LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Lock-in amplifier VI to sideband

Hello,

 

I have a quesiton about modifying the lock-in VI shown here: <link no longer exists> I saw some threads that discussed a similar problem, but there were no clear answers. 

 

We have a infrared signal that is modulated at 280KHz.  This signal is interfered with an equal wavelength infrared beam that is modulated at 260 Hz, which then creates a sideband with a better signal to noise compared to just when locked-in to the 280 KHz signal.  So, our signal is just one input, but our reference is ONE frequency, but comes from two mixed signals.  Here are some pictures of the frequency spectrum showing where we want to lock-in and the setup of the experiment (see attachements).  We actually want to lock-in at a sideband of the 3rd harmonic (3*280KHz+260Hz). 

 

I imagine we can input the two references into a daq device, but how should we modify the Lock-in amplifier VI that NI provides ("lock-in amplifier example") to mix the two reference to make the harmonic sideband reference?  Can I do this in Labview?  We would greatly appreciate any help!  The necessary hardware to do this operation cost $30k!!!!  Thank you.

-Eric
Download All
0 Kudos
Message 1 of 30
(5,499 Views)

Let me clarify that we do NOT have any hardware at this point.  We want to simulate the experiment first and develop our VI to be able to do this operation, then we will probably order one of the pieces of hardware on this page (which is recommended for lock-in amplifier applications):  <link no longer exists>

 

-Eric

0 Kudos
Message 2 of 30
(5,452 Views)

Eric,

 

The hardware devices on the page you linked are much too slow for your signal at 840260 Hz. The maximum sampling rate for those devices is 102.4 kHz or 204.8 kHz, both of which are far below the Nyquist rate for your signal (> 2*840260 = 1.68+ MHz). Sampling near the minimum frequency which meets the Nyquist criterion will work in theory but may require a large number of samples to get good information.

 

Is the signal you want to measure on the 280 kHz signal or on the 260 Hz signal? Are you looking for phase or amplitude information? How fast does that information change?

 

Lynn

 

0 Kudos
Message 3 of 30
(5,437 Views)

Hello Lynn,

 

Thanks for your response.  The originally modulated signal is at 280 KHz, but what we want to measure is a sideband of that, so we are actually measuring the sideband: 840260Hz or 839740Hz  (we are not using either as carrier wave purposely, but basically the same thing is happening, except we are not demodulating it back to 280 KHz and measuring at 280KHz).  At least that is my understanding of what we should be doing 🙂

 

I think the graph that I attached sums up why we want to measure at this sideband:  better signal/noise ratio. 

 

We are interested in amplitude and phase information (it is crucial).  That information will change on the order of 190times a second (we must be able to lock-in and get this information 192 different times a second).  Another way of saying this is we are getting 190 data points/second for phase and amplitude. 

 

Thanks for pointing out the sampling rate problem, it makes complete sense.  So it seems like utilizing a software based lock-in technique with a DAQ device can not really work in the 800s of KHz range that well.  It seems surprising there is no available hardware to do this.  I don't have a lot of training in this kind of thing, so could you ellaborate on the longer sampling time possibly being a solution?  It seems like it would not get around the fact that we are not sampling enough times within each cylce of the sideband, but, again, I don't know a whole lot about the hardware side of things.  

 

Thanks again for your input!

 

-Eric 

 

 

0 Kudos
Message 4 of 30
(5,414 Views)
Eric, Have you looked into the SRS 844 RF lock-in, http://thinksrs.com/products/SR844.htm It should be able to handle your needs. As far as the data acquisition goes it is easily programmable in LabVIEW. You should be able to achieve the 192 Hz DAQ rate with it. You can get Amplitude and Phase, or the real and imaginary components. I have use it in the past with LabVIEW and had not had any major problems. Cheers, mcduff
0 Kudos
Message 5 of 30
(5,405 Views)

Hello Mcduff,

 

Thanks for your response.  In looking at the manual for the SR844 it appears that it covers the right frequency range (all the way into the MHz range) and can lock-in at up to the 2nd harmonic, which would suffice and be about 560 KHz. Our sideband would be about 560,000 Hz+or-260Hz.

 

You were able to write a program in labview to have it lock in at a sideband such as this?    It seems like it can be done, I am just not sure how to put in those two references and mix them, then tell the lock-in to lock at a sideband.  We are sorely lacking in electronics/electrical engineering know-how...,but can do things in labview if we know where to start.  It seems like maybe you could make the internal oscillator and external input reference run at the same time and then just mix the signals in labview to generate your sideband reference?  Any tips would be greatly appreciated!  Thanks again! 

 

-Eric 

0 Kudos
Message 6 of 30
(5,396 Views)

Look at this thread for a program, http://forums.ni.com/t5/LabVIEW/Can-driver-for-SR830-and-SR844-be-used-interchangeably/td-p/1600796

There are some minor bugs in the program, I can give you an update if you are interested.

 

Also look at the instrument drivers page on NI's site, there are programs there also.

 

As far as reference goes, you cannot just input both of them to the lockin. You are combining them in a non-linear way to get the sum and difference frequency components that you are looking to detect. Your photodetector is doing this for you for your light signal. If you wanted a reference, you would need to do the same electronically. You would have to send your two references to a mixer and then filter out the frequency component that you want. I have no idea how easy this is to do.

 

I'll try to answer any further questions you have, but will not be able to unitl tomorrow.

 

Good luck.

 

Cheers,

mcduff

 

EDIT:

Look at the specs for some of NI's multifunction DAQs that have 4 or more Analog outputs, if all of the output are phase locked you may be able to make your reference and driving frequencies for your experiment. You will have to ask a NI rep how much phase jitter there is between the channels.

 

 

0 Kudos
Message 7 of 30
(5,389 Views)

Hey McDuff,

 

Thanks a lot for your feedback.  Indeed, mixing these signal electronically has been challenging to say the least.  It is certainly easy to mix them, but we have been told that filtering to specfic sidebands would be impossible since for example one side band is located at:  2*280KHz+260Hz while another is at:  2*280KHz-260Hz.  The good news is that the mixer we bought is doubly balanced, so that means the "carrier" or 280KHz wave is blocked AND additional harmonics are excluded as well.  However, it seems like we can not filter between those two sidebands and it may work out fine if we just measure both together...we're not sure and will check this out as soon as we get all the equipment that is needed.

 

 

However, do you think it's possible to mix these two analog reference signals (280KHz and 260 Hz) using some sort of NI daq device, then output the mixed frequency filtered to the sideband we want?  That seems possible right?  I guess we need something that can sample the 280KHz signal with enough sampling rate.  Let me know what you think!  Thanks again!

 

-Eric

0 Kudos
Message 8 of 30
(5,294 Views)

Eric,

 

Look at the math. You want to be able to resolve two signals at 280260 Hz and 279740 Hz.

 

f (Hz)    280260       279740

T (us)   3.568115    3.574748

 

The difference in the periods is 6.63 ns. To measure that directly you would need to sample faster than 150 MHz. If your mixer does not suppress the carrier sufficiently, then the problem is worse because you will also have a carrier frequency component.

 

How are you generating the frequencies, especially the 280 kHz?

 

Lynn

0 Kudos
Message 9 of 30
(5,284 Views)

Eric,

I'll leave the math to you, but here is an idea, no idea whether it would work. (You can try the math in LabVIEW, without using a DAQ to see if you can get all three frequency components.)

The simplest mixer just multiplies two signals. Assume we have two sinusoidal signals at frequencies f1 and f2, when we multiply them the possible frequencies are f1 + f2 and f1 - f2; the sum and difference components. This is true if we have pure, ideal sine signals.

Now assume we have two square waves at frequencies f1 and f2. If you multiply them together, once again you get sum and difference components f1 + f2 and f1 - f2; however, the square waves contain all of the ODD harmonics, that is, 3f1, 5f1, 7f1, etc and 3f2, 5f2, 7f2, etc. So now you have sum and difference between all these possible combinations, including 3*f1 + f2. (In your earlier post you used the third harmonic, if you need the even harmonics then this will not work.)

Assume we have a function generator with 4 synchronous outputs, see if NI has one. In LabVIEW I can create the signals I want. For one ouput I put a square wave at f1, for another output I put a square wave at f2, for another output I put f1 times f2. That is I am doing all of the math for these signals. I can electronically filter bandpass the mixer ouput for the frequency I want, (hard to have that narrow of a filter) or I can use a math filter in LabVIEW before outputing to analog out for the frequency I want (probably easier). Now you can have phase synchronous signals at three frequencies. (Try the math in LabVIEW.)

 

Just saw what Lynn said, who is a lot smarter than me. I may be mistaken here, but the time constant on a lock-in amplifier sets the frequency discrimination; for example a 100ms time constant would give you a 10 Hz frequency discrimination. (Here I am referring to a stand-alone instrument, such as the SRS 844.)

 

Hope this helps.

 

Regards,

mcduff

 

0 Kudos
Message 10 of 30
(5,282 Views)