Hi Bert,
I don't think we have any fundamental disagreement here. Maybe just a
difference in emphasis.
On 8/21/2014 4:34 AM, ewkeh...@aol.com wrote:
Sorry but I disagree. Having done extensive work with the M7 and M6 in
connection with the with GPSDO work we are doing we have characterized the
units extensively. First from what we can see the difference between a SSR-6T
and a $ 16 M6 is that one has a TCXO and outputs sawtooth correction data but
uncorrected both are the same.
I agree that uncorrected results will be basically the same. Note the
specs I stated for the navigation receiver (30ns rms, 60ns max) vs. the
timing receiver (15ns rms, <60ns max). Not a huge difference. But,
timing receivers do have features that can improve performance over
navigation receivers. Some that come to mind are position hold mode,
TRAIM, maintaining performance with only one satellite locked, sawtooth
correction, and precision survey. I haven't dug through the Ublox data
sheets to see which ones they support. Some are only useful during
degraded conditions so it might be difficult to test their
effectiveness. Some are automatic while some, like sawtooth correction,
require extra work to take advantage of.
You mentioned the TCXO in the timing receiver. Ublox says that it helps
with acquisition and maintenance of satellite lock, but I don't think it
has any significant effect on the quality of the 1 PPS or the variable
frequency which will still be limited by the timing resolution (i.e.
period) of the TCXO. Is that right?
Last year we did extensive work on Brooks GPSDO and it works well with
uncorrected M12's and ublox $ 16 M6's. With a Morion we got 1 E-12.
With a geometry shrink in the M7 silicon higher frequency is possible and
also lower power. Ublox most likely wants lower power and higher
performance but not necessarily lower sawtooth because those OEM's that need
it will get a version with sawtooth data. Basic engine is still the same. Time
nuts are not a big enough market. Sawtooth is smaller compared to the M6 doe
to the higher clock frequency and it is safe to assume that when they come out
with a M8 it will even be less.
I won't be surprised to find that sawtooth correction becomes irrelevent
due to higher clock speeds which results in small sawtooth size. The
Navsync CW-12 has been around for some years now. It runs at 'up to 120
MHz', whatever that means. I've measured its' 1 PPS jitter as 4 or 5 ns
(1 sigma) and about 20 to 25 ns max. It doesn't support sawtooth
correction, but it hardly needs to. I tied it to an HP Z3817A GPSDO.
That's not a typo. It's a strange beast that requires an external 1 PPS
input. It includes an E1938 oscillator. The result was a 1 PPS jitter
of < 100 ps (1 sigma) and < 1 ns max. That's better than my Z3801A or
Tbolt. It might be capable of even better performance. There's a
possibility that the E1938 oscillator is noisier than it should be. I
should repeat that test with other GPS receivers to see if the output
degrades due to the higher jitter on the older receivers. Another
project to add to the list!
Higher clock speeds will also allow more processing. I don't know if
that will allow improved performance or if the receivers have already
done everything that they can.
On the universal controller we have a GPS filter not correction on the
input that does improve performance.
I took a page out of Ulrich's work when I saw a picture of his GPSDO where
he thermally isolated his M12. With the FE 5680 work I made the M12 part of
the Rb by mounting it with metal stand offs to the backplate of the Rb.It
in turn is temperature controlled.
In the case of my FE 405B work I actually placed the M6 inside the OCXO
took the battery off. I think I have a picture if interested.
Yes, if you look in the manual for Jackson Lab's GPSTCXO it says that
shielding the board from drafts improves performance. You've taken that
to the next level.
Not knowing that it can not be done I did what I call a GPS-PLL using a M7.
Attached is the board layout on the right side is what we are presently
using with the Morion, on the left is a version for 5 V OCXO's so Hams can
use 12 Volt. The one on the right is driven by readily available parts for
any Ham and no adjustments. Total cost not counting GPS and OCXO below $ 10.
We are still fine tuning the filter but right out of the box we got 1
E-10. This is for Ham's not time nut standard. Data exceeds attachment
limitations but any one can contact me off list and I will send it. We
destroyed the M7 have not figured out how but a new one is on order and once
testing is completed schematics will also be available.
Every project requires a sacrifice to ensure success. In my case, it's
usually a blood sacrifice caused by stabbing or goring myself with some
tool. :(
Ed
I have the bad habit layouts
first documentation maybe second. Frustrating for the team, but I am
getting better. As I said before mainly for Ham's and one of our Australian
team
member will roll it out to the Ham community. But any body is free to use
it I just think time nuts can do better.
Bert Kehren
In a message dated 8/21/2014 1:30:50 A.M. Eastern Daylight Time,
ed_pal...@sasktel.net writes:
Thanks, Tony. That's good info.
So now we've confirmed that the neo-7M has an NCO and it appears that
it's resolution is 20 ns. The data sheet shows the 'Accuracy of time
pulse signal' is 30 ns RMS and 60 ns for 99%, but it isn't clear whether
they're referring to jitter or error with respect to GPS seconds.
The original question was whether the neo-7M would make a good GPSDO.
As we've seen, the answer is no. Cheap, yes. Good, no. Setting aside
the NCO issue, the neo-7M isn't a timing receiver, it's a navigation
receiver. That limits it's performance in many ways.
Ublox sells timing receivers, but they're still NCO-based. They're also
significantly more expensive than the navigation receivers. One example
is Synergy Systems' SSR-6Tr if it's still available. It was announced,
and discussed on this list, in 2012 but it still isn't listed on their
web site so I don't know what it's status is. It's based on the LEA-6T
timing receiver which has a spec for the 1 PPS is 'within 15 ns to
GPS/UTC (1 sigma)'. That can be further reduced with some extra work.
If the performance of an NCO-based unit isn't enough, you might want to
consider Jackson Labs GPSTCXO which is a real GPSDO. More expensive
than the NCO-based units, but you get what you pay for.
No, I'm not associated with Synergy or Jackson labs.
So Graham, if you survived the firestorm started by your simple
question, are you any wiser?
Ed
On 8/20/2014 7:56 PM, Tony wrote:
On 19/08/2014 16:11, Ed Palmer wrote:
Does anyone have a neo-7M and an HP 5371A or a 5372A Analyzer? Use
the Histogram Time Interval function to measure a block of samples.
That will show the length of the samples with a resolution of 200
ps. That's what I did a couple of years ago when I analyzed the
Navsync CW-12 with the old and new firmware.
FWIW, I just had a look at the timepulse on a NEO-7M. I configured it
to 10MHz, 50:50 duty cycle when locked, disabled when out of lock. I
don't have any of those Analyzers so I used an HP 54615B digital
scope. The period of the majority of cycles was 104ns with 'random'
cycles being 84ns. I did not observe any other cycle periods. I don't
know how accurate the time measurements are on the scope, but it looks
like the timing is derived from an approx 48MHz clock, and the timing
phase/frequency adjusted by periodically deleting 48MHz clock cycles.
Although I said random, I couldn't make any observations as to the
statistics of the short and long cycles or their distribution - I
guess I'll have to write some software for my STM32F4 discovery board
for that.
If I get time, I'll do the same with a Reyax RYN25AI receiver which
has a UBLOX MAX-7C module.
Tony
_______________________________________________
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
