[time-nuts] HP 58540A Phase Noise Improvements

[SAIDJACK at aol.com]

In a message dated 7/7/2006 09:15:28 Pacific Daylight Time,  
time.bandit at btinternet.com writes:

Hi  Paul,

While there are certainly oscillators out there approaching the  -160dBc at 
1KHz (OSA BVAs spring to mind), some of the phase noise  performance of the 
OCXO will be degraded by the GPS receiver itself. Not  sure from my 
experience that this sort of figure would be achievable in  practice.

I'm interested to see what other comments you  get.

Good luck.
Hello Paul, Rob,
 
yes you are right, the phase noise also depends very much on the design of  
the GPSDO itself, not just the oscillator. But it can be done!
 
Some weeks ago I sent a phase noise measurement of our new Fury GPSDO unit  
to this message board (let me know if you like to get a copy), in this you can  
see that our output has -155dBc phase noise at 1KHz offset, and we achieve  
this with an MTI 230 oscillator, an innovative power supply, filtering, and  
buffering circuitry. There are also no visible spurs above 20Hz carrier  offset.
 
Keeping in mind that the circuitry matters, it may not be straight forward  
to replace the 230 OCXO - you have to make sure the 260 unit you want to  use 
is compatible in Vdd, VEFC, control slope, output power/type/frequency,  power 
consumption (very unlikely) etc. Also, HP probably measured the thermal  
behavior of the entire unit at the factory (they mention this in some of their  
literature) and stored these calibration values in EEPROM - these  values won't 
work well for a different OCXO.
 
Also, to get the best performing 260 unit, you have to get a 5MHz one, and  
the HP unit probably uses a 10MHz one.
 
There is also another alternative you can use besides aquiring one of our  
$750 Fury units :)
 
This alternative is to use the 260 Oscillator (you may already have one) in  
a very simple, and highly effective PLL loop slaved to the HP GPSDO:
 
You can use a simple TI/Philips 74LVC86 Exor gate as the phase comparator  
between the outputs of the two OCXO's, then low-pass filter the output of the  
Exor (with a time constant of <1s), and feed this control voltage to the 260  
OCXO. Since the PLL phase comparison frequency will be 5 or 10MHz (depenging on 
 the 260 OCXO), you can use simple RC, or two cascaded RC circuits to 
low-pass  filter the output of the Exor Gate. The Exor gate will now control the 
phase of  the 260 unit to follow the HP GPSDO with 90 degree phase shift.
 
You may have to buffer or add a DC offset to the sine wave outputs from the  
GPSDO and 260 OCXO before feeding them to the Exor gate.
 
Run the Exor gate with a clean 5V power supply to get 0V - 5V EFC control  
voltage (2.5V being the desired steady state EFC voltage).  The 260 OCXO  will 
require a very clean supply as well (no switching regulators etc). For the  
loop filter, preferrably use Polyester caps, and make sure not to load the gate  
too heavily (use a 3.3K Metal Film resistor with a 47uF + 10nF cap for  
example).
 
The output of the 260 will now be perfectly aligned to the GPS reference,  
but any noise from the HP unit above the loop filter bandwidth (1Hz or less)  
will be completely filtered out. This includes all spurs etc that the HP  unit 
is generating.
 
Besides the 260 OCXO, all you need is <$2 in components to make this  PLL.
 
Hope this helps,
bye,
Said