[hpsdr] Welcome mail and different questions/ideas

[Robert M. Ganter HB9DNN]

Hello list,

I subscribed to the HPSDR list quite a while ago as I have been working
on this topic as well.
Lurking around has been great fun up to now and, although I am an
electrical engineer by profession I have been learning a lot.

I had a short discussion with Phil Covington about CORDIC a while ago. I
 was in the process of writing a CORDIC algorithm in VHDL. Unfortunately
it neither synthesized correctly nor could it be simulated. This has
changed now and, except for precision estimation for more than 12bits
(Excel can't do FFT for more than 4096 points...) the algorithm proved
to work correctly. As soon as I have documented the tests I will publish
it (it's under GNU licence). The same applies for the CIC and FIR
filters, which I designed with Excel and a DOS program called DISPRO25.
According to the design values the filters will provide a 2x20kHz (I and
Q) passband with less than 0.04dB ripple and an stopband attenuation of
more than 120dB at 2x24kHz. All mirrors from the CIC stage are
attenuated better than 120dB, therefore no out-of-band signals should
appear at the output (except for spurs from the ADC). The combination of
the two signals (I and Q, complex signal) into a real signal will lead
to 40kHz passband bandwidth.

There was a discussion on the list about the output sample rate to be
used (and the bandwidth/interface needed to transfer to/from a PC).
My understanding is, that for all but broadband modes (high bit rate
Packet, Video, etc.) a baseband bandwidth of 40kHz would be more than
enough for virtually all modes used in amateur radio. Using the approach
mentioned above an SP/DIF interface for 48kHz sampling rate would be ok
for this. Soundcards with higher sampling rates often have,
unfortunately, only drivers for Windows (and sometimes Mac OS). This way
it is not possible to run it under other OSes (e.g. Linux) and there is
no way to write specialized drivers (e.g. to send control data from/to
the PC). There are quite a few 48kHz soundcards and interfaces with
SP/DIF in out on the market where open source drivers are available.
Another solution could be a 100MBit Ethernet interface. This way even
more than one channel could be transferred to/from the PC or even more
than one host.

Two examples come to mind here:
- a two or even more channel receiver with backend processing anywhere
on the network
- a spectrum display channel. One question here is the bandwidth which
should be displayed. What are the requirements/whishes for such a
feature? Are 40kHz enough? Should it be more? In this case a different
filter set must be used and (probably) the FFT has to be done within the
FPGA to limit bandwith. How many points, which frequency resolution
would You expect/wish for this feature to be useful?

I found an interesting website today: http://www.wavefrontsemi.com
(maybe better known under the name Alesis). They have some interesting
chips (ADC/DAC and DSP) which might be interesting for baseband processing.

That much for now. I really appreciate Your ideas and answers to this
and other topics on this list. IMO, SDRs are one of the most thrilling
things of amateur radio today.

vy 73

Robert HB9DNN