I want to share my issue with a lock-in amplifier and get some help

[LIADuck]

I'm setting up an EPR (Electron Paramagnetic Resonance) instrument and encountering an issue with the lock-in amplifier. I’d like to share the details of the problem and receive any help or advice.

First, I’ll briefly explain the setup. (See Figure 1) On the left side of Figure 1, you can see the data measured without using the lock-in amplifier. When this data is input into the lock-in amplifier, it produces a differentiated form of data, as shown on the right side of Figure 1 in the X channel In a lock-in amplifier, if the phase is correctly adjusted, 100% of the signal should ideally appear in either the X or Y channel. I'm curious why there is still a signal remaining in the Y channel despite these adjustments.. (For a more detailed image, please refer to Figure 2.)

Ref: Zurich Instrument Hompage

This is a very ideal dataset; however, there are slight imperfections, as highlighted by the red circles.

I suspected this issue might be related to the phase-lock of the lock-in amplifier, so I changed the settings to R and Theta for the experiment. The resulting data is shown in Figure 3. In my view, Theta (Phase) should be relatively stable, with perhaps a bit of noise, but the experimental results show significant instability after the first peak. I’m wondering if this is normal data behavior or if it indicates a problem.

Another potential issue I’m considering is that our lock-in amplifier(SRS, Sr830) has a bandwidth of 100 kHz. However, the input to the lock-in amplifier consists of a 100 kHz modulated signal combined with a 100 MHz signal. Can the lock-in amplifier successfully lock onto the 100 kHz modulated signal in this case?

I thought I was familiar with lock-in amplifiers, but I realize there’s still much I don’t know. Even a small piece of advice would be a big help.

Thank you very much for reading

 

[tech99]

I think the 100kHz you are trying to measure actually exists as amplitude modulation of your 100MHz signal. I know nothing about LIAs but maybe it requires an envelope detector/LPF ahead of the LIA. This will then extract the 100kHz signal you are interested in.

 

[LIADuck]

I think the 100kHz you are trying to measure actually exists as amplitude modulation of your 100MHz signal. I know nothing about LIAs but maybe it requires an envelope detector/LPF ahead of the LIA. This will then extract the 100kHz signal you are interested in.

Thank you for your valuable comments. Maybe we can filter the high-freq component in the signal using LPF. I'll try it.

 

[tech99]

I think the 100 kHz component is modulated on to the 100MHz carrier, not just mixed with it, so filtering will not work - it requires a diode envelope detector first.

 

[f95toli]

Could you explain where the 100MHz signal is coming from? Is that the downmixed output signal from the EPR (i.e. you are using some form of heterodyne scheme)? Presumably your EPR frequency is is at least a few GHz (Say X-band)?

The SR830 does indeed have a BW of 100 kHz. Why are you using a modulation signal equal to the BW? Why not using something slightly lower, say a few tens of KHz?

I also don't quite understand why you think the phase should be constant, if you are measuring the slope of a complicated signal(such as a EPR spectra) I would expect the resulting phase to vary at least a bit when measured with a lock-in

 

[LIADuck]

Could you explain where the 100MHz signal is coming from? Is that the downmixed output signal from the EPR (i.e. you are using some form of heterodyne scheme)? Presumably your EPR frequency is is at least a few GHz (Say X-band)?

The SR830 does indeed have a BW of 100 kHz. Why are you using a modulation signal equal to the BW? Why not using something slightly lower, say a few tens of KHz?

I also don't quite understand why you think the phase should be constant, if you are measuring the slope of a complicated signal(such as a EPR spectra) I would expect the resulting phase to vary at least a bit when measured with a lock-in

Yes, 100 MHz signal is down-converted signal ( Local oscillator - Intermediate frequency). Signal frequency is 9 GHz , X-band region. In that exp, Local oscillator frequency is 9.13 GHz, and IF is 9.03 GHz.

SR830 actually has 102 kHz BW according to datasheet

The 100 kHz modulation frequency is the default frequency used in EPR, likely related to 1/f noise. However, it can be lowered as needed.

I thought it should remain consistent because, as seen in Fig. 3, the phase during the measurement of the left peak was maintained almost consistently. I expected the phase plot to at least appear as a mirror image on both sides. Of course, I could be wrong.

Thank you for reading.

 

[tech99]

100 kHz seems very close to the top end of the LIA response, so you might see phase distortion from the LPF at the LIA input. It depends on the sweep speed whether sidebands fall close to 102 kHz. As you are seeing asymmetry I think I would check this.

 

[LIADuck]

100 kHz seems very close to the top end of the LIA response, so you might see phase distortion from the LPF at the LIA input. It depends on the sweep speed whether sidebands fall close to 102 kHz. As you are seeing asymmetry I think I would check this.

I have one question. Does sweep speed refer to the measurement speed of the experiment? The right image in Fig. 3 shows results measured by varying the magnetic field strength in smaller steps compared to the left image.

Thanks for your reply. I will try experiment following your comment for BW of LIA

 

[tech99]

The sweep must be carried out slowly so that the bandwidth is not increased excessively. You have only 2kHz bandwidth available at best, so the sweep probably needs to take more than a few milliseconds. Also it sounds as if the steps are discontinuous, which might generate additional side frequencies which will have to be accommodated by the LIA bandwidth.