Trimble Thunderbolt GPS Disciplined Oscillator

[edpalmer42]

No, to discipline an Rb clock by GPS is not good at all.. this would spoil the very good short and midterm stability by the GPS jitter.

The Rb itself is so stable, that it makes no sense to discipline it further, it's better to calibrate / compare it once in a month or once in a 3 months.

Most of GPSDO time bases, like the Thunderbolt or the ones from HP have a quite good Holdover logic, that means they learn how the OCXO behaves over time and over temperature changes, and if the GPS signal fails, the Holdover circuit corrects the OCXO by monitoring the temperature.

TB works quite stable in this Holdover mode, HP GPSDOs seems to be better, due to a more stable OCXO. For performance figures, see various time-nuts sites.

Much more important, if the antenna is located freely on the roof of the house, maybe +/- 60° free sight to the sky, you will see at least 3-4 satellites all the time , and therefore never loose GPS track, maybe for a few minutes in thunderstorms only.

Btw.: Initially, an exact determination of the antenna position (< 1m) is important.
TB & LH offer a 24/48/72h survey function for that.

Frank

Normally, I would agree that there's no point in disciplining an Rb standard.  But in this case, I'm not sure.

Yes, a GPSDO learns how it's oscillator performs so it can control the oscillator during holdover.  But for the OP, where he might only have GPS for a few hours per day, will the learning algorithm be able to function?  What if the few daily hours of GPS are broken up into smaller intervals?  The learning part might go out the window and only the frequency-adjustment feature will work.  Under those conditions, disciplining an Rb might be the best option.

Yes, you may, depending on how good your OCXO and Rb are, degrade the short and medium stability by disciplining the Rb.  If you're using the signal as a reference for a multi-GHz transmitter that could be significant.  For other applications, it might not matter.

If you discipline the OCXO, and it drifts enough on holdover, you may have a relatively long way to go to get it back on frequency when GPS returns.  An Rb is unlikely to drift enough for that to be an issue. 

It's up to the OP to decide how he will be using the signal and what specs are most important for his application.  This might be a situation where testing is required to determine just how good or bad the performance of each solution will be.  If he's lucky, he might be able to borrow a GPSDO from someone for testing.  Another possibility would be to buy a timing-grade antenna which would be necessary anyway.  Add a timing-grade GPS receiver board like a Motorola M12 or Trimble Resolution T (relatively cheap at < ~$50) and start making some measurements of the 1 PPS output of the receiver.  This will tell how good or bad GPS reception will be.  Don't rely on the results from a cellphone!

Ed

[Bryan]

My sole purpose for the GPSDO is to tune and calibrate some test equipment, namely a frequency counter to start.

http://youtu.be/B_Y971uY3Nk

With a  accurate frequency counter I can than calibrate some other equipment, time mark generator, frequency generator etc. I suppose worst case I could move the equipment to another location and do the calibration there. 

Out of curiosity how many satellites are required to give a accurate 10Mhz, is it just one or are others needed? I thought the reason for more than one is because the satellite will eventually move out of the antenna reception depending on the time of the day etc, so for a continuous lock you would want more than one.  Looking more seriously at the Z38xx receivers Seems they have a  very high quality double oven, and lot's of information on the internet on using them. Seem more available right now than the Trimble Thunderbolts.

[VK5RC]

I thought it just needed one satellite (for any PPS signal) to discipline the oscillator of the GPSDO. I thought others are needed to get a precise GPS fix of the antenna to then get the PPS output EXACTLY in synch with UTC, this is where antenna position and antenna type and delays in amps/cables etc gets really critical if it is of importance to you.
The Thunderbolt will go through a long phase of determining its position or you can give it a rough fix to shorten this period.

[edpalmer42]

Although only 1 satellite is necessary to maintain timing, 3 or 4 satellites are necessary to establish a location.  This is important for timing applications because it minimizes or eliminates phase jumps when satellites leave or join the group that's being tracked.  There may also be situations where one satellite has a lower signal level or slightly less desirable timing characteristics than another.  Tracking multiple satellites allows the receiver more opportunity to choose the ones it wants to track.  I suspect that different receivers (particularly different ages) have different capabilities.  I know that newer receivers are capable of tracking more satellites simultaneously.  The HP Z series and the Tbolt can only track 6 or 8 satellites.  Newer receivers can track over 50, but that high a number is total overkill for timing applications.

Bryan, it sounds like for your application, either a disciplined quartz or Rb oscillator would work fine.  Disciplined Rb units are very hard to find and very expensive, but there are DIY versions around.  For quartz units either purchase or DIY are valid options.  Some are more towards the design end, some are more like kits.  This unit:  http://www.rt66.com/~shera/index_fs.htm can be tweaked to discipline either quartz or Rb.  This unit:  http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm is famous for giving really good performance while being dirt simple.

Ed

[Dr. Frank]

Out of curiosity how many satellites are required to give a accurate 10Mhz, is it just one or are others needed? I thought the reason for more than one is because the satellite will eventually move out of the antenna reception depending on the time of the day etc, so for a continuous lock you would want more than one. 

The first thing to do is a location survey, i.e. to precisely measure the antennas position.

When the location of the antenna is known (and fixed), then a single satellite is sufficient to determine the time precisely.
Simple mathematical reason: the triangulation needs 4 satellites to solve the equations for the 4 dimensions: X,Y,Z, and c*t.
Therefore, if 3 variables are known, the 4th variable can be calculated from one satellite.

As the GPS signal is disturbed (jitter) on the path to the receiver, more satellites will add stability to the time signal, anyhow.

8 channels are fully sufficient to get a very stable time signal.
Additionally the number of visible satellites is limited, especially, as you may want to filter out satellites which are lower than 20° over horizon.. this avoids reflected signals.
Instead you want to receive the signal directly, only.
Therefore, you receive rarely more than 6 satellites.

Frank

[Bryan]

Bryan, it sounds like for your application, either a disciplined quartz or Rb oscillator would work fine.  Disciplined Rb units are very hard to find and very expensive, but there are DIY versions around.  For quartz units either purchase or DIY are valid options.  Some are more towards the design end, some are more like kits.  This unit:  http://www.rt66.com/~shera/index_fs.htm can be tweaked to discipline either quartz or Rb.  This unit:  http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm is famous for giving really good performance while being dirt simple.

Ed

Yes, have seen a few links on making a GPSDO from a GPS receiver that outputs a 1pps or 10khz output Although it appears that 10Khz is much easier to implement.   

[johnny_canuck]

I want to use the 10 MHz output of my Thunderbolt as an external frequency reference for my frequency counter. The counter requires HCMOS logic levels. Does anyone have a schematic for a suitable interface? Tnx.

Ken

[VK4GHZ]

Does your Trimble Thunderbolt GPSDO report a temperature of -55° ?
It is easy to fix by replacing U9, a DS1620S SOIC-8.