Science Buddies: "Ask an Expert"

Hey,

As part of my project I need to compare the relative phases of two signals each from a different sensor. The device would need two BNC connections to input the two signals for comparison, and some screen or output from which I can retrieve the phase difference information.

I need to know what a device that does this comparison is called. So far, I have looked into phase detectors and phase comparators, but in searches these have all been chips or the equivalent for use in a phase-lock-loop or similar device. I need a standalone device like I described (with two inputs, etc), that will explicitly measure phase differences between signals.

I have tried using an oscilloscope to measure the phases of two signals and manually find their phase differences, but this has not been effective for my needs.

What device can I use to compare the phases of two signals?

amnc11 - Without understanding exactly what your experiment is, I would ask a few questions to narrow the problem.

1) Are the two signals being evaluated the same frequency, therefore your measurement is trying to determine a delay, difference, reflection or shift between the two sources?
2) Are the two signals at different, albeit fixed frequencies, looking to measure a heterodyne effect between sources?
3) Are the two signals each exhibiting variable frequencies such that this is a real-time DSP problem?

Try to define more precisely what you are trying to do, perhaps some simple phase-lock loop circuitry can be used to support your measurements.

Rick Marz

My problem is most similar to your first example.

The two signals both derive from the same output of a wave generator, only they travel through different paths that affect their relative phases. The two signals will always be the same frequencies as one another, but that 'same frequency' might fluctuate from 100-300khz. The actual phases of the two signals will not necessarily remain stagnant, but the phase difference between the two signals will stay constant for any given datapoint.

For each data point, I will be changing one of the two paths, thus changing the phase difference between the two signals. I will have to compile the phase difference measurements from hundreds of data points, so my method of measuring phase difference cannot be too time-intensive.

In my current setup, I can output each of the two signals via BNC, so I am hoping that the device or circuit I use takes two BNC inputs.

amnc11 - It sounds like you could implement this setup with a simple Phase Lock detector similar to a NE565 or LM565 integrated circuit or similar. Phase Lock Loop circuits evaluate two input signals and compare the phase angles between the two and create an output signal (sometimes called an error signal) that varies with the difference in phase angle. Traditionally, this 'difference' signal is used to drive a VCO (voltage controlled oscillator) that produces a synchronizing feedback signal. You might simply use the output of the phase detector to measure the difference between the two input signals. All this assumes you have some knowledge of using integrated circuits. You mentioned using an oscilloscope, so you sound proficient. The NE565 datasheet specs it's operation to 500 kHZ, so seems to be in range with your needs. The BNC connector is a common connector used in many coax cable to instrument setups, but not necessarily critical in this experiment. Here is a link to a block diagram.

http://www.circuitstoday.com/phase-locked-loop-ics

Rick Marz

amnc11 - Another approach you can use involves the oscilloscope. Instead of just inputting the two signals and looking at their respective 'y' values and noting phase shift differences on the horizontal axis, set the scope to X-Y input instead of the more familiar timebase setting for the horizontal sweep. As you are comparing phase-shift at the same frequency, the output will show the phase difference very easily. This phenomena is known as the Lissajous figure. You can find many references to explain it, here is one from the Wikipedia.

http://en.wikipedia.org/wiki/Lissajous_curve

Rick Marz

Thank you very much! That looks like a clever solution to my problem.

-Amnc11