Understanding the file Fast vs Detailed CSV Files for the GW Instek GDS-1054B Oscilloscope.

Thread Starter

MB107

Joined Jul 24, 2016
345
I'm having trouble understanding the files for the GW Instek GDS-1054B Oscilloscope. I believe I understand the Detailed files but I cant seem to make heads or tails of the Fast CSV format. I am trying to use this to test the output of an automotive speedometer speed sensor. I have taken voltage output at several speeds measuring the VRS speed senor output and stored the data in both detailed and fast CSV format.

I believe the detailed files are giving me exactly what I want but I would like to be able to use the Fast CVS files for when I need to take data at higher rates. The detailed files give me both voltage and time. The Fast CVS only gives the voltage data in an integer format. Both files give me a vertical scale. If I scale the data by the vertical scale it seems fix the data for the Fast CSV files but it mutilated the data for the detailed CSV files. OK so I get that the Detailed CSV files are OK as is and no need to scale. But with the Fast CSV data I scale the data by multiplying by the Vertical scale factor. That makes the data linear and looks like something that I can use, except its too high like 1000V pk-pk. I found that when I scale the Fast CSV and then divide by 25 the output is the same as the detailed CSV. I cannot find anything in the docs to get the correct voltage output from the fast CSV.

Attached are two files one is fast and one is detailed. The excel file shows what happens when you calculate the output at similar frequencies and scale using the vertical scale from each file.

I hope this all makes sens and I hope someone can help.
 

Attachments

spinnaker

Joined Oct 29, 2009
7,830
Can you point use to the documentation anyway? Maybe someone else can figure it out.

It might also help if you included a screen shot of what you are seeing on your scope to go witht he CSV files.
 

Thread Starter

MB107

Joined Jul 24, 2016
345
Can you point use to the documentation anyway? Maybe someone else can figure it out.

It might also help if you included a screen shot of what you are seeing on your scope to go witht he CSV files.
Thanks for the reply.

Documentation can be found here: First link is the user manual, second link is the programming manual.

Page 143-144 of the user manual talks about the detailed and Fast CSV file formats.
https://www.gwinstek.com/en-global/products/downloadSeriesDownNew/1676/130
https://www.gwinstek.com/en-global/products/downloadSeriesDownNew/1700/130

See attached screen shot:
 

Attachments

Thread Starter

MB107

Joined Jul 24, 2016
345
looks like the fast csv is just a non-floating point data format
Exactly, but some where in this header should be a factor to multiply by to get the real value. Vertical Scale appears to be half the answer.

Maybe there is a secret society that knows that tightly guarded factor to multiply by and it is consequently not published.
 

pmd34

Joined Feb 22, 2014
527
Hi MB107. I had a look at the files. If you take the "wave form data" number and divide it by the "vertical scale" and "extended Div" for your fast data your y axes for the 2 traces look very similar.

CSV file.jpg
 

Thread Starter

MB107

Joined Jul 24, 2016
345
Thanks for the reply PDM34

This is not working all of my files have a Vertical Units Extend Div of 16. So dividing by this makes the situation better but not as good at dividing my 25. The vertical scale changes from file to file. You have only seen one of each file so multiplying by the vertical scale factor only becomes apparent to linearize the Fast data when you plot them all. So I attached t excel workbooks with all of the detailed files and all of the Fast files. The last tab in each workbook is a summery that plots voltage pk-pk vs frequency. I have copied the Fast data summery plot to the last tab of the Detailed data workbook so comparisons can be made. I highlighted 2 frequencies and there associated Voltage, scaled for the fast date. These two voltages should be the close and they do come close when I put a 25 in the cell above the column that says divide by 25. There has to be more to this.
 

Attachments

Danko

Joined Nov 22, 2017
1,828
I found that when I scale the Fast CSV and then divide by 25 the output is the same as the detailed CSV.
8-bit oscilloscope has 2 ^ 8 = 256 data points vertically.
So, one vertical division contains 256 / 8 = 32 points, not 25.
Therefore, after scale vertical data, you need divide it by 32:
upload_2019-2-26_6-22-52.png
Used data from file DS0002C.CSV
 

Thread Starter

MB107

Joined Jul 24, 2016
345
8-bit oscilloscope has 2 ^ 8 = 256 data points vertically.
So, one vertical division contains 256 / 8 = 32 points, not 25.
Therefore, after scale vertical data, you need divide it by 32:
View attachment 170994
Used data from file DS0002C.CSV
Thank you

32 sounds a lot better than the 25, I just pulled out of an orifice. I did in fact twiddle with the numbers an found 27 was better than 25 but I understand this and it sound reasonable. A bit of statistics will flush it out.
 

Leokrgl

Joined Nov 20, 2019
1
Thank you

32 sounds a lot better than the 25, I just pulled out of an orifice. I did in fact twiddle with the numbers an found 27 was better than 25 but I understand this and it sound reasonable. A bit of statistics will flush it out.
Hello,
Lately I came across the problem you are mentionning and discussing: I've got a bunch of integers that are supposed to be the voltage at each period of measured time and I don't know how to find the real value of the voltage :D. I am sorry to ask you that again, but I'm afraid to misunderstand you... So if I got a serie of integers in a csv file(vertical scale 5*10e-2 V) like the one you showed, I need to multiply my integer by 5*10e-2 and divide it by 32? So i.e. the integer in the .csv is 36, then the real voltage is 36*(5*10e-2)/32 ??
I hope I made myself clear and I hope to receive an answer.
Thank you very much
 

ejonesve

Joined Jan 5, 2023
1
I found for the GDS that the divisions appear to be divided into 25 points. The 256/8 seems more than reasonable, but it must be that they chose 25 for some reason. I'm not certain if that applies to all models.

This snip is from the GDS1000 Series User Manual:

1672940321715.png

I've also found from looking at the OpenWave-2KE V1.05 source confirmation that there is a division by 25.

1672940686826.png

This doesn't completely get me to my desire to reproduce the dataset with real numbers, but I'm hopeful someone will see this and then provide a succinct solution for all of us.
 
Top