Forums › General Discussions › Limit Lines in Spike 3.1.12
- This topic has 10 replies, 3 voices, and was last updated 4 years, 12 months ago by miran.
- Volker February 20, 2018 at 6:51 am
Nice to have the Limt Lines now in the preset…
Still I cannot get my old limit lines running. This is what I had in Version 3.1.11:
It gave me the attached picture.
How can I get this in 3.1.12 ?
Andrew February 20, 2018 at 11:18 am
- This reply was modified 5 years, 1 month ago by Andrew.
I’m sorry for the inconvenience. The functionality has changed slightly. The old method had you specify both the min and max amplitude for each frequency (on each line in the file). Now you will want to create two files, a minimum and maximum line and load them into two different limit lines.
The file format has only changed slightly, instead of specifying two amplitudes for a min and max, the single amplitude will be interpreted as a min or max depending on your selection in the limit line menu.
I attached both files you would load to recreate your old setup and a picture of them working on my screen.
AndrewVolker February 20, 2018 at 11:45 pm
thanks for your immediate support. I thought that this is working like you explained, but I still cannot get it running.
I would like to switch to a more simple to replay example:
– Single carrier 250 MHz
I used limit line 3.
Setting to Lower Bound will always pass.
Setting to Upper Bound will always fail.
No matter what the aplitude is. Increasing RBW does also not solve the problem.
Two more problems that I would put on the bug list:
– it is not possible to enter these values in the application, since rounding of the frequency will happen. It has to be done in a text editor and you have to load it. Still you cannot view it because of roundings…
– Changes to the limit lines will not always reflect to the app. Sometimes you will have to switch on and off or switch to another line. I have not found any system behind this behavior.
VolkerAndrew February 21, 2018 at 10:49 am
So I played around with the single tone file you linked, and got it to work as I think you intended, as a test for whether a CW is present. I attached some images, where I zoomed in on the CW and I increased the Div to 15 to see more y-axis.
I did also change the point at 250M to -60 instead of +24 so it’s easier to play with a signal generator at my desk.
I attached three pictures, one where I increased the RBW and you can see it passing. Then two more where I shifted the CW freq and lowered the RBW and it fails both.
The limit line was configured as a min limit line.
Can you provide any more information or maybe a picture showing the scenario you see. If you are able to zoom in on the span as well, it could help me determine why you are unable to reproduce my results.
I will need to look into the roundings, I agree that it is an issue. I will try to get a fix for this for next release. I’ve been editing the files in a text editor for now and it seems to be working.
Edit: I think I see the issue you are experiencing with the line always passing sometimes. One of the issues here is that the spike that you are putting in the limit line is very narrow, and there are only a finite number of points in the sweep, so it is possible that the entirety of the spike and adjacent points fits between two of the sweep points. Basically the lowest RBW you can get away with is what you should use as it will increase the number of points in the sweep giving you more resolution to test again the limit line with.
I look forward to your response.
RegardsVolker February 22, 2018 at 12:47 am
I think you are right about the RBW. If I come down to a RBW of about 300 Hz it is working. Since this was for testing and quick checking of IP3 of an amplifier, I increased RBW to 5kHz. In this case I can see the two main carriers. With the low RBW they hide behind the limitlines ( picture lowrbw.jpg ).
If I increase the RBW to 3kHz it will pass ( picture rbw3khz.jpg ).
It will fail at a summ level somewhere below -73 dBm.
It gets worse if you increase the span…
Anyhow this was working in Version 3.1.11.
So the algorithm of checking has changed. If it is like your wrote that you are checking single samples against the limit you may run into trouble when you have a good generator. Should you not check against a interpolated line between two samples ?
I really hope you can get it running, because I ever wanted the limit lines stored with the preset as it is first done in this version.
The limits I used:
VolkerAndrew February 22, 2018 at 11:15 am
Thanks for the images, I was able to reproduce what you are seeing.
It took me a minute to determine why this is happening and have determined the cause.
In the latest release, we have changed how we plot the limit line. The algorithm we use to test the limit is still the same but instead of plotting what we test against (the interpolated points) we plot the limit line as requested. Since your limit line includes a Spike between two output bins it gives the visual appearance of failing the limit line but in fact the testing is being done on adjacent interpolated bins. If you were to load this limit line into 3.1.11 you will notice the limit line being displayed will not actually come up to -60dBm as we plot only what is being tested against. In fact I included a screen shot of 3.1.11 with the same -60dBm peak being loaded into the limit line and you can see it gets rendered at ~-68dBm because that is the interpolated value that is being tested against. You can also see my input signal below -60dBm.
On review of this, I think we will need to go back to plotting the values tested against rather than the requested limit line.
In conclusion, my current understanding is that we have changed how we are plotting limit lines in the new version but the pass/fail message will match what you would see in the older version 3.1.11.
It might be more beneficial to create a box like limit line that has at least a width of 1 bin (RBW), and then sharply drops off, I think this will create the result you are looking for with more consistency. Currently with a very narrow spike, the high point can easily fall between two bins at which we can’t test at.
If you want to load an older version of the software you can find the link here.
I will work on restoring the old way of plotting the limit line for the next release.
Let me know if you have follow up questions.
AndrewVolker March 28, 2018 at 5:25 am
limit testing in version 3.1.14 is now nearly working as expected. Thank you for great service.
Still there is something strange. In the picture you will see the two lower limit lines ( red needles ) that will check for minimum power level of two carriers. They should come up to +22 dBm. Instead they just show somthing about +18 dBm.
It is working when you fall below the graphical line, but the line is not corresponding the values in the Limit Line Table.
Limit2.jpg should not show pass !
VolkerAndrew March 28, 2018 at 11:12 am
Did you try making the limit more like a box rather than a Spike like I suggested in my earlier response? Since your spike is smaller than 1 RBW the limit test point is being interpolated between the high point and adjacent low points. Unless the frequency bin happens to fall exactly on your high limit line point, it will interpolate somewhere along the steep limit line. This latest version is now drawing what is being tested against so there is no confusion, (the last version improperly drew the limit line as entered rather than the actual interpolated test points)
Ultimately I think this spike in the limit line approach to check for a CW is not ideal with how we implements limit lines and it looks like you had a perfect setup back in 3.1.11 because I can’t even reproduce your original image using the original limit line you provided. I think going towards a box/square shape and being mindful of your RBW will solve this problem long term.
miranParticipantmiran March 28, 2018 at 11:15 am
If You have already discussion about limit line. What is with relative limit? I don’t find possibility to setup relative limit. After making the mask, I want to use this on any frequency.
I hope I don’t have to big wish.
Regards, MiranAndrew March 28, 2018 at 12:13 pm
You are talking about a relative frequency limit line? This is a good idea, but unfortunately we don’t have any implementation of this yet. I appreciate the feedback.
miranParticipantmiran March 28, 2018 at 12:35 pm
Yes, I mean a relative frequency limit line. Center frequency is relative. For mask I write only minus and plus frequencies offset from central frequency and level.
You must be logged in to reply to this topic.