LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

XY graph zoom problem

Hello,

 

There is a issue when you zoom in using a XY graph. Say you have a two points one at 0,0 and another at 10000,10. If you zoom in to say point 5000,5 the centre point the line will be way off nowhere near where it should be crossing. I was wondering if anyone knows a fix for this problem?

 

Thanks

0 Kudos
Message 1 of 9
(3,230 Views)

Since XY Graphs use floating point numbers this appears to be an issue with the numbers not being "exact" numbers. For instance I tested this with a line from 1,1 to 9,9 and zoomed in on 5,5. When the zoom level gets to a certain point, the fact that a number is not "perfectly" that number (off by a few millionths), makes the line move off of where it should be.

 

Work around? Just don't zoom that much (your data can't be that accurate anyway), or ensure that one of the points is within the zoom window.

 

Rob

0 Kudos
Message 2 of 9
(3,217 Views)

I'll explain what i'm doing, I'm placing a plot line on a xy graph and then having temperture follow that line maunally so the y side of the graph is temperature which needs to be accurate and the x is in seconds which can run up to a million seconds depending so I need to be able to zoom in to a specific point along the way. I'm aware I can plot every point a long the way as a work around but to me I shouldn't have to work around the graph it should just be accurate.

0 Kudos
Message 3 of 9
(3,215 Views)

I agree that this does not look right, but is probably due to limitations in the math used. Once you zoom into the area of iterest, the position of the line actually changes as a function of zoom level. (for example, the 5000 crossing of the line is at 5.00, then at 5.015, then at 5.16). I am sure the accuracy it could be improved, but maybe at a cost of performance.

 

It should definitely be investigated.

0 Kudos
Message 4 of 9
(3,206 Views)

Is there a reason that you can not calculate the temperature based on the seconds? It would be much more accurate to calculate the temperature, or at least, a few extra points in between.

 

Also, any temperature controller or temperature feedback will not be accurate to more than a couple of decimal places (perhaps a thousandth of a degree if you're using a calibrated platinum RTD).

 

Rob

0 Kudos
Message 5 of 9
(3,201 Views)

I have attached a image of the line this is with points (0,0) and (10000,10) zoomed in on x-axis 4900 to 5100. As you can see 5000 is at 5.15 not 5.00 which is 3% off. How can we can NI to investigate this?

 

Thanks,

 

example.jpg

0 Kudos
Message 6 of 9
(3,182 Views)

I put a note in the monthly bug thread. Hopefully a NI developer will look at it, and post a CAR here.

0 Kudos
Message 7 of 9
(3,175 Views)

Confirmed, CAR #147606.

Message 8 of 9
(3,126 Views)

@Jeff B wrote:

Confirmed, CAR #147606.


Above CAR tracks the problem for Windows. CAR 278004 covers the need to fix for Mac and Linux (the fix for Windows was unfortunately not platform independent). We don't preannounce bug fixes, but it might be reasonable to surmise from what I just said that this is "probably" fixed in 2012 on Windows.

 

Side Note: One of the reasons we're hesitant to talk about bug fixes before a release is that it is always possible a fix will have to be reverted prior to release. We really want to avoid telling someone a bug is fixed and then have to come back later and tell them it is not.

 

Roy

0 Kudos
Message 9 of 9
(3,114 Views)