Noise from Locking circuit
You mean with a laser as a frequency source? The noise comes about because you are beating the laser with a time delayed version of itself. The laser is jumping all over the place in frequency, probably +/- many GHz in frequency, so the output of the mixer will be some average DC term with lots of higher frequency junk on top of it. You lowpass that junk completely, and then just use the filtered DC output.
I agree with your solution biff, but not necessarily with the the problem - single frequency, narrow linewidth research lasers can have coherence lengths of many kilometres and thus won't contribute much towards noise within the locking servo loop bandwidth. For the bog-standard diode laser though, I'd whole-heartedly agree with you :)
Aye kikki, low pass filtering of the mixer output will go a *long way* towards improving PDH locking signals! At the very least, your mixer output will contain output frequencies of the sum and difference (which is the DC component you're after) of your modulation frequencies. While the sum term (at twice your Pound-Drever modulation frequency) will have a zero-mean value, it's magnitude will be significant and I've seen it wreak havoc with the loop stability if your modulation frequency is comparable to the locking loop bandwidth. A low pass filter after your mixer will greatly improve things.
Since you're using a minicircuits mixer, also be aware that the signal level at the IF output port will be *tiny* ... in the order of 10's to ~100 mV (if you're not overdriving the mixer).. You'll also probably need a stage of low noise amplification before/after filtering. Be aware of DC offsets of any active amplifier/filter stages, since these can swamp the tiny (desired) DC error signals if you're not careful - and in a temperature dependant fashion if you're unlucky ;)
Good luck!
I have not played with lasers in quite some time so you will have to enlighten me. I am still thinking that lasers line widths are like this:
Tunable laser - Wikipedia, the free encyclopedia
"In most lasers, this linewidth is quite narrow (for example, the 1064 -nm wavelength transition of a Nd:YAG laser has a linewidth of approximately 120 GHz, corresponding to a 0.45 -nm wavelength range"
On top of that, there is a general RIN noise floor in this method that will definately pop out of that mixer at something like -110 dBc/Hz from DC to a GHz or so.
The filter cavity bandwidth is probaby relatively large, which is a saving grace, so the group delay of the reflected port is probably < 10 nS? So a long coherence time will help somewhat.
Heya biff - good point about the RIN noise floor... I'd forgotten all about that one - I was always lucky that I was either working way above (or below) the dominant noise spur (which I guess was quite deliberate, come to think of it! ;)
Modern laser stability (can be) quite remarkable - I used a laser similar to this for years: http://www.npphotonics.com/files/pro...ower_Laser.pdf - which features an incredible 5 kHz linewidth! While I was absorbed with fixed frequency lasers, it seems that wideband tunable lasers also went through some amazing developements... I just stumbled across this one while searching - http://www.dmphotonics.com/Tekhnosca..._Photonics.pdf ... 15 kHz linewidth and 100 nm tunability? Wow!
Yes that is an impressive laser. I was assumiing something more like a DFB diode laser, where you can even get a mode hop. But, until we hear from the OP, i prefer to not speculate further. He may not even be talking about an optical system.