[time-nuts] Re: Tuning a GPSDO loop for optimal disturbance handling

[Bob kb8tq]

Hi

To convert 1.2 uV  / sqrt(Hz) to something else, first you would need to know
the target bandwidth. On a TCXO Vtune it could be almost anything. Just to
get started, let’s make a random guess of 2.5 KHz. Sqrt(4K) gives us 50 which 
is the main reason for picking that number.

50 x 1.2 = 60 uV which is also a nice round number. Note that there is no reason
to assume that the DVM bandwidth and TCXO bandwidth are “same/same”. 

Next up is the rather alarming fact that a lot of DAC’s spec noise at a single
frequency ( like 100 Hz or 1 KHz ). Some specs do this without even mentioning
what the frequency is.  They may or may not give you any idea of what happens 
at lower frequencies. At some point the DAC noise will hit a 1/F corner and start 
climbing off towards infinity. 

DAC’s tend to have “glitch energy”. More or less, when you change the output,
you get an extra “pop” of energy that would not be there if the part was ideal. 
Just how much of a pop you get depends on the code you go from and the code
you go to as well as the basic part you are using. This can lead to some very
“interesting” noise issues ….

Bob



> On Mar 23, 2022, at 2:19 PM, Erik Kaashoek <erik at kaashoek.com> wrote:
> 
> Hi All,
> 
> Some more work done
> There was a  calculation error in previous message. The minimum DAC step is 5 uV which is equal to a relative frequency change of 0.5e-10
> The frequency meter used for testing (Picotest U6200A), using a gate time of 1 s, measured a relative frequency "noise" in the order of 2e-10 so a 1 step DAC change is below the frequency measurement noise.
> The 6.5 digit DVM used to measure the Vtune was directly connected to the Ground and Vtune of the VC-TCXO and is able to resolved all 6.5 digits in the 2 V range (checked with a 1.5 V battery slowly draining) thus anything down to 1 uV should be visible.
> However the  DVM measured on the Vtune about 20 uV RMS noise (consistent with the measured frequency noise), largely  independent of the high DAC output voltage. Only when the total DAC output is set to zero the noise reduced
> This all was consistent with the minimum observable DAC change of 4 steps (20 uV change Vtune and 2e-10 relative frequency change)
> Moving to a separate voltage regulator for the DAC and TCXO did not make a clear difference in the frequency or voltage noise levels.
> The DAC used has an internal Vref with a noise level of 290 uVp-p but as the DAC output noise does not scale to the set output voltage the reference noise does not seem to be the problem.
> The DAC output noise is specified as 1.2 uV/sqrt(Hz) , but I do not understand how to translate this number to the current measurements
> 
> Looking at the noise patterns there seems to be a positive correlation between voltage going up and frequency going up but noise is a bit difficult to interpret.
> 
> To test if the DAC output voltage noise was causing the frequency noise the DAC was replaced by a 1.5 V battery. The DVM showed no noise at all at Vtune with the battery, just a slow draining (1 uV steps) of the battery. This also proved there is no ground plane problem for the GND/ Vtune of the VC-TCXO.
> With the battery the frequency noise remained similar thus proving the DAC noise was not the major contributor to the frequency noise.
> 
> It is possible the 2e-10 noise level is the lower measurement limit of the frequency meter used (U6200A) with 1 s gate time but I have no way to check this.
> 
> The ADEV of the GPSDO 10MHz output measured over 60 minutes (attached) stays below 1e-10 and nicely follows the GPS PPS ADEV down starting at 100s
> 
> Any suggestions?
> 
> There is still a lot to learn.
> Erik.
> 
> On 22-3-2022 21:29, Dana Whitlow wrote:
>> Eric,
>> 
>> Do the observed DAC steps correspond in polarity to the observed frequency
>> changes, or just the reverse?  That's a key determination to make in placing
>> blame, for it tells you whether the DAC steps are causative to the frequency
>> changes, or rather the PLL's reaction to problems in, say, the VCO itself.
>> 
>> Dana
>> 
>> 
>> On Tue, Mar 22, 2022 at 2:10 PM Erik Kaashoek <erik at kaashoek.com> wrote:
>> 
>>> Hi Bob,
>>> By your advice I went for a 6.5 digit DVM and after logging and plotting
>>> the DAC output its clear there are some stability issues in the DAC
>>> output. The voltage is wandering around at about the level of frequency
>>> wandering observed.
>>> A different supply topology for the DAC and VCXO will have to be
>>> created.. The DAC resolution is 50 uV (2.048 V / 400000 steps) but the
>>> random variations are about 4 times p-p larger.
>>> Once this is done I hope longer term logging of frequency and voltage
>>> will make sense.
>>> Or is it better to have a low pass filter between the DAC and the VCXO
>>> Vtune input so the loop is fast enough to remove the remaining drift?
>>> If so, what should the time constant of the filter be compared to the
>>> intersect of the ADEV of the GPS and the VCXO (100 s)?
>>> Having a big time constant (10s?) will be a pain in the initial tuning
>>> and will require active components.
>>> Does owning a 6.5 digit DVM qualify one as a volt-nut? Or should first
>>> some voltage references be added?
>>> Erik.
>>> 
> <Capture.JPG>_______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.