[time-nuts] Water on Enceladus - What does this imply about NASA'a ability to measure frequency?

[Jim Lux]

On 4/4/14 9:34 AM, Chris Albertson wrote:
> On Fri, Apr 4, 2014 at 6:19 AM, Jim Lux <jimlux at earthlink.net> wrote:
>>
>> Radio science and navigation measurements are quite impressive in their
>> accuracy and attention to detail. measuring range to cm (out of a billion
>> km, i.e 1 part in 1E14) and velocity to mm/s is sort of standard.
>
> Looks to be about one order of magnitude better than standard.
> They claimed 90 microns/sec velocity in this case.   I was looking for
> a document that shows the design of the radio science system in the
> spacecraft but did not find it.

One of the links in the other message I posted has some of the links

  Best I can tell is they use a phase
> locked receiver transmitter as a kind of transponder so the high
> precision clock is on Earth.

Exactly.  Cassini carries a "deep space transponder" or DST, which is 
the predecessor of the Small Deep Space Transponder(SDST) which has been 
flying on most missions from JPL.  APL has their version on Messenger 
(Mercury) and New Horizons (Pluto).

The basic technique is to have a phase locked receiver with very narrow 
loop bandwidth (a few Hz) that locks to the uplink carrier. 
Traditionally that would be at around 9 MHz (called f0 in the coherent 
transponder world), and the receiver LO would be at 748 * f0, so the IF 
is at f0.  That same oscillator is then multiplied up by 880 to generate 
the X band transmit signal.  (hence the 880/749 ratio). S-band 
transponders use a similar scheme with 240/221 as the ratio.  Early 
X-band radios used S-band designs with an added x4, just as many modern 
Ka-band radios use a x4 on the output of a Xband transmitter (e.g. 
Cassini is a X up/X down and X up/Ka-down system.

the DST and SDST use DROs as the microwave oscillator.  The SDST uses a 
VCXO that's around 80 MHz (8f0), but other than that, it's pretty much 
the same design approach.  In the SDST, the carrier tracking loop and 
data demodulator is implemented in a digital ASIC which drives a DAC to 
control the VCXO.

Newer coherent transponders do things a bit differently. They use the 
same stable XO, but then use a pair of NCO/DDSs to generate the 
reference oscillator for the multiplier/PLL for Rx and Tx.  The tracking 
loop (and its filters) is implemented in digital hardware (FPGA). 
There's some cleverness in setting things up so spurs don't bite you, 
and that changes in the crystal frequency don't propagate through.

By the way, for the best performance, you want to actually move the 
receiver LO to keep the signal at the same place in the IF, as opposed 
to doing some sort of block conversion and tracking entirely in 
software.  That way, you don't worry about the phase vs frequency 
characteristic of the IF filters: you're always at the same place.  All 
you have to worry about is phase vs temperature at one frequency.
That said, the Electra UHF proximity radios use a LO that goes in big 
steps, so their coherent turnaround performance isn't as good (although, 
conceivably, one could characterize the IF group delay characteristics 
and build an equalizer in software)

We're also going to GaAs VCOs because they have wider turning ranges.

Historically, transponders are made in extremely limited quantities (3-4 
units every few years) and they have a lot of "touch labor" for tuning 
(e.g. you get your frequency assignment years in advance, and you order 
crystals at the right frequency, etc.).  There's a fair amount of reuse 
of spare transponders (e.g. a mission which is flying 1 or 2 will buy an 
extra, and then when they successfully launch, will hand off that spare 
to the next mission) which leads to all sorts of channel assignment 
issues (Opportunity and MRO have the same DSN channel, for instance), so 
there's been interest in designs which can have their channel selected 
after manufacturing.

DROs don't have the tuning range to cover the whole 50MHz X-band, and 
certainly not the 500 MHz Ka-band.  We spent a couple years trying to 
make a dual control input DRO with a coarse and fine inputs, but it 
didn't work out so well.  As readers of this list will appreciate, a 
quiet oscillator has high Q resonators, and that is the opposite of what 
you want for wide tuning range.  We developed some prototypes using GaAs 
VCOs whcih seem to work quite well, and that's the direction we'll 
probably go in the future.  Personally, I will be happy if I never have 
to fool with making DROs again. High performance DROs are the epitome of 
touch labor, and being basically a mechanical resonator, have 
microphonics, temperature coefficients, picky alignment during assembly, 
etc..  A monolithic solution which uses lithography is FAR better, if 
you can get it to work.





   They say this is the first time they
> are able to detect mechanical movement in the ground station antenna
> in the Doppler data.

Yes. they've made a big effort to do this in preparation for Juno (and 
Bepi-Colombo) because they're trying to push the radio/gravity science 
performance by a couple orders of magnitude over Cassini.  Get a few 
more spherical harmonics out of the model.


   I guess 90 uM/sec sensitivity just about
> everything is a noise source.

Very much so.