That date says '98! GPSDO is probably a TCXO, or even VCTCXO. Usually you'd guess the GPSDO has worse short term drift, but long term average is 'perfect'; while the rubidium is better short term drift. Many rubidiums have an option to be disciplined by a GPS PPS signal.
I can't be 100% sure which model IFR Comm's test set it came from. It was something like a 120, 1600 or 1900. I do remember getting a pallet load of government surplus parts units. All those models are the same form factor. Actually, the programable step attenuator I sent you likely came from one of them also.
For timing purposes it's better to run a timing mode on the GPS receiver so it will just lock to the single strongest satellite. It removes phase jitter from jumping from one satellite to another. The GPS system uses caesium ground references which are about 2 orders of magnitude more accurate than your rubidium std. They then use them to re-sinc twice per day 2 to 4 rubidium clocks found in each satellite.
GPS satellites have used both Caesium and Rubidium clocks (since Block II in the 1980s). The European Galileo system uses Rubidium and Passive Hydrogen masers for its satellites' clocks. GLONASS seems to use only Caesium. BeiDou-3 may also be using Hydrogen masers too.
@@Brian_Of_Melbourne, I did some more reading on the present status of the GPS as found on the web. I do contract work for both the scientists who work on improving the system's frequency standards, so I really should ask them what's up with it these days. They are changing it behind my back. :) In any event, the timing produced by the GPS system is more accurate than a single, used cell-system rubidium reference, such as mine or IMSAI Guy's.
The diode between the op amp inputs is there for temp compensation. About -2.1 mV/*C depending on the current in the diode. The parallel R1 and R2 is a tolerance tweak for offset bias.
Yeah the short term accuracy is based on the Crystal accuracy and the longer term accuracy is that of the GPS atomic clocks...some "GPS locked" clocks actually vary quite a bit over the short term.. The Leo Bodnar device is pritty good especially for the price.. Alot of Microwave folks use it for the reference clock!
I used to run the old Z3801A GPSDO continuously feeding the bench but with the 48 VDC power supply and the unit itself they actually created a bit of heat in the shack (not a bad thing in the winter time!). Then moved to one of the old cell site Rb oscillators which I only turn on when needed at the bench for accurate measurements. Have thought about the Leo Bodnar now for a while and may revisit it. 73 - Dino KLØS
8:30 "It [the GPSDO] does it with a Phase Lock Loop." Typically, they're implemented with a *Frequency* Lock Loop. i.e. Essentially just counting pulses, comparing the counts (may be inherent, with one count vice clock), and then ever-so-subtly tweaking the OCXO control voltage. So technically, any temporary phase locking is merely a pleasing outcome of the FLL and its relative stability. But perhaps this one is different ! Cheers. 🙂
I seen a guy walking down the street today wearing nothing but socks. I think he was tweaking some Rubidium. I DO NOT ADVISE based on what my eyes had to endure today.
I think the Rb is more accurate. The PLLs in the GPS detector have a limit to how well their ratio of two integers can multiply the GPS signals to equal 1 Hz. The Rb source's 5.8 GHz reference is closer to 10 MHz and can approximate 10 MHz more precisely. But, the definition of 10 MHz is based on Cs instead of Rb.
Its an interesting subject. For a Rb standard, it very much depends if the C-Field adjustment on the Rb is set up correctly. Excitation of a Rb87 at 6.8347GHz causes the Rb87 atom to move from its lower to upper ground state to allow optical pumping to continue. But that nominal 6.8347GHz can be "trimmed" due to Zeeman splitting by an applied magnetic field. (The C-Field). A Rb frequency standard is very frequency stable. But not that does not mean the derived frequency is accurate. Don't get me wrong..... It will be VERY close and probably way better than say 1E-9. But a well set up Rb will be better than 1E-11 with its C-Field optimally set. Even a cheap GPSDO will have a medium term accuracy of better than 1E-10 once warmed up so better than a non optimal Rb standard. Example..... On our LF transmitters, we used free running Rb standards. (HP5065A). They were monitored by the UK NPL off-air and every month, they would mail me (by post) how far we had drifted w.r.t UTC. From the data presented, I would trim the C-Field adjustment to bring the Rb into line with the NPL Caesiums. We kept the Rb standards circa 5E-12 using this method. The Original Rb standards were upgraded to GPS Disciplined Rb standards. There was nothing to adjust. We got monthly reports from NPL that we were circa 3E-14 consistently. The GPS disciplining engine "trimmed" the Rb Science package C-Field. But your comments on the 1Hz signal hold true. There are GNSS chipsets and there are GNSS chipsets!! The ones used in commercial GPSDOs are optimised for timing applications where the 1PPS output is very phase/jitter stable. The GPSDOs found on eBay for a $100 will not be using a particularly good GNSS chipset. A minefield.
I have two Rb units and a GPS Osc -- I found comparing all three the "breathing" drift is the GPS. Over a long term the GPS should have higher absolute accuracy but short term not so much.
Curious to see the phase noise / jitter off the leo bodnar. If correct it's based on a u-blox m8c which has a 26 mhz internal crystal. since that not a whole number division with 10 mhz there probably will be some jitter. Tried this with a m8t before, which has a 24 mhz crystal. There was noticeable more jitter on 10 mhz compared to 12-8-6 mhz
I would suggest the C-Field adjustment on your Rb is way off. An un-disciplined Rb is circa 5E-11 if it is set up correctly and the Leo Bodnar locked up will be way better than 1E-11 and the drift between the pair of devices was many orders of magnitude more. My locked Datum GPDO in the lab here is running at 3E-12 at the moment.
Rubidium need few hour to make very acurate frequency, the first few minute on is few 1/100 hz offset. I adjust TCOCXO versu rubidium and after a day, the difference are 0.0002hz! I use HP 53131 with 5 seconde time base and TCOCXO are in frequenecy generator, output at 1ghz to the frequenecy counter.
from Stanford Research Systems: The Rubidium Oscillator status LED (either Unlocked or Locked) will be on whenever power is applied to the unit. The status will change from Unlocked to Locked within six minutes after power is applied if the module is started from room temperature. Warm-up and locking will take longer if the module is started from a lower temperature. Locked indicates the unit is warmed up and its crystal oscillator has been locked to the rubidium physics package. The output frequency will be about 200ppm below nominal when power is first applied to a cold module. The frequency will gradually converge on 10MHz as the module warms up. After about 4 minutes (typically) the unit will reach operating temperature and lock the crystal oscillator to the rubidium hyperfine transition. Within seven minutes,the module will be locked to within 0.001ppm of 10MHz.
@@IMSAIGuy I have old Efratom 100323-003. Compare to my ocxo up for many days, my rubidium have very very very little offset first minute after lock and is stable after hour later with full body warmup. This is why my comment, power up few hour before use for optimum performance, but juste after lock, is far more I need in precision in 99.9% of case. My Rubidium have voltage steering input to make ultrafine adjustments .
Is that an OCXO (oven-controlled) ? Did it get warm ? How much power did it draw ? Merely curious. No data on the web for that PN that Google can find.
Theorem 7, "Two things are not the same." Its a truism. Just another data point showing the truth. If they look the same, you are not looking close enough.
![เป็นไปได้ไหม - WanMai [Official MV]](http://i.ytimg.com/vi/_6lZ6b3_7Vo/mqdefault.jpg)
Hey, IMSAIGuy, I sent Leo Bodnar unit quite a while ago, and I'm glad it made its way to another video. I hope you're well -- thanks for the video!
Thanks again!
That date says '98!
GPSDO is probably a TCXO, or even VCTCXO. Usually you'd guess the GPSDO has worse short term drift, but long term average is 'perfect'; while the rubidium is better short term drift. Many rubidiums have an option to be disciplined by a GPS PPS signal.
1998? week 27? great content as always! can we make a two transistors oscilator for 88-108Mhz?
I can't be 100% sure which model IFR Comm's test set it came from. It was something like a 120, 1600 or 1900. I do remember getting a pallet load of government surplus parts units. All those models are the same form factor. Actually, the programable step attenuator I sent you likely came from one of them also.
For timing purposes it's better to run a timing mode on the GPS receiver so it will just lock to the single strongest satellite. It removes phase jitter from jumping from one satellite to another.
The GPS system uses caesium ground references which are about 2 orders of magnitude more accurate than your rubidium std. They then use them to re-sinc twice per day 2 to 4 rubidium clocks found in each satellite.
GPS satellites have used both Caesium and Rubidium clocks (since Block II in the 1980s). The European Galileo system uses Rubidium and Passive Hydrogen masers for its satellites' clocks. GLONASS seems to use only Caesium. BeiDou-3 may also be using Hydrogen masers too.
@@Brian_Of_Melbourne, I did some more reading on the present status of the GPS as found on the web. I do contract work for both the scientists who work on improving the system's frequency standards, so I really should ask them what's up with it these days. They are changing it behind my back. :)
In any event, the timing produced by the GPS system is more accurate than a single, used cell-system rubidium reference, such as mine or IMSAI Guy's.
The diode between the op amp inputs is there for temp compensation. About -2.1 mV/*C depending on the current in the diode. The parallel R1 and R2 is a tolerance tweak for offset bias.
Yeah the short term accuracy is based on the Crystal accuracy and the longer term accuracy is that of the GPS atomic clocks...some "GPS locked" clocks actually vary quite a bit over the short term.. The Leo Bodnar device is pritty good especially for the price.. Alot of Microwave folks use it for the reference clock!
It's Leo Bodnard
@@willthecat3861 Leo Bodnah even
I used to run the old Z3801A GPSDO continuously feeding the bench but with the 48 VDC power supply and the unit itself they actually created a bit of heat in the shack (not a bad thing in the winter time!). Then moved to one of the old cell site Rb oscillators which I only turn on when needed at the bench for accurate measurements. Have thought about the Leo Bodnar now for a while and may revisit it. 73 - Dino KLØS
8:30 "It [the GPSDO] does it with a Phase Lock Loop."
Typically, they're implemented with a *Frequency* Lock Loop.
i.e. Essentially just counting pulses, comparing the counts (may be inherent, with one count vice clock), and then ever-so-subtly tweaking the OCXO control voltage.
So technically, any temporary phase locking is merely a pleasing outcome of the FLL and its relative stability.
But perhaps this one is different ! Cheers. 🙂
Nice demo of XY mode being useful!
I seen a guy walking down the street today wearing nothing but socks. I think he was tweaking some Rubidium. I DO NOT ADVISE based on what my eyes had to endure today.
Super fantastic video
I think the Rb is more accurate. The PLLs in the GPS detector have a limit to how well their ratio of two integers can multiply the GPS signals to equal 1 Hz. The Rb source's 5.8 GHz reference is closer to 10 MHz and can approximate 10 MHz more precisely. But, the definition of 10 MHz is based on Cs instead of Rb.
Its an interesting subject. For a Rb standard, it very much depends if the C-Field adjustment on the Rb is set up correctly.
Excitation of a Rb87 at 6.8347GHz causes the Rb87 atom to move from its lower to upper ground state to allow optical pumping to continue. But that nominal 6.8347GHz can be "trimmed" due to Zeeman splitting by an applied magnetic field. (The C-Field).
A Rb frequency standard is very frequency stable. But not that does not mean the derived frequency is accurate. Don't get me wrong..... It will be VERY close and probably way better than say 1E-9. But a well set up Rb will be better than 1E-11 with its C-Field optimally set.
Even a cheap GPSDO will have a medium term accuracy of better than 1E-10 once warmed up so better than a non optimal Rb standard.
Example.....
On our LF transmitters, we used free running Rb standards. (HP5065A). They were monitored by the UK NPL off-air and every month, they would mail me (by post) how far we had drifted w.r.t UTC. From the data presented, I would trim the C-Field adjustment to bring the Rb into line with the NPL Caesiums. We kept the Rb standards circa 5E-12 using this method.
The Original Rb standards were upgraded to GPS Disciplined Rb standards. There was nothing to adjust. We got monthly reports from NPL that we were circa 3E-14 consistently. The GPS disciplining engine "trimmed" the Rb Science package C-Field.
But your comments on the 1Hz signal hold true. There are GNSS chipsets and there are GNSS chipsets!! The ones used in commercial GPSDOs are optimised for timing applications where the 1PPS output is very phase/jitter stable. The GPSDOs found on eBay for a $100 will not be using a particularly good GNSS chipset. A minefield.
I have two Rb units and a GPS Osc -- I found comparing all three the "breathing" drift is the GPS. Over a long term the GPS should have higher absolute accuracy but short term not so much.
Curious to see the phase noise / jitter off the leo bodnar. If correct it's based on a u-blox m8c which has a 26 mhz internal crystal. since that not a whole number division with 10 mhz there probably will be some jitter. Tried this with a m8t before, which has a 24 mhz crystal. There was noticeable more jitter on 10 mhz compared to 12-8-6 mhz
I would suggest the C-Field adjustment on your Rb is way off. An un-disciplined Rb is circa 5E-11 if it is set up correctly and the Leo Bodnar locked up will be way better than 1E-11 and the drift between the pair of devices was many orders of magnitude more. My locked Datum GPDO in the lab here is running at 3E-12 at the moment.
Awesome numbers !
@@andymouse The GPS Disciplined Rb we used at work were 2 orders of magnitude better. 3E-14 on average.
@@stevengwilliam8096 :)
Rubidium need few hour to make very acurate frequency, the first few minute on is few 1/100 hz offset.
I adjust TCOCXO versu rubidium and after a day, the difference are 0.0002hz!
I use HP 53131 with 5 seconde time base and TCOCXO are in frequenecy generator, output at 1ghz to the frequenecy counter.
my rubidium has a signal that shows PLL lock. It does take a view minutes. I don't think any more time that that is necessary.
from Stanford Research Systems:
The Rubidium Oscillator status LED (either Unlocked or Locked) will be on whenever power is applied to the unit. The status will change from Unlocked to Locked within six minutes after power is applied if the module is started from room temperature. Warm-up and locking will take longer if the module is started from a lower temperature. Locked indicates the unit is warmed up and its crystal oscillator has been locked to the rubidium physics package. The output frequency will be about 200ppm below nominal when power is first applied to a cold module. The frequency will gradually converge on 10MHz as the module warms up. After about 4 minutes (typically) the unit will reach operating temperature and lock the crystal oscillator to the rubidium hyperfine transition. Within seven minutes,the module will be locked to within 0.001ppm of 10MHz.
@@IMSAIGuy
I have old Efratom 100323-003.
Compare to my ocxo up for many days, my rubidium have very very very little offset first minute after lock and is stable after hour later with full body warmup.
This is why my comment, power up few hour before use for optimum performance, but juste after lock, is far more I need in precision in 99.9% of case.
My Rubidium have voltage steering input to make ultrafine adjustments .
Curious to see how much drift it has if you wave a heat-gun by it.
Is that an OCXO (oven-controlled) ? Did it get warm ? How much power did it draw ?
Merely curious. No data on the web for that PN that Google can find.
Theorem 7, "Two things are not the same." Its a truism. Just another data point showing the truth. If they look the same, you are not looking close enough.