[SC.LUG] Damn!

[Ian Molton]

On Sun, 8 Dec 2002 14:50:39 -0000
"Martyn Kinder" <martyn@czd.org.uk> wrote:

> <snip>
> > - Setting a frequency reference accurate to about 1 part in 10^9
> > accessible to both
> 
> Not likely ;-)
> 
> I think this is the easiest bit. Using GPS, you can generate a very
> accurate 10MHz reference signal that should be in phase no where you
> are in the world...

yeah, but actually getting that 10MHz signal out of the GPS into the PC
and through the software is going to skew it BADLY...

> larger. To achieve reasonable resolution at 10Ghz, you really need to
> know where you are to within 1/10 wavelength - approx 30mm....

indeed :)

What Im wondering is if we can infer this from our data by basically
observing something, and then 'tuning' our system to give the best
image?

> There are a few people (amateurs) in the US experimenting with this.
> The usual technique is to place a delay on the Rx that is "nearest"
> the source and then combine the signals - you will probably need to
> draw this out and you will see how you can cancel wavefronts to achive
> the resolution. Its very impressive. The americans have done it (on a
> much smaller scale) by running identical lengths of coax (also using a
> common local oscillator) back to the receiver.

thats what I was going to do :)

> Now this will still
> only work if the radio source is at or near the zenith. Imagine that
> the source is at an elevation of 1 degree and due south.

zenith = horizon ? (Im new to the terms used in astronomy really...)

> I am making the assumption that we are not going to complicate this
> further by trying to track and will rely on a drift scan.

I think its going to be complicated enough to align our dishes so we
both point at the same area of sky ;-)

> Your Rx is 10km due north of me. The radio front will reach you a mere
> 10/300,000 seconds after it reaches me. Therefore we need a delay at
> my end 1/30,000 second.

thats approaching the minimum sample period we can do on commodity
hardware...

Maybe I should move house so we have exactly 1/44100 of a second delay?
;-)

> Not much you may say but you can't rely on the
> public internet services to generate that delay and dynamically mix.
> Therefore we need a common time source, again compensated for to allow
> for the distance beween us recorded onto a second channel. i.e. the
> left channel of a stereo pair.

Too complex (and wasteful of bandwidth). we need to use packets.

I assume you arent going to buy a dedicated internet connection for
this, so we're going to lose packets of data from time to time.

My machine can read from its dish constantly, so no problem there. if
mine is the server, it'll need to receive packts from you with a
timestamp, so it can reintegrate the data.

also, our bandwidth is limited - we cant do 20KHz continuous, as thats
about 50K/second!

what we can do is take chunks of 20KHz bandwidth and use them. if we did
one in 4 chunks, thats 12K/sec which is the most our DSL can sustain
(12K/sec out, 50 in).

Alternatively, we could drop our bandwidth to 5KHz and run continuous,
but I think thats going to be a worse option, as our LNBs will certainly
be drifting by more than 5KHz, I think.

We will need a frequency reference so we can calibrate our dishes too...
any massive discrepancies, we can fix by tuning the FM radio (in my
case) to a slightly higher or lower band. My cheap one will tune in 5KHz
increments.

> analyse the data. Now this will give us the effective resolution of an
> eliptical dish  radio telescope 10km major axis :-> and probably 600mm
> minor axis :-<  So we need to do the same with another volunteer to
> the east or west of us - now we are in buiness

Hehehe :)

any takers? you can get old 60cm dishes by just offering to remove them
from peoples houses, so the initial equipment cost is low (3 dishes and
a radio that can hit 1GHz, about 60ukp)
 
> ...It'd probably suck, but it'd be fun to try :)
> 
> Oh I agree... in fact this discussion has fired me up to have a go at
> a SBLI )i.e. Garden sized as a pilot. If we can get the  problems
> sorted here it should be straightforward to scale it upwards.

I need to get some connectors to solder onto my LNBs...

Hmm. I wonder... many LNBs have active mixers - perhaps one of those
types will give a better mix than simply shorting two other dishes
together...