Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Keenan Tims]

I had some of these boards made:
http://openbsc.osmocom.org/trac/wiki/osmo-lea6t-gps from the Gerbers
in the git repo. I'm running the cheaper LEA-6S modules for
prototyping but the boards seem to work well and were designed for the
pin compatible LEA-6T you're after.

On 25 May 2016 at 23:30, Keenan Tims  wrote:
> I had some of these boards made:
> http://openbsc.osmocom.org/trac/wiki/osmo-lea6t-gps from the Gerbers
> in the git repo. I'm running the cheaper LEA-6S modules for
> prototyping but the boards seem to work well and were designed for the
> pin compatible LEA-6T you're after.
>
> On 20 May 2016 at 01:27, Angus  wrote:
>>
>> On Wed, 18 May 2016 05:31:45 +, you wrote:
>>
>> >There is what looks like a decent carrier board for Ublox modules on 
>> >OSHPARK.COM's shared project library.  It has a voltage regulator and 
>> >RS-232 interface on it... would be nicer if it had a prototype area and 
>> >swoopty PPS driver,  but I'm too lazy to lay out a better one.  Three 
>> >boards will cost you $30.
>> >https://oshpark.com/shared_projects/8QR7ymk8
>>
>> Just in case anyone thinking of it has not noticed - that one is for
>> the smaller NEO version rather than the LEA version.
>>
>> So if anyone knows of a LEA version, or has one that they want to
>> upload or whatever... ideally with sawtooth correction  :)
>>
>> Angus.
>>
>>
>> ___
>> 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.
___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Angus]

On Wed, 18 May 2016 05:31:45 +, you wrote:

>There is what looks like a decent carrier board for Ublox modules on 
>OSHPARK.COM's shared project library.  It has a voltage regulator and RS-232 
>interface on it... would be nicer if it had a prototype area and swoopty PPS 
>driver,  but I'm too lazy to lay out a better one.  Three boards will cost you 
>$30.
>https://oshpark.com/shared_projects/8QR7ymk8

Just in case anyone thinking of it has not noticed - that one is for
the smaller NEO version rather than the LEA version.

So if anyone knows of a LEA version, or has one that they want to
upload or whatever... ideally with sawtooth correction  :)  

Angus.


___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Bert Kehren via time-nuts]

That is why we stay with Ublox 5T for saw tooth correction and 6M for what  
I call PLL applications where you can change the 1 pps to higher frequency 
to be  used in a PLL.
Maybe we will one of these days see 1 ns on ebay. I am sure NDA's prohibit  
the company from resale of chip
Bert Kehren
 
 
In a message dated 5/19/2016 8:00:59 P.M. Eastern Daylight Time,  
kb...@n1k.org writes:

Hi

Well, the first step is to talk to the outfits about you  weekly usage rate 
of the 
chip sets. In some cases anything under 10K chips  a day simply isn’t 
interesting to 
them. After that you sign their NDA and  they sign your NDA. After both of 
the
lawyer teams are happy that the right  signatures are in the right places, 
you 
get access to the basic info. From  there you go back and forth a bit about 
what
can or can’t be done.  

Next step is to get a spec on an actual chip set and the design data  for 
it. You then
are off to source this and that odd part needed to  complete the design. 
After that you
do a board layout. Almost inevitably  that’s an HDI board. Spend $5K or 
more on the first
panel of boards and  wait 30 to 60 days to get them (in one case it was 90 
days, they  goofed
twice in a row). 

Boards come in. Send them off to pick and  place for assembly.  Then you 
find out what 
you did or didn’t miss in  the spec and if your choice of odd parts to 
complete the design 
actually  work or not. Figure on a re-spin of the board as a real 
possibility.  

Assuming it passes a basic smoke test, off to see how it does as a  timing 
receiver. Maybe
it works. Maybe there are bugs. Maybe the bugs can  be fixed …. maybe they 
can’t. Likely
that part is a few months at least  before all the emails are sorted out 
and there is a decision. 

Off to  the next vendor and the same process on their chip set … then the 
next one  ….
then the next one …. then the next one.  Same process every time.  Same NDA’
s same
gag order as a result. Ultimately you can get a part that  will give you 
about +/- 1 ns 
without sawtooth correction and who knows how  much better with sawtooth. 
Since the 
sawtooth is part of the NDA stuff, I  can’t even tell you who that is… 
Here I’m simply
talking about resolution  and the range of the un-corrected pulse. 

Bob



> On  May 19, 2016, at 4:21 PM, David  wrote:
>  
> On Thu, 19 May 2016 15:30:05 -0400, you wrote:
> 
>>  ...
>> 
>>> You can see this phenomena with the LEA  modules quite clearly: 
>>> The LEA-4, -5, -6, -7 and -8 modules  all use an internal 48MHz clock. 
>>> Even though there were 2  complete overhauls of the system in this ~13 
year
>>>  timespan.
>> 
>> Which is why a lot of outfits have  abandoned uBlox and moved on to other
>> outfits that didn’t stall  out.
>> 
>> Bob
> 
> Could you give some  examples?  I have been reviewing the uBlox and
> other modules for  my own project but maybe there are better options
> that I have  missed.
> ___
>  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.

___
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.
___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Bob Camp]

Hi

Well, the first step is to talk to the outfits about you weekly usage rate of 
the 
chip sets. In some cases anything under 10K chips a day simply isn’t 
interesting to 
them. After that you sign their NDA and they sign your NDA. After both of the
lawyer teams are happy that the right signatures are in the right places, you 
get access to the basic info. From there you go back and forth a bit about what
can or can’t be done. 

Next step is to get a spec on an actual chip set and the design data for it. 
You then
are off to source this and that odd part needed to complete the design. After 
that you
do a board layout. Almost inevitably that’s an HDI board. Spend $5K or more on 
the first
panel of boards and wait 30 to 60 days to get them (in one case it was 90 days, 
they goofed
twice in a row). 

Boards come in. Send them off to pick and place for assembly.  Then you find 
out what 
you did or didn’t miss in the spec and if your choice of odd parts to complete 
the design 
actually work or not. Figure on a re-spin of the board as a real possibility. 

Assuming it passes a basic smoke test, off to see how it does as a timing 
receiver. Maybe
it works. Maybe there are bugs. Maybe the bugs can be fixed …. maybe they 
can’t. Likely
that part is a few months at least before all the emails are sorted out and 
there is a decision. 

Off to the next vendor and the same process on their chip set … then the next 
one ….
then the next one …. then the next one.  Same process every time. Same NDA’s 
same
gag order as a result. Ultimately you can get a part that will give you about 
+/- 1 ns 
without sawtooth correction and who knows how much better with sawtooth. Since 
the 
sawtooth is part of the NDA stuff, I can’t even tell you who that is… Here I’m 
simply
talking about resolution and the range of the un-corrected pulse. 

Bob



> On May 19, 2016, at 4:21 PM, David  wrote:
> 
> On Thu, 19 May 2016 15:30:05 -0400, you wrote:
> 
>> ...
>> 
>>> You can see this phenomena with the LEA modules quite clearly: 
>>> The LEA-4, -5, -6, -7 and -8 modules all use an internal 48MHz clock. 
>>> Even though there were 2 complete overhauls of the system in this ~13 year
>>> timespan.
>> 
>> Which is why a lot of outfits have abandoned uBlox and moved on to other
>> outfits that didn’t stall out.
>> 
>> Bob
> 
> Could you give some examples?  I have been reviewing the uBlox and
> other modules for my own project but maybe there are better options
> that I have missed.
> ___
> 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.

___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[David]

On Thu, 19 May 2016 15:30:05 -0400, you wrote:

>...
>
>> You can see this phenomena with the LEA modules quite clearly: 
>> The LEA-4, -5, -6, -7 and -8 modules all use an internal 48MHz clock. 
>> Even though there were 2 complete overhauls of the system in this ~13 year
>> timespan.
>
>Which is why a lot of outfits have abandoned uBlox and moved on to other
>outfits that didn’t stall out.
>
>Bob

Could you give some examples?  I have been reviewing the uBlox and
other modules for my own project but maybe there are better options
that I have missed.
___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[David]

On Thu, 19 May 2016 19:17:17 +0200, you wrote:

>...
>
>The node size of microcontrollers is also kind of stuck, mostly due
>to power consumption constraints. If you want a transistor to switch
>off well and not just becomming a high valued resistor, you have to make
>it big. Hence most microcontrollers have a process node size between 130nm
>and 250nm. Few use 90nm (i'm only aware of the high end STM32 uC). Low power
>microcontrollers can even use larger node sizes (350nm and larger).
>
>...
>
>Attila Kinali

I thought this was do to the need to support NOR Flash or EEPROM,
mixed signal logic like converters, and relatively high interface
voltages.
___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Bob Camp]

Hi

> On May 19, 2016, at 1:17 PM, Attila Kinali  wrote:
> 
> On Thu, 19 May 2016 09:08:40 -0400
> Bert Kehren via time-nuts  wrote:
> 
>> Today GPS receivers are getting better by the year due to the fact that in  
>> order to save silicon the chips are getting smaller and the clock frequency 
>> goes  up reducing saw tooth excursion. If it was not for hanging bridge 
>> filtering it  on the input of a GPSDO would be simple without info from the 
>> receiver
> 
> This assumptions do not hold true. GPS timing receivers are almost always
> just standard receivers with special firmware and slightly modified hardware.
> The normal GPS receivers are optimized for low power consumption as they
> are integrated into mobile devices. As such, the clock frequency with which
> the baseband processor runs will hardly change, as the frequency is the
> biggest knob with which the power consumption can be tuned. 
> 
> The node size of microcontrollers is also kind of stuck, mostly due
> to power consumption constraints. If you want a transistor to switch
> off well and not just becomming a high valued resistor, you have to make
> it big. Hence most microcontrollers have a process node size between 130nm
> and 250nm. Few use 90nm (i'm only aware of the high end STM32 uC). Low power
> microcontrollers can even use larger node sizes (350nm and larger).
> 
> You can see this phenomena with the LEA modules quite clearly: 
> The LEA-4, -5, -6, -7 and -8 modules all use an internal 48MHz clock. 
> Even though there were 2 complete overhauls of the system in this ~13 year
> timespan.

Which is why a lot of outfits have abandoned uBlox and moved on to other
outfits that didn’t stall out.

Bob

> 
>   Attila Kinali
> -- 
> It is upon moral qualities that a society is ultimately founded. All 
> the prosperity and technological sophistication in the world is of no 
> use without that foundation.
> -- Miss Matheson, The Diamond Age, Neil Stephenson
> ___
> 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.

___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Attila Kinali]

On Thu, 19 May 2016 09:08:40 -0400
Bert Kehren via time-nuts  wrote:

> Today GPS receivers are getting better by the year due to the fact that in  
> order to save silicon the chips are getting smaller and the clock frequency 
> goes  up reducing saw tooth excursion. If it was not for hanging bridge 
> filtering it  on the input of a GPSDO would be simple without info from the 
> receiver

This assumptions do not hold true. GPS timing receivers are almost always
just standard receivers with special firmware and slightly modified hardware.
The normal GPS receivers are optimized for low power consumption as they
are integrated into mobile devices. As such, the clock frequency with which
the baseband processor runs will hardly change, as the frequency is the
biggest knob with which the power consumption can be tuned. 

The node size of microcontrollers is also kind of stuck, mostly due
to power consumption constraints. If you want a transistor to switch
off well and not just becomming a high valued resistor, you have to make
it big. Hence most microcontrollers have a process node size between 130nm
and 250nm. Few use 90nm (i'm only aware of the high end STM32 uC). Low power
microcontrollers can even use larger node sizes (350nm and larger).

You can see this phenomena with the LEA modules quite clearly: 
The LEA-4, -5, -6, -7 and -8 modules all use an internal 48MHz clock. 
Even though there were 2 complete overhauls of the system in this ~13 year
timespan.

Attila Kinali
-- 
It is upon moral qualities that a society is ultimately founded. All 
the prosperity and technological sophistication in the world is of no 
use without that foundation.
 -- Miss Matheson, The Diamond Age, Neil Stephenson
___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Bert Kehren via time-nuts]

Today GPS receivers are getting better by the year due to the fact that in  
order to save silicon the chips are getting smaller and the clock frequency 
goes  up reducing saw tooth excursion. If it was not for hanging bridge 
filtering it  on the input of a GPSDO would be simple without info from the 
receiver. Back to  the Motorola days the receiver was a larger contributor to 
the timing error now  it is external to the receiver but removing receiver 
error, depending on the  application does make sense. We have done tests of 
different units using a  Cesium and a HP5372A and the Tbolt stand out with 100 
psec +- 500.  That is  short term 278 samples speaking only for the unit 
performance, long term other  factors external to the unit degrades time by a 
factor 100.
Back to ublox if you use a T the question is why. If it is GPSDO it can use 
 the saw tooth data in the software I am sure commercial units do it. 
Richard MCC  was working on his GPSDO incorporating that info. For pure time 
application if  you want hardware correction a programmable timing element 
makes 
sense specially  since the cost has come down. DS1124 250 psec. does the 
job and less than $ 5  works for us. Micrel  at 10 psec will take two, again 
does it make sense?  Since we are frequency nuts not time nuts DS1124 is our 
choice. Chips with  larger steps are obsolete and would be significantly 
more expensive since it  takes more esilicon. 
In our GPSDO performance tests we found no difference between using Tbolt  
and M6.It does incorporate an adjustable GPS filter on the input.
Bert Kehren
 
 
In a message dated 5/18/2016 8:00:34 P.M. Eastern Daylight Time,  
csteinm...@yandex.com writes:

Bert  wrote:

> Our first tests are with a 500 ps
> DS  chip  because I did have them, but  200 ps will work with a T5 or 6. 
We
>  are  also considering using two Micrel SY 89295UTG in series with 10  ps
> resolution,  but with the limitation of a single frequency GPS  receiver 
and
> ionosphere delay  variations one has to ask does it  make sense?

Think first about what the GPS engine has to work with --  nothing but 
the timing generated by its current GPS timing solution.   It seems very 
doubtful that sub-nS accuracy is possible.  This  provisional conclusion 
is supported if we look at commercial GPSDOs using  single-frequency GPS 
engines and sawtooth correction (or using local  oscillators that divide 
evenly by 100nS, like the Tbolt).  The best  commercial units seem to 
place the PPS within 5nS or so on a routine  basis.

So, even 500pS appears to be considerably better than necessary,  given 
the limitations imposed by the timing engine, atmospheric  dispersion, 
and the GPS system.

Best  regards,

Charles


___
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.

___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Charles Steinmetz]

Bert wrote:


Our first tests are with a 500 ps
DS  chip because I did have them, but  200 ps will work with a T5 or 6. We
are  also considering using two Micrel SY 89295UTG in series with 10 ps
resolution,  but with the limitation of a single frequency GPS receiver and
ionosphere delay  variations one has to ask does it make sense?


Think first about what the GPS engine has to work with -- nothing but 
the timing generated by its current GPS timing solution.  It seems very 
doubtful that sub-nS accuracy is possible.  This provisional conclusion 
is supported if we look at commercial GPSDOs using single-frequency GPS 
engines and sawtooth correction (or using local oscillators that divide 
evenly by 100nS, like the Tbolt).  The best commercial units seem to 
place the PPS within 5nS or so on a routine basis.


So, even 500pS appears to be considerably better than necessary, given 
the limitations imposed by the timing engine, atmospheric dispersion, 
and the GPS system.


Best regards,

Charles


___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Bob Camp]

Hi

While it may not be the only parameter worth considering: 

The standard deviation of the delta between adjacent pps outputs goes down to 
below 500 ps on the LEA-6T when sawtooth is applied. 

Bob


> On May 18, 2016, at 7:23 PM, Charles Steinmetz  wrote:
> 
> Bert wrote:
> 
>> Our first tests are with a 500 ps
>> DS  chip because I did have them, but  200 ps will work with a T5 or 6. We
>> are  also considering using two Micrel SY 89295UTG in series with 10 ps
>> resolution,  but with the limitation of a single frequency GPS receiver and
>> ionosphere delay  variations one has to ask does it make sense?
> 
> Think first about what the GPS engine has to work with -- nothing but the 
> timing generated by its current GPS timing solution.  It seems very doubtful 
> that sub-nS accuracy is possible.  This provisional conclusion is supported 
> if we look at commercial GPSDOs using single-frequency GPS engines and 
> sawtooth correction (or using local oscillators that divide evenly by 100nS, 
> like the Tbolt).  The best commercial units seem to place the PPS within 5nS 
> or so on a routine basis.
> 
> So, even 500pS appears to be considerably better than necessary, given the 
> limitations imposed by the timing engine, atmospheric dispersion, and the GPS 
> system.
> 
> Best regards,
> 
> Charles
> 
> 
> ___
> 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.

___
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.

Re: [time-nuts] Last Call Group Buy Ublox LEA-6T

[Bert Kehren via time-nuts]

Attached is a picture of our ublox board that we plan on making available  
to time-nuts once we have the FE 5680/50 GPSDO under control so it does not 
wipe  out under any circumstances the memory. We will if any one is 
interested in a  board also make the board only available for less than $ 10 
shipping and  handling included.
With out saw tooth error correction I see no reason to get a T version. We  
are also debating what delay chip to use. Our first tests are with a 500 ps 
DS  chip because I did have them, but  200 ps will work with a T5 or 6. We 
are  also considering using two Micrel SY 89295UTG in series with 10 ps 
resolution,  but with the limitation of a single frequency GPS receiver and 
ionosphere delay  variations one has to ask does it make sense? Any thoughts on 
that  subject?
 
 
In a message dated 5/18/2016 2:00:35 A.M. Eastern Daylight Time,  
hol...@hotmail.com writes:

https://oshpark.com/shared_projects/8QR7ymk8

___
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.