[time-nuts] Thunderbolt controllers

[Arnold Tibus]

Sorry, I don't like to contradict,
but I have different experiences made concerning 
some statements:

On Sun, 6 Jul 2008 20:48:27 +0000, Mark Sims wrote:


>The Thunderbolt default config is to not save the survey position.  Unless you use some software to save the position,  every time you power it up it will do a new survey.  This takes from 1 hour to several days  to >
complete. [...] 


How that?
My Thunderbolt does no self surveying after switch on 
and does recover in a few minutes, here my report.

If I disconnect the antenna and reconnect a few min. later 
or I disconnect the pwr fully from my Thunderbolt device, 
let it cool down for about 15 min,
after reconnecting Thundertbolt restarts the internal 
recovery procedure autonomously with my saved position 
in the EEPROM.

the major steps: 
mode: power up, activity: inactive for about 60 seconds
date and time show up, then, 
mode: power up, activity: osc warm up 
terminating after 2 min 15 sec
then does start the recovery loop function
filter init performing after around 3 min 15 sec.
recovery loop and filter init ready after 3 min 20 sec. 
Thunderbolt does work then
fully locked and is disciplining again, working nominal, 
just the almanac alarm is still on for some minutes more.

mode (0) normal
activity (0) phase locking
self survey process 0% 
it was at no time executed

I am running rcvr mode (7) overdet. clock (time)
GPS status (0) doing fixes
After I had once determined my exact antenna position 
several times with the self survey function (try a value 
of more than 2000, perhaps 5000 fixes) 
I stored this value in 'setup', 'position' as 'accurate position' 
and did save by klicking 'save segment'.
I don't know if it is of importance in this context, 
I did set the self surveying parameters 
'self survey enable' ticked  to 'enable' and 
'save position flag' to 'don't save'.
In this window you can modify the survey length of 
2000 fixes (default value).

A few minutes later the oscillator is working at about the full 
precision...


> After that you really don't need a controller except for peace of mind that it is working.   Mayby once or twice a week I have noticed mine going into holdover mode because of really crappy satellite geometry and 
signal levels.  Without a controller giving a realtime display you would never know your oscillator is undisciplined. 



I do contemplate with a small control panel, 
because it is time to think about, that consuming 
just 100 W during the year does result in 2.4 kWh 
times 365 equal 876 kWh (not far from 1 MWh!).
A few watts shall suffice.
And I do not want to take always the Laptop with 
we (I would need severals). The Laptop or PC shall 
only be necessary for more detailed setup and control.
Such a control panel should me enable to read 
date and time, the actual coordinates, the DAC-Voltage, 
the most important warning flags, the correct sat reception,
the pps tick and if the oscillator is correctly disciplined.
The display may be a 4 line LCD, swapped pages if necessary,
together with a few LEDs (blinking or not).

As a wish I would like to have the discipline function, fast 
recovery, man. holdover and restart self survey switchable.



>A very minimal controller might be an AVR Butterfly.  It only has a 6 character display and joyswitch.  Rather not up to the task,  but dirt cheap (around 20 bucks).  It could display a minimal go/nogo type of 
indication.

>Best is a dedicated cheap old laptop.  You can get them for 50 bucks or so with a 1024x768 full color screen, keyboard, and real serial port.  I have a nice controller program in the works that displays full unit and 
satellite status,  graphs the DAC voltage, temperature, OSC error,  PPS error, and Allan variances.  Allows you to control and configure the unit.   Disadvantage is size.  The Thunderbolt and power supply do fit nicely 
under a laptop and add an inch or so the the height.

>Next best is a dedicated controller based upon the MegaDonkey microcontroller or similar device.  It has a 160x80 monochrome LCD  with a touch screen and two real serial ports (see mega-donkey.com).   
Advantage is it allows the Thunderbolt/power supply/controller to be built into a single enclosure to make a stand-alone freq reference with enough display and user input resources to be quite useable.  
Disadvantage compared to a cheap laptop is cost and the small display.  You would have to page though several display screens to get a full take on the unit status.  Also the graphs would not be nearly as nice and 
there is not enough on-chip RAM to do ADEVs, etc.  Once I get the laptop program done I'll probably do a controller on the MegaDonkey.



>> I find it a good idea to design some (small?) h/w to command the
>> Thunderbolt without the need of a PC. 

>>>Why?  Is anything needed?

>>>What happens if you just apply power with no PC?  I'd expect it would power 
>>>up and self-survey and after a while, make a good clock.  I'm pretty sure 
>>>mine was working correctly before I got the software working.



As I reported already, Thunderbold seem to restart 
perfectly alone once set properly, until I get a problem....
I have no real control over it, no information about, 
and for portable use I need always a PC.

Did I forget or misunderstand something (I am sure about)?
I still do not understand all the details for best setup 
to optimize the stability...

Regards
Arnold T.