LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Lock-in amplifier VI to sideband

Eric,

 

It sounds like we are very close. A few differences in wording and terminology but not enough to change what I am working on. 

 

Thanks.

 

Lynn

0 Kudos
Message 21 of 30
(1,055 Views)

Sorry I was not able to get back to this thread sooner. Sometimes I have to work on my paying job.

 

The executive summary is here and a more detailed description is in the attached file.

 


1. Lock-in is not suitable. Requires reference not easily obtained and only recovers AM.

2. Need I/Q detection to get both AM and PM. (Lock-in is morning "person"? AM but  not PM)

3. Proposed solution: 

* Heterodyne desired signal component to low IF.

* Low pass filter to remove most undeisred components.

* Sample IF at sufficiently high rate to get phase info.

* Do software I/Q detection to recover AM and PM.

4. Remaining issues:

* Heterodyne mixer and LPF require external hardware - analog switch, RC or active filter, and dutycyle correction.

* May require separate detection of chopper frequency

5. The 192 Hz update requirement may be incompatible with the 260 Hz chopper frequency.


 

 

Lynn

0 Kudos
Message 22 of 30
(1,033 Views)

Lynn,

 

Thanks for all your work on coming up with a solution.  This sort of takes things in a different direction as far as using something other than a lock-in which may be needed .  I will have to do some reading up on I/Q detection because I am not familiar with it.  I'll give an update on my progress at some point (other projects are starting to pile up so this has been progressing slowly).  

-Eric

0 Kudos
Message 23 of 30
(963 Views)

Hey Edgar,

 

Sorry for the delayed response.  I think I have seen your post on the Comsol forum as well (that forum has helped me a lot too)! Is it possible to make such a filter tunable because the 280 KHz signal varies a little from measurement to measurement?  That seems like it would be a problem so I have been looking into the "phasing method" for single sideband modulation.  

 

Does anyone know anything about SSB modulation for these frequency ranges (and using the phasing method since it seems the filtering method won't work very well)?  I can understand the math, but I haven't been able to see anything about practical implementation for SSB modulation (like where to buy the hardware and/or how to build something like this).  

 

-Eric

0 Kudos
Message 24 of 30
(958 Views)

Eric,

 

this kind of very narrowband filters is typically not tunable. If you need tunable frequency you would need to mix to a higher filter frequency, e.g. 9 MHz and mix back. On 9 MHz or other standard IF frequencies very high quality off-the shelf quartz filter are available.

The I/Q approach as suggested by Lynn is certainly an option as well.

 

Cheers

Edgar

0 Kudos
Message 25 of 30
(945 Views)

Eric,

 

I think there may be another problem. I had been working on a simulation of the signals in an attempt to identify the best frequencies for the heterodyne and I/Q signals. The requirement for making 192 measurements per second and the 260 Hz chppoer frequency create some timing issues. With either a lock-in or the I/Q approach some filtering is required to separate the desired signal from the reference frequency components. Typically such filters are chosen so that the outputs include data from multiple cycles of the frequencies being rejected. Since one of the frequencies to be rejected is the chopper frequency, the filter bandwidth is likely to be an at least octave lower and preferably a decade lower. However, such filters will make it impossible to get meaningful data at 192 Hz.

 

The graph below shows a low amplitude square wave at 96 Hz, representing the data signal from the AFM.  A measurement in each half cycle will meet the 192 Hz requirement. The chopper signal is a 260 Hz square wave. Consider that the combined signal at the detector will have a phase and amplitude change at each transition of the chopper signal. The AFM signal wil not be a square wave, of course, but that is easy to use for visualization. The problem is that the timing of the AFM transitions is probably not known and may not be easily extracted from the data.  In about 1/3 of the chopper half cycles shown, an AFM transition occurs, sometimes near the center and other times close to the edges. Any averaging or filtering should not cross the AFM transition boundaries to avoid compromising the data, but this means that the filtering will cover about one cycle of the chopper frequency, which is not enough to separate the components.

 

Chopped AFM data.png

 

Lynn

0 Kudos
Message 26 of 30
(897 Views)

Hey Lynn,

 

Thanks for the input.  It seems like the lock-in method will be difficult to implement mainly because it will be difficult to define a reference frequency.  However, are you sure we have to worry a lot about the 192 measurement per second because I think we can consider those seperate measurements.  There will be no correlation between each of those 192 measurements/second actually and we can think of those as a seperate 260Hz+280 KHz mixed signal waves (or whatever mixed signal we choose to measure with the lock-in) which have somewhat random phases and amplitudes (they are random except for the fact that they are defined spatially by the structure we are measuring).  For the the lock-in method the 192 measurements per second happens at a much slower rate than the signal we want to actually measure.  Is this is more an issue with the IQ approach you are suggesting?  I still need to get up to speed with the IQ approach.

 

I have been reading up on IQ modulation.  It seems to avoid the need to use filters to select the upper or lower sidebands and it looks like this is a method for generating single sideband modulated signals directly.  Do you think this could be used to generate the reference frequency for the lock-in amplifier?  

 

-Eric

0 Kudos
Message 27 of 30
(891 Views)

Hello,

 

I am now seeing on this page that the NI 6115 can be used as a Lock-in amplifier with appropriate Labview code.   

 

http://www.ni.com/white-paper/5613/en/

 

It has a sampling rate that is 10MS/s, which I think is high enough for my application.  Also, it has multiple inputs and outputs:

 

http://sine.ni.com/nips/cds/view/p/lang/en/nid/11886#specifications

 

Is there any reason why this wouldn't work for me?  I have figured out that really I can make this measurement at ~240KHz*2+400 Hz, which puts the signal of interest at 480.4 KHz.  Let me know how that sounds.

 

I think if we can get this to work with NI interface card/Daq device, then this would be easier for us. Thanks!

 

-Eric

0 Kudos
Message 28 of 30
(867 Views)

Eric,

 

I had not read that white paper before.  The key VIs are password protected so we cannot see what is inside. My guess is that they perform some magic in software to get around the limitations of the hardware (up to a point).

 

These may be issues for your application:

1. The data must include a reference signal. The AFM does not generate a separated signal at the sideband frequency.

2. It calculates the filter settings automatically. If these are not compatible with the AFM update rate (192 Hz) or the chopper frequency (260 Hz or 400 Hz now?), it is not clear how you could change them. The white paper does not document the information. The context help refers to the "Lock-in Demo Users Manual." That does not appear to be available on the NI web site any longer.

3. It appears to determine both real and imaginary parts so the amplitude and phase information may be available. The trade-offs between data block size - which affects the filtering - and the timing requirements may constrin performance.

 

The NI 6115 has only 12-bit resolution which will limit the ability to recover small signals in the presence of much larger ones (the carrier). You have not indicated how big you expect the signals to be compared to the largest component at the A/D converter input.

 

It certainly has some possibilites, but I would not buy the 6115 until more analysis of the capabilities and requirements is complete.

 

Lynn

0 Kudos
Message 29 of 30
(856 Views)

I need to catch a bus in a few minutes, but ...

 

if you are now using 400 Hz as your chopper frequency, you may be in luck. It may be possible to use 400 Hz as your reference signal, and measure the 1201 harmonic. I did something similar to this where I locked multiple function generators together where one was a common divisor frequency of both of other function generators. I used that as a reference and detected the higher harmonics. The SR844 RF lockin can only detect the 2nd harmonic, its lower frequency sibling can detect higher harmonics . . . but you may be able to detect higher harmonics in software that NI provides.

 

cheers,

mcduff

0 Kudos
Message 30 of 30
(849 Views)