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

[linrad] Re: Linrad via Ethernet

> Have been thinking again on this topic
> There seem to be a lot of embedded ethernet cards around nowadays - with
> 100MB cards /webservers at <US$200
> The Linrad software already supports network capability.
The current network implementation is very inefficient.
I am sure someone can give me the hints I need to do it better the day
some hardware that uses a network cleverly becomes available.

> It would get past all the audio/noise issues and the SNR and strong
> signal handling could be determined by the front end.
I do not see this as the problem. I do not think it would be very
difficult to make a card for the PC that would not pick up interference.

> You could have the radio unit remote to the computer to reduce noise -
> it could even be done over the net!
The problem is the interference from the PC that is picked up by the antenna.
Surely a remote A/D converter and a network (fibre or WLAN) would give
a beautiful solution for such problems. By providing a sync tone (for example
on 1296 MHz it would be possible to have several such receiver inputs
and route them all to the same computer via the network. Linrad currently
forms an adaptive antenna from two channels but in principle there is no limit
to the number of channels one might use. By using more channels one can form
adaptive lobes with a directive gain proportional to the number of channels
but the advantage is very much larger than that because Linrad can form such
a directive lobe for each interference source and make interference suppression
far more efficient. At the moment, only the algorithms for narrowband spurs
and wideband pulses are in place but in principle one could add routines for
any interference for which it is possible to build a computer model.

> USB seems too hard and there is no
> fundamental reason why it should be used
USB 1 as used in the SDR-14 is not fast enough. It gives serious limitations
to performance.

> The question is - what sort of data rate would you realize with these
> ethernet cards if you put a raw data stream through from say a serial
> ADC such as the AD9874?
I think one can come close to 100MB/s with a card that broadcasts its data.
The AD9874 would be a piece of cake but I do not think this chip is suitable
for Linrad because although this chip gives 24 bit output I do not think
the dynamic range in the output is very good. The chip has a 16 dB front end
attenuator and 12 dB of AGC. It uses a (probably noisy) synthetizer to
mix the signal to baseband and I can just not imagine this is something
I would want to use on a network. Seems just right for the USB or perhaps the
serial port....

My dream is a unit with one or perhaps two Analog Devices AD6644 (or better)
combined with a GC4016 that would produce 2*20 bit complex samples up to
perhaps 2 MHz and send the data stream as broadcast packages on a 100 MB
ethernet link.

The GC4016 is far superiour to AD6620, it contains four receivers that can
be cascaded pairwise to produce the very steep filters needed to make the
PC processing efficient but yet without alias spurs. The target price in
1K quantities is about USD 40 both for the A/D and for the four-fold decimation
chip so the unit will not be very cheap - but performance would be good,
very good if the sampling clock were well designed:-)


Leif / SM5BSZ

This message is sent to you because you are subscribed to
  the mailing list <linrad@xxxxxxxxxxxxxxxxxxxxx>.
To unsubscribe, E-mail to: <linrad-off@xxxxxxxxxxxxxxxxxxxxx>
To switch to the DIGEST mode, E-mail to <linrad-digest@xxxxxxxxxxxxxxxxxxxxx>
To switch to the INDEX mode, E-mail to <linrad-index@xxxxxxxxxxxxxxxxxxxxx>
Send administrative queries to  <linrad-request@xxxxxxxxxxxxxxxxxxxxx>