@Vincent Lecoeuche

Thanks for your reply! You mentioned scan but I just left the continue wave laser on and let the light enters into the system. What do you mean by the ‘spectral scan speed’? Do you mean the wavelength/frequency instability of the source? The wavelength/frequency stability is typical 2.5pm/0.3GHz in an hour according to the user guide, so I don’t think it’s the reason.

You also mentioned temperature changes. According to my calculation, the optical length of the cavity changes nearly 4.5um/K (I took refractive index temperature coefficient and linear thermal expansion into consideration), which means about 0.3k variation of temperature may result in a length change by one wavelength. It seems reasonable, but I placed a heating plate (it’s absolutely stronger than my radiation) under the cavity, there was no sign that the shifting resonant dips getting dense or sparse.

Like VIncent, I suspect drift of the laser or fibre cavity.  Have you measured or calculated the free spectral range and line width of the resonance?

Temperature coefficient of refractive index plus linear expansion of silica gives a net temperature coefficient of roughly 9e-6 K-1, so I estimate your cavity length is about 0.5 m, and free spectral range approximately 0.4 GHz.  https://www.researchgate.net/post/How_should_I_make_the_PZT_driven_fiber_phase_modulation_stable_in_fiber_optical_resonator#view=58bc61ab93553b259102dea0

Could your laser drift by a few MHz or  tens of MHz over the period of observation?

Another possibility is that there is some birefringence in the resonator cavity.  Could change in birefringence of the cavity, or drift in the launch polarisation be responsible?

@Vincent Lecoeuche

It’s my fault that I didn’t make it clear.

I’ve measured the cavity length, which is 2m. The laser wavelength is set to 1550nm and the display of oscilloscope is as below. The time span of the whole screen is 1s.

The resonant dip’s appearance is totally random.

Greeting, @ Alan Robinson.

I haven’t measured the FSR and line width of the resonance. But you remind me that, the estimated FSR should be 0.1GHz, (sorry, the accurate length of cavity is 2m), but the frequency instability is 0.3GHz. Does it mean this source can’t be used in my experiment?

I’ve posted the display on oscilloscope, and I tried to bend the fiber in different radius but it seems make no sense.

Besides, how should I drift the laser frequency? Most of resonant cavity experiments involved frequency scanning method, though.

@Vincent Lecoeuche

Thanks for your suggestions. I’ll try to lock the laser frequency to the cavity.

@Chenge Wang

What are you trying to measure in your experiment?  The frequency stability needed will depend on what you hope to achieve, and how fast the wanted signal changes, compared with the laser drift and noise.

What type of laser are you using?  Semiconductor lasers are often tuned by changing the temperature or the current.  For one type of DFB (distributed feedback) laser, I measured tuning rates around 12 GHz / K and 0.6 GHz / mA.  External cavity lasers may have lower tuning rates.

Depending on the splice losses and polarisation control in your resonator, your resonance bandwidth could be less than 1 MHz, in which case 1 μA fluctuation in drive might cause the spikes you see.  I cannot tell from your oscilloscope traces whether you suffer from short noise spikes where you are biased near the edge of a relatively broad transmission dip, or have slower frequency fluctuations which sweep through a narrow resonance.

Current tuning is convenient if you only need to tune the laser over a few hundred MHz or less. Applying a low amplitude sinusoidal current dither can be useful if you wish to lock onto the cavity resonance, as Vincent suggests. Phase sensitive detection at either the fundamental or the 3rd harmonic of the dither frequency will find the peak. A second harmonic lock is useful for operation where the rate of change of signal with phase is greatest.

@Alan Robinson

I just want to figure out the reason of the phenomenon. In my expectation, without any modulation, the display on oscilloscope should be a stable line.

The laser I’m using is Keysight N7714, a tunable laser source. According to your suggestions, I’ll check the influence of drive current, and try to lock the frequency to the cavity resonance.

Thanks for your reply again!

Hi it could be due to some back reflections (due to connections ..)?

@Letizia DE MARIA

Thanks for your reply!

It is an all-fiber system and I think the splice connection of single-mode fibers should not bring too much back reflections.

But I suspected this factor earlier before, so I placed an isolator after the source and another in the resonator, it didn’t work.

@Chenge Wang "But I suspected this factor earlier before, so I placed an isolator after the source and another in the resonator, it didn’t work."

With 99%/1% coupler split ratio, I would expect the insertion loss of the isolator to make a significant difference to the width and depth of the transmission dip when you scan through resonance.  If you see exactly the same transmission dips with an extra 0.5 dB attenuation in the loop, the problem might not be frequency stability of the laser.  It could be worth checking whether you have a stable output without the resonator, or with the loop open, and ports 2 and 4 terminated.

Have you tried fine-tuning the laser frequency to estimate the linewidth?  Alternatively, can you apply a slow FM modulation over the 100 MHz free spectral range?  It looks as if either might be an option from the N7714A data sheet?  http://www.keysight.com/main/redirector.jspx?action=obs&nid=927328.3.00&lc=eng&cc=GB&ckey=1864084&pubno=5990-5512EN<ype=LitStation&ctype=EDITORIAL&ml=eng

Could the problem be related to polarisation instability in the connection between laser source and resonator.  Ideally, you should align the input polarisation with one of the polarisation eigenmodes of the resonator loop.

The optical path difference of 1-3 and 1-4-2-3 may be the key.  Eliminating the resonator can also be possible with the path length difference longer than the coherence length.

@Alan Robinson

I continually observed the transmissions dips of the system with/without inserting the isolator over 4h yesterday, it seems that there’s no change between them. Does it mean that the frequency stability is not the reason?

 Besides, I cutoff the connection between port 2 and port 4, and check the output of port 2,3,4, all of them present stable line, no resonant dips.

 I tried to apply a FM modulation on my laser so I checked the user-guide of it, but it states this function is not supported. So sad :(

 Actually, what I’m constructing is the system mentioned in this paper (Stokes L F, Chodorow M, Shaw H J. All-single-mode fiber resonator[J]. Optics Letters, 1982, 7(6): 288-290.). I haven’t used polarization maintaining fibers yet, because I think single mode fiber can also be qualified according to that reference.

https://www.osapublishing.org/ol/abstract.cfm?uri=ol-7-6-288

@Diana Tsou

Thanks for your reply! I agree with your point, so I’m seeking the reason causing this “random difference”. The linewidth of the laser which I used is about 100KHz, so the coherence length should be >1km. I don’t think the path length difference can be competitive.

My hypothesis is that your observation was resulted from optical destructive interference.  Expectation with using PMF instead of SMF would be more pronounced of dips and environment sensitive.  I am curious if you have come to this point experimentally.

@Diana Tsou

Thanks for your suggestion! Recently I’ve been busy with my thesis. I will perform the experiment with PMF and PMF couplers then. If it is proved that SMF introduced destructive interference, I’ll update the test result in this post.