Power system frequency not measured correctly

Hello,

I used a Dewesoft Sirius i-HS HV to measure the voltage during a load step test of a generator. When I checked the recorded data (slow and fast recording), I realized the frequency value is not measured during the step when a voltage dip occurs. This was not an isolated incident, it happened during the whole measurement set. I have to mention that I used a virtual star point connection for an IT system, LP filter of 30 kHz.
What could be the reason? I did the same type of measurements tenths of times before and I never had the problem.

Cheers,

Andrei Ghiran

When I export the average and min frequency I get the same values in both columns. Should be some difference between them.

Hi,

Can you please send a data file. It is hard to figure out what could be wrong just from that picture.

Kind regards,

Urban Baš

I attached it. Thanks for the fast reply!

Do you have a data file that was stored at full rate?

This one is almost completely at full rate. I split the archive in two volumes in order to be under the 10 MB limit.

Can you upload the files on the ftp server without zipping them? We can't extract the files.

I attached you the manual for it.

I just uploaded it. Maybe I should mention that the load steps were 50% of the generator power so the frequency deviation should be 1-2Hz.

We analyzed your data file and found out the cause of this problem. The line frequency you have chosen is 60 Hz and Dewesoft will calculate the frequency when it is 60 Hz +/- 2%. You can change this in the settings (see attached file), but setting this on to high can effect the calculation accuracy, so you have to be careful when changing this.

Thank you very much for the help! I will keep in mind for the future load step tests. Do you have an indication about how much would the parameter influence the accuracy?

We tested the accuracy with a simulated signal and we found out that if you have a clean sin signal the accuracy is still very good even if you set the frequency threshold to 10%. I saw that you have clean sin signals so this shouldn't effect the accuracy of your test.

I attached you the setup that we used for our test.