Photodiode SPICE models?
in [Design]
|
Prev: Interesting TIA design
Next: FS: NEW Minicircuits ZKL-1R5 RF MICROWAVE AMPLIFIER(s) 0.01-1.5GHz. $50 on Ebay (LOT of TEN)
From: Phil Hobbs on 30 Oct 2009 17:32 Phil Hobbs wrote: > George Herold wrote: >> On Oct 28, 12:40 pm, Phil Hobbs >> <pcdhSpamMeSensel...(a)electrooptical.net> wrote: >>> Yuri Belenky wrote: >>>> Phil Hobbs wrote: >>>>> Dear all: >>>>> I'm on a bit of a roll with TIA designs at the moment, but I've just >>>>> been using current sources with caps in parallel to model >>>>> fully-depleted photodiodes, and I need a better model to hack up. >>>>> The LTSPICE Yahoo group's files area is broken at the moment. Do any >>>>> of you have a decent PD model with 4 terminals (anode, cathode, and >>>>> two to represent the light input)? >>>>> Nonlinear capacitance would be a plus but isn't required. >>>>> (I've used SPICE more in the last two weeks than in my entire previous >>>>> career--I'm more of an algebra guy when it comes to circuit design, >>>>> but some things really need sims.) >>>>> Thanks >>>>> Phil Hobbs >>>> I've used this model several years ago (MRD500 photodiode): >>>> --- model starts here ------------------------ >>>> * PSpice Model Editor - Version 10.5.0 >>>> *$ >>>> ***************************************************************************�**** >>>> >>>> * A = Anode >>>> * C = Cathode >>>> * P = (light)Power as voltage input >>>> ..SUBCKT MRD500 A C P PARAMS: Spec=1.0 Sensy=0.3894 >>>> * MRD500 from Motorola * RAP 7/97 >>>> DPD A C PhotoDet >>>> RDUMMY P 0 1G >>>> GPD C A TABLE {V(P)* Spec * Sensy} (0,0) (1,1) >>>> ..MODEL PhotoDet D IS=1.02p RS=10 N=0.841978 BV=100 IBV=1u >>>> + CJO=6.03p VJ=0.202968 M=0.146605 TT=1n ISR=17p >>>> ..ends >>>> *$ >>>> --- end of model ------------------------------ >>>> Let me know if you need it as OrCAD library file. >>>> Yuri. >>> Thanks. >>> >>> A couple of people have sent models (Thanks, guys), but there is one >>> remaining puzzle: shot noise. I'd have thought that the .asc file below >>> would generate exactly the shot noise of Iphoto--diodes have full shot >>> noise, so connecting it to a DC current source should turn D1 into a >>> full shot noise current source, which ought to get picked off via the >>> infinite capacitor into the 0V source, and so into the behavioural >>> current source at the output. >>> >>> However, it doesn't do anything of the sort--it's all over the map. >>> Different diodes and diode-connected transistors give all sorts of >>> different values. >>> >>> Any ideas what's going on? One possible clue is that any resistance at >>> all in series with V2 reduces the noise enormously. >>> > >>> <snip> >> >> Phil, this is most likely a stupid idea, but does model have the >> correct source impedance for the 'diode as shot noise source'. R = kT/ >> (eI) > > > Hmm. That's an interesting point...perhaps the diode models aren't > implemented the way I thought, with the shot noise being a current > source internally. > >> >> (I loved the 'link' that I clicked on an it loaded the whole >> schematic.. cute) > > LTSpice rides again--flat ASCII schematic files. Go, Mike! > > Cheers > > Phil Hobbs > Okay, I solved the problem. There's something weird about the diode model in LTSPice--it really really doesn't like zero impedance loads. I've attached a hacked-up version of the SFH2400 model from the LTSpice group on Yahoo, with switchable shot noise--it takes a parameter called ShotNoiseOn (default=1) which does what you'd expect. It uses a balanced pair of diodes and forms the shot noise contribution by summing their terminal voltages and multiplying by I_dc/(sqrt(2)*(25.8 mV)). This makes sure that only the uncorrelated part (i.e. noise) winds up in the output. Single-ended voltage versions cause signal feedthrough as well. Cheers Phil Hobbs
From: George Herold on 30 Oct 2009 23:51
On Oct 30, 5:32 pm, Phil Hobbs <pcdhSpamMeSensel...(a)electrooptical.net> wrote: > Phil Hobbs wrote: > > George Herold wrote: > >> On Oct 28, 12:40 pm, Phil Hobbs > >> <pcdhSpamMeSensel...(a)electrooptical.net> wrote: > >>> Yuri Belenky wrote: > >>>> Phil Hobbs wrote: > >>>>> Dear all: > >>>>> I'm on a bit of a roll with TIA designs at the moment, but I've just > >>>>> been using current sources with caps in parallel to model > >>>>> fully-depleted photodiodes, and I need a better model to hack up. > >>>>> The LTSPICE Yahoo group's files area is broken at the moment. Do any > >>>>> of you have a decent PD model with 4 terminals (anode, cathode, and > >>>>> two to represent the light input)? > >>>>> Nonlinear capacitance would be a plus but isn't required. > >>>>> (I've used SPICE more in the last two weeks than in my entire previous > >>>>> career--I'm more of an algebra guy when it comes to circuit design, > >>>>> but some things really need sims.) > >>>>> Thanks > >>>>> Phil Hobbs > >>>> I've used this model several years ago (MRD500 photodiode): > >>>> --- model starts here ------------------------ > >>>> * PSpice Model Editor - Version 10.5.0 > >>>> *$ > >>>> ******************************************************************************* > > >>>> * A = Anode > >>>> * C = Cathode > >>>> * P = (light)Power as voltage input > >>>> ..SUBCKT MRD500 A C P PARAMS: Spec=1.0 Sensy=0.3894 > >>>> * MRD500 from Motorola * RAP 7/97 > >>>> DPD A C PhotoDet > >>>> RDUMMY P 0 1G > >>>> GPD C A TABLE {V(P)* Spec * Sensy} (0,0) (1,1) > >>>> ..MODEL PhotoDet D IS=1.02p RS=10 N=0.841978 BV=100 IBV=1u > >>>> + CJO=6.03p VJ=0.202968 M=0.146605 TT=1n ISR=17p > >>>> ..ends > >>>> *$ > >>>> --- end of model ------------------------------ > >>>> Let me know if you need it as OrCAD library file. > >>>> Yuri. > >>> Thanks. > > >>> A couple of people have sent models (Thanks, guys), but there is one > >>> remaining puzzle: shot noise. I'd have thought that the .asc file below > >>> would generate exactly the shot noise of Iphoto--diodes have full shot > >>> noise, so connecting it to a DC current source should turn D1 into a > >>> full shot noise current source, which ought to get picked off via the > >>> infinite capacitor into the 0V source, and so into the behavioural > >>> current source at the output. > > >>> However, it doesn't do anything of the sort--it's all over the map. > >>> Different diodes and diode-connected transistors give all sorts of > >>> different values. > > >>> Any ideas what's going on? One possible clue is that any resistance at > >>> all in series with V2 reduces the noise enormously. > > >>> <snip> > > >> Phil, this is most likely a stupid idea, but does model have the > >> correct source impedance for the 'diode as shot noise source'. R = kT/ > >> (eI) > > > Hmm. That's an interesting point...perhaps the diode models aren't > > implemented the way I thought, with the shot noise being a current > > source internally. > > >> (I loved the 'link' that I clicked on an it loaded the whole > >> schematic.. cute) > > > LTSpice rides again--flat ASCII schematic files. Go, Mike! > > > Cheers > > > Phil Hobbs > > Okay, I solved the problem. There's something weird about the diode > model in LTSPice--it really really doesn't like zero impedance loads. > > I've attached a hacked-up version of the SFH2400 model from the LTSpice > group on Yahoo, with switchable shot noise--it takes a parameter called > ShotNoiseOn (default=1) which does what you'd expect. > > It uses a balanced pair of diodes and forms the shot noise contribution > by summing their terminal voltages and multiplying by > I_dc/(sqrt(2)*(25.8 mV)). This makes sure that only the uncorrelated > part (i.e. noise) winds up in the output. Single-ended voltage versions > cause signal feedthrough as well. > > Cheers > > Phil Hobbs > > sfh2400_shotnoise_hy.asy > 1KViewDownload > > sfh2400_shotnoise_hy.asc > 4KViewDownload- Hide quoted text - > > - Show quoted text - Excellent! Now you've got something with the correct impedance. I was 'burned' by the diode impedance when I first tried to measure the shot noise from a diode. I biased it with a battery and large resistor and fed the bottom end into a TIA opamp. NO noise??? (well except for opamp and resistor noise.) I had to add a capacitor to allow for a low impedance path for the shot noise. Funny, how you usually have to screw something up before you really learn it. George H. |