On Monday, May 22, 2017 at 3:12:32 PM UTC-4, Phil Hobbs wrote:> On 05/22/2017 10:33 AM, George Herold wrote: > > On Sunday, May 21, 2017 at 8:10:45 PM UTC-4, Winfield Hill wrote: > >> Cursitor Doom wrote... > >>> > >>> Jim Thompson wrote: > >>> > >>>> AKA Flicker Noise > >>> > >>> Yup, pink noise. No jittering/jitterbugging involved. ;) > >> > >> Actually, I'd say you're wrong. The zener noise that we > >> see in avalanche mode (greater then 7 volts or so) is due > >> to step-wise "microplasma" ns-scale changes in current. > >> A decade or so ago we fully discussed it here on s.e.d., > >> complete with measurements, waveforms, detailed physics > >> paper references and the works. Nailed it down. As it > >> happens, the jitterbug analogy isn't so far off. > > > > Those are some nice threads, I went back and read them a few years > > ago. Turning lemons into lemonade, a 20V zener baised at ~10uA > > makes a nice noise source out to ~100's of kHz. > > > > I've also been reading Hamamatsu app notes on their Silicon PMT. > > Which is really a Si avalanche photodiode in "Geiger-mode" > > (biased above the breakdown voltage.) Fun stuff. > > > > George H. > > I'm in conversation with a big biosci company about cost-reducing their > detection subsystems, so this has come up a lot lately. "Si PMTs" are > basically an array of APDs wired in parallel and run in Geiger mode. > The idea is that having multiple pixels allows you to extend Geiger mode > to pulses containing more photons without losing resolution. > > AFAIK the biggest ones are about 120x120 pixels. Trouble is, even when > uniformly illuminated they start to become nonlinear at depressingly low > photon numbers--the linearity error is 10% once 10% of the pixels have > fired, because only 90% are available. > > You can apply a correction for this that might work adequately up to > about 25%, but it starts to run out of gas pretty fast compared with a > real PMT. Their dynamic range requirements were much too large for this > to work for them. > > Also the dark counts are ridiculously worse--more than 10**6 times worse > on an area basis. > > The real advantage of MPPCs is that you can do thresholding, which helps > reject single-pixel events at the expense of sensitivity. That reduces > the effective dark count rate by a lot.The only app I would have at the moment is doing Bell inequality experiments with ~800 nm correlated photons. (I don't really know all the details of the experiment.) So I don't think I'd care about linearity. All the small pixels means smaller gain, as each photon triggers a smaller capacitance discharge. George H.> > 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
Combating Zener Jitter
Started by ●May 21, 2017
Reply by ●May 22, 20172017-05-22
Reply by ●May 23, 20172017-05-23
On Mon, 22 May 2017 15:57:13 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>On 05/22/2017 03:48 PM, John Larkin wrote: >> On Mon, 22 May 2017 15:12:20 -0400, Phil Hobbs >> <pcdhSpamMeSenseless@electrooptical.net> wrote: >> >>> On 05/22/2017 10:33 AM, George Herold wrote: >>>> On Sunday, May 21, 2017 at 8:10:45 PM UTC-4, Winfield Hill wrote: >>>>> Cursitor Doom wrote... >>>>>> >>>>>> Jim Thompson wrote: >>>>>> >>>>>>> AKA Flicker Noise >>>>>> >>>>>> Yup, pink noise. No jittering/jitterbugging involved. ;) >>>>> >>>>> Actually, I'd say you're wrong. The zener noise that we >>>>> see in avalanche mode (greater then 7 volts or so) is due >>>>> to step-wise "microplasma" ns-scale changes in current. >>>>> A decade or so ago we fully discussed it here on s.e.d., >>>>> complete with measurements, waveforms, detailed physics >>>>> paper references and the works. Nailed it down. As it >>>>> happens, the jitterbug analogy isn't so far off. >>>> >>>> Those are some nice threads, I went back and read them a few years >>>> ago. Turning lemons into lemonade, a 20V zener baised at ~10uA >>>> makes a nice noise source out to ~100's of kHz. >>>> >>>> I've also been reading Hamamatsu app notes on their Silicon PMT. >>>> Which is really a Si avalanche photodiode in "Geiger-mode" >>>> (biased above the breakdown voltage.) Fun stuff. >>>> >>>> George H. >>> >>> I'm in conversation with a big biosci company about cost-reducing their >>> detection subsystems, so this has come up a lot lately. "Si PMTs" are >>> basically an array of APDs wired in parallel and run in Geiger mode. >>> The idea is that having multiple pixels allows you to extend Geiger mode >>> to pulses containing more photons without losing resolution. >>> >>> AFAIK the biggest ones are about 120x120 pixels. Trouble is, even when >>> uniformly illuminated they start to become nonlinear at depressingly low >>> photon numbers--the linearity error is 10% once 10% of the pixels have >>> fired, because only 90% are available. >>> >>> You can apply a correction for this that might work adequately up to >>> about 25%, but it starts to run out of gas pretty fast compared with a >>> real PMT. Their dynamic range requirements were much too large for this >>> to work for them. >>> >>> Also the dark counts are ridiculously worse--more than 10**6 times worse >>> on an area basis. >>> >>> The real advantage of MPPCs is that you can do thresholding, which helps >>> reject single-pixel events at the expense of sensitivity. That reduces >>> the effective dark count rate by a lot. >>> >>> Cheers >>> >>> Phil Hobbs >> >> If your customer is gigantic enough, maybe Hamamatsu would sell them >> the cute little MEMS PMTs. > >Wouldn't that be nice! Of course the designer has probably died of old >age by now.... > >Cheers > >Phil HobbsOne cool idea is to use a microchannel plate in front of a 2D delay-line detector. That can log the X-Y-T hits of single electrons or photons at a pretty high rate. My only patent is for a delay-line detector, US7019307. -- John Larkin Highland Technology, Inc lunatic fringe electronics
Reply by ●May 23, 20172017-05-23
On Monday, May 22, 2017 at 2:46:27 PM UTC-4, Jon Elson wrote:> George Herold wrote: > > > > > > Huh, I didn't know about Sens-L. Thanks! Hey they make one that is > > sensitive in the NIR! > > http://sensl.com/products/r-series/ > > I can afford $78 too... if they were in stock. > > (Oh I guess not released yet. I had to register to look at the data > > sheets.... R-series has ~25% PDE at 800 nm.) I wonder if they have > > to make them thin, which is why the are not so good in the NIR? > > > > Do you notice any problem with after pulsing? It's sorta like dark count, > > but a different noise source. > We haven't really seen after pulsing in our (very limited so far) testing. > As long as you don't turn the bias up too high, the dark current pulses are > well below the type of signals we are using, so they just about drop out of > the system. That's way better than our results with vacuum PMTs, where the > dark current is typically enough that it triggers our discriminators.Hi Jon, what sort of dark current do you see? The dark current between individual PMT's can change by more than an order of magnitude. Do you have an adjustable or fixed threshold level? (With fixed you can change the pmt bias voltage to change the pulse height.) But maybe you've moved on too... just thinking last night. George h.> > SensL also has SiPMs that bring out a fast output that is capacitively > coupled from each micropixel. You get really good time response from that. > Each micropixel has about an 18 K Ohm quench resistor in series for the > "slow" output, which we thought would give a horrible time response. But, > as we typically fire hundreds of micropixels at a time, it actually is > nowhere that bad. > > Yeah, they have stupid rules about registering to read advertising copy, I > have NO IDEA why outfits do this. SiPMs are not exactly the newest concept > out there. > > Jon
Reply by ●May 23, 20172017-05-23
On 05/23/2017 09:28 AM, George Herold wrote:> On Monday, May 22, 2017 at 2:46:27 PM UTC-4, Jon Elson wrote: >> George Herold wrote: >> >> >>> >>> Huh, I didn't know about Sens-L. Thanks! Hey they make one that is >>> sensitive in the NIR! >>> http://sensl.com/products/r-series/ >>> I can afford $78 too... if they were in stock. >>> (Oh I guess not released yet. I had to register to look at the data >>> sheets.... R-series has ~25% PDE at 800 nm.) I wonder if they have >>> to make them thin, which is why the are not so good in the NIR? >>> >>> Do you notice any problem with after pulsing? It's sorta like dark count, >>> but a different noise source. >> We haven't really seen after pulsing in our (very limited so far) testing. >> As long as you don't turn the bias up too high, the dark current pulses are >> well below the type of signals we are using, so they just about drop out of >> the system. That's way better than our results with vacuum PMTs, where the >> dark current is typically enough that it triggers our discriminators. > > Hi Jon, what sort of dark current do you see? > The dark current between individual PMT's can change by more than an > order of magnitude. > Do you have an adjustable or fixed threshold level? > (With fixed you can change the pmt bias voltage to change the > pulse height.) > > But maybe you've moved on too... just thinking last night. > > George h.If any APD seems to have lower input-referred dark current than any PMT, something is badly wrong. APDs are a good six orders of magnitude worse on an area basis. 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
Reply by ●May 23, 20172017-05-23
On 05/23/2017 12:07 AM, John Larkin wrote:> On Mon, 22 May 2017 15:57:13 -0400, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> On 05/22/2017 03:48 PM, John Larkin wrote: >>> On Mon, 22 May 2017 15:12:20 -0400, Phil Hobbs >>> <pcdhSpamMeSenseless@electrooptical.net> wrote: >>> >>>> On 05/22/2017 10:33 AM, George Herold wrote: >>>>> On Sunday, May 21, 2017 at 8:10:45 PM UTC-4, Winfield Hill wrote: >>>>>> Cursitor Doom wrote... >>>>>>> >>>>>>> Jim Thompson wrote: >>>>>>> >>>>>>>> AKA Flicker Noise >>>>>>> >>>>>>> Yup, pink noise. No jittering/jitterbugging involved. ;) >>>>>> >>>>>> Actually, I'd say you're wrong. The zener noise that we >>>>>> see in avalanche mode (greater then 7 volts or so) is due >>>>>> to step-wise "microplasma" ns-scale changes in current. >>>>>> A decade or so ago we fully discussed it here on s.e.d., >>>>>> complete with measurements, waveforms, detailed physics >>>>>> paper references and the works. Nailed it down. As it >>>>>> happens, the jitterbug analogy isn't so far off. >>>>> >>>>> Those are some nice threads, I went back and read them a few years >>>>> ago. Turning lemons into lemonade, a 20V zener baised at ~10uA >>>>> makes a nice noise source out to ~100's of kHz. >>>>> >>>>> I've also been reading Hamamatsu app notes on their Silicon PMT. >>>>> Which is really a Si avalanche photodiode in "Geiger-mode" >>>>> (biased above the breakdown voltage.) Fun stuff. >>>>> >>>>> George H. >>>> >>>> I'm in conversation with a big biosci company about cost-reducing their >>>> detection subsystems, so this has come up a lot lately. "Si PMTs" are >>>> basically an array of APDs wired in parallel and run in Geiger mode. >>>> The idea is that having multiple pixels allows you to extend Geiger mode >>>> to pulses containing more photons without losing resolution. >>>> >>>> AFAIK the biggest ones are about 120x120 pixels. Trouble is, even when >>>> uniformly illuminated they start to become nonlinear at depressingly low >>>> photon numbers--the linearity error is 10% once 10% of the pixels have >>>> fired, because only 90% are available. >>>> >>>> You can apply a correction for this that might work adequately up to >>>> about 25%, but it starts to run out of gas pretty fast compared with a >>>> real PMT. Their dynamic range requirements were much too large for this >>>> to work for them. >>>> >>>> Also the dark counts are ridiculously worse--more than 10**6 times worse >>>> on an area basis. >>>> >>>> The real advantage of MPPCs is that you can do thresholding, which helps >>>> reject single-pixel events at the expense of sensitivity. That reduces >>>> the effective dark count rate by a lot. >>>> >>>> Cheers >>>> >>>> Phil Hobbs >>> >>> If your customer is gigantic enough, maybe Hamamatsu would sell them >>> the cute little MEMS PMTs. >> >> Wouldn't that be nice! Of course the designer has probably died of old >> age by now.... >> >> Cheers >> >> Phil Hobbs > > One cool idea is to use a microchannel plate in front of a 2D > delay-line detector. That can log the X-Y-T hits of single electrons > or photons at a pretty high rate. > > My only patent is for a delay-line detector, US7019307.Yeah, I haven't filed one since I left IBM. They cost money, and I couldn't afford to defend it anyway. Hopefully after this year I'll have more leisure to work on stuff of interest to me personally. We'll see. 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
Reply by ●May 23, 20172017-05-23
On Tuesday, May 23, 2017 at 10:38:14 AM UTC-4, Phil Hobbs wrote:> On 05/23/2017 09:28 AM, George Herold wrote: > > On Monday, May 22, 2017 at 2:46:27 PM UTC-4, Jon Elson wrote: > >> George Herold wrote: > >> > >> > >>> > >>> Huh, I didn't know about Sens-L. Thanks! Hey they make one that is > >>> sensitive in the NIR! > >>> http://sensl.com/products/r-series/ > >>> I can afford $78 too... if they were in stock. > >>> (Oh I guess not released yet. I had to register to look at the data > >>> sheets.... R-series has ~25% PDE at 800 nm.) I wonder if they have > >>> to make them thin, which is why the are not so good in the NIR? > >>> > >>> Do you notice any problem with after pulsing? It's sorta like dark count, > >>> but a different noise source. > >> We haven't really seen after pulsing in our (very limited so far) testing. > >> As long as you don't turn the bias up too high, the dark current pulses are > >> well below the type of signals we are using, so they just about drop out of > >> the system. That's way better than our results with vacuum PMTs, where the > >> dark current is typically enough that it triggers our discriminators. > > > > Hi Jon, what sort of dark current do you see? > > The dark current between individual PMT's can change by more than an > > order of magnitude. > > Do you have an adjustable or fixed threshold level? > > (With fixed you can change the pmt bias voltage to change the > > pulse height.) > > > > But maybe you've moved on too... just thinking last night. > > > > George h. > > If any APD seems to have lower input-referred dark current than any PMT, > something is badly wrong.Right, maybe a noisy pmt power supply? Or a crappy preamp. I redesigned a pmt preamp. A colleague used an Amptek A111. It was not the right choice. (spendy too.) George H.> > APDs are a good six orders of magnitude worse on an area basis. > > 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
Reply by ●May 23, 20172017-05-23
On 05/23/2017 11:11 AM, George Herold wrote:> On Tuesday, May 23, 2017 at 10:38:14 AM UTC-4, Phil Hobbs wrote: >> On 05/23/2017 09:28 AM, George Herold wrote: >>> On Monday, May 22, 2017 at 2:46:27 PM UTC-4, Jon Elson wrote: >>>> George Herold wrote: >>>> >>>> >>>>> >>>>> Huh, I didn't know about Sens-L. Thanks! Hey they make one that is >>>>> sensitive in the NIR! >>>>> http://sensl.com/products/r-series/ >>>>> I can afford $78 too... if they were in stock. >>>>> (Oh I guess not released yet. I had to register to look at the data >>>>> sheets.... R-series has ~25% PDE at 800 nm.) I wonder if they have >>>>> to make them thin, which is why the are not so good in the NIR? >>>>> >>>>> Do you notice any problem with after pulsing? It's sorta like dark count, >>>>> but a different noise source. >>>> We haven't really seen after pulsing in our (very limited so far) testing. >>>> As long as you don't turn the bias up too high, the dark current pulses are >>>> well below the type of signals we are using, so they just about drop out of >>>> the system. That's way better than our results with vacuum PMTs, where the >>>> dark current is typically enough that it triggers our discriminators. >>> >>> Hi Jon, what sort of dark current do you see? >>> The dark current between individual PMT's can change by more than an >>> order of magnitude. >>> Do you have an adjustable or fixed threshold level? >>> (With fixed you can change the pmt bias voltage to change the >>> pulse height.) >>> >>> But maybe you've moved on too... just thinking last night. >>> >>> George h. >> >> If any APD seems to have lower input-referred dark current than any PMT, >> something is badly wrong. > Right, maybe a noisy pmt power supply?Or ridiculously high gain, or a gassy tube, or an IR PMT running uncooled, or a radium-dial clock on the bench, or horrible capacitive pickup due to being unshielded, or badly wrong dynode voltages causing arcing, or boatloads of stray light, or.... Something, anyway.> Or a crappy preamp. I redesigned a pmt preamp. A colleague used > an Amptek A111. It was not the right choice. (spendy too.)PMTs don't actually need super nice front ends. Having a nearly noiseless 100-120 dB of current gain can easily make a 50-ohm amp competitive--a million electrons arriving in 20 ns is 8 uA, so a 50 ohm amp with a 300K noise temperature and 25 MHz BW (20 ns equivalent width) would have a SNR of SNR = 10 log[(8uA)**2 * 50R]/(-203.8dBW + 74 dB) = 45 dB for a single-photon pulse. Minus 3 dB or something due to smearing out the pulse by sqrt(2). Still a really easy job for a 50-ohm amp. Cheers Phil Hobbs> > George H. >> >> APDs are a good six orders of magnitude worse on an area basis. >> >> 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 >-- 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
Reply by ●May 23, 20172017-05-23
On Tue, 23 May 2017 10:40:49 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>On 05/23/2017 12:07 AM, John Larkin wrote: >> On Mon, 22 May 2017 15:57:13 -0400, Phil Hobbs >> <pcdhSpamMeSenseless@electrooptical.net> wrote: >> >>> On 05/22/2017 03:48 PM, John Larkin wrote: >>>> On Mon, 22 May 2017 15:12:20 -0400, Phil Hobbs >>>> <pcdhSpamMeSenseless@electrooptical.net> wrote: >>>> >>>>> On 05/22/2017 10:33 AM, George Herold wrote: >>>>>> On Sunday, May 21, 2017 at 8:10:45 PM UTC-4, Winfield Hill wrote: >>>>>>> Cursitor Doom wrote... >>>>>>>> >>>>>>>> Jim Thompson wrote: >>>>>>>> >>>>>>>>> AKA Flicker Noise >>>>>>>> >>>>>>>> Yup, pink noise. No jittering/jitterbugging involved. ;) >>>>>>> >>>>>>> Actually, I'd say you're wrong. The zener noise that we >>>>>>> see in avalanche mode (greater then 7 volts or so) is due >>>>>>> to step-wise "microplasma" ns-scale changes in current. >>>>>>> A decade or so ago we fully discussed it here on s.e.d., >>>>>>> complete with measurements, waveforms, detailed physics >>>>>>> paper references and the works. Nailed it down. As it >>>>>>> happens, the jitterbug analogy isn't so far off. >>>>>> >>>>>> Those are some nice threads, I went back and read them a few years >>>>>> ago. Turning lemons into lemonade, a 20V zener baised at ~10uA >>>>>> makes a nice noise source out to ~100's of kHz. >>>>>> >>>>>> I've also been reading Hamamatsu app notes on their Silicon PMT. >>>>>> Which is really a Si avalanche photodiode in "Geiger-mode" >>>>>> (biased above the breakdown voltage.) Fun stuff. >>>>>> >>>>>> George H. >>>>> >>>>> I'm in conversation with a big biosci company about cost-reducing their >>>>> detection subsystems, so this has come up a lot lately. "Si PMTs" are >>>>> basically an array of APDs wired in parallel and run in Geiger mode. >>>>> The idea is that having multiple pixels allows you to extend Geiger mode >>>>> to pulses containing more photons without losing resolution. >>>>> >>>>> AFAIK the biggest ones are about 120x120 pixels. Trouble is, even when >>>>> uniformly illuminated they start to become nonlinear at depressingly low >>>>> photon numbers--the linearity error is 10% once 10% of the pixels have >>>>> fired, because only 90% are available. >>>>> >>>>> You can apply a correction for this that might work adequately up to >>>>> about 25%, but it starts to run out of gas pretty fast compared with a >>>>> real PMT. Their dynamic range requirements were much too large for this >>>>> to work for them. >>>>> >>>>> Also the dark counts are ridiculously worse--more than 10**6 times worse >>>>> on an area basis. >>>>> >>>>> The real advantage of MPPCs is that you can do thresholding, which helps >>>>> reject single-pixel events at the expense of sensitivity. That reduces >>>>> the effective dark count rate by a lot. >>>>> >>>>> Cheers >>>>> >>>>> Phil Hobbs >>>> >>>> If your customer is gigantic enough, maybe Hamamatsu would sell them >>>> the cute little MEMS PMTs. >>> >>> Wouldn't that be nice! Of course the designer has probably died of old >>> age by now.... >>> >>> Cheers >>> >>> Phil Hobbs >> >> One cool idea is to use a microchannel plate in front of a 2D >> delay-line detector. That can log the X-Y-T hits of single electrons >> or photons at a pretty high rate. >> >> My only patent is for a delay-line detector, US7019307. > >Yeah, I haven't filed one since I left IBM. They cost money, and I >couldn't afford to defend it anyway. >I think I got that one after a short brainstorming session with Tye. Imago was in the VC/file-lots-of-patents phase, so they filed it. I discovered the patent accidentally during a web search. I suppose I signed something at some point, but I'd forgotten. https://dl.dropboxusercontent.com/u/53724080/Gear/Anodes.JPG If you let the electron cloud expand a bit before it hits the delay lines, you can filter and interpolate, and get X-Y resolution of a small fraction of the line pitch. http://www.cameca.com/image.asbx?AttributeFileId=dd44ee36-708e-4cb5-8313-6c5ce95e8c66 Those dots are single atoms in a 3D reconstruction.>Hopefully after this year I'll have more leisure to work on stuff of >interest to me personally. We'll see. > >Cheers > >Phil HobbsHave some fun! -- John Larkin Highland Technology, Inc lunatic fringe electronics
Reply by ●May 23, 20172017-05-23
On 05/23/2017 12:06 PM, John Larkin wrote:> On Tue, 23 May 2017 10:40:49 -0400, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> On 05/23/2017 12:07 AM, John Larkin wrote: >>> On Mon, 22 May 2017 15:57:13 -0400, Phil Hobbs >>> <pcdhSpamMeSenseless@electrooptical.net> wrote: >>> >>>> On 05/22/2017 03:48 PM, John Larkin wrote: >>>>> On Mon, 22 May 2017 15:12:20 -0400, Phil Hobbs >>>>> <pcdhSpamMeSenseless@electrooptical.net> wrote: >>>>> >>>>>> On 05/22/2017 10:33 AM, George Herold wrote: >>>>>>> On Sunday, May 21, 2017 at 8:10:45 PM UTC-4, Winfield Hill wrote: >>>>>>>> Cursitor Doom wrote... >>>>>>>>> >>>>>>>>> Jim Thompson wrote: >>>>>>>>> >>>>>>>>>> AKA Flicker Noise >>>>>>>>> >>>>>>>>> Yup, pink noise. No jittering/jitterbugging involved. ;) >>>>>>>> >>>>>>>> Actually, I'd say you're wrong. The zener noise that we >>>>>>>> see in avalanche mode (greater then 7 volts or so) is due >>>>>>>> to step-wise "microplasma" ns-scale changes in current. >>>>>>>> A decade or so ago we fully discussed it here on s.e.d., >>>>>>>> complete with measurements, waveforms, detailed physics >>>>>>>> paper references and the works. Nailed it down. As it >>>>>>>> happens, the jitterbug analogy isn't so far off. >>>>>>> >>>>>>> Those are some nice threads, I went back and read them a few years >>>>>>> ago. Turning lemons into lemonade, a 20V zener baised at ~10uA >>>>>>> makes a nice noise source out to ~100's of kHz. >>>>>>> >>>>>>> I've also been reading Hamamatsu app notes on their Silicon PMT. >>>>>>> Which is really a Si avalanche photodiode in "Geiger-mode" >>>>>>> (biased above the breakdown voltage.) Fun stuff. >>>>>>> >>>>>>> George H. >>>>>> >>>>>> I'm in conversation with a big biosci company about cost-reducing their >>>>>> detection subsystems, so this has come up a lot lately. "Si PMTs" are >>>>>> basically an array of APDs wired in parallel and run in Geiger mode. >>>>>> The idea is that having multiple pixels allows you to extend Geiger mode >>>>>> to pulses containing more photons without losing resolution. >>>>>> >>>>>> AFAIK the biggest ones are about 120x120 pixels. Trouble is, even when >>>>>> uniformly illuminated they start to become nonlinear at depressingly low >>>>>> photon numbers--the linearity error is 10% once 10% of the pixels have >>>>>> fired, because only 90% are available. >>>>>> >>>>>> You can apply a correction for this that might work adequately up to >>>>>> about 25%, but it starts to run out of gas pretty fast compared with a >>>>>> real PMT. Their dynamic range requirements were much too large for this >>>>>> to work for them. >>>>>> >>>>>> Also the dark counts are ridiculously worse--more than 10**6 times worse >>>>>> on an area basis. >>>>>> >>>>>> The real advantage of MPPCs is that you can do thresholding, which helps >>>>>> reject single-pixel events at the expense of sensitivity. That reduces >>>>>> the effective dark count rate by a lot. >>>>>> >>>>>> Cheers >>>>>> >>>>>> Phil Hobbs >>>>> >>>>> If your customer is gigantic enough, maybe Hamamatsu would sell them >>>>> the cute little MEMS PMTs. >>>> >>>> Wouldn't that be nice! Of course the designer has probably died of old >>>> age by now.... >>>> >>>> Cheers >>>> >>>> Phil Hobbs >>> >>> One cool idea is to use a microchannel plate in front of a 2D >>> delay-line detector. That can log the X-Y-T hits of single electrons >>> or photons at a pretty high rate. >>> >>> My only patent is for a delay-line detector, US7019307. >> >> Yeah, I haven't filed one since I left IBM. They cost money, and I >> couldn't afford to defend it anyway. >> > > I think I got that one after a short brainstorming session with Tye. > Imago was in the VC/file-lots-of-patents phase, so they filed it. I > discovered the patent accidentally during a web search. I suppose I > signed something at some point, but I'd forgotten. > > https://dl.dropboxusercontent.com/u/53724080/Gear/Anodes.JPG > > If you let the electron cloud expand a bit before it hits the delay > lines, you can filter and interpolate, and get X-Y resolution of a > small fraction of the line pitch. > > http://www.cameca.com/image.asbx?AttributeFileId=dd44ee36-708e-4cb5-8313-6c5ce95e8c66 > > Those dots are single atoms in a 3D reconstruction. > > >> Hopefully after this year I'll have more leisure to work on stuff of >> interest to me personally. We'll see. >>> > Have some fun!Yup. I want to build more instruments and fewer TIAs. That's one reason for offering TIAs and noise cancellers for sale. I'm also hoping to offer a super-wideband temperature control system at some point--the proto works okay. 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
Reply by ●May 23, 20172017-05-23
On Tuesday, May 23, 2017 at 11:37:07 AM UTC-4, Phil Hobbs wrote:> On 05/23/2017 11:11 AM, George Herold wrote: > > On Tuesday, May 23, 2017 at 10:38:14 AM UTC-4, Phil Hobbs wrote: > >> On 05/23/2017 09:28 AM, George Herold wrote: > >>> On Monday, May 22, 2017 at 2:46:27 PM UTC-4, Jon Elson wrote: > >>>> George Herold wrote: > >>>> > >>>> > >>>>> > >>>>> Huh, I didn't know about Sens-L. Thanks! Hey they make one that is > >>>>> sensitive in the NIR! > >>>>> http://sensl.com/products/r-series/ > >>>>> I can afford $78 too... if they were in stock. > >>>>> (Oh I guess not released yet. I had to register to look at the data > >>>>> sheets.... R-series has ~25% PDE at 800 nm.) I wonder if they have > >>>>> to make them thin, which is why the are not so good in the NIR? > >>>>> > >>>>> Do you notice any problem with after pulsing? It's sorta like dark count, > >>>>> but a different noise source. > >>>> We haven't really seen after pulsing in our (very limited so far) testing. > >>>> As long as you don't turn the bias up too high, the dark current pulses are > >>>> well below the type of signals we are using, so they just about drop out of > >>>> the system. That's way better than our results with vacuum PMTs, where the > >>>> dark current is typically enough that it triggers our discriminators. > >>> > >>> Hi Jon, what sort of dark current do you see? > >>> The dark current between individual PMT's can change by more than an > >>> order of magnitude. > >>> Do you have an adjustable or fixed threshold level? > >>> (With fixed you can change the pmt bias voltage to change the > >>> pulse height.) > >>> > >>> But maybe you've moved on too... just thinking last night. > >>> > >>> George h. > >> > >> If any APD seems to have lower input-referred dark current than any PMT, > >> something is badly wrong. > > Right, maybe a noisy pmt power supply? > > Or ridiculously high gain, or a gassy tube, or an IR PMT running > uncooled, or a radium-dial clock on the bench, or horrible capacitive > pickup due to being unshielded, or badly wrong dynode voltages causing > arcing, or boatloads of stray light, or.... Something, anyway. > > > Or a crappy preamp. I redesigned a pmt preamp. A colleague used > > an Amptek A111. It was not the right choice. (spendy too.) > > PMTs don't actually need super nice front ends. Having a nearly > noiseless 100-120 dB of current gain can easily make a 50-ohm amp > competitive--a million electrons arriving in 20 ns is 8 uA, so a 50 ohm > amp with a 300K noise temperature and 25 MHz BW (20 ns equivalent > width) would have a SNR of > > SNR = 10 log[(8uA)**2 * 50R]/(-203.8dBW + 74 dB) = 45 dB > > for a single-photon pulse. Minus 3 dB or something due to smearing out > the pulse by sqrt(2). Still a really easy job for a 50-ohm amp.Right, I didn't say it was hard. (after all, *I* got it to work first try.) GH> > Cheers > > Phil Hobbs > > > > > George H. > >> > >> APDs are a good six orders of magnitude worse on an area basis. > >> > >> 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 > > > > > -- > 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
