Bad noise performance of LT1028 in LTSpice?

[Stuart Brorson]

Greetings, LTSpice mavens,

I am having a problem with evaluating the noise performance
of an LT1028. This part is supposed to be extremely
quiet (1nV/rtHz @ 1kHz). However, when I simulate a
simple circuit using LTSpice, I get a noise density of
a little more than 1uV/rtHz at the output of the circuit
(i.e. plotting onoise when doing a .noise analysis).
Please take a look at the attached LTSpice schematic
to see what I am simulating.

I can't believe that LTSpice is providing me with wrong
numbers; LT wouldn't have bad values for its own part.
However, I am puzzled about what I might be doing wrong
in this circuit. Does anybody have any idea why the
output noise is so high?

Thanks,

Stuart



Version 4
SHEET 1 900 960
WIRE -928 96 -928 128
WIRE -928 208 -928 240
WIRE -1056 96 -1056 128
WIRE -1056 208 -1056 240
WIRE -432 304 -432 352
WIRE -432 432 -432 496
WIRE -432 496 -544 496
WIRE -560 240 -480 240
WIRE -640 240 -720 240
WIRE -720 240 -736 240
WIRE -176 240 -208 240
WIRE -304 352 -304 240
WIRE -304 240 -384 240
WIRE -304 416 -304 496
WIRE -304 496 -432 496
WIRE -736 512 -784 512
WIRE -784 512 -784 592
WIRE -784 592 -208 592
WIRE -208 592 -208 240
WIRE -208 240 -256 240
WIRE -1088 384 -1088 352
WIRE -1088 352 -1152 352
WIRE -1088 464 -1088 480
WIRE -1088 592 -1088 624
WIRE -736 480 -864 480
WIRE -256 240 -304 240
WIRE -16 240 -96 240
WIRE -208 640 -208 592
WIRE -208 720 -208 768
WIRE -208 880 -208 848
WIRE -704 464 -704 432
WIRE -704 432 -688 432
WIRE -704 528 -704 560
WIRE -704 560 -688 560
WIRE -1088 480 -1088 512
WIRE -960 528 -960 480
WIRE -960 480 -1088 480
WIRE -864 528 -864 480
WIRE -864 480 -912 480
WIRE -864 592 -864 624
WIRE -960 592 -960 624
WIRE -1152 352 -1264 352
WIRE -1264 400 -1264 352
WIRE -1264 480 -1264 512
WIRE -912 480 -960 480
WIRE -544 496 -672 496
WIRE -16 272 -16 240
WIRE -16 416 -16 352
FLAG -1056 96 VCC
FLAG -928 96 VEE
FLAG -928 240 0
FLAG -1056 240 0
FLAG -720 240 VCC
FLAG -1088 624 0
FLAG -256 240 VREF_758V
FLAG -208 880 0
FLAG -688 432 VEE
FLAG -688 560 VCC
FLAG -864 624 0
FLAG -960 624 0
FLAG -1152 352 VREF_10V
FLAG -1264 512 0
FLAG -912 480 VLT1028In
FLAG -544 496 VLT1028Out
FLAG -16 416 0
SYMBOL res -80 224 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R3
SYMATTR Value 115
SYMBOL Opamps\\LT1028 -704 560 M180
SYMATTR InstName U2
SYMBOL npn -480 304 R270
SYMATTR InstName Q1
SYMATTR Value 2N2222
SYMBOL voltage -1056 112 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V2
SYMATTR Value 12
SYMBOL voltage -928 224 R180
WINDOW 0 24 104 Left 0
WINDOW 3 24 16 Left 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V3
SYMATTR Value 12
SYMBOL res -416 448 R180
WINDOW 0 36 76 Left 0
WINDOW 3 36 40 Left 0
SYMATTR InstName R5
SYMATTR Value 100
SYMBOL res -544 224 R90
WINDOW 0 0 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName R6
SYMATTR Value 50
SYMBOL cap -320 352 R0
SYMATTR InstName C1
SYMATTR Value .015F
SYMBOL res -1104 368 R0
SYMATTR InstName R7
SYMATTR Value 1.5K
SYMBOL res -1104 496 R0
SYMATTR InstName R8
SYMATTR Value 4.75K
SYMBOL res -224 624 R0
SYMATTR InstName R9
SYMATTR Value 2K
SYMBOL res -224 752 R0
SYMATTR InstName R10
SYMATTR Value 400
SYMBOL cap -976 528 R0
SYMATTR InstName C2
SYMATTR Value 475F
SYMATTR SpiceLine Rser=.1
SYMBOL cap -880 528 R0
SYMATTR InstName C3
SYMATTR Value .015F
SYMATTR SpiceLine Rser=.1
SYMBOL voltage -1264 384 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value 10V
SYMBOL res -32 256 R0
SYMATTR InstName R1
SYMATTR Value 10K
TEXT -792 672 Left 0 !.op
TEXT -800 704 Left 0 !.noise V(VLT1028Out) V1 dec 10 1Hz 1MegHz

 

[Mike Engelhardt]

Stuart,

I am having a problem with evaluating the
noise performance of an LT1028. This part
is supposed to be extremely quiet (1nV/rtHz
@ 1kHz).

The LT1028 and later versions of that part
do have 1nV/Hz½ equiv. input noise voltage,
but the opamp macromodels don't model noise.

I recommend that you use the UniversalOpamp
symbol and set the parameters for en, in,
enk and ink for the equiv. input noise voltage,
current, and corner frequencies.

--Mike

 

[Stuart Brorson]

: Stuart,

:> I am having a problem with evaluating the
:> noise performance of an LT1028. This part
:> is supposed to be extremely quiet (1nV/rtHz
:> @ 1kHz).

: The LT1028 and later versions of that part
: do have 1nV/Hz? equiv. input noise voltage,
: but the opamp macromodels don't model noise.

: I recommend that you use the UniversalOpamp
: symbol and set the parameters for en, in,
: enk and ink for the equiv. input noise voltage,
: current, and corner frequencies.

Mike,

Thanks. I have done just that & I am now happy with my simulation.

BTW: I went looking around the LTSpice group on Yahoo, and couldn't
find any documentation about what the parameters for the universal
opamp were, or what their units were. I did find some comments on
this newsgroup about some of the parameters. Therefore, I took the
liberty of creating a .doc table with this info on it, and I uploaded
it to the /temp folder on the Yahoo groups website. Hopefully, this
table will be useful to others, and might even make it into the next
version of the LTSpice manual!

Meanwhile, if any of my info on the table is wrong, or if there are
other parameters which should be included, please do not hesitate to
include them in the table.

Thanks again,

Stuart

 

[Stuart Brorson]

: : Stuart,

: :> I am having a problem with evaluating the
: :> noise performance of an LT1028. This part
: :> is supposed to be extremely quiet (1nV/rtHz
: :> @ 1kHz).

: : The LT1028 and later versions of that part
: : do have 1nV/Hz? equiv. input noise voltage,
: : but the opamp macromodels don't model noise.

: : I recommend that you use the UniversalOpamp
: : symbol and set the parameters for en, in,
: : enk and ink for the equiv. input noise voltage,
: : current, and corner frequencies.

: Thanks. I have done just that & I am now happy with my simulation.

Oh, I almost forgot to ask: Simulating with the universal opamp
component only converges with the level 1 model. Levels 2, and 3
fail. Any idea why?

Stuart

 

[Mike Engelhardt]

Stuart,

: Thanks. I have done just that & I am now happy with my simulation.

Oh, I almost forgot to ask: Simulating with the universal opamp
component only converges with the level 1 model. Levels 2, and 3
fail. Any idea why?

Of course not without seeing the circuit. Usually you can use a
different DC solution method to get the solution. There's an
e-mail address on the help about box.

BTW: I went looking around the LTSpice group on Yahoo, and couldn't
find any documentation about what the parameters for the universal
opamp were, or what their units were.

See the example circuit ./examples/Educational/UniversalOpamp.asc
Units are MKSA.

--Mike

 

[Dietmar Warning]

Stuart,


The LT1028 and later versions of that part
do have 1nV/Hz½ equiv. input noise voltage,
but the opamp macromodels don't model noise.

I recommend that you use the UniversalOpamp
symbol and set the parameters for en, in,
enk and ink for the equiv. input noise voltage,
current, and corner frequencies.

--Mike

On this point I must contradict, Mike. If we use the LT1028N model from
your lib you have an about 20% accurate 1/f nose modelling, which is most
enough for simulation purpose. It is a little bit to optimistic, but in
this noise range you must make hand calculations in any case.
(To honour Klaus Milzner.)

Dietmar

 

[Helmut Sennewald]

On this point I must contradict, Mike. If we use the LT1028N model from
your lib you have an about 20% accurate 1/f nose modelling, which is most
enough for simulation purpose. It is a little bit to optimistic, but in
this noise range you must make hand calculations in any case.
(To honour Klaus Milzner.)

Hello Dietmar,
thanks for your hint. I had also used the LT1028 and was firstly
disappointed that it doesn't have noise modelled.
I simulated now again with your suggested LT1028N and indeed, it models
noise as specified.

There are four models for the LT1028 in the ltc.lib file.

Name Noise model
LT1028 no
LT1028A no
LT1028CS no
LT1028N yes

Best Regards,
Helmut

LTSPICE is a GUI based free and unlimited SPICE from www.linear.com/software
..
Another great feature is that you can run any of your old text based
netlists as well. LTSPICE is also 99% compatible to PSPICE regarding
the analog circuit models. Digital models are different.

There is also an independent newsgroup dedicated to it.
http://groups.yahoo.com/group/LTspice
GUI=Graphical User Interface





Test circuit/schematic "lt1028n_test.asc" . Just copy the following
text into a file named lt1028n_test.asc. OPEN this file with
LTSPICE and press RUN. You will see the voltage noise density from
0.1Hz to 1MEG. Adding a source resistor and feedback resistors
allows you to simulate your whole application.


Version 4
SHEET 1 900 960
WIRE -352 176 -352 208
WIRE -352 288 -352 320
WIRE -480 176 -480 208
WIRE -480 288 -480 320
WIRE -432 496 -560 496
WIRE -736 512 -896 512
WIRE -704 464 -704 432
WIRE -704 528 -704 560
WIRE -976 512 -1088 512
WIRE -1088 560 -1088 512
WIRE -1088 640 -1088 672
WIRE -736 480 -800 480
WIRE -800 480 -800 368
WIRE -800 368 -560 368
WIRE -560 368 -560 496
WIRE -560 496 -672 496
WIRE -432 528 -432 496
WIRE -432 608 -432 656
FLAG -480 176 VCC
FLAG -352 176 VEE
FLAG -352 320 0
FLAG -480 320 0
FLAG -1088 672 0
FLAG -704 432 VCC
FLAG -704 560 VEE
FLAG -432 656 0
FLAG -432 496 out
SYMBOL voltage -480 192 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V2
SYMATTR Value 12
SYMBOL voltage -352 304 R180
WINDOW 0 24 104 Left 0
WINDOW 3 24 16 Left 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V3
SYMATTR Value 12
SYMBOL voltage -1088 544 R0
WINDOW 123 24 132 Left 0
WINDOW 39 0 0 Left 0
SYMATTR Value2 AC 1
SYMATTR InstName V1
SYMATTR Value PULSE(-5 5 0 10n 10n 240n 500n)
SYMBOL res -992 528 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 0 56 VBottom 0
SYMATTR InstName R1
SYMATTR Value 1
SYMBOL res -448 512 R0
SYMATTR InstName R2
SYMATTR Value 10k
SYMBOL Opamps\\LT1028N -704 432 R0
SYMATTR InstName U1
TEXT -1128 208 Left 0 !.op
TEXT -1136 256 Left 0 !.noise V(out) V1 dec 100 0.1 1MEG

 

[Mike Engelhardt]

Dietmar,

On this point I must contradict, Mike. If we use
the LT1028N model from your lib you have an
about 20% accurate 1/f nose modelling, which is
most enough for simulation purpose. It is a little
bit to optimistic, but in this noise range you must
make hand calculations in any case.

Yes, I'm aware of the attempt to model noise
in that model, but it's the only opamp amp
macro model in the library with that attempt.
The UniversalOpamp gives exact noise from input
voltage and current noises and even the corner
frequencies. It's really the way to go. I'm very
comfortable contradicting your advice. I don't
think people should follow it. Besides, none of
Klaus Milzner's versions of 1028 models
meaningfully model the small signal reponse.

--Mike