[time-nuts] Tbolt disciplined LPRO Rubidium

WarrenS warrensjmail-one at yahoo.com
Fri May 13 16:16:29 UTC 2011


In response to request on how I set up my Tbolt with a long time constant to 
directly disciplined a LPRO 101 Rubidium;

I use a standard LPRO 101 Rb as an external replacement for the Tbolt's 
internal OCXO,
and connect the LPRO's external C_Field adjustment input to the Tbolt's Dac 
out thru a 1K ohm resistor (added as protection for the Tbolt.)

Due to Tbolt's software limitations, an Extended_TC setting method needs to 
be used for setting Time Constants above 1000 sec.
To calculate the Tbolt's Effective TC setting using this extended TC setting 
method

a) The Effective_TC of the TBolt is Actual_TC setting times the 
Multiplier_Factor
b) The Effective_Damping factor of a Tbolt is Actual_Damping setting divided 
by the square root of the Multiplier_Factor
c) The Multiplier_Factor is equal to "Dac_Gain used / Actual_EFC_Gain"


Example for setting Tbolt's Effective_TC = 20,000 sec,  Effective_Damping 
factor = 0.6 for a LPRO.
a) Use Multiplier_Factor = 100
b) Dac Gain = +0.90 Hz/V
c) TC = 200 sec
d) Damping = 6.0
e) Set Initial Dac value to +2.5V,

After adjusting  the LPRO's to 10 MHz using it's internal freq C field pot 
(with the external C_Field input floating  at 2.5V)     .
f) It is a good idea to limit the Tbolt's Dac output swing by setting the 
Max_Dac out to + 3.0 V, and the Min_Dac out to + 2.0 V.


Example of setting Tbolt's Effective_TC = 5,000 sec,  Effective_Damping 
factor = 0.7,   that I use to discipline a very stable external dual oven HP 
10811.
a) Use Multiplier factor = 10
b) Dac Gain = -3.24 Hz/V    (Actual Gain for this 10811 is -0.324 Hz/V)
c) TC = 500 sec
d) Damping = 2.2


The effective TC can be measured by how long of time that it takes for a 
Freq offset error to return to zero, best measured by using LH's Dac voltage 
plot.
The effective Damping can be calculated by how much Freq error overshoots 
there is, which in turn determines how long of time the Phase error takes to 
return to zero.
For a damping =  0.6  the Freq overshoot is 33% and Phase correction time is 
about 3.5 times the Freq correction time  (10%  phase overshoot)
Using a damping of  0.7,  the Freq overshoot is 25%, & phase correction time 
is  5x the TC setting. (no phase overshoot)
With a damping factor of 1.0, there will be very little Freq overshoot, but 
that causes the phase correction time to be very long. (overdamped Phase 
response)

Using Lady Heather, the effective TC and Damping_Factor settings can be 
verified by plotting the offset freq recovery time to an intentional Dac 
step error.
a) Disable the control loop (D D)
b) Change the Dac voltage (D S ... ) by an amount equal to 1e-10 or 1e-9 
(verify on Osc plot)
c) Enable the control loop (D E)
d) Plot the Phase, Dac_Voltage and Osc_Freq over at least 5 times the 
effective TC setting.

Before replacing a Tbolt's internal Osc with a cheap rubidium, one should 
understand the noise tradeoffs, especially at low taus.
Lots of plots available on request

ws

******************* 




More information about the time-nuts mailing list