Unknown large linewidth of the fringes of a cavity

[kelly0303]

Hello! I am building a bow-tie cavity and for some reason the peaks I see when scanning the laser frequency are extremely wide and with a very long tail to the right. However, if I put a 50 Ohm resistance in between the BNC cable from the diode and the oscilloscope (using a BNC T connector) the peaks become narrow (as narrow as I would expect from the calculations) and the tail is significantly reduced (although not gone completely). Why is this happening? I expected that adding the 50 Ohm would reduce the amplitude (which happens, too), but why the linewidth? The issue is that now the amplitude is so much reduced that it is too small to send it to the servo, and if I don't use the 50 Ohm I have the original problem. Any advice would be appreciated. Thank you!

 

[Baluncore]

It seems you have a resonant length of cable that is not correctly terminated.
How long is the coax cable between the diode and the oscilloscope?
Please provide a diagram.

 

[Baluncore]

Maybe the oscilloscope input is set to 1 MΩ or 10 MΩ input impedance. Current from the diode charges the coax cable capacitance, then it takes too long to discharge.

 

[kelly0303]

Maybe the oscilloscope input is set to 1 MΩ or 10 MΩ input impedance. Current from the diode charges the coax cable capacitance, then it takes too long to discharge.

Thank you for your reply. The cable is about 50 cm long (normal BNC to BNC cable, also tried a few), I am not sure what diagram you mean.

The oscilloscope is indeed in the 1M$\Omega$ impedance setup, but that is the impedance input to my servo, too (and I can't change it there). I was hoping to find a way for it to work with this impedance settings.

 

[Baluncore]

If the diode is a current source, then you need to convert the current to a voltage for display and input to the servo. That should be done with a resistor, with or without a buffer amplifier. A diode detector current source has an infinite impedance, that will charge the coaxial line capacitance, which makes a low-pass filter when paralleled with the oscilloscope. The exponential decay recovery of the circuit gives it the long tail.

Try replacing the 50 ohm termination on the 'T' connector with a BNC having an R = 1k, or 10 k, resistor across the signal. R = signal voltage / signal current.

1. What type of diode detector do you use?
2. What signal current do you expect?
3. What type of 50 ohm coaxial cable is used?

4. What is the minimum input voltage to the servo detector?
5. To resolve the real line width, what response recovery time do you require?

 

[Twigg]

I had this same issue with a different kind of cavity, and solved it by adding a variable-gain voltage amplifier after the photodetector. (The "photodetector" in my case was a standard commercial photodiode + transimpedance amplifier in a single package.) I increase the gain until the servo locks well and I see good signal-to-noise on the error signal, but I make sure that the gain isn't so high that I see a non-linear response to changes in the optical power.

In my case, I used a SR560 voltage preamplifier from Stanford Research Systems. It's seriously overkill, and I used it because I had it on hand and couldn't be bothered to find anything else. You could make a much cheaper solution with any decent op amp and a potentiometer. If you go that route, make sure to keep your amplifier's output impedance well below the 1MOhm (or whatever) input impedance of your oscilloscope.