Post date: Oct 20, 2009 4:47:51 AM
Update: This is a legacy item that applies to VLBI that I was trying to do in the past. A GPS derived clock isn't necessary for the interferometry that I'm doing at a single location. I can confirm though, that when I run my SDRs on separate clocks, that I can do VLBI with Thunderbolts and the Valon 5007/3008 synthesizers.
VLBI requires very high quality oscillators. 'Classical' interferometers share a common oscillator that is used to drive two radios negating the need for a high quality clock source. For VLBI, two oscillators that beat in unison are required necessitating the use of atomic clocks or GPS-disciplined clocks. A shift in one cycle between each clock would result in a 360 degree rotation of the correlation product in the interferometer. For instance, if two atomic clocks have an offset of 1 Hz at the frequency being observed, the interferometer correlation product would spin around one turn per second.
It sounds like a rather daunting problem. Some solutions...
- Hydrogen maser clocks: They have very good short term stability which is most important for interferometry. However, they are too expensive and exotic for amateur use.
- Cesium clocks: A good alternative, but they are still expensive and not something one would want to try out VLBI on a whim.
- Rubidium clocks: They are also cheap and plentiful on Ebay. I opted for two LPRO-101 clocks originally but they are not nearly as good as a pair of Thunderbolt GPS clocks. They can work, but their stability and phase noise are issues that are hard to deal with. Two LPRO-101s can match o.k. when in the same environment, but in different environments they drift apart quickly such that they are impossible to control adequately.
-GPS disciplined clocks: They are cheap and easy to find on Ebay. I have upgraded to Thunderbolt clocks and this is something that I should have done a long time ago. I ordered two Thunderbolts/antenna/power supply kits from Ebay, making sure to get the ones with the OXCO and Firmware version 3.0 and fired them up. After fiddling around with Tboltmon.exe, I reconfigured each unit to my location and was amazed by how much better these units are than the LPRO-101s.
Frequency multiplication to generate 125 MHz signal
The next challenge (and a big one): Multiplying the 10 MHz reference up to 125 MHz.
Those obsessed with phase noise know that a good crystal oscillator can produce a very clean signal that is hard to compete with. DDS and PLL synthesizers are much dirtier, but improved products are coming out that are going to make our lives a lot easier in this regard. High cost out-of-the-box solutions such as a HP signal generator are one way to get to 125 MHz that may interest some people, but I was looking for something less costly and with less bulk.
TAPR Clock Blocks
Early on, I used TAPR clock blocks www.tapr.org/kits_clock-block.html. The Clock Block was a gamble on my part because of the warning that this PLL clock multiplier had high phase noise that would likely make it less than ideal. However, the QS1R is not and ordinary radio it was unclear how phase noise was going to degrade the receiver's performance. My assumption was/is that phase noise would be ignored by the QS1R, while any clock errors will be related to clock jitter. The file "Phase noise plot.jpg" below shows how a TAPR Clock Block driven QS1R performs. It is certainly not ideal, but may satisfy some users. However, this issue is now moot, given a new product that is now available (see below).
Valon Technology Synthesizers
I have now upgraded to Valon Technology 5007 and 3008 synthesizers (www.valontechnology.com). Phase noise is -125 dB @ 100 kHz and it has two frequency outputs that should make it possible to do VLBI at much higher frequencies such as at 1420 MHz. A diagram of the Thunderbolt/Valon Clock system is on my main page and is pretty self explanatory. To me, this seems to be an ideal out-of-the-box clock solution for a wide range of uses. I can generate two separate signals ranging from 5 Mhz up to 4400 MHz that is "bang-on" the right frequency without calibration/fiddling.