Magnettech MS5000 transient measurements - wrong x-axis

[Silvio]

Hello,

eprload returns an incorrect x-axis for transient measurements at fixed B0 for the Magnettech MS5000 spectrometer. The x-axis is the magnetic field instead of time.

The identifier for this type of transient measurement is in the xml file: Type="kinetic" as shown below
<Recipe Name="LOGSTest_006" Author="" Type="kinetic" MeasurementCount="1" CreationDate="2018-08-29T09:19:32.1987361Z">

For field sweeps the type is "single".

The total time of the transient is stored in the parameter: KineticTotalTransientTime
From this, the x-axis can be constructed as a linearly spaced array from 0 to KineticTotalTransientTime with the number of points of the signal array.

Attached are two files: one field sweep for comparison and a transient measurement

Best,
Silvio

[Stefan Stoll]

Interestingly, the time axis not stored in the transient file. There appear to be 9955 points, but it is not obvious from the file what the time axis is. What is the expected time axis?

[Silvio]

Attached is the spectrum of LOGSTest_006 as shown in ESRStudio (MS5000 software). The time axis is from 0 to almost 10.

However, I just saw the following. If the file is loaded with eprload I think the time is in Params.Curves.MW_Absorption.x and the EPR signal in Params.Curves.MW_Absorption.data.

As far as I understand, the EPR signal and the magnetic field in MS5000 xml files are stored as a function of time. Since the magnetic field is read out by the MS5000 by a much lower rate than the EPR signal, eprload interpolates the magnetic field in order to obtain the same amount of points at the correct time to be able to convert the time axis to a magnetic field axis. Is that correct?
In the case of transient measurements, the interpolation could be skipped and the x-axis could be just the time taken from Params.Curves.MW_Absorption.x.

I also found another way to obtain transient data with the MS5000. In the data files this can be seen from parameter (xml file and spectrum as seen in ESRStudio attached):
<Param Name="BHoldEnabled">True</Param>

If Type="kinetic" BHoldEnabled is still saved in the xml but ignored by the MS5000 software.

Summarised: if Type == "kinetic" or BHoldEnabled == True then time as x-axis.

I hope this helps.

[Stefan Stoll]

Thanks for the additional details. We will look into this. Would be great to have a general implementation that works for all types of data from this spectrometer.

[Silvio]

I found another small bug in the eprload() regarding the x-axis.

The last about 70 values of the field axis are at least in the example of LOGSTest_001.xml from the first post are NaN (I also checked other files. All show the same behaviour).

The reason for that is that the x-axis of the magnetic field (BField vs. time, Params.Curves.MW_Absorption.x) is ends earlier than the x-axis of the EPR signal (MW_Absorption vs. time, Params.BField.x). If only an interpolation is performed, the last field values cannot be calculated, because an extrapolation would be necessary.

For the uploaded dataset to check:

Code: Select all

[x, y, Params] = eprload('LOGSTest_001.xml');
Params.Curves.BField.x(end) % = 61
Params.Curves.MW_Absorption.x(end) % = 61.0740

Params.Curves.MW_Absorption.x(60979) % = 61.0

% Last field point with number
x(60979)
% First occurence of NaN
x(60980:end) % = NaN

I hope this helps!

[Stefan Stoll]

An improved version of eprload is in the works. Question regarding LOGSTest_006_transient: the first time point in the file is at 0.022 s, but the screenshot appears to show the data starting at 0 s. Is ESRStudio extrapolating to 0?

[Silvio]

Apparently, the y data at t=0 s is just a copy of the value at t=0.022 s. Please see the attached screenshots.

LOGSTest_014_zoom.png (22.43 KiB) Viewed 5032 times

LOGSTest_006_zoom.png (22.24 KiB) Viewed 5032 times

[Stefan Stoll]

Thanks for this! Check out the new version 5.2.21, where these issues should be resolved.

[niconeuman]

Greetings, Dr. Stoll,
I'm writing because of another possible bug in loading Magnettech .xml spectra.
I noticed that the spectra, which originally had 30000 points, was loaded as a 1550 point vector.
Also, this loaded spectrum had a non constant spacing of x-points, which caused discrepancies in the fitting. I noticed that because when I simulated the spectra with pepper I needed a certain g|| value, which I tried to constrain manually. But in the fitting that g-value clearly didn't match the experimental spectrum. So I plotted Bexp-linspace(Bexp(1),Bexp(end),length(Bexp))' vs. the same linspace and showed a highly nonlinear deviation.
I can send the files for you to check, if needed.
Thank you very much
Best regards,
Nicolas

[Stefan Stoll]

Please post the file.