EMRFD Message Archive 3851

Message Date From Subject
3851 2009-12-12 07:55:46 Tayloe Dan-P26412 Re: The most noiseless opamp available
> __________________________________________________________
> 3. The most noiseless opamp available
> Posted by: "maltiserracin" pawlud@gmail.com
maltiserracin
> Date: Thu Dec 10, 2009 2:29 pm ((PST))
>
> The AD797 is the most noiseless opamp available !
>
>
http://www.analog.com/en/other-products/militaryaerospace/ad797/products
/product.html
<http://www.analog.com/en/other-products/militaryaerospace/ad797/product
s/product.html>
>

As far as ultra low noise op-amps, I have usually used the LT1115 (0.9
nV/sqrt Hz), but for a couple of dollars more you can get the LT1028 or
LT1128. and move down to 0.85 nV/SqrtHz.

In my better DC receiver designs, I almost think that ultra low
distortion might almost be a better goal (IP2/IP3 performance). The
ultra low distortion parts include:

LT1115 (0.9 nV/sqrt Hz) - THD+N (0.0002%) Use as a reference

AD797 (0.9 nV/sqrt Hz) - THD+N (0.0001%) Also use as a reference

LM4562 (2.7 nV/Sqrt/Hz) - THD+N (0.00003%)

LME49713 (1.9 nV/SqrtHz) - THD+N (0.00008%)

OPA1611 (1.1 nV/SqrtHz) - THD+N (0.000015%)

OPA211 (1.1 nV/SqrtHz) - THD+N (0.000016%)

Try to do that with a single transistor!

When you look at the voltage noise alone, it is a bit misleading. The
total noise takes into account both the voltage noise and the current
noise. As the voltage noise goes down, the current noise tend to go up.
You have to really watch the source impedance to really made best use of
these guys.

Someone gave me a spreadsheet that takes both into account and computes
the optimum source impedance. In my case, using the pre-amp as the
first amplifier out of the receiver detector, you can play with the
source impedance level by playing with the RF input transformer
step-up/step down ratio to vary the source impedance.

There are lots of games and trades that can be done in that space.

- Dan, N7VE


[Non-text portions of this message have been removed]
3852 2009-12-12 11:22:54 Andy Re: The most noiseless opamp available
> Someone gave me a spreadsheet that takes both into account and computes
> the optimum source impedance.

I doubt it takes a spreadsheet. A slide rule, or pencil-and-pencil,
does just fine.

If I'm not mistaken, the optimum source impedance (giving you the
lowest effective input noise power or noise figure) is found by
dividing the input voltage noise by the input current noise. Each
tends to vary by frequency so you may want to look at that. Maybe
that's what the spreadsheet was for.

Of course this only works if you can transform your signal source to
that optimum impedance, lossless-ly; say by changing the turns ratio
of a matching transformer (as we used to do for low-noise audio
preamps). It doesn't work to add a series or shunt resistor to your
source to bring it to the optimum impedance.

Andy
3853 2009-12-12 11:50:03 Tayloe Dan-P26412 Re: The most noiseless opamp available
> Someone gave me a spreadsheet that takes both into account and
computes
> the optimum source impedance.

I doubt it takes a spreadsheet. A slide rule, or pencil-and-pencil,
If I'm not mistaken, the optimum source impedance (giving you the
lowest effective input noise power or noise figure) is found by
dividing the input voltage noise by the input current noise. Each
tends to vary by frequency so you may want to look at that. Maybe
that's what the spreadsheet was for.

Of course this only works if you can transform your signal source to
that optimum impedance, lossless-ly; say by changing the turns ratio
of a matching transformer (as we used to do for low-noise audio
preamps). It doesn't work to add a series or shunt resistor to your
source to bring it to the optimum impedance.

Andy


It is true that the optimum point is indeed the noise voltage divided by
the noise current. However, the value of the spreadsheet was to
calculate the resulting noise figure of various op-amps over a range of
impedances, which produced a family of curves that allowed you to pick
the best op-amp for a particular source impedance. Since we typically
don't have infinite freedom in the choices of impedance transformations
(such as a limited of number of turns ratio combinations), I found these
curves useful.

- Dan, N7VE



[Non-text portions of this message have been removed]
3855 2009-12-12 12:49:11 victor Re: The most noiseless opamp available
You do need to do more analysis than just divide the OpAmp Vnoise to Inoise. You should input to the spreadsheet also the gain defining resistors and any other resistor in the circuit that affects the amplifier stage noise (it is a different case for an inverting and non-inverting gain stage) and then you get the equivalent input (or output) noise and the optimum generator noise for the whole circuit. Another case would be for a current feedback amplifier which has different Inoise for the inverting and non-inverting inputs.
Victor - 4Z4ME

3863 2009-12-13 15:30:12 ehydra Re: The most noiseless opamp available
So, is a single transistor more effective than a OpAmp if someone is
interested in lowest noise? I wonder because an OpAmp always include in
the input stage at least two separate transistors. And because it runs
in closed-loop condition, the second transistor should contribute factor
sqrt(2).

I think for a low-if (<100kHz) or baseband processing receiver the
NE5534 is a good decision in ratio between price to performance. I don't
have a practical experience there.


A single transistor claimed for lowest-NF is the IF3601. Ref:
http://www.sg-acoustics.ch/analogue_audio/parts/index.html


What others think?


regards -
Henry


--
ehydra.dyndns.info


victor schrieb:
> You do need to do more analysis than just divide the OpAmp Vnoise to Inoise. You should input to the spreadsheet also the gain defining resistors and any other resistor in the circuit that affects the amplifier stage noise (it is a different case for an inverting and non-inverting gain stage) and then you get the equivalent input (or output) noise and the optimum generator noise for the whole circuit. Another case would be for a current feedback amplifier which has different Inoise for the inverting and non-inverting inputs.
> Victor - 4Z4ME
3864 2009-12-13 15:43:10 Dave - WB6DHW Re: The most noiseless opamp available
A number of direct conversion receivers using DBM's use a common base
amplifier. 2N4401's are very low noise and also very cheap. That is
what I use on my UHFSDR.

Dave - WB6DHW
<http://wb6dhw.com>

ehydra wrote:
> So, is a single transistor more effective than a OpAmp if someone is
> interested in lowest noise? I wonder because an OpAmp always include in
> the input stage at least two separate transistors. And because it runs
> in closed-loop condition, the second transistor should contribute factor
> sqrt(2).
>
> I think for a low-if (<100kHz) or baseband processing receiver the
> NE5534 is a good decision in ratio between price to performance. I don't
> have a practical experience there.
>
>
> A single transistor claimed for lowest-NF is the IF3601. Ref:
> http://www.sg-acoustics.ch/analogue_audio/parts/index.html
>
>
> What others think?
>
>
> regards -
> Henry
>
>
>
3874 2009-12-14 07:01:07 ehydra Re: The most noiseless opamp available
Hello Dave -

Your UHFSDR is nice looking design work. Especially the 4-phase circuit.

Thanks -
Henry


PS:
> ehydra wrote:
>> A single transistor claimed for lowest-NF is the IF3601. Ref:
>> http://www.sg-acoustics.ch/analogue_audio/parts/index.html
>>

I must correct that lowest-NF is not lowest voltage noise as common for
audio apps.


--
ehydra.dyndns.info



Dave - WB6DHW schrieb:
> A number of direct conversion receivers using DBM's use a common base
> amplifier. 2N4401's are very low noise and also very cheap. That is
> what I use on my UHFSDR.
3877 2009-12-16 15:05:36 ep_mand Re: The most noiseless opamp available
Just for fun, I also have been searching devices for lowest AF noise figure, especially at 50 ohms input. Some results (1 kHz):
2SC3329 npn: << 1 dB [Best bipolar?]
2SA1084 pnp: 1 dB
2SC2546 npn: < 1 dB.

Then, at 100 ohms (50 ohms data missing, must be over 1 dB):
2SK170 nFET: 0.5 dB
2SJ74 pFET: 0.5 dB
(2SK369 nFET 2 dB, only)

Only voltage noise data found, 100Hz (even better at 1 kHz):
IF3601 nFET: 0.3 nV / (sq.root Hz) [Best FET?]

1 kHz noise voltages (divided by sq.root Hz):
SSM2220 and MAT-3 pnp pairs, 2SA1084, 2SC2546: all 0.5 nV
(2SC3329 must have about 0.4 nV)
2SK369: 0.7 nV
2SJ74: < 0.8 nV
2SK170: 0.9 nV

If input impedance is variable, it might be best to use FETs as they behave well at constant drain current, and get even better at higher impedance, especially IF3601 and 2SK369. Best bipolars are quite good at high Z but only if collector current is reduced to 1 - 10 microamps. (50 - 100 ohms needs about 6 - 10 mA.)

As there are "zillions" of semiconductors (and more coming), somebody may already have found better devices than these.
3878 2009-12-16 16:07:19 w4zcb Re: The most noiseless opamp available
As there are "zillions" of semiconductors (and more coming), somebody may already have found better devices than these.

Charley Wenzel, who has made a company and a career out of low noise, uses the 2SK369 for the lowest noise. I talked with him one day, perhaps 5-6 years ago, and asked him if there was a better transistor out to use in his phase noise test amplifier which is shown schematically on his web page. His response was that if there was, he hadn't found it yet. I have a couple dozen if you'd like to evaluate it.

W4ZCB

[Non-text portions of this message have been removed]
3879 2009-12-16 17:08:15 Chris Trask Re: The most noiseless opamp available
>
> As there are "zillions" of semiconductors (and more coming), somebody may
> already have found better devices than these.
>
> Charley Wenzel, who has made a company and a career out of low noise, uses
> the 2SK369 for the lowest noise. I talked with him one day, perhaps 5-6
> years ago, and asked him if there was a better transistor out to use in
his
> phase noise test amplifier which is shown schematically on his web page.
His
> response was that if there was, he hadn't found it yet. I have a couple
> dozen if you'd like to evaluate it.
>

Sure glad you made mention of that device. I was looking at an
interference canceller and looking for ways to improve it. I'll have to
locate some of these and see how well they would play as a current-boosted
source follower. Capacitances are a bit high for HF applications, but then
the device is oriented towards low-level audio. If used cascode with a
2N4401 you could improve the frequency response and still have the low
noise.

There's some available you-know-where, but they're not exactly giving
them away at $1.00 to $1.80 each.

There's a number of low-noise bipolar devices made by NEC that are
readily available from Digi-Key and Mouser. I use the NE46134 and NE68133
fairly regularly for low noise HF/VHF designs.

Chris

,----------------------. High Performance Mixers and
/ What's all this \ Amplifiers for RF Communications
/ extinct stuff, anyhow? /
\ _______,--------------' Chris Trask / N7ZWY
_