Forums

Does the noise density function increase if an OpAmp goes into slew rate limit?

Started by RobertMacy May 28, 2014
On Fri, 30 May 2014 13:26:50 -0700, Jim Thompson  
<To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote:

>> ...snip... > It'll be a subcircuit library... what you do with it from there is up > to you... LTspice, OrCAD, what-have-you. > > ...Jim Thompson
Righteo. Will you know the answer to the question as to what the noise floor WITH and WITHOUT slew rate limiting will be? Example: power from +/-12Vdc, Rinput=Rfdbk=1k, Cfdbk=22pF, Rload=10k, Cload=10pF, compare driving with 7mVpk at 1MHz to driving with 7Vpk sinewave. The first should be easy with .noise; but the other...?
On Fri, 30 May 2014 15:35:19 -0700, RobertMacy
<robert.a.macy@gmail.com> wrote:

>On Fri, 30 May 2014 13:26:50 -0700, Jim Thompson ><To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote: > >>> ...snip... >> It'll be a subcircuit library... what you do with it from there is up >> to you... LTspice, OrCAD, what-have-you. >> >> ...Jim Thompson > > >Righteo. > >Will you know the answer to the question as to what the noise floor WITH >and WITHOUT slew rate limiting will be? >Example: >power from +/-12Vdc, Rinput=Rfdbk=1k, Cfdbk=22pF, Rload=10k, Cload=10pF, >compare driving with 7mVpk at 1MHz to driving with 7Vpk sinewave. The >first should be easy with .noise; but the other...?
I have no idea... you're supposed to elucidate me >:-} ...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.
On Fri, 30 May 2014 16:35:45 -0700, Jim Thompson  
<To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote:

>> ...snip... > > I have no idea... you're supposed to elucidate me >:-} > > ...Jim Thompson
We all know how painful that can be. We'll see what pops out of this analysis.
On Fri, 30 May 2014 15:35:19 -0700, RobertMacy <robert.a.macy@gmail.com>
wrote:

>On Fri, 30 May 2014 13:26:50 -0700, Jim Thompson =20 ><To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote: > >>> ...snip... >> It'll be a subcircuit library... what you do with it from there is up >> to you... LTspice, OrCAD, what-have-you. >> =09 >> ...Jim Thompson > > >Righteo. > >Will you know the answer to the question as to what the noise floor WITH=
=20
>and WITHOUT slew rate limiting will be? >Example: >power from +/-12Vdc, Rinput=3DRfdbk=3D1k, Cfdbk=3D22pF, Rload=3D10k, =
Cload=3D10pF, =20
>compare driving with 7mVpk at 1MHz to driving with 7Vpk sinewave. The =20 >first should be easy with .noise; but the other...?
OK, but i still have questions about how to separate the distortion products from the noise. I suppose that there may be a way with only moderate to heavy levels of serious analysis. There may be a way to find and account for the distortion products and subtract that from the final waveform and analyze the result for noise. May take some rather fancy tools though. ?-) =20
On Fri, 30 May 2014 19:24:06 -0700, josephkk  
<joseph_barrett@sbcglobal.net> wrote:

> On Fri, 30 May 2014 15:35:19 -0700, RobertMacy <robert.a.macy@gmail.com> > wrote: > >>> ...snip.... >> Will you know the answer to the question as to what the noise floor WITH >> and WITHOUT slew rate limiting will be? >> Example: >> power from +/-12Vdc, Rinput=Rfdbk=1k, Cfdbk=22pF, Rload=10k, Cload=10pF, >> compare driving with 7mVpk at 1MHz to driving with 7Vpk sinewave. The >> first should be easy with .noise; but the other...? > > OK, but i still have questions about how to separate the distortion > products from the noise. I suppose that there may be a way with only > moderate to heavy levels of serious analysis. There may be a way to find > and account for the distortion products and subtract that from the final > waveform and analyze the result for noise. May take some rather fancy > tools though. > > ?-) >
The principle is simple. First, drive the system with a pure sinewave. Next, assume that distortion of a repetitive waveform can ONLY be related to the frequency of that sinewave. For example, drive with 1MHz and without any distortion caused by the slew rate limit you should get 7 Vpk, but what comes out is a really weird 2Vpk signal. Now the ONLY energy you should have in that signal is at 1MHz, 2MHz, 3MHz, 4MHz, 5MHz etc. It is NOT possible to get anything else. [I think] Where would say 349kHz come from? You can't get there from here. Small note: if by examination the waveform appears to be symmetrical about the axis [top looks like mirror image of bottom]; you will have a dominance of ODD harmonics. The EVEN harmonics will be VERY small. To get rid of weird startup effects, throw away a little at the start. To simplify [but not necessary] I set up the tran statement like this: .tran 0 1.1m 0.1m 0.01us Now after making all the steps uniform using ltsputil.exe for a total of 100001 samples of your output waveform remember has a step size of 0.01 uS, the output data is the output from 0.1m to 1.1m, where the new t=0 was at 0.1m and the new t=1m was at 1.1m, and throw away t=0 data point. That leaves you with a sampled packet length of 100000. In 'FFT' signal processing that translates into Nyquist rate of 50MHz with a BW of 1kHz - very adequate for what we're doing. To compare apples to apples, run the .noise from 1kHz to 50MHz. Now take the FFT of your data and the first harmonic is down, third is close by, and so on. Since you KNOW each harmonic, simply remove it. Just as you write a batch file to run ltsputil.exe, write some functions to get rid of ALL harmonics. I purposely have noise in the signals so removing signal harmonics leaves only the noise WITHOUT the signal and I can see how the noise density function has changed. Try on something and you can convince yourself it is simple. If interested, I think I could mock up a little OpAmp circuit with some feedback diodes to clip the output, then you can see the effects on the noise caused by the non-linearity and see the 1/f effects, too. There is no real need to remove the signals and all harmonics, because the noise floor looks just like you're using a spectrum analyzer. Do you have octave, free Matlab clone?
>"RobertMacy" wrote in message news:op.xgpfkxzy2cx0wh@ajm...
> > ?-) >
>The principle is simple.
>First, drive the system with a pure sinewave. Next, assume that distortion >of a repetitive waveform can ONLY be related to the frequency of that >sinewave. For example, drive with 1MHz and without any distortion caused >by the slew rate limit you should get 7 Vpk, but what comes out is a >really weird 2Vpk signal. Now the ONLY energy you should have in that >signal is at 1MHz, 2MHz, 3MHz, 4MHz, 5MHz etc. It is NOT possible to get >anything else.
Highly unlikely, but not strictly true.
>[I think] Where would say 349kHz come from? You can't get there from >here.
The system could be chaotic. http://en.wikipedia.org/wiki/Chaos_theory The assumption of a Taylor series expansion is the problem. One bistable divider will nullify that. Kevin Aylward B.Sc. www.kevinaylward.co.uk www.anasoft.co.uk - SuperSpice
On Fri, 30 May 2014 12:34:25 -0700, Jim Thompson
<To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:

>On Fri, 30 May 2014 12:09:29 -0700, RobertMacy ><robert.a.macy@gmail.com> wrote: > >>On Thu, 29 May 2014 07:48:08 -0700, Jim Thompson >><To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote: >> >>>> ...snip... >>> >>> I'll try to find you an all-device-level OpAmp model to play with. >>> >>> ...Jim Thompson >> >>Did you find one? > >I'll design you an all-device-level OpAmp over the weekend. How's >that for service with a smile ?:-} > > ...Jim Thompson
The OpAmp I designed when I was 23 years old... .SUBCKT MC1530 IN+ IN- OUT VCC GND VEE R_R9 N_1 GND 3.2K R_R5 N_2 VEE 1.5K R_R3 N_3 VEE 2.2K R_R6 N_4 GND 1.5K R_R10 GND N_5 3.3K R_R10A N_6 VEE 100 R_R10B N_7 VEE 100 Q_Q38 N_9 N_8 N_4 Q2N3904 Q_Q37 N_10 N_10 VEE Q2N3904 Q_Q40 N_5 N_5 N_6 Q2N3904 Q_Q31 N_11 N_11 N_2 Q2N3904 Q_Q30 N_1 N_1 N_11 Q2N3904 Q_Q29 N_12 N_1 N_3 Q2N3904 Q_Q42 OUT N_10 VEE Q2N3904 5 Q_Q33 N_8 IN- N_12 Q2N3904 R_R8 VCC N_13 5K R_R4 VCC N_9 3K Q_Q35 VCC N_9 N_14 Q2N3904 Q_Q41 VCC N_13 OUT Q2N3904 5 Q_Q34 N_13 N_15 N_10 Q2N3904 Q_Q39 N_15 N_5 N_7 Q2N3904 R_R2 VCC N_8 7.75K Q_Q36 VCC N_16 N_4 Q2N3904 Q_Q32 N_16 IN+ N_12 Q2N3904 R_R1 VCC N_16 7.75K R_R11 N_15 OUT 30K R_R7 N_14 N_15 6K C_C1 N_17 N_16 33nF R_R26 N_17 N_8 3 .model Q2N3904 NPN(Is=6.734f Xti=3 Eg=1.11 Vaf=74.03 Bf=416.4 Ne=1.259 + Ise=6.734f Ikf=66.78m Xtb=1.5 Br=.7371 Nc=2 Isc=0 Ikr=0 Rc=1 + Cjc=3.638p Mjc=.3085 Vjc=.75 Fc=.5 Cje=4.493p Mje=.2593 Vje=.75 + Tr=239.5n Tf=301.2p Itf=.4 Vtf=4 Xtf=2 Rb=10) .ENDS MC1530 Watch the wrap in the .MODEL area ...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.
On Mon, 02 Jun 2014 10:28:04 -0700, Jim Thompson  
<To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote:

> ...snip... > Watch the wrap in the .MODEL area > > ...Jim Thompson
Got it, thanks. will let you know what I find out.
On Fri, 30 May 2014 16:35:45 -0700, Jim Thompson  
<To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote:

>> ...snip.... > > I have no idea... you're supposed to elucidate me >:-} > > ...Jim Thompson
I had hoped somebody had done such a test using a spectrum analyzer for confirmation. However, staying with circuit simulations and using your MC1530 OpAmp containing those 2N3904's thrown in placed into a simple inverting configurations [GND the IN+] using 10K external resistors for GAIN=-1 powered by +/-6Vdc... To find the noise density function at the output, run .noise which shows 24.1 nVrms/rtHz, as reference value. As a 'walk before run' ONLY added the noise from R7, 6K, and R11, 30k, and the external Rinput, 10k, and Rfdbk, 10k. and drive the inverter with a 2mVpk 400kHz sine signal. .tran shows very 'clean' signal, but if I run ".tranoise" the output looks more 'realistic' with an approprite noise floor for the FFT. Just these four sources produce a huge level of noise, 16.4nVrm/rtHz, [Remember, that's from just from those four sources, not including any other resistors nor NPN transistor noise!]. Next the input signal was increased to put the OpAmp into slew rate limit using 2Vpk sine wave at 400kHz. Get about 1.2 to 1.5Vpk out and looks very 'sawtooth'. The .tranoise now shows the background noise has indeed increased to more than 50nVrms/rtHz! increasing more than 3 times. from memory the same thing happened with the OpAmp macromodels which increase theirnoise floor around 10dB, also. Since I'm now comfortable with your text netlist being accurately transferred into LTspice schematic form, matching the style on the Data Sheet, I'll soon send you a copy for your website info. If you want, to be really complete, tell me which Pins go to which LEAD/LAG compensation internal components, and I'll include a 'complete' module, too.
On Fri, 30 May 2014 16:35:45 -0700, Jim Thompson  
<To-Email-Use-The-Envelope-Icon@on-my-web-site.com> wrote:

>> ..snip.... > I have no idea... you're supposed to elucidate me >:-} > > ...Jim Thompson
Didn't see any comment to yesterday's posting, showing preliminary modleing of slew rate limiting effects. Also, I have the subckt model AND the discrete model all done to send to you. just found your teaching aid .pdf with the mask photo, didn't see the CAP yet, but at least have the pin out now, so can do a 'complete' MC1530 subckt model including the lead/lag terminals. Do those 'minimum size' NPNs really act like the 3904 model? seems their areas would have to affect the model's terms in some way.