Yet another kick at the cheap-hardware can

GNU Radio
1

http://www.sbrac.org/files/digital_receiver_cheap.pdf

This has everything in one place–commit to a single host I/O, and go
cheaper as a result.

The estimated BOM cost for this, including PCB would be under $100.00.

If you sacrifice very-fine tunability, then you don’t need a DDC in the
FPGA, and only need
a CIC decimator chain, and you only need Rx logic in the FPGA, so you
can get away with
the smaller EP1C6 FPGA. There’s a 9K-LE Xilinx Spartan-6 which is
marginally cheaper
($16.44 vs $17.50) than the Altera, but only available in larger
quantities from Digikey.
Also, I think the Altera toolchain is cheaper (free??) – I dunno, I’m
not an FPGA guy.

Note the use of ultra-cheap 8-bit ADCs. This design isn’t going to win
any awards for
dynamic range, but it helps keep the BOM cost down, and as someone
else observed, you
get processing gain every time you reduce the bandwidth. So at 5MHz
bandwidth, you’ve
added a couple of effective bits. For the types of wide-band
science-radio experiments
one might want to do with this, a handful of bits is just fine.

Now, I want to emphasize again that I have no interest in physically
producing such a thing,
but I’m always willing to contribute my engineering wisdom, for
whatever that’s worth.

Also, to set a ground rule for future discussions. If this turns,
yet-again, into an Ettus-bashing
fest, I’m dropping out of the thread, and not participating in any
further discussions. Such
nonsense isn’t productive, or even fair or reasonable. Matt and his
employees (and part-time
contractors, like me) are good, hard-working people with an excellent
product, and who have
pioneered reasonably-priced hardware that works well with Gnu
Radio.

The question I think this discussion can answer is fairly simple: are
there design choices that can
be made, with significant compromises in functionality, that can
produce a design that is practically
producible by an open-source hardware community, and will such a
device be useful-enough over
the types of hobbiest uses-cases we’re interested in. Further, will
such a device meet the
delivered-price goals.

If the answer to the above is “yes”, then the next question is: is
there a community of interested
volunteers to bring the project to fruition? Such an interested
community would involve:

 o High-level hardware design
 o Detailed schematic capture and PCB layout
 o FPGA firmware design
 o Host-interface (FX2?) firmware design
 o Host driver software design and implementation
 o Small-scale financial investment for initial PCBs, components,

etc

Once such a board works, then someone needs to be found to distribute
either kits or finished product.

Something that vaguely compares to this effort is the FunCube Dongle,
which is a quadrature
receiver covering 64MHz to 1.7GHz, but with 96KHz host-side bandwidth.
That project is
selling fully-built units for about USD 170.00.

–
Principal Investigator
Shirleys Bay Radio Astronomy Consortium

2

On Wed, Jan 19, 2011 at 11:23 PM, Marcus D. Leech [email protected]
wrote:

o Small-scale financial investment for initial PCBs, components, etc

I have no knowledge of radio design beyond block diagrams, but I’m
very interested in this project as the sort of device every community
workshop or school should be able to get hold of. I’m happy to
prototype PCBs and devices locally and help on the software
interfacing side.

Mark

3

Well, may be an option for someone , but we are trying to get a cheaper,
and
open-source, hardware and I hope we can do this.

Euripedes

2011/1/20 William C. [email protected]

4

The hardware looks somewhat pricey–roughly $1000.00 for a
receiver+sampler+network “stack”.

But since this is supposed to be an open-source project, I don’t think
licensing FPGA IP is the correct direction, either.

5

I don’t know if this is kosher, but has anyone looked at the (vast array
of)
offerings from Comblocks (comblock.com)? They sell FPGA IP cores for all
of
their hardware, and it seems like it might be a good match for building
a
basic I/Q acquisition system. Here’s a full product list:
Product list
http://comblock.com/product_list.htmlThe block diagram at the bottom
of
this page gives an idea of how things could work:
High-Speed Data Acquisition 256MB 1Gbit/s
http://comblock.com/com8002.html-William

6

On 20.01.2011 16:16, William C. wrote:

I don’t know if this is kosher, but has anyone looked at the (vast array of)
offerings from Comblocks (comblock.com http://comblock.com)? They sell
FPGA IP cores for all of their hardware, and it seems like it might be a good
match for building a basic I/Q acquisition system. Here’s a full product
list: Product list

Not really.
Open Projects like GNU need open IP like in:

There is Ethernet controller code (up to GBit/s), USB2 code,
OpenRISC, PIC, Microcontrollers etc.

The DSP section has special signal processing blocks,
like this project of a hardware FFT processor:
http://opencores.org/project,pipelined_fft_128
What about a Gnuradio emitting FFT samples instead of time domain
samples,
for applications like spectrum analyzers or cognitive radio monitoring
modes?

Gnuradio would be a good software/hardware platform for further
development
of OpenCores. I like the idea. The SDR section is still empty
http://opencores.org/project,software_defined_radio
Why not contributing SDR code to OpenCores?

Btw, USRP2 is also using OpenCores verilog code in the FPGA directory.

7

On 20.01.2011 00:23, Marcus D. Leech wrote:

If the answer to the above is “yes”, then the next question is: is
there a community of interested
volunteers to bring the project to fruition? Such an interested
community would involve:

I’m interested in developing a SDR teaching platform, especially
for introducing electronics to children. I tried it with simple
logic circuits and blinking LED, a simple FM radio (surely no Hifi).
But kids today grow up with MP3 players, computers and other high tech
electronics. They expect electronics to be more fascinating.
So, what’s more attractive than stupid blinking LED experiments?
The Gnuradio RADAR experiment? Looking into the wide radio spectrum?
Visualizing radio emissions of electronic devices, the microwave oven,
car key, mobile phone, … ?
As a child, I was very fascinated with my analogous oscilloscope,
analyzing and repairing all kind of devices. I think nowadays, after
so much progess in technology, it should be a device like Gnuradio.

 o High-level hardware design

I’m thinking of a scope frontend, with switchable voltage levels.
Maybe my old scope schematics (was an appendix to the user manual)
could give inspiration. At least I would use a circuit for overvoltage
protection and an attenuator (in front of the first LNA).
It does not have to be integrated on the PCB, possibly only an
additional box in front of the antenna input.
I don’t mind if there is no DC coupling.
The power levels could be calibrated in software.

 o Detailed schematic capture and PCB layout

I installed KiCad (http://kicad.sourceforge.net/).
My first impression was very positive, especially with
the FPGA example project.

 o FPGA firmware design

used it for glue logic circuits before, in VHDL …

 o Host-interface (FX2?) firmware design
 o Host driver software design and implementation

I suppose there is also public domain software available.

 o Small-scale financial investment for initial PCBs, components, etc

I don’t mind if a board of $100 explodes during debugging.
As I child I burned many semiconductor devices.
Our local PCB manufacturer asks 50

8

Hi Marcus,
Who works on this project now?
Nobody, really, except that I’ve posted a few “straw man” designs.

Why choose USB as the interface to host. The USB interface became the
bandwidth bottleneck
in USRP1, so why use network interface?
USB-2.0 is relatively cheap to implement, which is one of the project
goals.

Using 8-bit samples, you can achieve roughly 16Msps. Using 4-bit
samples, you can do better.
4-bit samples reduces your dynamic range, but for certain
high-bandwidth applications, that’s
OK.

9

Hi Marcus,
In your design there is only a single RX. I think it is better to build
an
expandable board which can expand 2 RX 3 RX…
That will only introduce a little more cost but will meet much more
people’s
need.

On Thu, Jan 20, 2011 at 11:39 AM, James Jordan

10

Hi,
I’m very intersted in this project and will help to make things happen.
I
can help in any part of the design, but I have more experience in FPGA
developing.

Euripedes

2011/1/20 Marcus D. Leech [email protected]