-
Rubidium add on to my GPSDO?
Posted by
FriedMule
on 09 Feb, 2022 00:17
-
I have a relative fine GPSDO that some of you recommended me buying. I would like to get even better accuracy. And understand that a Rubidium, connected to my GPSDO 1pps should be the best, except if I have thousands and thousands of dollars.
My frequency counter is a 12 digit counter, I would love for most of them to be zero's, not jumping between 1-5 and so on. :-)
A person told me to use two GPSDO's and feet both to some sort of Rubidium device, that calculated the correct frequency from these two signals.
What would you recommend? What would you connect to my GPSDO?
https://www.ebay.com/itm/332390729098
-
#1 Reply
Posted by
rfclown
on 09 Feb, 2022 00:54
-
"Rubidium, connected to my GPSDO 1pps" makes no sense. 1pps is an output. Rubidium produces a high frequency (10MHz or whatever) output.
[edit]... I just read another thread that discussed a GPS disciplined Rubidium. I hadn't thought of that. But it doesn't change the rest of what I wrote. If you do that, the standard is the GPSDO....
Advantage of Rubidium is that it just works (unless it's broken). Don't have to wonder if you are seeing satellites. But GPSDOs are taking advantage of many Cesiums in the sky and will give you better long term accuracy. I started with an eBay auction that gave me two Rubidium modules. They were very close in frequency, but which one is more right??? Who knows. A man with one watch always knows what time it is. A man with two watches is never quite sure. More recently a got an eBay GPSDO and tweaked my Rubidiums with it (yes, Rubidiums can be adjusted). I personally don't need this kind of accuracy for anything.
A 12 digit counter. Probably something like the cheap FA-2 I have. Just because something shows 12 digits doesn't mean that all those digits are meaningful. If you connect your GPSDO 10 MHz output to the reference and the input of your counter you should get a perfect reading. My FA-2 has a decent OCXO in it, and I also tweaked that according to my GPSDO. (Feed 10 MHz into counter and adjust counter TCXO).
I don't know what you are trying to do. I don't think you need a Rubidium. Your only reference is your GPSDO, which if it is indicating that it sees lots of satellites and has been on for a while should be fine. Feed that into the reference input of your counter, and then you have an accurate frequency counter.
-
#2 Reply
Posted by
FriedMule
on 09 Feb, 2022 01:34
-
As I understand it, and as you yourself are writing, the GSDO is very stable over longer time, several minutes to hours and even days. The Rubidium should be perfect for the short time readings, frequencies in a minute or so.
The reason for a 1pps GPSDO should be to correct the Rubidium every second, while the Rubidium keeps a more precise timing between than a GPSDO would ever be able to.
Yes I know 12 digits is extreme and rarely usable, but I hope to build some crystal oscillated clocks that shall run independent and fairly precise, so to be able to test them, would it be great to have something that shows exactly what deviation a crystal have, without having to let them run for days or weeks.
-
#3 Reply
Posted by
rfclown
on 09 Feb, 2022 02:08
-
A rubidium gets it short term stability from a crystal oscillator inside. The rubidium disciplines a crystal oscillator, otherwise the rubidium output would have lousy phase noise. When you are wanting higher long term accuracy, you want the disciplining to have LONG time constants.
"...I hope to build some crystal oscillated clocks that shall run independent and fairly precise, so to be able to test them, would it be great to have something that shows exactly what deviation a crystal have, without having to let them run for days or weeks..."
If you have your counter referenced by your GPSDO, you are already there. You can see the of deviation from any kind of crystal oscillator immediately. You're reference will be orders of magnitude better than what you're measuring.
-
#4 Reply
Posted by
FriedMule
on 09 Feb, 2022 03:09
-
Sounds great, so what I can measure is good enough to let the clock run for a year with only a few seconds error?
-
#5 Reply
Posted by
rfclown
on 09 Feb, 2022 03:51
-
Sounds great, so what I can measure is good enough to let the clock run for a year with only a few seconds error?
I don't know what "the clock" is. If you want a precise clock, just read the GPS time. Most any GPS module will spit out a NMEA message every second telling you the time. 30 years later you won't have gained or lost a single second (unless the US government has collapsed and the GPS system isn't working anymore). I'm not a time nut. Hopefully someone else will chime in to help here. I'm don't know what you are trying to do.
-
#6 Reply
Posted by
tkamiya
on 10 Feb, 2022 18:26
-
I've actually tried this.
My opinion is that Rubidium is SO stable, GPSDO is not needed. In fact, with both interfaced, Adev of total system degraded.
What would be useful is, should GPS go offline, RB will keep constant timing, and drift rate will be far less than GPSDO alone. Other than that, it isn't needed.
Keep in mind GPSDO is superior stability is 24 hour average. On minute to minute basis, it does drift somewhat. I can't imagine that will have a meaningful effect on home lab though.
-
#7 Reply
Posted by
testpoint1
on 10 Feb, 2022 21:57
-
which Rubidium oscillator you use?
-
#8 Reply
Posted by
tkamiya
on 10 Feb, 2022 23:28
-
I have all kinds, but the experiment with GPSDO was done with PRS-10.
-
#9 Reply
Posted by
nctnico
on 10 Feb, 2022 23:46
-
The PRS-10 looks like a nice device. The problem with Rubidium oscillators is that their short term stability and phase noise isn't very good. It is better to have an oscillator behind a Rubidium to improve phase noise and short term stability. So in the end having a Rubidium locked to a GPSDO and using the Rubidium output is not a good idea.
Using 2 GPSDOs doesn't make sense to me; they both see the same sattelites so will have the same offset.
-
#10 Reply
Posted by
tkamiya
on 10 Feb, 2022 23:53
-
PRS-10 as well as many commercially made rubidium oscillator modules already have a high grade OCXO inside.
I actually experimented by measuring Adev between PRS-10 alone and with GPSDO. Former won by factor of 10.
-
-
SRS advertises the PRS10 as having "<-130 dBc/Hz phase noise (10 Hz)". I've been using one I got off ebay as a reference for a 10 GHz synthesizer and so far its phase noise seems to be living up to the claim.
-
#12 Reply
Posted by
Zenwizard
on 11 Feb, 2022 17:52
-
PRS-10 as well as many commercially made rubidium oscillator modules already have a high grade OCXO inside.
I actually experimented by measuring Adev between PRS-10 alone and with GPSDO. Former won by factor of 10.
What was the integration time on the PPL for the RB. I have a PRS10 based unit and the response time of the RB PLL can be adjusted for up to 18 Hours.
My point of reference on this is an Agilent Mobil calibration lab that uses a GPSDO in to an SRS 725. They have stated this is good enough for calibration and adjustments for everything that did not require Cesium.
For the SRS units the manual even calls out the adjusting the integration time higher for the RB unit IF the incoming 1pps is jumpy. Standered from the factory is 2.5 ish hours and even SRS says this might be a little short for some GPSDO's.
I was going to pull the trigger on a new 13 digit FC for some Adev experimentation however it seems my vehicles have had other plans.
-
#13 Reply
Posted by
Echo88
on 11 Feb, 2022 18:17
-
Which 13 digit FC do you have in mind and wont a DMTD-device be more suitable than a FC for Adev-measurement Zenwizard?
-
#14 Reply
Posted by
Zenwizard
on 11 Feb, 2022 20:54
-
Which 13 digit FC do you have in mind and wont a DMTD-device be more suitable than a FC for Adev-measurement Zenwizard?
I was looking at a well optioned key sight 53230A. I don't know where I got the 13 digit. I am looking from some high digit count FC's But the 53230A can do direct Adev measurements.
Zen
-
#15 Reply
Posted by
nctnico
on 12 Feb, 2022 00:31
-
Which 13 digit FC do you have in mind and wont a DMTD-device be more suitable than a FC for Adev-measurement Zenwizard?
I was looking at a well optioned key sight 53230A. I don't know where I got the 13 digit. I am looking from some high digit count FC's But the 53230A can do direct Adev measurements.
But not at the resolution a DMTD offers. Besides that, when looking at a frequency counter be sure to take a look at Tektronix FCA3100 aka Pendulum CNT-91 as well. I like it better compared to the Keysight 53230A.
-
#16 Reply
Posted by
Zenwizard
on 12 Feb, 2022 01:13
-
If one was going to look for a DMTD what should the target be Is this a off the shelf or is this a build up solution? I am thinking of crashing the output of my srs725 and looking at frequency stability of some clocks. I do realize i am getting dangerously close to time nutting territory here.
-
#17 Reply
Posted by
martinr33
on 12 Feb, 2022 04:57
-
1) GPS signals have extreme long term stability, but drift quite a lot in the short term. A couple of digits on my 12-digit counter.
2) Rubidium oscillators do need to be tweaked to be on frequency, but are very stable once done.
3) A high-end ovenized oscillator in a counter is probably the best short term source. I have a 53132A unit with an aftermarket oscillator.
You can get the counter characterized by measuring the GPSDO over a few hours. The average of the GPSDO frequency will represent the counter.
Now, you can tune your rubidium to the same average count.
Then you can recalibrate your counter using the rubidium source / GPS as necessary.