[time-nuts] Thunderbolt reception problems

WarrenS warrensjmail-one at yahoo.com
Wed Nov 25 20:00:04 UTC 2009


John

Thanks, that puts things a bit more in perspective.
Sometimes One persons 'A lot' is another persons 'almost Great'.

5 min = 300 sec
As much as 30 ns / 300 sec = 0.1ns /sec = less than  1e-10 freq shift

A lot of NON broken things can cause that kind of  small drift.

With under 1e-10 shifts over 5 minutes, 
see below for some addition comments.

ws

----- Original Message ----- 
From: "John Green" <wpxs472 at gmail.com>
To: <time-nuts at febo.com>
Sent: Wednesday, November 25, 2009 9:47 AM
Subject: Re: [time-nuts] Thunderbolt reception problems


> Warren, Thanks for the Tbolt tips. 
>The power supply was a good linear supply so I doubt that was causing what I see. 
>
ws) Make sure it is not line sensitive, OR Temperature sensitive. 
A few mv change on the +12 V could cause that kind of error. 
Plug the PS into a variac and blow a hair driver at it to make sure that  + -  15 % line change & 15+ deg 
does not effect the Tbolt over the 5 min time span.

**
> The room temp was cycling a degree, maybe two.
>
ws) 1 deg F can cause a 1e-10 freq shift, More important is how fast it is changing.
Keeping the changes down to about 1 deg C per hr AT The Tbolt sensor 
is a conservative overkill that will keep temp from being an issue.

**
> I did sometime see some quick shifts that were not coincident
> with temperature but they were all less than 10 nS. 
>
ws) This Sounds like the typical Tbolt satellite switching problem.
Solution is one or all of the following:
Better signal or antenna view, better Position setting, lower AMU,  Higher TC, higher elevation setting.
The problem is that although these small fast phase jumps do not effect the Peak Phase much, 
They can cause a great deal of short term Freq noise.  

**
> The largest swings always followed temperature excursions. Even after putting it in a box and
> lowering the temp. drift to fractions of a degree, I would see swings if as much as 30 nS in 5 minutes. 
>
ws) When you first put the Tbolt in a new environment (a box) it will take a couple hours for it to adjust to its new environment.
 After That, the important thing is what kind of temp change per time does the Tbolt see.
Unfortunately, due to the two different kind of sensors used in Tbolts, 
(one type with high resolution, one is low resolution)  this may not always be easy to see.
One solution that gives you less phase shift error with 'quick' temp change is to lower the TC, 
The compromise is Less PP pahse error but you get more ADEV noise at 1 and 10 second tau.
With slow changes at 'a fraction of a degree' (a fraction is 1/10 F or less),  NOT likely it is the Tbolt drfting
unless yours does not have a working Oven Osc at all. Check the current draw of the +12 volts with changing temp to see.
More likely it is the Power supply changing IF you are holding the TBOLT constant while the room is changing.
OR ANY of your other test equipment that is outside the BOX, and subject ot the quick temp change.

**
> I thought that this was just too much to fool
> with and decided to use it as a reference for a 900 MHz ham repeater. It
> will be interesting to see how it does in a room with no heat or air
> conditioning. I hope to get on next week. I have another one and may, at
> some time try putting a double oven oscillator to see how it does. I thought
> about trying an LPRO but I think the Tbolt would just make it worse rather
> than better.
> 
ws)  There are easier ways to get the performance of a double oven Osc without the cost and trouble ...


********************************
John Green said:
> mine got a lot worse with TC set to 500 sec.
Depends if  that was Short term Osc Freq noise or long term Phase drift?  
Short answer:  
A TC of 500 can causes a LOT more (x10 +) long term Phase shift error if the temp is changing, 
especially when the Dac_Gain and Damping are not also set correctly.


Here comes those darn Tbolt setting trade offs again.
First point is: 
If you are a battery backed cell site or the average Ham that just wants to know that the  
in house freq reference is within 1e-9 (1ns drift per second) the default Tbolts settings are fine and make an OK plug and play unit.

On the other hand if you are a NUT & would like 10 ns per day (1e-13 freq)  then the factory defaults are not so good (Said).
Another major trade off is if you want the best 1 sec ADEV numbers OR the lowest long term Phase errors. Takes different settings.

Using the  TC and Damping settings, different trade offs or compromises can be made.
BUT until the Dac_Gain is set correctly (which the default setting is NOT), 
their proper setting And interaction is just a shot in the dark.

Dac_gain is the sensitivity of the OXCO EFC input in Hz per Volt.
Basic way to find the correct value is to disable the tracking, 
then output a + and - 0.1 volt Dac difference from its nominal tracking value using Tboltmon S/W,
Average the measured + - HZ freq change of the 10 MHZ,  multiply by 10, and update the Dac_Gain AND SAVE (sign is negative) 
If you don't have a external counter with enough resolution, You can find the Dac_Gain using the Tbolt 'turnover test'.

> Even putting them in a box to shield them from room temperature variations, they seem to wander around a lot. 
My answer depends on what you mean by 'a lot' ,  but some general comments.
It would seem that there are many different OXCO used in Tbolts, from no oven to double oven.
Even with my very small sampling, I've seen units where case temp change does not matter much at all 
to where the temp should be held to under 0.1 deg C change per hr.  

You may have one of the poor Temperature performing units, BUT more likely what is causing it to wonder around a lot is the GPS.
We know the GPS wonders around a lot short term, much more than Temperature can, over short periods of time like 10 seconds.
It is safe to say that there is something 'broken' somewhere, if a TC setting of 10 secs works better than a setting of 100 + secs.
Just a guess on what is broken is "the Data or setup".  
BUT if you are using a common PS on both units, check it, especially the +12 volts, It should be stable to better than 1mv.
Yours may be changing by a volt, to cause the type of errors you are seeing.  
Also be sure the units are not jumping in and out of holdover due to a high AMU setting and low GPS signal level or indoor antenna.
If you can post or send me a plot of what your errors look like, I could make a better guess of what is broken.


For those that do not want to fiddle, or customize their settings, 
but want a better performing Tbolt than the Default settings,
For an average NON cell site unit that is in a nice environment, 
using an outdoor antenna, inside with stable temp, not under the heater outlet, and not being bounced or moved around,
A compromise set of values better than the Defaults are:  

1) Dac_Gain  -3.5 Hz/Volt (default is -5)  BUT best to set it to the correct measured value. (If off, the values below will not be correct)
Range, unknown, 

2) TC of  500 nominal, (default is 100), 
 Range 250 to 750 seconds,  Lower numbers give better results if the temp is changing and not controlled, Higher numbers give better weak signal answers.

3) Damping 0.7 nominal  (Default is 1.2)  
Range 0.5 to 1.5  Lower numbers give less long term phase error, Higher numbers lower short term Osc ADEV noise.

4) AMU  2.5 (default is 4),  
Range 1 to 4, Lower numbers if weak GPS signal.

5) Elevation mask to 15 deg (default is 5) A band-aid compromise setting so that the unit does not lock onto low level signals.
( better to use elevation than AMU because of the behavior of Tbolts control loop)


Also important, be sure the unit is in fixed position and not 3D, position tracking or survey mode, 
AND that it has the antenna's correct location loaded and saved. 

ws

**********************
----- Original Message ----- 
From: "John Green" <wpxs472 at gmail.com>
To: <time-nuts at febo.com>
Sent: Tuesday, November 24, 2009 7:45 AM
Subject: Re: [time-nuts] Thunderbolt reception problems


>I have been following the discussion regarding Tbolt performance closely. I
> have 2 of them and have not seen performance even close to what others
> report. I don't seem to have sensitivity related issues that others have
> reported. I am feeding Tbolt and Z3801 from a common antenna through a
> purpose made GPS splitter. The Tbolt seems to see more sats quicker than the
> Z3801. The Tbolts I have are extremely temperature sensitive. Even putting
> them in a box to shield them from room temperature variations, they seem to
> wander around a lot. This sounds crazy and is not in line with what others
> have reported but when I changed the TC to 500 seconds, mine got a lot
> worse. I found that a TC of around 10 seconds with a damping of 1 made for
> the least amount of phase slip compared to a Z3801.
> 
>





More information about the time-nuts mailing list