Re: [Discuss-gnuradio] Re: A Humble Request.... - "Open-Hardware"

[Moeller]

On 10.01.2011 00:44, J.D. Bakker wrote:
> The SSRP doesn't look too bad. Another option is the OpenHPSDR 
> (http://openhpsdr.org/); I believe people are working on GNUradio drivers.
I like the idea. But this one is even more complex (more expensive?) than the 
USRP. I believe that it's necessary for amateur radio applications,
because USRP does not offer sensitive RF/HF hardware with preselection filters. 
I would be happy with just a simple solution for signal analysis in
electronics devices at home.
>> This is the commercial point of view. I compare this to the 
>> Microsoft-approach. They need lots of money for producing, testing and 
>> supporting
>> operating systems and Office.
>
> Don't take this the wrong way, but you're sounding like a theorist with zero 
> experience in hardware design.
I did electronics design with Mentor Graphics EDA tools, also with Protel and 
other tools in later projects (including PowerPC controllers). At the
moment I'm on a more abstract and theoretical level. I know that sophisticated 
hardware projects are really expensive. SSRP and USRP are not that
complex, not even a real CPU inside.
> When I designed the LART (a decade-old software radio experimentation 
> platform), it took me two revisions of the six-layer PCB prototype (at >$500
> per revision) to get the board working. A few months 
But a simple one like SSRP is not that complex. You're mixing two different 
worlds. Complex SDR have a real CPU and bigger FPGA like the Virtex 4 or 5
for real signal processing. This cannot be done in a small Spartan chip (only 
DDC, not much more).
> later it turned out that the voltage regulator chips I'd used were suddenly 
> impossible to get, so I had to do two more board re-spins. The proto
> boards were assembled at a professional facility for >$1000, until I got a 
> reasonable yield soldering 0.5mm pitch LQFP packages myself (destroying
> several PCBs and $150 processor chips in the process). To test the boards I 
> needed access to several tens of thousands of $ worth in logic
> analyzers, signal generators and spectrum analyzers. And then there was the 
> global tantalum capacitor shortage...
SSRP does not have a single processor chip. He sells the bare board for $15. I 
assume that he did not have such costs.

> Needless to say, none of this would have happened if I would have had to pay 
> for it out of my own pocket. Luckily, this was a University project,
> and I managed to persuade the powers that be to allow me to release the 
> design files.
Ok, you lucky one found somebody to pay for it. Many hobbyists have to pay all 
themselves. They need a cheaper solution.
>
> Once again: software development and debugging is essentially free, with a 
> 'make' of a new revision costing nothing but time (and some electricity).
> None of that is true for hardware development. Calling it the 
> "Microsoft-approach" doesn't change any of that.
The SSRP cost of $120 assembled board is "almost" for free. Of course the BOM 
bill you have to pay.
> If you were talking software, I'd say "please post your patch".
The software cycles are much faster indeed. But it's not impossible to discuss 
hardware patches in an open project. An open project would help "poor"
hobbyists and students to enter the world of SDR, create an do-it-yourself SDR. 
A start from zero is difficult, but modifying and realizing open
source hardware projects is much easier.