and, like Chris trying to catch my breath and survive
the dizziness. The recent contributions and Projects like Gibralter sort of
insure that the initial concept of motherboard - project board will survive
and the excitement prevail. I did sort of expect this to happen!

However, I think it's time to step back and do what Alex suggested, and take
an overall look! Just with the projects proposed I see multiple assembled
finished products, from a TAC to a Quad 2 input Codec for connecting 4 high
quality low cost Softrocks. The Atlas will allow that, and a buss design in
a lot of recent stand alone products has prevented 'lower' cost
experimentation.

In the ancient days of this project laced with Mythology, about 2 months
ago, I asked Phil_C why he needed a FPGA when we had one on the proposed
Lionheart? (Now Ozy) He gave perfectly logical answer and truly justifies
the use of a second FPGA for down conversion. My hope at the time was that
we would leverage the distributed resources of the buss structure on
different cards. It is STILL my hope that we do this.

Without dropping back a little and considering the overall resources
exploiting the integrated environment I fear that we are headed for some
rather chaotic trouble. Let's not end up with a bunch of stand alone
projects sitting on a connector on a buss whose main purpose is to supply
power.

The power of this project, whether Phil_C, and I want to play with PLC's, or
sequencers and the like, is well though out distributed resources from the
Atlas. I fear 4 USB ports, 3 Serial Ports, Multiple GPS inputs etc.

Dunno but this is fun, and it's nice to see the abundance of messages again.
I'm in it for keeps no matter what. Since I am a closet Time-Nut I see an
Atlas>Ozy>Gibralter just for TAC with 10 - (13) or so precision. Another
Atlas for sound card elimination for SDR-1000. Since I promised myself some
test equipment so another Atlas with a Ozy and a Mercury!

Let's INTEGRATE!

Eric - AA4SW
 



-----Original Message-----
From: hpsdr-bounces at hpsdr.org [mailto:hpsdr-bounces at hpsdr.org] On Behalf Of
Lyle Johnson
Sent: Thursday, May 11, 2006 12:47 PM
To: Alex
Cc: HPSDR List
Subject: Re: [hpsdr] [OZY] QFP or BGA?

***** High Performance Software Defined Radio Discussion List *****

> Also board assembly costs are getting pretty cheap for small runs.
> Personally I'd rather have machine assembled units, I'm getting too old to
hand solder SMT.

It takes me about 20 minutes to solder up a DSPx module when I need to. 
  It costs me close to $10 each to have them assembled in multi-hundred 
lots, and I have to wait a few weeks to get them.  And I had to pay a 
non-trivial cost to get the vendor set up to produce the assemblies.

I suspect Ozy would cost about the same.  Will we be able to "sell" 200 
or so Ozy?  Who is going to foot the bill and take the risk?  Figure 
about a grand for setup costs, and the price of 3 or 4 Alpha boards 
becomes pretty high.

In the end, we have to balance risk and cost against system needs.  If 
the system can be designed with parts that have leads, risk is minimized 
and up front cost (time and dollars/euros/pesos) becomes negligible.

No one has trouble manufacturing with TQFP/PQFP, and they are 
straightforward to hand assemble for those who prefer to do so.

If Ozy *needs* the extra IO lines then the answer is clear: go BGA and 
pay for an assembly house to make them.  There will be NRE set up costs 
in addition to the basic cost of assembly. We'll likely spend about $150 
to $250/board for the Alphas to get them assembled if done by "BGA 
hand/rework" techniques.  The NRE will still have to be paid for production.

If Ozy does not *need* the extra IO lines, then the choice is clear: 
TQFP/PQFP.  No NRE for Alphas.  No delays.  Easier, faster, cheaper...

If Phil decides to go BGA and someone really, really wants PQFP, they 
can make a derivative.   Or vice-versa.  It is, after all, open source :-)

Finally, if the cost of Ozy-with-BGA + JANUS + ATLAS exceeds the cost of 
a *good* soundcard, then...

73,

Lyle KK7P

_______________________________________________
HPSDR Discussion List
To post msg: hpsdr at hpsdr.org
Subscription help: http://lists.hpsdr.org/listinfo.cgi/hpsdr-hpsdr.org
HPSDR web page: http://hpsdr.org
Archives: http://lists.hpsdr.org/pipermail/hpsdr-hpsdr.org/


 1147388510.0