4

I have a modern microprocessor (Cortex-M3) with a decent ADC (10-bit 400kSPS or 12-bit 200kSPS).

How to design and build a simple DSSS system for learning purposes? What is the simplest RF front-end that would be needed for this?

The bandwidth is not very important but the system should demonstrate high noise resistance since this is one of the big selling points of DSSS.

EDIT:

I hoped I would get more information on how to proceed on the software side. Like should I do blind correlation of the whole buffer at every sample or are there any "magic" tricks. Maybe I should ask another, more focused question.

jpc
  • 5,302
  • 1
  • 24
  • 45

3 Answers3

1

There are licensing problems associated with operating your own spread spectrum transceiver. A special dispensation was granted to some US radio amateurs several years ago, but it was rescinded:

http://www.tapr.org/ss_tapr_sta.html

You should approach your regulatory authority about the relevant legislation. Unless you are a licensed radio amateur, it is very unlikely that you will get permission to operate such equipment.

Leon Heller
  • 38,774
  • 2
  • 60
  • 96
  • Thanks for the tip. I think the the cited FCC statement was replaced with a little more permissive one: http://www.tapr.org/ss_fcc.html But your point is still valid. – jpc Mar 24 '11 at 11:21
1

The simplest thing to do would be to do all the modulation and carrier shifting in the baseband and then upconvert directly to desired transmitting frequency using one of the many IQ upconverters/downconverters from ADI, TI, Linear Technologies etc.

You will however need much better ADC/DAC then 10 bits (e.g. 24 bits audio codec chip) to get decent SNR, and also low phase noise oscillator for the up/down conversion.

Another approach would be to do the modulation at some low intermittent frequenty and then heterodyne that up. This approach needs more filtering to get rid of unwanted images that result from mixing.

Jaroslav Cmunt
  • 1,450
  • 9
  • 10
  • So basically I/Q or at least phase modulation is the best way to go? When demodulating I should demodulate the phase and then perform the correlation on the result? Is simple amplitude or maybe frequency modulation also possible or would I lose all the benefits of DSSS? – jpc Mar 26 '11 at 01:33
  • The IQ up/down conversion technique is modulation agnostic, e.g. you use it to shift the signal from one frequency (carrier) to another, e.g. from baseband to transmit freqency or from receive freqency back to baseband. Of course having both I and Q channels makes modulation/demodulation easier if you know the algorithms. I don't have much practical experience with these, which is why I commented only on up/down part, which I know. Look for open source software for software defined radio for study of modulation techniques - GNU radio will probably have plenty of code on these. – Jaroslav Cmunt Mar 26 '11 at 18:15
1

While having 24-bits would be great lots of useful work can be done with only 14-bits. As an example, see the USRPs from Ettus Research used worldwide by researchers in conjunction with GnuRadio software.

http://www.ettus.com/downloads/ettus_ds_usrp_n200series_v3.pdf

user3574
  • 11
  • 1