ultra-wideband FM
in [Design]
|
Prev: Precision wirewound pot wiper bounce?
Next: retard mike shows us how retarded his mother is again
From: Joerg on 13 Jul 2010 16:10 John Larkin wrote: > On Tue, 13 Jul 2010 12:56:10 -0400, Phil Hobbs > <pcdhSpamMeSenseless(a)electrooptical.net> wrote: > >> John Larkin wrote: >>> Hi, >>> >>> One of the nasty things about cheap fiber-coupled lasers is that they >>> have terrible amplitude stability and linearity, full of mode jumps >>> and such. Given that, sending a signal over a fiberoptic link using >>> amplitude modulation is usually done with a stable CW laser feeding a >>> lithium-niobate modulator. The modulator itself is nonlinear and >>> expensive and a nuisance to drive and bias. >>> >>> Digitizing and sending samples is OK, up to a point. It gets messy at >>> some point from a sheer speed standpoint. >>> >>> So the idea of using FM pops up. If my baseband analog signal were, >>> say, DC to 150 MHz, and I picked the highest carrier center frequency >>> that's reasonably easy to work with, say 1 GHz, it could maybe be >>> done. The laser driver and receiver aren't too difficult. The issues >>> are the modulator, the demodulator, and the pure signal theory >>> necessary to turn the time-domain behavior of the link into classic >>> measures like s/n and distortion of the recovered baseband signal. >>> Asymmetrically bandlimiting an FM signal is computationally messy. >>> >>> I'd expect that commercial VCOs wouldn't have anything like this sort >>> of fractional modulation bandwidth. And if they did, a varicap >>> modulating an LC oscillator would probably distort like mad. (Faint >>> echoes of the capacitor charge debate?) The modulator may have to be >>> some EclipsLite version of a 555 on steroids. Or a multi-GHz VCO >>> heterodyned down. Yuk: sounds like RF. >>> >>> On the theory side, does anyone know of (or have?) one of the high-end >>> math tools that could do a quantitative signal-quality analysis of >>> such a link, given, say, approximate experimental data on the >>> time-domain behavior of the laser link? Hiring a consultant to do this >>> would be a desirable alternate to getting and learning this stuff >>> ourselves. >>> >>> Any thoughts? >>> >>> John >>> >>> >> Most Fabry-Perot type diode lasers will current-tune by a wave number or >> two (30-60 GHz) between mode hops, and can be modulated up to about 1 >> GHz. They're very sensitive to back-reflections--to be sure of good >> behaviour, you need at least 60 dB of optical isolation. >> >> Choosing the right temperature will help keep the mode hops at bay, but >> you'll have to keep tweaking T as the laser ages and the phase of the >> inevitable fibre feedback changes. >> >> DFB lasers, as used in telecom, are much better behaved, but modern ones >> hardly current-tune at all (like 100-300 MHz/mA). This is so that you >> can AM them in a dense WDM network without the resulting chirp >> scribbling all over the neighbouring channels. AM is a better idea with >> these guys--like a class B amplifier, they're nice and linear once you >> get above threshold, so some AC-coupled scheme should work fine. >> >> For current tuning, your best bet would be a FP laser in the 750-830 nm >> band, with some automatic scheme for avoiding mode hops. >> >> How are you planning on demodulating the FM? This usually needs an >> interferometer. >> >> Cheers >> >> Phil Hobbs >> >> (Stuck in Atlanta due to weather in NYC) > > I wasn't planning to modulate the frequency of the light. I was > thinking of driving the laser on/off at around 1 GHz, add FMing that. > > Rob thinks I should just digitize the signal and ship it 8b/10b. > If 10 bits is enough you could also just PWM, as someone had already suggested. -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: Tim Wescott on 13 Jul 2010 16:31 On 07/13/2010 08:59 AM, John Larkin wrote: > On Tue, 13 Jul 2010 08:43:13 -0700, Tim Wescott<tim(a)seemywebsite.com> > wrote: > >> On 07/13/2010 08:29 AM, John Larkin wrote: >>> >>> >>> Hi, >>> >>> One of the nasty things about cheap fiber-coupled lasers is that they >>> have terrible amplitude stability and linearity, full of mode jumps >>> and such. Given that, sending a signal over a fiberoptic link using >>> amplitude modulation is usually done with a stable CW laser feeding a >>> lithium-niobate modulator. The modulator itself is nonlinear and >>> expensive and a nuisance to drive and bias. >>> >>> Digitizing and sending samples is OK, up to a point. It gets messy at >>> some point from a sheer speed standpoint. >>> >>> So the idea of using FM pops up. If my baseband analog signal were, >>> say, DC to 150 MHz, and I picked the highest carrier center frequency >>> that's reasonably easy to work with, say 1 GHz, it could maybe be >>> done. The laser driver and receiver aren't too difficult. The issues >>> are the modulator, the demodulator, and the pure signal theory >>> necessary to turn the time-domain behavior of the link into classic >>> measures like s/n and distortion of the recovered baseband signal. >>> Asymmetrically bandlimiting an FM signal is computationally messy. >>> >>> I'd expect that commercial VCOs wouldn't have anything like this sort >>> of fractional modulation bandwidth. And if they did, a varicap >>> modulating an LC oscillator would probably distort like mad. (Faint >>> echoes of the capacitor charge debate?) The modulator may have to be >>> some EclipsLite version of a 555 on steroids. Or a multi-GHz VCO >>> heterodyned down. Yuk: sounds like RF. >>> >>> On the theory side, does anyone know of (or have?) one of the high-end >>> math tools that could do a quantitative signal-quality analysis of >>> such a link, given, say, approximate experimental data on the >>> time-domain behavior of the laser link? Hiring a consultant to do this >>> would be a desirable alternate to getting and learning this stuff >>> ourselves. >>> >>> Any thoughts? >> >> Frequency modulate what? > > A carrier. With the baseband signal. That's how FM is usually done. Yes it is. I just wanted you to confirm that you weren't thinking of modulating the color of the laser and calling it "FM". It's theoretically possible, although it would probably be insanity to attempt with today's lasers. >> >> I assume you're going to frequency modulate the 1GHz carrier, then >> amplitude modulate the laser with that -- correct? > > Yup. The laser would actually run on/off at the (modulated) carrier > frequency. > >> >> I think the most important "high-end" math tool in this case is the >> consultant's brain -- Scilab you can get off the web for free and go buy >> a nice car with the money you would have spent to buy Matlab. With >> either Scilab or Matlab you still need a nice squishy pile of neurons >> that knows how to feed in the questions the right way and interpret the >> results. > > Hence the option to have a consultant, a real RF signals guy familiar > with the tools, furnish the neurons. It all sounds interesting. I assume (well, gather from other posts in the thread) the ultimate goal is to be able to send nice clean analog signals while dodging any need to digitize? IFAIK this is done: there are (or used to be) off-the-shelf fiber-optic links that you can stick analog video into and get analog video out. I vaguely remember a hand-waving explanation that included mention of a high-frequency FM signal that then amplitude modulated the light. Are you trying to replicate one of these, only cheaper? Ditto, only better? Reinventing the wheel because it feels good? -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
From: Vladimir Vassilevsky on 13 Jul 2010 16:44 Joerg wrote: > John Larkin wrote: > >>>>One of the nasty things about cheap fiber-coupled lasers is that they >>>>have terrible amplitude stability and linearity, >>>>So the idea of using FM pops up. >> >>Rob thinks I should just digitize the signal and ship it 8b/10b. Rob is right. > > If 10 bits is enough you could also just PWM, as someone had already > suggested. For ~10 bit accuracy, the PWM rate must be ~20 times higher then the highest frequency of the signal. Generating 10-bit linear ramp at 2 GHz is nontrivial. (This accounts for negative feedback in transmitter. Without NFB, the results are going to be several times worse). Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
From: Joerg on 13 Jul 2010 16:45 Tim Wescott wrote: > On 07/13/2010 08:59 AM, John Larkin wrote: >> On Tue, 13 Jul 2010 08:43:13 -0700, Tim Wescott<tim(a)seemywebsite.com> >> wrote: >> >>> On 07/13/2010 08:29 AM, John Larkin wrote: >>>> >>>> >>>> Hi, >>>> >>>> One of the nasty things about cheap fiber-coupled lasers is that they >>>> have terrible amplitude stability and linearity, full of mode jumps >>>> and such. Given that, sending a signal over a fiberoptic link using >>>> amplitude modulation is usually done with a stable CW laser feeding a >>>> lithium-niobate modulator. The modulator itself is nonlinear and >>>> expensive and a nuisance to drive and bias. >>>> >>>> Digitizing and sending samples is OK, up to a point. It gets messy at >>>> some point from a sheer speed standpoint. >>>> >>>> So the idea of using FM pops up. If my baseband analog signal were, >>>> say, DC to 150 MHz, and I picked the highest carrier center frequency >>>> that's reasonably easy to work with, say 1 GHz, it could maybe be >>>> done. The laser driver and receiver aren't too difficult. The issues >>>> are the modulator, the demodulator, and the pure signal theory >>>> necessary to turn the time-domain behavior of the link into classic >>>> measures like s/n and distortion of the recovered baseband signal. >>>> Asymmetrically bandlimiting an FM signal is computationally messy. >>>> >>>> I'd expect that commercial VCOs wouldn't have anything like this sort >>>> of fractional modulation bandwidth. And if they did, a varicap >>>> modulating an LC oscillator would probably distort like mad. (Faint >>>> echoes of the capacitor charge debate?) The modulator may have to be >>>> some EclipsLite version of a 555 on steroids. Or a multi-GHz VCO >>>> heterodyned down. Yuk: sounds like RF. >>>> >>>> On the theory side, does anyone know of (or have?) one of the high-end >>>> math tools that could do a quantitative signal-quality analysis of >>>> such a link, given, say, approximate experimental data on the >>>> time-domain behavior of the laser link? Hiring a consultant to do this >>>> would be a desirable alternate to getting and learning this stuff >>>> ourselves. >>>> >>>> Any thoughts? >>> >>> Frequency modulate what? >> >> A carrier. With the baseband signal. That's how FM is usually done. > > Yes it is. I just wanted you to confirm that you weren't thinking of > modulating the color of the laser and calling it "FM". It's > theoretically possible, although it would probably be insanity to > attempt with today's lasers. > Guilty. Done it, modulated the wavelength. But not at anywhere close to the speed John would need although I wouldn't see what would have prevented that if using a fat RF BJT or an LDMOS. >>> >>> I assume you're going to frequency modulate the 1GHz carrier, then >>> amplitude modulate the laser with that -- correct? >> >> Yup. The laser would actually run on/off at the (modulated) carrier >> frequency. >> >>> >>> I think the most important "high-end" math tool in this case is the >>> consultant's brain -- Scilab you can get off the web for free and go buy >>> a nice car with the money you would have spent to buy Matlab. With >>> either Scilab or Matlab you still need a nice squishy pile of neurons >>> that knows how to feed in the questions the right way and interpret the >>> results. >> >> Hence the option to have a consultant, a real RF signals guy familiar >> with the tools, furnish the neurons. > > It all sounds interesting. I assume (well, gather from other posts in > the thread) the ultimate goal is to be able to send nice clean analog > signals while dodging any need to digitize? > I think Paul's suggestion to do PWM was right on. John, IMHO math tools are less important here. What I feel is more important is to find the parts that can do the job with enough margins. The only tool I used on my last laser project was an HP11C. And that wasn't even mine, technically it belongs to my wife, I just confiscated it around 15 years ago :-) > IFAIK this is done: there are (or used to be) off-the-shelf fiber-optic > links that you can stick analog video into and get analog video out. I > vaguely remember a hand-waving explanation that included mention of a > high-frequency FM signal that then amplitude modulated the light. > I've seen FM ones but those are 10MHz BW, or less, usually. [...] -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: Vladimir Vassilevsky on 13 Jul 2010 16:56
Joerg wrote: > I think Paul's suggestion to do PWM was right on. It depends. John didn't post any requirements except bandwidth. > John, IMHO math tools are less important here. Math is very important there, as it will clearly illustrate the limitations of PWM or FM. > What I feel is more important is to find the > parts that can do the job with enough margins. Besides the parts nonideality, there are fundamental problems. > > I've seen FM ones but those are 10MHz BW, or less, usually. Well, analog VCRs record FM signal to tape (THD ~ 1%, SNR ~40dB). You only have to do x50 frequency upscaled version :-) VLV |