[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Low cost hardware option

From: Marcus D. Leech
Subject: Re: [Discuss-gnuradio] Low cost hardware option
Date: Wed, 12 Jan 2011 17:06:07 -0500
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv: Gecko/20101209 Fedora/3.1.7-0.35.b3pre.fc14 Thunderbird/3.1.7

Hi, i'm watching all discussion about poor students and the evil Mr Ettus who don't play like Santa Claus and whant to make some profit :). I'm also watching all topics and discussion regarding a low cost solution for use with GNURADIO. I guess we can have a cheap option to us and I'm
I don't think anyone *seriously* thinks Matt is evil for wanting to make a profit from his hard labour. That makes as much sense as insisting that Henry Ford should have produced cars at materials cost, and his labour force should have worked for free, "for the greater good". The fact is that Henry Ford contributed to "the greater good" *while* making a significant profit, and giving his workers a living wage, etc, etc.
  OMG, did I just compare Matt to Henry Ford?  Hmmmmm.  :-) :-)

But, I think the question "is there an approach to higher-bandwidth Gnu Radio compatible hardware that is simpler (and thus hopefully cheaper) than what is currently on the market, and can we reasonably totally-open-source the result" is a reasonable one.

very interested in work in such a solution. What I'm suggeting here is to take all people who want's to take the job and start a small project. I'm a embedded systems enthusiatic and was a "starving student", now I'm a starving engineer, since I'm unemployed, that have some time to work on this project. The first question is: Ok, we need a low cost solution with some possible applications but what are the limits?
Well, I posted a first-cut, potential block-diagram for the Rx side earlier today:


It covers a goodly swath of the bands of interest for hobbiests (68.75MHz to 2.2GHz). One could substitute a different PLL synthesizer to cover different bands of interest. In fact, if one added another level of output divider on the output of the ADF4350, one could go down to lower frequencies. Hmmm. Thinkage. Looking at the datasheet for the AD5387, the part is only specced to go down to 50MHz, but looking at the plots, it will likely go down much lower (perhaps as low as 5MHz). So adding another configurable, bypassable, divider stage on the output of the ADF4350 yields lower-frequency (in the HF bands) capability, and still provides coverage up to 2.2GHz. Actually, a single, fixed divider that is bypassable might give the required coverage, for example, a single /16 or something.
  Have to play with the numbers.

It will handle up to 20Msps, in 4 discrete steps (2.5Msps, 5Msps, 10Msps, 20Msps).

It specifies a generic FMC connector, so you could use it with an FPGA board that uses the FMC
  (FPGA Mezzanine Connector).

FPGA evaluation boards are available *relatively* cheaply, since I think folks like Xilinx sell them through distributors as "loss leaders" to get you into the mood to design-in their technology to mass-market devices. But hey, if the Open Source community wants to leverage that little
  bit of market anomaly, that's just fine :-)

The approach is to combine the RF front-end with a modestly-capable ADC, and use the ADC output as the "demarc" point where you plug it into some kind of FPGA+GiGe motherboard (like the Xilinx eval boards), or perhaps an EZ-FX2 type board giving a USB-2.0 interface. The USB board would be the cheaper way of getting data out to the PC, with some provisos, like it won't do DDC, so you're limited to whatever reasonable frequency resolution you can get out of the PLL synthesizer. With an FPGA board, you'll likely be able
  to do a DDC.

Perhaps somebody wants to work on a comparable block-diagram for the Tx side? With roughly-comparable capabilities? One might make simplifying assumptions, like ADC and DAC clock rates are always common, which implies that the analog baseband filter selection would
  be common, etc, etc.

The FMC connector standard apparently provides for 68 signal paths. That should be plenty.

Marcus Leech
Principal Investigator
Shirleys Bay Radio Astronomy Consortium

reply via email to

[Prev in Thread] Current Thread [Next in Thread]