[hpsdr] Some things worthy of your attention

[jeff millar]

Krister Wikström wrote:
> ***** High Performance Software Defined Radio Discussion List *****
> 
> Some of you are no doubt familiar with the following, but I'd like to
> point your attention to two books that I've found very valuable and in
> my humble opinion should be on every serious SDR developer's bookshelf:
> 
> 1. Peter B. Kenington: RF and Baseband Techniques for Software Defined
> Radio, ISBN 1-58053-793-6 (Artech House)
> This tome covers practically all there is to SDR.
> 
> 2. Jouko Vankka: Digital Synthesizers and Transmitters for Software
> Radio, ISBN-10 1-4020-3194-7 (Springer)
> Great detail regarding transmitters with high linearity and high
> efficiency, two things that usually don't go hand in hand.

thanks for the tips

> Regarding the direction for future developments in HPSDR and other
> projects, I think it might be a good idea to have a look at the new LXI
> standard for interconnecting measuring instruments via the Ethernet.
> Ethernet is ubiquitous, cheap and broadband, and lends itself very well
> for remote control, wired or wireless. 

It's certainly possible to connect an SDR that uses an FPGA to a 
processor via Layer 2 Ethernet.  With a Cyclone II FPGA,the Ethernet 
will take up about 20% of the logic.  The software can use "RAW" 
Ethernet packet format to perform I/O to the FPGA with good performance. 
  As a bonus, the packets can route through a standard Ethernet switch, 
which enable scaling to an array of processors and RF devices

It would be nice to jump on an industry standard for that.  but from 
looking at the LXI standard, it seems all wrapped up in stuff that 
doesn't matter.  Also it's aimed at a market that charges a lot for 
simple things...which means little free software.

One design challenge with Ethernet comes from it's legacy of software 
implementations.  That makes the protocol stack design complex, and 
implementation in hardware becomes more difficult.

USB has a much better design for software talking to hardware.  And the 
industry has produced a lot of nice USB client chips.

> Another relevant standard could be the RP3-01 specification from the
> OBSAI  base station standard. This spec defines a digital interface
> between a remote RF head and the baseband signal processing unit.

 From what I've seen in the industry, that's a high cost interface with 
a lack of free software.  IEEE-1394 Firewire probably makes a better 
choice if you really want several hundred megabits/sec of streaming data.

> I think the present SDR-1000 3-board stack with separate audio and
> control interfaces is not very elegant, to put it mildly. I can
> understand the initial rationale for using a sound card for the A/D and
> D/A conversion, after all most PC:s and all laptops have a sound card.
> BUT, as it turned out, most sound cards aren't good enough, a fact that
> rather negates the initial idea. We definitely need an all-digital
> interface between the RF head and the PC, the sooner the better.

Using sound cards provided a stop gap technology when the only decent 
dynamic range came from parts with kHz bandwidths.  Now, A/D's with MHz 
bandwidths (and the CPU's to process the data) make sound cards look 
very limited.

Fast A/D's also provide digital I/Q which completely eliminates analog 
balance problems.

So, just about any time now all the radios will have Digital I/O through 
a common PC compatible Interface, could be either USB or Ethernet, but I 
bet USB wins.

jeff, wa1hco

 1167976708.0