On Tue, 11 Aug 2015 13:40:45 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>On 08/11/2015 12:52 PM, Jim Thompson wrote: >> On Tue, 11 Aug 2015 07:45:28 -0700, Jim Thompson >> <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote: >> >>> On Mon, 10 Aug 2015 22:37:04 -0400, Phil Hobbs >>> <pcdhSpamMeSenseless@electrooptical.net> wrote: >>> >>>> On 8/10/2015 10:18 PM, Jim Thompson wrote: >>>>> On Mon, 10 Aug 2015 21:11:51 -0400, Phil Hobbs >>>>> <hobbs@electrooptical.net> wrote: >>>>> >>>>>> On 8/10/2015 9:04 PM, Jim Thompson wrote: >>>>>>> On Sun, 9 Aug 2015 16:24:53 -0700 (PDT), Phil Hobbs >>>>>>> <pcdhobbs@gmail.com> wrote: >>>>>>> >>>>>>>> Hi, Jim, >>>>>>>> >>>>>>>> If you've got some spare bandwidth, I could really use an improved >>>>>>>> OPA2140 model. >>>>>>>> >>>>>>>> It's a very good part for moderate speed, low current, high >>>>>>>> linearity TIAs, but it's a bear to simulate because the model won't >>>>>>>> converge on an operating point in LTspice. >>>>>>>> >>>>>>>> I took the TINA model TI supplies and replaced the VSWITCH cards >>>>>>>> with LTspice SW cards with negative hysteresis (more or less like >>>>>>>> your fave tanh curve), but it still won't converge on ac or noise >>>>>>>> sims. >>>>>>>> >>>>>>>> Thanks >>>>>>>> >>>>>>>> Phil Hobbs >>>>>>> >>>>>>> How do you set the parameters of the SW in LTspice? ...Jim Thompson >>>>>>> >>>>>> >>>>>> RON, ROFF, which do what you'd expect; >>>>>> VT, which is the switching threshold (the switch looks like RON for V > >>>>>> VT, and ROFF for V < VT); >>>>>> and VH, the hysteresis. >>>>>> >>>>>> If VH is positive, then the upward-going transition occurs at V = VT + >>>>>> VH and the downward-going one at V = VT - VH. (IOW it does what you >>>>>> expect except that VH is the half-width of the deadband, not the full >>>>>> width.) >>>>>> >>>>>> Negative values of VH cause a smooth transition between RON and ROFF, so >>>>>> that the switching begins at VT + VH (which is less than VT in this >>>>>> case) and is complete at VT - VH (which is more than VT). IIRC the >>>>>> function is some smooth polynomial approximation to the arctan, so at >>>>>> least the first few derivatives are continuous. >>>>>> >>>>>> Cheers >>>>>> >>>>>> Phil Hobbs >>>>> >>>>> I'm used to PSpice, click on a symbol and you see a table where you >>>>> fill in the blanks. I click on SW in LTspice, I see several blank >>>>> lines... >>>>> >>>>> SpiceLine >>>>> SpicelIne2 >>>>> >>>>> Where do I enter Ron, Roff, Von, Voff, etc? >>>>> >>>>> ...Jim Thompson >>>>> >>>> You have to put in a .model statement: >>>> >>>> .model simplesw sw(Vt=1 Vh=-.5 Ron=1 Roff=1T) >>>> >>>> and put 'simplesw' on the 'value' line in the dialogue box. >>>> >>>> To make a NC switch, you can interchange the values of Ron and Roff, i.e. >>>> >>>> .model simpleNC sw(Vt=1 Vh=-.5 Ron=1T Roff=1) >>>> >>>> Cheers >>>> >>>> Phil Hobbs >>> >>> Checked it out, LTspice switch with negative hysteresis looks exactly >>> same V-I curve as in PSpice... PSpice runs the OPA2140, LTspice will >>> not, so it looks like a major makeover. >>> >>> If you need only GBW and noise, that can be done simply, otherwise the >>> long winter ;-) >>> >>> <http://www.analog-innovations.com/SED/OP2140_JT_2015-08-11_07-39-39.jpg> >>> >>> ...Jim Thompson >> >> BTW: PSpice won't converge on asymmetric supplies either. So the TI >> model is crap :-( >> >> ...Jim Thompson >> > >If I were more motivated I'd try it on TINA, but given the complexity of >the model I'd be surprised if they'd fix it even if it's broken on TINA >as well. > >Cheers > >Phil HobbsThe SUBCKT looks to be an extremist's attempt at obfuscation. All kinds of voltage sources, indicating either attempts at obfuscation or total ignorance of how to write proper behavioral controlled sources. Also a bunch of unbounded current sources that probably pull to infinity at the slightest provocation >:-} Maybe you should choose another brand and let TI know? When I get the time I'll consider writing a model from the datasheet. The TI SUBCKT is grotesque, to put it mildly ;-) ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
AN: TL598 Spice Model
Started by ●August 9, 2015
Reply by ●August 11, 20152015-08-11
Reply by ●August 11, 20152015-08-11
On Tue, 11 Aug 2015 11:09:10 -0700, Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:>On Tue, 11 Aug 2015 13:40:45 -0400, Phil Hobbs ><pcdhSpamMeSenseless@electrooptical.net> wrote: > >>On 08/11/2015 12:52 PM, Jim Thompson wrote: >>> On Tue, 11 Aug 2015 07:45:28 -0700, Jim Thompson >>> <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:[snip]>>>> >>>> If you need only GBW and noise, that can be done simply, otherwise the >>>> long winter ;-) >>>> >>>> <http://www.analog-innovations.com/SED/OP2140_JT_2015-08-11_07-39-39.jpg> >>>> >>>> ...Jim Thompson >>> >>> BTW: PSpice won't converge on asymmetric supplies either. So the TI >>> model is crap :-( >>> >>> ...Jim Thompson >>> >> >>If I were more motivated I'd try it on TINA, but given the complexity of >>the model I'd be surprised if they'd fix it even if it's broken on TINA >>as well. >> >>Cheers >> >>Phil Hobbs > >The SUBCKT looks to be an extremist's attempt at obfuscation. All >kinds of voltage sources, indicating either attempts at obfuscation or >total ignorance of how to write proper behavioral controlled sources. >Also a bunch of unbounded current sources that probably pull to >infinity at the slightest provocation >:-} > >Maybe you should choose another brand and let TI know? > >When I get the time I'll consider writing a model from the datasheet. >The TI SUBCKT is grotesque, to put it mildly ;-) > > ...Jim ThompsonHere's a funny for you, a current flowing FROM V- INTO the chip... IS3 40 V- -500F Another... obsolete H controlled source... HCCVS2 50 11 VCCVS2_in 1K ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 11, 20152015-08-11
On 08/11/2015 02:09 PM, Jim Thompson wrote:> On Tue, 11 Aug 2015 13:40:45 -0400, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> On 08/11/2015 12:52 PM, Jim Thompson wrote: >>> On Tue, 11 Aug 2015 07:45:28 -0700, Jim Thompson >>> <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote: >>> >>>> On Mon, 10 Aug 2015 22:37:04 -0400, Phil Hobbs >>>> <pcdhSpamMeSenseless@electrooptical.net> wrote: >>>> >>>>> On 8/10/2015 10:18 PM, Jim Thompson wrote: >>>>>> On Mon, 10 Aug 2015 21:11:51 -0400, Phil Hobbs >>>>>> <hobbs@electrooptical.net> wrote: >>>>>> >>>>>>> On 8/10/2015 9:04 PM, Jim Thompson wrote: >>>>>>>> On Sun, 9 Aug 2015 16:24:53 -0700 (PDT), Phil Hobbs >>>>>>>> <pcdhobbs@gmail.com> wrote: >>>>>>>> >>>>>>>>> Hi, Jim, >>>>>>>>> >>>>>>>>> If you've got some spare bandwidth, I could really use an improved >>>>>>>>> OPA2140 model. >>>>>>>>> >>>>>>>>> It's a very good part for moderate speed, low current, high >>>>>>>>> linearity TIAs, but it's a bear to simulate because the model won't >>>>>>>>> converge on an operating point in LTspice. >>>>>>>>> >>>>>>>>> I took the TINA model TI supplies and replaced the VSWITCH cards >>>>>>>>> with LTspice SW cards with negative hysteresis (more or less like >>>>>>>>> your fave tanh curve), but it still won't converge on ac or noise >>>>>>>>> sims. >>>>>>>>> >>>>>>>>> Thanks >>>>>>>>> >>>>>>>>> Phil Hobbs >>>>>>>> >>>>>>>> How do you set the parameters of the SW in LTspice? ...Jim Thompson >>>>>>>> >>>>>>> >>>>>>> RON, ROFF, which do what you'd expect; >>>>>>> VT, which is the switching threshold (the switch looks like RON for V > >>>>>>> VT, and ROFF for V < VT); >>>>>>> and VH, the hysteresis. >>>>>>> >>>>>>> If VH is positive, then the upward-going transition occurs at V = VT + >>>>>>> VH and the downward-going one at V = VT - VH. (IOW it does what you >>>>>>> expect except that VH is the half-width of the deadband, not the full >>>>>>> width.) >>>>>>> >>>>>>> Negative values of VH cause a smooth transition between RON and ROFF, so >>>>>>> that the switching begins at VT + VH (which is less than VT in this >>>>>>> case) and is complete at VT - VH (which is more than VT). IIRC the >>>>>>> function is some smooth polynomial approximation to the arctan, so at >>>>>>> least the first few derivatives are continuous. >>>>>>> >>>>>>> Cheers >>>>>>> >>>>>>> Phil Hobbs >>>>>> >>>>>> I'm used to PSpice, click on a symbol and you see a table where you >>>>>> fill in the blanks. I click on SW in LTspice, I see several blank >>>>>> lines... >>>>>> >>>>>> SpiceLine >>>>>> SpicelIne2 >>>>>> >>>>>> Where do I enter Ron, Roff, Von, Voff, etc? >>>>>> >>>>>> ...Jim Thompson >>>>>> >>>>> You have to put in a .model statement: >>>>> >>>>> .model simplesw sw(Vt=1 Vh=-.5 Ron=1 Roff=1T) >>>>> >>>>> and put 'simplesw' on the 'value' line in the dialogue box. >>>>> >>>>> To make a NC switch, you can interchange the values of Ron and Roff, i.e. >>>>> >>>>> .model simpleNC sw(Vt=1 Vh=-.5 Ron=1T Roff=1) >>>>> >>>>> Cheers >>>>> >>>>> Phil Hobbs >>>> >>>> Checked it out, LTspice switch with negative hysteresis looks exactly >>>> same V-I curve as in PSpice... PSpice runs the OPA2140, LTspice will >>>> not, so it looks like a major makeover. >>>> >>>> If you need only GBW and noise, that can be done simply, otherwise the >>>> long winter ;-) >>>> >>>> <http://www.analog-innovations.com/SED/OP2140_JT_2015-08-11_07-39-39.jpg> >>>> >>>> ...Jim Thompson >>> >>> BTW: PSpice won't converge on asymmetric supplies either. So the TI >>> model is crap :-( >>> >>> ...Jim Thompson >>> >> >> If I were more motivated I'd try it on TINA, but given the complexity of >> the model I'd be surprised if they'd fix it even if it's broken on TINA >> as well. >> >> Cheers >> >> Phil Hobbs > > The SUBCKT looks to be an extremist's attempt at obfuscation. All > kinds of voltage sources, indicating either attempts at obfuscation or > total ignorance of how to write proper behavioral controlled sources. > Also a bunch of unbounded current sources that probably pull to > infinity at the slightest provocation >:-} > > Maybe you should choose another brand and let TI know? > > When I get the time I'll consider writing a model from the datasheet. > The TI SUBCKT is grotesque, to put it mildly ;-)Well, it's about 10 times longer than your normal op amp model--I was hoping that was a good thing. ;) The OPA140 is pretty well your perfect low-power TIA part, except for the 10 pF input capacitance, which is on the high side. 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 ●August 11, 20152015-08-11
On Tue, 11 Aug 2015 14:26:30 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>On 08/11/2015 02:09 PM, Jim Thompson wrote: >> On Tue, 11 Aug 2015 13:40:45 -0400, Phil Hobbs >> <pcdhSpamMeSenseless@electrooptical.net> wrote: >> >>> On 08/11/2015 12:52 PM, Jim Thompson wrote: >>>> On Tue, 11 Aug 2015 07:45:28 -0700, Jim Thompson >>>> <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:[snip]>>>>> >>>>> Checked it out, LTspice switch with negative hysteresis looks exactly >>>>> same V-I curve as in PSpice... PSpice runs the OPA2140, LTspice will >>>>> not, so it looks like a major makeover. >>>>> >>>>> If you need only GBW and noise, that can be done simply, otherwise the >>>>> long winter ;-) >>>>> >>>>> <http://www.analog-innovations.com/SED/OP2140_JT_2015-08-11_07-39-39.jpg> >>>>> >>>>> ...Jim Thompson >>>> >>>> BTW: PSpice won't converge on asymmetric supplies either. So the TI >>>> model is crap :-( >>>> >>>> ...Jim Thompson >>>> >>> >>> If I were more motivated I'd try it on TINA, but given the complexity of >>> the model I'd be surprised if they'd fix it even if it's broken on TINA >>> as well. >>> >>> Cheers >>> >>> Phil Hobbs >> >> The SUBCKT looks to be an extremist's attempt at obfuscation. All >> kinds of voltage sources, indicating either attempts at obfuscation or >> total ignorance of how to write proper behavioral controlled sources. >> Also a bunch of unbounded current sources that probably pull to >> infinity at the slightest provocation >:-} >> >> Maybe you should choose another brand and let TI know? >> >> When I get the time I'll consider writing a model from the datasheet. >> The TI SUBCKT is grotesque, to put it mildly ;-) > >Well, it's about 10 times longer than your normal op amp model--I was >hoping that was a good thing. ;) > >The OPA140 is pretty well your perfect low-power TIA part, except for >the 10 pF input capacitance, which is on the high side. > >Cheers > >Phil HobbsWhen they spec 7pF common-mode input capacitance, is that from _each_ input pin, or combined? ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 11, 20152015-08-11
On 8/11/2015 3:38 PM, Jim Thompson wrote:> On Tue, 11 Aug 2015 14:26:30 -0400, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> On 08/11/2015 02:09 PM, Jim Thompson wrote: >>> On Tue, 11 Aug 2015 13:40:45 -0400, Phil Hobbs >>> <pcdhSpamMeSenseless@electrooptical.net> wrote: >>> >>>> On 08/11/2015 12:52 PM, Jim Thompson wrote: >>>>> On Tue, 11 Aug 2015 07:45:28 -0700, Jim Thompson >>>>> <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote: > [snip] >>>>>> >>>>>> Checked it out, LTspice switch with negative hysteresis looks exactly >>>>>> same V-I curve as in PSpice... PSpice runs the OPA2140, LTspice will >>>>>> not, so it looks like a major makeover. >>>>>> >>>>>> If you need only GBW and noise, that can be done simply, otherwise the >>>>>> long winter ;-) >>>>>> >>>>>> <http://www.analog-innovations.com/SED/OP2140_JT_2015-08-11_07-39-39.jpg> >>>>>> >>>>>> ...Jim Thompson >>>>> >>>>> BTW: PSpice won't converge on asymmetric supplies either. So the TI >>>>> model is crap :-( >>>>> >>>>> ...Jim Thompson >>>>> >>>> >>>> If I were more motivated I'd try it on TINA, but given the complexity of >>>> the model I'd be surprised if they'd fix it even if it's broken on TINA >>>> as well. >>>> >>>> Cheers >>>> >>>> Phil Hobbs >>> >>> The SUBCKT looks to be an extremist's attempt at obfuscation. All >>> kinds of voltage sources, indicating either attempts at obfuscation or >>> total ignorance of how to write proper behavioral controlled sources. >>> Also a bunch of unbounded current sources that probably pull to >>> infinity at the slightest provocation >:-} >>> >>> Maybe you should choose another brand and let TI know? >>> >>> When I get the time I'll consider writing a model from the datasheet. >>> The TI SUBCKT is grotesque, to put it mildly ;-) >> >> Well, it's about 10 times longer than your normal op amp model--I was >> hoping that was a good thing. ;) >> >> The OPA140 is pretty well your perfect low-power TIA part, except for >> the 10 pF input capacitance, which is on the high side. >> >> Cheers >> >> Phil Hobbs > > When they spec 7pF common-mode input capacitance, is that from _each_ > input pin, or combined? > > ...Jim Thompson >I've been trying to get a straight answer to that question from various manufacturers for a long time, so far without success. My guess (after discussion with JL) is that they just hang a 3-terminal capacitance meter from one input to the supply common and call that "common mode", and then do the same between inputs, and call that "differential mode". Measuring those values in normal circuit operation can be a bit subtle. Normally what I care about is the total capacitive loading on the summing junction of an inverting stage, which might be C_CM, or C_diff + C_CM/2, or something else. The best method is probably to put a pot between the SJ and the actual inverting input, and measure the phase margin with various values of resistance. I should have bugged Jim Williams for an app note while I had the chance. :( 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 ●August 13, 20152015-08-13
On Sun, 9 Aug 2015 16:24:53 -0700 (PDT), Phil Hobbs <pcdhobbs@gmail.com> wrote:>Hi, Jim, > >If you've got some spare bandwidth, I could really use an improved OPA2140 model. > >It's a very good part for moderate speed, low current, high linearity TIAs, but it's a bear to simulate because the model won't converge on an operating point in LTspice. > > I took the TINA model TI supplies and replaced the VSWITCH cards with LTspice SW cards with negative hysteresis (more or less like your fave tanh curve), but it still won't converge on ac or noise sims. > >Thanks > >Phil HobbsFit to the AC and slew-rate characteristics... <http://www.analog-innovations.com/SED/Fit_OPA2140.pdf> No swing limits yet, but I'm only up to 35 lines >:-} ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 13, 20152015-08-13
On Thu, 13 Aug 2015 16:19:09 -0700, Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:>On Sun, 9 Aug 2015 16:24:53 -0700 (PDT), Phil Hobbs ><pcdhobbs@gmail.com> wrote: > >>Hi, Jim, >> >>If you've got some spare bandwidth, I could really use an improved OPA2140 model. >> >>It's a very good part for moderate speed, low current, high linearity TIAs, but it's a bear to simulate because the model won't converge on an operating point in LTspice. >> >> I took the TINA model TI supplies and replaced the VSWITCH cards with LTspice SW cards with negative hysteresis (more or less like your fave tanh curve), but it still won't converge on ac or noise sims. >> >>Thanks >> >>Phil Hobbs > >Fit to the AC and slew-rate characteristics... > ><http://www.analog-innovations.com/SED/Fit_OPA2140.pdf> > >No swing limits yet, but I'm only up to 35 lines >:-} > > ...Jim ThompsonVersus 467 lines in TI's model. And I'm just fitting graphs in the datasheet, so no relation to internal components and no obfuscation needed. ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 15, 20152015-08-15
On Fri, 14 Aug 2015 20:15:04 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>On 8/13/2015 7:38 PM, Jim Thompson wrote: >> On Thu, 13 Aug 2015 16:19:09 -0700, Jim Thompson >> <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote: >> >>> On Sun, 9 Aug 2015 16:24:53 -0700 (PDT), Phil Hobbs >>> <pcdhobbs@gmail.com> wrote: >>> >>>> Hi, Jim, >>>> >>>> If you've got some spare bandwidth, I could really use an improved OPA2140 model. >>>> >>>> It's a very good part for moderate speed, low current, high linearity TIAs, but it's a bear to simulate because the model won't converge on an operating point in LTspice. >>>> >>>> I took the TINA model TI supplies and replaced the VSWITCH cards with LTspice SW cards with negative hysteresis (more or less like your fave tanh curve), but it still won't converge on ac or noise sims. >>>> >>>> Thanks >>>> >>>> Phil Hobbs >>> >>> Fit to the AC and slew-rate characteristics... >>> >>> <http://www.analog-innovations.com/SED/Fit_OPA2140.pdf> >>> >>> No swing limits yet, but I'm only up to 35 lines >:-} >>> >>> ...Jim Thompson >> >> Versus 467 lines in TI's model. >> >> And I'm just fitting graphs in the datasheet, so no relation to >> internal components and no obfuscation needed. >> >> ...Jim Thompson >> >Nice, thanks. I'll have a whack at it when I get out of my current C++ >programming blitz. > >Cheers > >Phil HobbsIt's fun... as I have a spare time gap I develop another function to add to my toolkit. I'll explain how I did that phase dip later ;-) ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 18, 20152015-08-18
On Thu, 13 Aug 2015 16:38:08 -0700, Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:>On Thu, 13 Aug 2015 16:19:09 -0700, Jim Thompson ><To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote: > >>On Sun, 9 Aug 2015 16:24:53 -0700 (PDT), Phil Hobbs >><pcdhobbs@gmail.com> wrote: >> >>>Hi, Jim, >>> >>>If you've got some spare bandwidth, I could really use an improved OPA2140 model. >>> >>>It's a very good part for moderate speed, low current, high linearity TIAs, but it's a bear to simulate because the model won't converge on an operating point in LTspice. >>> >>> I took the TINA model TI supplies and replaced the VSWITCH cards with LTspice SW cards with negative hysteresis (more or less like your fave tanh curve), but it still won't converge on ac or noise sims. >>> >>>Thanks >>> >>>Phil Hobbs >> >>Fit to the AC and slew-rate characteristics... >> >><http://www.analog-innovations.com/SED/Fit_OPA2140.pdf> >> >>No swing limits yet, but I'm only up to 35 lines >:-} >> >> ...Jim Thompson > >Versus 467 lines in TI's model. > >And I'm just fitting graphs in the datasheet, so no relation to >internal components and no obfuscation needed. > > ...Jim ThompsonFurther improvement... <http://www.analog-innovations.com/SED/OPA2140_LG_LP_Noise.pdf> Figured out how to nail 1/f noise... I can even twiddle the exponent ;-) I miss matching the TI noise model above ~300MHz but, given the rest of the TI model is not well-behaved, I not sure I'm the one in the wrong. ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 18, 20152015-08-18
On 08/18/2015 02:02 PM, Jim Thompson wrote:> On Thu, 13 Aug 2015 16:38:08 -0700, Jim Thompson > <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote: > >> On Thu, 13 Aug 2015 16:19:09 -0700, Jim Thompson >> <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote: >> >>> On Sun, 9 Aug 2015 16:24:53 -0700 (PDT), Phil Hobbs >>> <pcdhobbs@gmail.com> wrote: >>> >>>> Hi, Jim, >>>> >>>> If you've got some spare bandwidth, I could really use an improved OPA2140 model. >>>> >>>> It's a very good part for moderate speed, low current, high linearity TIAs, but it's a bear to simulate because the model won't converge on an operating point in LTspice. >>>> >>>> I took the TINA model TI supplies and replaced the VSWITCH cards with LTspice SW cards with negative hysteresis (more or less like your fave tanh curve), but it still won't converge on ac or noise sims. >>>> >>>> Thanks >>>> >>>> Phil Hobbs >>> >>> Fit to the AC and slew-rate characteristics... >>> >>> <http://www.analog-innovations.com/SED/Fit_OPA2140.pdf> >>> >>> No swing limits yet, but I'm only up to 35 lines >:-} >>> >>> ...Jim Thompson >> >> Versus 467 lines in TI's model. >> >> And I'm just fitting graphs in the datasheet, so no relation to >> internal components and no obfuscation needed. >> >> ...Jim Thompson > > Further improvement... > > <http://www.analog-innovations.com/SED/OPA2140_LG_LP_Noise.pdf> > > Figured out how to nail 1/f noise... I can even twiddle the exponent > ;-) > > I miss matching the TI noise model above ~300MHz but, given the rest > of the TI model is not well-behaved, I not sure I'm the one in the > wrong. > > ...Jim Thompson >Nice. Can you post the model? 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
