Empathy List Archives

MS

Mark Sims

Tue, Jul 8, 2008 3:49 AM

The Thunderbolt attempts to align the specified PPS and OSC edges with the GPS satellite time signal. The PPB parameter is its estimate of the error of the oscillator edge from where it needs to be. They specify it in mysterious parts-per-billion, but never really say billions of what. They do say positive values mean the 10MHz signal is running slow relative to GPS time.

My ADEV plotter takes the PPS error in nanoseconds and adds 1.0 to it to get a per-second phase error. To get comparable values from the OSC PPB value, divide it by 1E9 and multiply by 100. Add that to 1.0 seconds to get a per-second phase errorof the OSC signal. The resulting ADEV plots of PPS and OSC PPB are off by a decade below 20 tau, and match very closely above 100 tau. The OSC PPB is worse below 100 tau (4.5e-9 vs 4.5 e-10 at 1 tau, both around 4e-11 at 100 tau) by 10,000 tau the OSC PPB is slightly better than the PPS signal (both in the 5e-13 range).

As far as messing with the discilining constants, I figure Trimble has them fairly well optimized for their OXCO. They probably only need to be changed if you graft a different OCXO onto the unit.

It’s a talkathon – but it’s not just talk.
http://www.imtalkathon.com/?source=EML_WLH_Talkathon_JustTalk

The Thunderbolt attempts to align the specified PPS and OSC edges with the GPS satellite time signal. The PPB parameter is its estimate of the error of the oscillator edge from where it needs to be. They specify it in mysterious parts-per-billion, but never really say billions of what. They do say positive values mean the 10MHz signal is running slow relative to GPS time. My ADEV plotter takes the PPS error in nanoseconds and adds 1.0 to it to get a per-second phase error. To get comparable values from the OSC PPB value, divide it by 1E9 and multiply by 100. Add that to 1.0 seconds to get a per-second phase errorof the OSC signal. The resulting ADEV plots of PPS and OSC PPB are off by a decade below 20 tau, and match very closely above 100 tau. The OSC PPB is worse below 100 tau (4.5e-9 vs 4.5 e-10 at 1 tau, both around 4e-11 at 100 tau) by 10,000 tau the OSC PPB is slightly better than the PPS signal (both in the 5e-13 range). As far as messing with the discilining constants, I figure Trimble has them fairly well optimized for their OXCO. They probably only need to be changed if you graft a different OCXO onto the unit. ---------------------------------------- _________________________________________________________________ It’s a talkathon – but it’s not just talk. http://www.imtalkathon.com/?source=EML_WLH_Talkathon_JustTalk

TV

Tom Van Baak

Tue, Jul 8, 2008 1:55 PM

Mark Sims writes:

My ADEV plotter takes the PPS error in nanoseconds and
adds 1.0 to it to get a per-second phase error. To get

I'm not sure why you would add 1.0 here? The PPS error is
a per-second time-interval (phase) error already, no?

comparable values from the OSC PPB value, divide it by
1E9 and multiply by 100. Add that to 1.0 seconds to get a
per-second phase error of the OSC signal. The resulting
ADEV plots of PPS and OSC PPB are off by a decade
below 20 tau, and match very closely above 100 tau. The
OSC PPB is worse below 100 tau (4.5e-9 vs 4.5 e-10 at
1 tau, both around 4e-11 at 100 tau) by 10,000 tau the
OSC PPB is slightly better than the PPS signal (both in the
5e-13 range).

It's not clear what you're doing here. Are you comparing your
TBolt to an external reference (Rb or Cs or other GPSDO)?

Or are you just processing the internal TBolt numbers and
assuming you can derive ADEV stability numbers of GPS
or the GPSDO or the OCXO itself from that?

As far as messing with the discilining constants, I figure
Trimble has them fairly well optimized for their OXCO. They
probably only need to be changed if you graft a different
OCXO onto the unit.

Yes, it's quite flexible and you can have a lot of fun replacing
the OCXO. I'm assuming a number of you will be trying this
in the coming months.

But even with the onboard OCXO, it allows you to change the
time constant which can make a noticeable difference in the
mid-term stability of the GPSDO. I'll dig out plots I made last
year if you're interested.

/tvb

Mark Sims writes: > My ADEV plotter takes the PPS error in nanoseconds and > adds 1.0 to it to get a per-second phase error. To get I'm not sure why you would add 1.0 here? The PPS error is a per-second time-interval (phase) error already, no? > comparable values from the OSC PPB value, divide it by > 1E9 and multiply by 100. Add that to 1.0 seconds to get a > per-second phase error of the OSC signal. The resulting > ADEV plots of PPS and OSC PPB are off by a decade > below 20 tau, and match very closely above 100 tau. The > OSC PPB is worse below 100 tau (4.5e-9 vs 4.5 e-10 at > 1 tau, both around 4e-11 at 100 tau) by 10,000 tau the > OSC PPB is slightly better than the PPS signal (both in the > 5e-13 range). It's not clear what you're doing here. Are you comparing your TBolt to an external reference (Rb or Cs or other GPSDO)? Or are you just processing the internal TBolt numbers and assuming you can derive ADEV stability numbers of GPS or the GPSDO or the OCXO itself from that? > As far as messing with the discilining constants, I figure > Trimble has them fairly well optimized for their OXCO. They > probably only need to be changed if you graft a different > OCXO onto the unit. Yes, it's quite flexible and you can have a lot of fun replacing the OCXO. I'm assuming a number of you will be trying this in the coming months. But even with the onboard OCXO, it allows you to change the time constant which can make a noticeable difference in the mid-term stability of the GPSDO. I'll dig out plots I made last year if you're interested. /tvb

AT

Arnold Tibus

Tue, Jul 8, 2008 4:01 PM

On Tue, 8 Jul 2008 06:55:04 -0700, Tom Van Baak wrote:
[...]

Yes, it's quite flexible and you can have a lot of fun replacing
the OCXO. I'm assuming a number of you will be trying this
in the coming months.

But even with the onboard OCXO, it allows you to change the
time constant which can make a noticeable difference in the
mid-term stability of the GPSDO. I'll dig out plots I made last
year if you're interested.

/tvb

Perhaps there is not much to earn with it, but I thought as well about
to replace the OCXO with a 10811 for test purpose etc.

I am very much interested as well to see how to set the time
constant in conjunction with the damping factor properly with the
original onboard oscillator and, if already performed by somebody,
with an other type.
Would be great if you and others would share your/ their experience!

What are the parameters to be watched with TBoltmon,
what are the goals, to get the lowest variation on the 10 MHz ppb
indication?
What about the PPS value in ns UTC?
If I do slight modifications specially on the time constant I get
the indication drifting very slowly away, but the 10 MHz ppb
does not vary much...
Under this circumstance I do not see how to set the ns offset for pps.

Is it necessary to compensate for the antenna cable delay and
how I can do it? (pps offset?)

I guess that the outcome will be to not make much modifications !?

Arnold T.

On Tue, 8 Jul 2008 06:55:04 -0700, Tom Van Baak wrote: [...] >Yes, it's quite flexible and you can have a lot of fun replacing >the OCXO. I'm assuming a number of you will be trying this >in the coming months. >But even with the onboard OCXO, it allows you to change the >time constant which can make a noticeable difference in the >mid-term stability of the GPSDO. I'll dig out plots I made last >year if you're interested. >/tvb Perhaps there is not much to earn with it, but I thought as well about to replace the OCXO with a 10811 for test purpose etc. I am very much interested as well to see how to set the time constant in conjunction with the damping factor properly with the original onboard oscillator and, if already performed by somebody, with an other type. Would be great if you and others would share your/ their experience! What are the parameters to be watched with TBoltmon, what are the goals, to get the lowest variation on the 10 MHz ppb indication? What about the PPS value in ns UTC? If I do slight modifications specially on the time constant I get the indication drifting very slowly away, but the 10 MHz ppb does not vary much... Under this circumstance I do not see how to set the ns offset for pps. Is it necessary to compensate for the antenna cable delay and how I can do it? (pps offset?) I guess that the outcome will be to not make much modifications !? Arnold T.

time-nuts@lists.febo.com

Thunderbolt Loop Damping

As far as messing with the discilining constants, I figure Trimble has them fairly well optimized for their OXCO. They probably only need to be changed if you graft a different OCXO onto the unit.