[LinuxPPS] Oncore Refclock interface & thermostat

[Hal V. Engel]

On Wednesday 07 January 2009 06:56:02 Udo van den Heuvel wrote:
> Hal V. Engel wrote:
> > Oncore modules like the UT+ units I am using have a power consumption of
> > 155mA and can handle antenna loads of up to 80mA (most other Oncore
> > models are lower).  So a typical unit with the antenna load will consume
> > about 175mA to 190mA.   With a higher power consumption antenna amp or if
> > using additional in line amps this could be as high as 235mA (for the
> > Oncore UT+) without exceeding what the module's antenna power circuit can
> > handle.  These numbers are significantly higher than the 60mA power
> > consumption of the Garmin 18LVC but this is in most cases less than 1
> > watt.
>
> Could I power them from the 5V / 12V lines of the PSU in my Epia?
> (i.e. mount the GPS module there, connect it to an internal serial port
> and make a coaxial antenna input in the case of the Epia.
>
> [...]
>
> Ah:
> > If you can mount the GPS module in the computer chassis then there is
> > lots of 5V power available right from one of the internal PSU connectors
> > and the limits of the USB interface become a non-issue.  The only concern
> > I would have is that the inside of a PC chassis is a high RF noise
> > environment and some GPS modules may not handle this very well.  In
> > addition, there are processors that run at frequencies that are close to
> > the L1 GPS frequency of 1.575GHz that could create significant problems. 
> > One example is the Intel Atom operates at at a nominal speed of 1.6GHz
> > and under these conditions you might need additional shielding around the
> > GPS module.
>
> How to shield? (maybe not so on-topic here...)
> But maybe it is if it is gps-related.

Again this may not be needed but it is something that you need to be aware of 
in case it is an issue with your setup.

If you do have an issue with this the shield can be almost anything that will 
block the RF frequencies of interest.  A metal enclosure or a metal mesh 
enclosure that is grounded to the computer chassis (IE. a Faraday cage for the 
GPS module) for example.  L1 has a wave length of 19cm so all you need is to 
have any openings in the enclosure be significantly smaller than that size 
(say less than 1/4 of a wave length which is a little under 5cm but the 
smaller the better) and it will block most of the spurious RF.   

When the GPS is mounted outside of the computer chassis the computers 
enclosure forms a Faraday cage that contains most of the RF generated by the 
computer inside the case (actually it shunts it to ground).  In addition as 
the GPS module is moved away from the computer the strength of the spurious RF 
that does make it outside of the case and to the GPS module is reduced in 
proportion to the square of the relative distance.  So a GPS mounted outside 
of the computer case could easily have the spurious RF reduced by 20db or 30db 
or more compared to the levels inside the computer case.

>
> > more shielding and higher levels of selectivity than other Oncore models.
> >  The ones you see for sale on ebay and other places are for the most part
> > working pulls from cell sites that are being replaced with newer units
> > like the M12T and M12M Timing Oncores (higher precision and lower power
> > consumption).
>
> I read about (older) oncores that they have a bug in the binary
> protocol. (somewhere in the ntp source?)
> It is some kind of limit in the time that could be represented in the
> chosen procotol. (not nmea)

I don't know about this and I have read about everything I could find on the 
Oncores.  The protocol used by the driver is the Oncore binary protocol since 
none of the timing specific models other than the VP support NMEA.  Reg might 
know what the bug is or if it exists or what Oncore models have the issue.

Also what does older mean?   There have been 5 generations of Oncores.  The 
UT+ models are third generation devices.  The first were the 6 channel models 
which date from before 1995.  Then came the 8 channel models (VP, GT, UT...) 
from 1995 to 1998 and then the second generation 8 channel models like the UT+ 
and GT+ which were in production until 2004 (I think - I have one that dates 
from 2003).  So which of these are "older" Oncores?  

The first generation 6 channel models are very rare now and the only ones I 
have seen for sale have been 6 channel VPs.  VPs in general are coveted, even 
the 6 channel models, and some consider these the best Oncores ever produced.    
These typically fetch higher prices that other models. You will see some GT 
models on ebay and an occasional UT.     But most of those you will see are 
third generation units.  I have yet to see an M12 or M12M in the secondary 
market.  

It is also common for these to be misidentified by the vendor (they will 
advertise a GT as a GT+ for example and the vendor I just got my last batch of 
units from thought that his UT+ models were UTs until I corrected him) so you 
need to make sure that you know what you are buying.  This either means using 
a vendor you can trust to give correct info and/or knowing how to parse Oncore 
model numbers.  For timing stuff you should stick to UT+ and VP models unless 
you can get your hands on an M12T or M12M Timing unit.  The M12x models are 
3.3 volt modules which makes them a little harder to interface to since you 
can not just use a simple 5V power feed from the computer.

>
> > One of them arrived DOA but was promptly replaced by the vendor.  At the
> > price he is selling them and with the free shipping on additional units
> > it is worth while to get more than one unit IMO (IE. spares and for
> > additional projects). These are a mix of firmware versions and one of the
> > units I got was version 2.2 and all of the others were version 3.2.  My
> > understanding is that he also has some that are version 3.1.
>
> How much does the firmware matter?

It appears not too much unless you have an exceptionally stable oscillator in 
your time server or the firmware version was a buggy one.   By the time 
Motorola got to the third generation units like the UT+ the firmware was 
fairly mature.  Versions of the firmware that had significant bugs tend be 
rare since fixes where made available very quickly and proactively and most of 
these buggy units were reflashed.   The third generation + models started out 
with version 2.x firmware.  Any UT or GT with version 1.x firmware is a second 
generation unit.  The last firmware version for VP modules was 10.0 and I am 
not sure which versions are preferred for the VP.     

Version 3.2 was the only version of the UT+ firmware that was released after 
SA was disabled.  Earlier versions (at least 3.1) had a bug that caused about 
a 100ns offset of the PPS pulse every 17 seconds.  The UT+ model specification 
calls for PPS accuracy to be +-50ns WITH SA and based on post SA tests I have 
seen on the net I suspect that without SA this is reduced to +-20ns.  In 
addition the Oncore protocol includes sawtooth correction information and if 
the ntp driver uses the sawtooth correction info the sawtooth corrected 
accuracy is much better.  This problem was masked by the time dither in SA and 
did not become apparent until SA was disabled.  This issue is fixed in 3.2.  
3.1 also had some issues with the 100PPS pulse that were fixed in 3.2.  2.2 
does not have a 100PPS feature.  From what I can tell any issues with specific 
common firmware versions are so small that other factors like interrupt 
latency and the affects of temperature on the computer oscillator are orders 
of magnitude bigger.  UT+ firmware versions 2.2, 3.1 and 3.2 will all work 
nicely for our applications.  These are the common versions for the UT+ 
firmware and you will not find much else out there.

The firmware version is encoded into the last digit of the model number and on 
UT+ models 

2 = version 2.2
3 = version 3.0 (rare)
4 = version 3.1 
5 = version 3.2.  

I don't know what version number a 1 would equate to but if I had to guess I 
would say 2.1 but these are likely to be very rare since most of these would 
have been reflashed to version 2.2.  

Reflashed units will have a sticker indicating that they have been reflashed 
and will normally also have a new model number/serial number sticker that 
reflects the upgraded firmware.  Synergy Systems will also reflash Oncores for 
$25 but even if you get a unit with a buggy version for free it is likely 
cheaper to just buy a unit with a known good firmware version.

Also all UT+ model numbers look like this: 

R5<battery>22U11<connector><firmware version>

where <battery> = 

1 no on-board backup battery 
2 on-board battery 

and <connector> = 

1 right angle OSX/MCX 

5 straight OSX/MCX.  

Most of the UT+ units on ebay are R5122U115x units since this was what was 
commonly used in cell sites.

Hopefully you found this info useful.

Hal