Forums

Frequency Detection Suggestions

Started by Bradley Kimbrough March 8, 2017
Hello All,

I am currently building a stabilized Helium Neon laser that uses a 3 mode tube and stabilizes the optical frequency by monitoring a beat frequency between the modes. The beat frequency is 450-600 MHz and is sensed from the waste beam at the back of the laser tube. For a prototype I used an amplified moderately high speed Si detector that fed into a 12 bit 125 MSPS DAC.  This data was then sent to the computer , bandpass filtered and Fourier transformed to determine the signal frequency. This frequency was then fed into a PID algorithm whose output controlled a heater that would heat the tube to stabilize its length.

Question:  Can you suggest and analog solution to frequency detection that could work in this scenario?  It does not need to be exceedingly fast since the PID loop time constant is on the order of seconds.  I've noticed that Analog Devices has an IC for Freq to Voltage conversion.  A PLL may be useful as well.  Not looking for specifics, just suggestions as to the best path forward.

Thank you,

Brad Kimbrough
On Wed, 8 Mar 2017 17:16:27 -0800 (PST), Bradley Kimbrough
<bradley.kimbrough@gmail.com> wrote:

>Hello All, > >I am currently building a stabilized Helium Neon laser that uses a 3 mode tube and stabilizes the optical frequency by monitoring a beat frequency between the modes. The beat frequency is 450-600 MHz and is sensed from the waste beam at the back of the laser tube. For a prototype I used an amplified moderately high speed Si detector that fed into a 12 bit 125 MSPS DAC. This data was then sent to the computer , bandpass filtered and Fourier transformed to determine the signal frequency. This frequency was then fed into a PID algorithm whose output controlled a heater that would heat the tube to stabilize its length. > >Question: Can you suggest and analog solution to frequency detection that could work in this scenario? It does not need to be exceedingly fast since the PID loop time constant is on the order of seconds. I've noticed that Analog Devices has an IC for Freq to Voltage conversion. A PLL may be useful as well. Not looking for specifics, just suggestions as to the best path forward. > >Thank you, > >Brad Kimbrough
You could divide the frequency down to something that an HC7046 PLL chip can handle, a few MHz maybe. Apply a clock as a reference. It might not actually phase lock, but the frequency detector part will work. -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
 
> You could divide the frequency down to something that an HC7046 PLL > chip can handle, a few MHz maybe. Apply a clock as a reference. It > might not actually phase lock, but the frequency detector part will > work. > > > -- > > John Larkin Highland Technology, Inc > picosecond timing precision measurement > > jlarkin att highlandtechnology dott com > http://www.highlandtechnology.com
John, Thanks for your input. I however gave you the wrong units. The frequency range is 400 to 650 kHz, not MHz. Regards, Brad Kimbrough
On Wed, 8 Mar 2017 17:37:14 -0800 (PST), Bradley Kimbrough
<bradley.kimbrough@gmail.com> wrote:

> >> You could divide the frequency down to something that an HC7046 PLL >> chip can handle, a few MHz maybe. Apply a clock as a reference. It >> might not actually phase lock, but the frequency detector part will >> work. >> >> >> -- >> >> John Larkin Highland Technology, Inc >> picosecond timing precision measurement >> >> jlarkin att highlandtechnology dott com >> http://www.highlandtechnology.com > >John, > >Thanks for your input. > >I however gave you the wrong units. The frequency range is 400 to 650 kHz, not MHz. > >Regards, > >Brad Kimbrough
Then you needn't divide. A v/f converter would work too, and an analog control loop. If you have a clean signal, a one-shot does v/f cheap. -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
On Thursday, March 9, 2017 at 1:07:17 PM UTC+11, John Larkin wrote:
> On Wed, 8 Mar 2017 17:37:14 -0800 (PST), Bradley Kimbrough > <bradley.kimbrough@gmail.com> wrote: > > > > >> You could divide the frequency down to something that an HC7046 PLL > >> chip can handle, a few MHz maybe. Apply a clock as a reference. It > >> might not actually phase lock, but the frequency detector part will > >> work. > >> > >> > >> -- > >> > >> John Larkin Highland Technology, Inc > >> picosecond timing precision measurement > >> > >> jlarkin att highlandtechnology dott com > >> http://www.highlandtechnology.com > > > >John, > > > >Thanks for your input. > > > >I however gave you the wrong units. The frequency range is 400 to 650 kHz, not MHz. > > Then you needn't divide. > > A v/f converter would work too, and an analog control loop. > > If you have a clean signal, a one-shot does v/f cheap.
Unfortunately the beat signal is a frequency, and needs to be turned into a voltage to drive the heater that controls the length of the tube. The problem is that slope of the voltage to frequency calibration line depends on at least one resistor and at least one capacitor, both of which have tolerances. If you know what beat frequency you want, and can generate it as accurately as you need (which shouldn't be difficult) you can use 4046-style phase and frequency detection hardware to compare the beat frequency with the desired reference frequency, and use the output to control the heater. http://assets.nexperia.com/documents/data-sheet/74HCT9046A.pdf doesn't have the dead spot at small phase offsets which can complicate the use of regular 4046-stye phase and frequency comparison outputs. Phase-locked loops can work fine with pure "product" - phase-only - detectors, but they tend to start up rather slowly. Floyd M. Gardner's "Phaselock Techniques" ISBN 0-471-04294-3 is old - 1979 - but does cover the basics. -- Bill Sloman, Sydney
On 03/08/2017 08:16 PM, Bradley Kimbrough wrote:
> Hello All, > > I am currently building a stabilized Helium Neon laser that uses a 3 > mode tube and stabilizes the optical frequency by monitoring a beat > frequency between the modes. The beat frequency is 450-600 MHz and > is sensed from the waste beam at the back of the laser tube. For a > prototype I used an amplified moderately high speed Si detector that > fed into a 12 bit 125 MSPS DAC. This data was then sent to the > computer , bandpass filtered and Fourier transformed to determine > the signal frequency. This frequency was then fed into a PID > algorithm whose output controlled a heater that would heat the tube > to stabilize its length. > > Question: Can you suggest and analog solution to frequency > detection that could work in this scenario? It does not need to be > exceedingly fast since the PID loop time constant is on the order of > seconds. I've noticed that Analog Devices has an IC for Freq to > Voltage conversion. A PLL may be useful as well. Not looking for > specifics, just suggestions as to the best path forward. > > Thank you, > > Brad Kimbrough >
We seem to have two threads going on this topic. It would be worth picking just one. From my followup to the other one:
> Interesting--I wouldn't have thought that a very well-posed problem. > How do you correct for frequency pulling due to the natural line > width and spectral hole burning? > > (The HP one was a single mode Zeeman-split laser where the beat > frequency between the two polarizations was phase-locked to a 2-MHz > crystal oscillator.)
I'd amplify this by saying that a barefoot HeNe is a much better frequency standard than the beat frequency. For one thing, the laser line is c/632.8 nm = 474 THz so you're effectively trying to frequency-multiply by a factor of a million or so, which will multiply the phase noise by 120 dB. For another, adjacent modes in a HeNe get pulled around by hundreds of kilohertz just by temperature gradients and stuff like that. You can see this by shining a normal 5 mW cylindrical HeNe into a photodiode and looking at the output on a spectrum analyzer. If you put your hand on one end of the case, you'll break the locking between adjacent modes, and you'll get spurs sweeping back and forth between ~200 kHz and 1 MHz. These are fourth-order mixing products, specifically the beat between one mode and a third-order IM product of two others. This shows that the mode spacing is uncertain at the level of hundreds of ppm. Multiplying that up by a million times would get you a frequency uncertainty in the hundreds of gigahertz, which is far, far worse than the stability of a free-running HeNe, even if it could oscillate over that wide a range, which it can't. What are you trying to do that needs better stability than a free-running HeNe? Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
> > > > Question: Can you suggest and analog solution to frequency > > detection that could work in this scenario? It does not need to be > > exceedingly fast since the PID loop time constant is on the order of > > seconds. I've noticed that Analog Devices has an IC for Freq to > > Voltage conversion. A PLL may be useful as well. Not looking for > > specifics, just suggestions as to the best path forward. >
Use the phase / frequency digital detector circuit in any modern PLL chip. Supply it your variable frequency and a suitable reference frequency and a small analog filter with op amp and it will give you a control voltage with any range you choose. m
Go over to Sam's laser FAQ, find the commercial stabilized laser section. Then take a look at the details for the "Laboratory for Science"  Model 220. If you look at the patents and description you should find everything you need except for the loop filter constants.

Steve 
Sam's Laser FAQ references US patent 4,468,773  for the main control loop, which gives you a pretty good block diagram and waveforms for the controller..

J.L. Hall of the University of Colorado also did some work with HENEs and F->V converters.

Steve 
On Thursday, March 9, 2017 at 7:29:59 AM UTC-7, Phil Hobbs wrote:
> On 03/08/2017 08:16 PM, Bradley Kimbrough wrote: > > Hello All, > > > > I am currently building a stabilized Helium Neon laser that uses a 3 > > mode tube and stabilizes the optical frequency by monitoring a beat > > frequency between the modes. The beat frequency is 450-600 MHz and > > is sensed from the waste beam at the back of the laser tube. For a > > prototype I used an amplified moderately high speed Si detector that > > fed into a 12 bit 125 MSPS DAC. This data was then sent to the > > computer , bandpass filtered and Fourier transformed to determine > > the signal frequency. This frequency was then fed into a PID > > algorithm whose output controlled a heater that would heat the tube > > to stabilize its length. > > > > Question: Can you suggest and analog solution to frequency > > detection that could work in this scenario? It does not need to be > > exceedingly fast since the PID loop time constant is on the order of > > seconds. I've noticed that Analog Devices has an IC for Freq to > > Voltage conversion. A PLL may be useful as well. Not looking for > > specifics, just suggestions as to the best path forward. > > > > Thank you, > > > > Brad Kimbrough > > > We seem to have two threads going on this topic. It would be worth > picking just one. > > From my followup to the other one: > > > Interesting--I wouldn't have thought that a very well-posed problem. > > How do you correct for frequency pulling due to the natural line > > width and spectral hole burning? > > > > (The HP one was a single mode Zeeman-split laser where the beat > > frequency between the two polarizations was phase-locked to a 2-MHz > > crystal oscillator.) > > I'd amplify this by saying that a barefoot HeNe is a much better > frequency standard than the beat frequency. For one thing, the laser > line is > > c/632.8 nm = 474 THz > > so you're effectively trying to frequency-multiply by a factor of a > million or so, which will multiply the phase noise by 120 dB. > > For another, adjacent modes in a HeNe get pulled around by hundreds of > kilohertz just by temperature gradients and stuff like that. You can > see this by shining a normal 5 mW cylindrical HeNe into a photodiode and > looking at the output on a spectrum analyzer. If you put your hand on > one end of the case, you'll break the locking between adjacent modes, > and you'll get spurs sweeping back and forth between ~200 kHz and 1 MHz. > These are fourth-order mixing products, specifically the beat between > one mode and a third-order IM product of two others. This shows that > the mode spacing is uncertain at the level of hundreds of ppm. > > Multiplying that up by a million times would get you a frequency > uncertainty in the hundreds of gigahertz, which is far, far worse than > the stability of a free-running HeNe, even if it could oscillate over > that wide a range, which it can't. > > What are you trying to do that needs better stability than a > free-running HeNe? > > Cheers > > Phil Hobbs > > -- > Dr Philip C D Hobbs > Principal Consultant > ElectroOptical Innovations LLC > Optics, Electro-optics, Photonics, Analog Electronics > > 160 North State Road #203 > Briarcliff Manor NY 10510 > > hobbs at electrooptical dot net > http://electrooptical.net
Phil, I appreciate your feedback. Here is a link to a good reference paper on the topic of three mode HeNe stabilization. https://www.vsl.nl/en/about-vsl/publications/frequency-stabilized-hene-gas-laser-35-mw-single-mode The goal is increased power from a stabilized HeNe laser. For me, small, slow drift is fine, I'm not as interested in the absolute frequency Thank you, Brad