From: pcdhSpamMeSenseless@electrooptical.net
On 01/21/2014 11:11 AM, haiticare2011@gmail.com wrote:
> On Monday, January 20, 2014 4:56:57 PM UTC-5, Phil Hobbs wrote:
>> On 01/20/2014 11:57 AM, haitica1@gmail.com wrote:
>>
>>> On Thursday, January 16, 2014 11:44:28 AM UTC-5, Phil Hobbs
>>> wrote:
>>
>>>> On 1/16/2014 11:35 AM, hai2011@gmail.com wrote:
>>> If you can actually get a MhZ-scale BW out of a generic Si PD, as
>>> my reading of your paper suggests, then could you get 100,000
>>> measurements per second done with a fast TIA and reasonably fast
>>> ADC? I wonder if that has any practical use?
>>
>> 100 kHz isn't fast--you can go way faster than that, if you have
>> enough light. I have a photodiode box whose 3 dB bandwidth is over
>> 30 GHz (a Tektronix SD-48).
>>
>> Of course that's in InGaAs, with very small diodes. Ordinary
>> silicon PIN diodes excel at having low capacitance, as low as 50
>> pF/cm**2, whereas InGaAs comes in about 100 times higher than
>> that.
>>
>> The down side is that the low capacitance comes from a very thick
>> depletion zone, so silicon diodes have much longer transit times,
>> and that's what limits their speed at low impedance. Large
>> diameter diodes are limited by the diffusion time in the epi, which
>> doesn't get depleted and hence doesn't have that helpful E field.
>> My fave BPW34 (Osram, not Vishay) tops out at around 10-20 ns rise
>> and fall times.
>>
>>
>> (In large diodes, transit time and capacitance both go like the
>> square of the diameter, but for different reasons.)
>>
>> Having enough light is really the key--that way you can use
>> smaller diodes with shorter transit time.
>
> All right, allow me to re-phrase that. Why would you want to have
> high speeds of photodetection, say mhz or above?
Well, the main reason is that what you're trying to measure is that
fast. Fast scanning measurements, streak cameras, communications,
pump-probe measurements, any number of things.
Another reason would be that you need some fancy filtering, and it's
easier to do digitally.
It can also help the SNR if you're sufficiently careful with the photon
budget. For instance, I did a bit of work last year for a small outfit
in New Mexico who were doing tunable diode laser spectroscopy in the
atmosphere, via heterodyne detection of sunlight. Sunlight is pretty
bright to the eye, but not nearly as bright as a laser beam, so any time
you try this you're always fighting a SNR war from the beginning. In
this sort of measurement, your signal is actually just a modulation of
the noise floor, so you win by increasing the measurement bandwidth.
They needed a photon budget and a laser noise canceller with a 100-MHz
bandwidth. (It was a success, and they're beginning to build the
systems commercially, I'm told. Coming soon to a nuclear enrichment
plant near you.)
The down side to making things faster is the current noise caused by the
diode capacitance, which in 1 Hz is
i_N = omega C_d e_NAmp,
where e_NAmp is the 1-Hz voltage noise of the amplifier or bootstrap.
Once this contribution takes over, the noise power increases as the cube
of the bandwidth, which gets ugly pretty fast. Having lots of light
means that you can use low noise bipolar amplifiers and smaller
photodiodes, so i_N goes down, and you also get more shot noise, so that
the amp noise is less important in a given bandwidth.
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
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
