[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Re: DSP based High Speed Data

The radio described in the article uses a DSP evaluation card with the same
type of codec as sound cards so it has the same limitations. To acheive low
error rates, the data needs to be sampled at at least 10 times the baud rate
so that the modem software can determine where the center of the bit cell is
and sample data there. Most audio codecs are limited to 48 ksps and newer
parts do 96 ksps. This limits DSP evaluation boards and sound cards to 9600

For high-speed data, the DSP needs to be preceeded by a faster codec. The
best way to do this today is to use an ADC that samples at 10-60 MSPS and
decimate down to the actual data rate. Decimation is just low pass filtering
(averaging) the samples and throwing away most of them. This can be
accomplished most inexpensively with digital down converter (DDC) and
digital upconverter (DUC) chips that were developed for cell phone base
stations. The IF is digitized directly and fed into these chips where it is
filtered, decimated and complex processing like taking the arctangent of the
phase to demodulate FM can be done in dedicated hardware.

These chips are much faster than general purpose DSP chips but more
specialized to radio reception and transmission. They implement the required
operations in parallel rather than requiring a programmer to specify these
serially as a set of instructions. They are about three times as expensive
as general purpose DSP chips but, they don't require programming other than
loading configuration data to set filter bandwidths and select the
modulation type (AM, FM, PSK, SSB, etc.).

Peter Anderson wrote a series of articles in QEX starting in March 1994
about a receiver using a DDC from Harris (now spun off and back to the
original name of Intersil). He just fed the 16-bit digital output to a DAC
to generate audio for SSB reception. When combined with a microprocessor to
do CRC calculations or implement error correcting codes, this type of
hardware should easily demodulate 200 kbps data streams. However, it needs
to attach to the receiver and transmitter IF or the output of a transverter.
I've purchased an Intersil HSP50214 DDC and 50215 DUC and will see what can
be done. The hardest part is going to be soldering the 380 pins on three big
quad flat packs (DDC, DUC and MCU). They use 0.026" lead spacing -- about
1/4 of the lead spacing on a DIP package.

>The original article was suggesting "high-speed" packet and did some
>calculations for data throughput.  The original article also assumed a
>base band bandwidth.
>I doubt if the "soundcard" solution can get anywhere near the speeds
>mentioned in the original article and I do not know any commercial sat-able
>amateur radios that have more than 5kHz baseband signal bandwidth.
>If the idea is really "high speed in DSP" I would be forgetting about sound
>card and intel libraries designed to run on a PC.  I would be trying to use
>something that can do some DSP on the full 20kHz baseband signal.
>Such a radio was described as a kit in QST last year (Sept-Nov1999).  THis
>would be my option if someone in the USA was prepared to do another kit
>Although originally a normal FM/CW/SSB weak signal radio this radio >should
>be able to handle betwen 56k&115kbaud (by my estimates only) with only
>writing some software and adding a serial interface.

Via the amsat-bb mailing list at AMSAT.ORG courtesy of AMSAT-NA.
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org