[time-nuts] How do I know my GPS stabilized oscillatorisworking?

[Ulrich Bangert]

Hi Said,

first i suggest you download the tool again because a lot of changes
(improvements?) have been made in the last days.

> * It would be good to have the tool normalize the vertical 
>  axix for every 
> column when showing more than one column superimposed. It 
> does  this if you 
> only view a single column, but if you show two or more with  
> different scales 
> (say frequency accuracy and temperature versus time)  the 
> tool doesen't create 
> multiple scales to fit the two data columns inside the  same window.

This is possible, however it is not completely automatic, with some
reason behind it. Open the Axis dialog in the Chart editor. In the right
part of the window there is a list of the predefined axes. Note the "+"
and "-" signs below the list! You can create as many additional custom
axes as you like. Use the "Position" rider and the "Position" value to
shift the position of the custom axis. Please note that the new axis is
not displayed until a data column is related to it. To relate a data
column to a certain axis, choose a column in the series dialog and use
the rider "General". With "Vertical axis" choose whatever axis you may
like for this data column. With only two data columns you do not need to
generate a new axis, just choose the right axis as the vertical axis for
the temperature data. Every axis is normalized automatically unless you
change this at your will.

>  * There are so many options to choose from that sometimes 
> it  is a bit 
> hard to find what one is looking for. For example, if I want  
> to change the 
> vertical scale from scientific notation to scalar notation to 
> see  more than two 
> digits of resolution, I have to go quite deep into the Axis 
> menu.  It would be 
> great if the axis could be clicked, and the scale changed 
> similar to  how Excel 
> does it.

I understand your point very well. Somtimes you really have to dive deep
into some sub-menu or dialog. But for reasons that are too difficult to
explain here, this is something that will surely not change. We will
have to live with the Chart Editor window, whether we like it or not. 

> * Once zoomed-in, it would be nice to have a zoom-out 
> feature.  I figured 
> out how to do it with the right mouse button and selecting 
> from right  to left 
> on the window, but maybe there is a way to just click 
> somewhere to zoom  out 
> again to see all data?  

The next version will have this feature.

> * On the Allan deviation table shown next to the plot, I 
> think  the tool 
> goes up only to show up to 4098 (?) s worth of data, even if 
> data worth  100000 
> seconds has been captured and is shown on the plot.  A couple 
> more  entries 
> in the table would be nice: maybe at time intervalls where a 
> major  vertical 
> decimal is broken (say for example 15500s if at that point 
> the measure  is 
> exactly 1.0E-014, then 102300s if it reaches 1.0E-015 at that 
> point  etc)

a) May it be that the vertical size of the window was too small to
display the data >4096 ???

b) The 1, 2, 4, 8 X Tau0 for the horizontal axis is quite usual and
generates nice looking equidistant points in the plot. Exactly as you
suggest, i guess it cannot be made because i think there is no "reverse"
function for the Allan-Deviation which could be used to compute "When
does it cut the 1.0E-015" barrier. However I agree that it is possible
to compute the function at more and different points. I will have a
thought on that.

> * My HP 5334B counter, when set to A-B time mode will  
> sometimes generate 
> negative time measurements (for example -6ns, -13ns  etc), 
> but then for some 
> reason measure across one second so it jumps to  +999999993ns 
> etc. Don't know 
> why the 5334B firmware does this. The effect  is that the 
> Plotter tool gets 
> jammed by the large difference in the two  scales. I cannot 
> filter the larger 
> values out with the "delete outlier" button.  Could this be 
> added so the tool can 
> recognize this issue and handle it  correctly?

Yep, the 5370 does the same. They want to indicate that the stop slope
has appeared BEFORE the start slope. I have been thinking on handling
something like this automatically but have not solved the problem
completely. Can you send me test data? Files of some MB are ok because
with automatics it is always a problem of performance too.

> * While drift can be removed, could you also add a feature 
> that shows the 
> least-mean-square fit (average line or something similar) through  a 
> drifting, noisy plot?

The current version has this feature, not only linear but also second
and third degree polynominal.

> * Could you please add a help button to explain the many features  
> (buttons) of the tool?

It is planned to improve the documentation and display it by means of a
"Help" button. But when it comes to explaining everything that can be
done with the chart editor: No, i don't have the time to do so.

> * Is there a way to select colors for columns? One of  the 
> traces I have is 
> always white on the plot, and it's very hard to see on  the 
> screen, as well 
> as almost impossible to printout onto paper.

Open the chart editor and choose the series dialog. In the right part of
the window you see the available series names and left from them the
checkboxes for switching the display of the series off and on. Between
the names and the checkbox you see small horizontal lines in the color
of the series, which (with close to white colors) MAY be difficult to
see. Double click on the line to open a standard Windows color dialog
for the series.

> * An "oscilloscope" type marker tool would be great, where 
> you  drag a 
> vertical line right and left with the mouse across the plot, 
> and you have  a 
> readout of the vertical value of the selected column data at 
> that exact  point.

In the "time stability measures" window, when you come near a data point
with the cursor, the point's data is displayed in a small window. I can
arrange the same for the main graphics window. When you think about
interpolating between points: Yes, this is possible but will need more
programming work. Will think about it!

> * It seems that the Allan Deviation etc. is always 
> calculated  over the 
> entire data set (is that true?), even if you zoom into a 
> small portion  of the 
> dataset. It would be great to only do the stability math on 
> the data  that's 
> actually visible on the screen (or even better: choose from 
> the two  options: 
> math over full dataset, or selected subset).

a) Yes, it is true!

b) Ok, you can get the result even now by deleting data before and
behind the region of interest but surely your suggestion is a
improvement over that. Again let me think about it.  

> * It would be nice if the tool remembers the last dataset file (and
maybe 
> even auto-loads it if selected), and the last settings preferences.
These 
> get lost now.

Until now i tried to get over without "remembering" anything in a
ini-file or in the registry. Perhaps i start now to do so.
 
> * A pointer to your website would be great somewhere in the  tool to
get 
> updates (maybe under a 'Help/About' button?). I had a very hard time
to find 
> your website over Google... 

That is surely a good idea!

Reagrds

Ulrich Bangert

> -----Ursprüngliche Nachricht-----
> Von: time-nuts-bounces at febo.com 
> [mailto:time-nuts-bounces at febo.com] Im Auftrag von SAIDJACK at aol.com
> Gesendet: Sonntag, 13. August 2006 03:04
> An: time-nuts at febo.com; mccorkle at ptialaska.net
> Betreff: Re: [time-nuts] How do I know my GPS stabilized 
> oscillatorisworking?
> 
> 
> Hi Ulrich,
>  
> I've been using your plotter tool for a while now, and 
> Congrats! It's quite  
> good, I was very impressed. It's a wonderfull tool.
>  
> Here are some suggestions that came out of every-day real 
> life experience,  
> and may help to make it even better:
>  
>    * It would be good to have the tool normalize the vertical 
>  axix for every 
> column when showing more than one column superimposed. It 
> does  this if you 
> only view a single column, but if you show two or more with  
> different scales 
> (say frequency accuracy and temperature versus time)  the 
> tool doesen't create 
> multiple scales to fit the two data columns inside the  same window.
>  
>    * There are so many options to choose from that sometimes 
> it  is a bit 
> hard to find what one is looking for. For example, if I want  
> to change the 
> vertical scale from scientific notation to scalar notation to 
> see  more than two 
> digits of resolution, I have to go quite deep into the Axis 
> menu.  It would be 
> great if the axis could be clicked, and the scale changed 
> similar to  how Excel 
> does it.
>  
>    * Once zoomed-in, it would be nice to have a zoom-out 
> feature.  I figured 
> out how to do it with the right mouse button and selecting 
> from right  to left 
> on the window, but maybe there is a way to just click 
> somewhere to zoom  out 
> again to see all data?
>  
>    * On the Allan deviation table shown next to the plot, I 
> think  the tool 
> goes up only to show up to 4098 (?) s worth of data, even if 
> data worth  100000 
> seconds has been captured and is shown on the plot.  A couple 
> more  entries 
> in the table would be nice: maybe at time intervalls where a 
> major  vertical 
> decimal is broken (say for example 15500s if at that point 
> the measure  is 
> exactly 1.0E-014, then 102300s if it reaches 1.0E-015 at that 
> point  etc)
>  
>    * My HP 5334B counter, when set to A-B time mode will  
> sometimes generate 
> negative time measurements (for example -6ns, -13ns  etc), 
> but then for some 
> reason measure across one second so it jumps to  +999999993ns 
> etc. Don't know 
> why the 5334B firmware does this. The effect  is that the 
> Plotter tool gets 
> jammed by the large difference in the two  scales. I cannot 
> filter the larger 
> values out with the "delete outlier" button.  Could this be 
> added so the tool can 
> recognize this issue and handle it  correctly?
>  
>  * While drift can be removed, could you also add a feature 
>  that shows the 
> least-mean-square fit (average line or something similar) through  a 
> drifting, noisy plot?
>  
>   * Could you please add a help button to explain the many features  
> (buttons) of the tool?
>  
>   * Is there a way to select colors for columns? One of  the 
> traces I have is 
> always white on the plot, and it's very hard to see on  the 
> screen, as well 
> as almost impossible to printout onto paper.
>  
>    * An "oscilloscope" type marker tool would be great, where 
> you  drag a 
> vertical line right and left with the mouse across the plot, 
> and you have  a 
> readout of the vertical value of the selected column data at 
> that exact  point.
>  
> * It seems that the Allan Deviation etc. is always 
> calculated  over the 
> entire data set (is that true?), even if you zoom into a 
> small portion  of the 
> dataset. It would be great to only do the stability math on 
> the data  that's 
> actually visible on the screen (or even better: choose from 
> the two  options: 
> math over full dataset, or selected subset).
>  
> Hope not to have rambled too much; again this is a great tool!
>  
> Thanks for sharing it,
> bye,
> Said
>  
>  
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com 
> https://www.febo.com/cgi-> bin/mailman/listinfo/time-nuts
>