Definition of amp noise

[Michael A. Terrell]

Don Lancaster wrote:

There's even an ugly and unsubstantiated rumor that someone once wrote a
book on this.

Don't you just hate rumors like that? ;-)

--
Link to my "Computers for disabled Veterans" project website deleted
after threats were telephoned to my church.

Michael A. Terrell
Central Florida

 

[Don Lancaster]

Michael A. Terrell wrote:

Don Lancaster wrote:

There's even an ugly and unsubstantiated rumor that someone once wrote a
book on this.



Don't you just hate rumors like that? ;-)

There's only been around 1,300,000 of them so far.

--
Many thanks,

Don Lancaster
Synergetics 3860 West First Street Box 809 Thatcher, AZ 85552
voice: (928)428-4073 email: don@tinaja.com

Please visit my GURU's LAIR web site at http://www.tinaja.com

 

Kevin wrote:

Be aware that the quoted gain of an op-amp is the DC open loop gain and
normally the feedback is selected to bring the closed loop gain to very
much less - typically lower than 1000 per stage.

The gain will typically start dropping off at a very low frequency in
the 1Hz region.

Good input. I found the graph of Voltage Gain vs Frequency. At DC its
~151 dB (35 million). Now at 40Hz it's ~125 dB (1.8 million). That's
with a load R of 2K at 25 C. Still doesn't seem to bad. If I knew a
better chip I'd gladly use it.

Chips have a temperature range. I can understand the upper limit but
often wonder why the lower. Perhaps the material properties change too
much. Too bad, I could pack some dry ice around the components to
lower the noise. This chip is good for -50 C. Dry Ice is around -79
C.

Does anyone understand the cause of 1/f (flicker noise)? I read about
it but still haven't seen any explanation for the cause. Is there any
way to lower 1/f noise? I thought about lowering thermal noise by
packing dry ice around the RF coils and R's. I wonder if it would be
worth it. Too bad components such as op-amps can't take dry ice
temperatures. Maybe if there's a device, perhaps some optical
amplifier that could take such temperatures then one could take
advantage of the dry ice. Hey, whatever works.

Paul

 

Don Lancaster wrote:

The fact that the signal phase is known suggests that you have
access to the generator of the signal, which would make your
problem a piece of cake for a commercial lock-in amplifier.

Or, to use a much more modern term, synchronous demodulation.

CMOS 4016 ferinstance.

Don, that would be great, but are you sure that's the correct part #?
I looked it up and found a switch.

Paul

 

[Bret Ludwig]

Don Lancaster wrote:

Huh?

One quarter of a ten cent CMOS 4016 hung on any opamp to make +1 -1 gain.

There's even an ugly and unsubstantiated rumor that someone once wrote a
book on this.

But it was poorly distributed and severely in need of editing.

 

[Winfield Hill]

pmlonline@gmail.com wrote...

Don Lancaster wrote:
The fact that the signal phase is known suggests that you have
access to the generator of the signal, which would make your
problem a piece of cake for a commercial lock-in amplifier.

Or, to use a much more modern term, synchronous demodulation.

CMOS 4016 ferinstance.

Don, that would be great, but are you sure that's the correct part #?
I looked it up and found a switch.

That's correct, it's a switch, as Don intended.

A cd4066 is better than a 4016, and a 74hc4053 is even better yet.


--
Thanks,
- Win

 

Winfield Hill wrote:

That's correct, it's a switch, as Don intended.

A cd4066 is better than a 4016, and a 74hc4053 is even better yet.

Well, I wonder how you can make a single 4016 switch chip into a
lock-in amplifier or synchronous demodulation.

Paul

 

[Winfield Hill]

pmlonline@gmail.com wrote...

Winfield Hill wrote:
That's correct, it's a switch, as Don intended.

A cd4066 is better than a 4016, and a 74hc4053 is even better yet.

Well, I wonder how you can make a single 4016 switch chip into a
lock-in amplifier or synchronous demodulation.

Easy as pie, piece of cake.


--
Thanks,
- Win

 

[Adrian Tuddenham]

Bret Ludwig <bretldwig@yahoo.com> wrote:

Don Lancaster wrote:

Huh?

One quarter of a ten cent CMOS 4016 hung on any opamp to make +1 -1 gain.

There's even an ugly and unsubstantiated rumor that someone once wrote a
book on this.


But it was poorly distributed and severely in need of editing.

How do you edit a rumour?

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

 

Winfield Hill wrote:

pmlonline@gmail.com wrote...

Winfield Hill wrote:
That's correct, it's a switch, as Don intended.

A cd4066 is better than a 4016, and a 74hc4053 is even better yet.

Well, I wonder how you can make a single 4016 switch chip into a
lock-in amplifier or synchronous demodulation.

Easy as pie, piece of cake.

You're going to make me think? I guess you make a mixer, but not sure
how you'd make a lock-in amp. For mixer I would put switch directly
across signal output and an oscillator would control the switch. So
the 37Hz signal with short the output half the cycle. LOL, that's a
really cheesy mixer though, no?

Paul

 

[Winfield Hill]

pmlonline@gmail.com wrote...

Winfield Hill wrote:
pmlonline@gmail.com wrote...

Winfield Hill wrote:
That's correct, it's a switch, as Don intended.

A cd4066 is better than a 4016, and a 74hc4053 is even better yet.

Well, I wonder how you can make a single 4016 switch chip into a
lock-in amplifier or synchronous demodulation.

Easy as pie, piece of cake.

You're going to make me think? I guess you make a mixer, but not sure
how you'd make a lock-in amp. For mixer I would put switch directly
across signal output and an oscillator would control the switch. So
the 37Hz signal with short the output half the cycle. LOL, that's a
really cheesy mixer though, no?

No.

But clearly you didn't read the 74hc4053 datasheet. All early lock-ins
used balanced signals with reversing switches operated by a synchronous
square wave. Tens of millions of these are still made, and work well.
This approach has the advantage of being highly linear as well as being
very inexpensive.

More advanced lock-in amps use multipliers with synchronous sine-waves,
thereby avoiding the demodulation of harmonics. Even more sophisticated
lock-in amps use a DSP to perform the multiplication, with a digitally-
generated sine wave.


--
Thanks,
- Win

 

[Dave]

pmlonline@gmail.com wrote:

Winfield Hill wrote:

pmlonline@gmail.com wrote...

Could someone so kindly tell me exactly what 1.3nV / Sqrt(Hz) @ 1KHz
means on a differential amp spec? What would the noise be between
37.00 Hz and 37.01 Hz? Would it be 1.3nV / Sqrt(37.00 - 37.01) or
1.3nV / Sqrt(37)

That's 1.3nV times sqrt 0.01Hz.



Thanks, it makes sense. So at 0.01 Hz bandwidth it comes to 130pV.
Where does the noise come from? Is it purely thermal or just a
byproduct of the amplification process?

To a very high degree of accuracy, thermal noise power is

P = k T B.

(where k is Boltmann's constant and T tempeature in Kelvin, B is
bandwidth in Hz).


There are corrections you need to apply at very high frequencies (over 1
THz) or very low tempeatures (1 K or less), but that is the thermal
noise power. Contry to popular belive, even at absolute zero there is
thermal noise, but it is insignficant except at opetical frequencies.

The amplifier will add some of its own noise.

Here's my problem. I am trying to get a rough guestimate of the values
to see if it is even plausible. I am estimating that a signal will be
picking up ~ 1pV of RF at 37Hz on a coil.

As somone else has said, 37Hz is audio not RF. If the system is RF
modulated at 37Hz, you will need to demodulate first, but a lock-in can
still be used, but I suggest a dual phase unit is purchased.

Others have given you links to articles on lock-ins, but here are a few
commments of mine. They are no doubt in the articles too, but they are
important so I will restate a few.

* You need to decide if you need a single or dual phase lock-in amp. If
you know the phase of the detected signal, a single phase unit is OK.
But for a lab instrument, I would *seriously* look at buying a dual
phase unit. They are a lot more useful and remove a lot of the
uncertainty that you get with a single phase unit.

* Single phase lock-in amps are often on eBay with a 'buy it now' of
under $100. Needless to say, dual phase ones are more expensive. The
dual phase units tend to be more modern too, which again adds to their
price.

* EG&G and Stanford are the two big manufactuers of them. There are a
few more.

* I would suggest if you are demodulating RF to get the 37Hz, a dual
phase one will be safer, as you don't need to consider what phase shifts
might have occured in the demodulation process.

* Most moden lock-ins will have GPIB and/or RS232 serial so you can read
the data down. Old ones will lack this.

* You obviously need a source of modulation to drive your signal with.
Some lock-ins have a signal generator built in, which you can use if you
wish, although you can always use an external one. Some units have no
internal signal geneator.

* One of the big lock-in manufacturers (not sure if EG&G or Stanford)
make a single channel unit with no controls on the panel. It is all done
via a PC. I've not used one.

* There may well be a book written about how to build a lock-in by
switching an op-amp to have a gain or +1 or -1, but there is certainly
an old paper on it written in India. I don't have a reference to it
off-hand, but could find it. That used FETs across the resistors in the
op-amp configuration, effecitvely making the resistors shorts or of the
required resistance. It did work. I built it and got a dynamic reserve
of around 70 dB.

* Systems using a square wave are suseptable to odd harmonis of the
signal. So if there is any 3*27=81Hz around, then that will be detected
too. Systems working on sine waves don't have that problem.

* Stanford is the only manufacturer of a lockin that works above 3MHz.

Also I would know the phase of the 37Hz signal. Could that be
an aid? The end goal is to design a device that can tell if the 37Hz
1pV signal is on or off. Is this doable? I think time is on our side.
That is, if we build a filter with high enough Q then it could detect
the signal. The problem is that the Q would probably have to be so
high that it may take a long time for the signal to rise above the
noise.

You will not avoid that with the lock-in. The lock-in is basically a
high Q tunable bandpass filter. The output is converted to a DC level. T
o reduce the noise, you must make the filter have a low cutoff
frequency. The time constant of the filters is set on the lock-in.
Normally (assuming a 6dB/octave filter) you have to wait 5x the time
constant for the measurement to settle to make an accurate reading (that
gets it to 99%). In your case, you can do a bit less than that, as you
only need an on/off status.

 

[Richard the Dreaded Liber]

On Fri, 12 Aug 2005 07:35:59 -0400, Spehro Pefhany wrote:

On Fri, 12 Aug 2005 05:15:57 +0100, the renowned Pooh Bear
pmlonline@gmail.com wrote:
Winfield Hill wrote:
pmlonline@gmail.com wrote...
Could someone so kindly tell me exactly what 1.3nV / Sqrt(Hz) @ 1KHz
means on a differential amp spec? What would the noise be between
37.00 Hz and 37.01 Hz? Would it be 1.3nV / Sqrt(37.00 - 37.01) or
1.3nV / Sqrt(37)

That's 1.3nV times sqrt 0.01Hz.

Thanks, it makes sense. So at 0.01 Hz bandwidth it comes to 130pV.
Where does the noise come from? Is it purely thermal or just a
byproduct of the amplification process?

Here's my problem. I am trying to get a rough guestimate of the values
to see if it is even plausible. I am estimating that a signal will be
picking up ~ 1pV of RF at 37Hz on a coil.

37Hz isn't an RF frequency. It's in the audio band !

As are the 76 and 82Hz ELF/SLF frequencies known to be used by the US
and Russian navies respectively to communicate with submerged nuclear
attack submarines.

How much money would it free up for social programs, space exploration,
and international relations, if we dismantled the nuke subs and used the
material to make stuff like power plants and other peaceful stuff? Y'know,
the old "beat your swords into plowshares" thing?

Thanks,
Rich

 

[Apostrophe Police]

On Thu, 11 Aug 2005 19:02:05 -0700, Winfield Hill wrote:

pmlonline@gmail.com wrote...
....
... a transformer should also be fairly noise free-- ...

Correct, it's low-frequency noise is given by its dc resistance.
But that won't abrogate the reality of ambient E-M signal noise.

One Demerit to Winfield Hill, for one extraneous apostrophe in an
otherwise possessive "its", and FIVE Punitive Demerits, because
Win Hill is one of our Gurun.
--
Rich Grise, Self-Appointed Chief,
Apostrophe Police

 

[Dave]

pmlonline@gmail.com wrote:

Hey everyone thanks for the great information. I found one site that
listed terrestrial noise down to 100KHz
http://www.broadcastpapers.com/radio/HarrisLowBandVHF04.htm

There are some errors/details not expalined too well there. When that
article talks about tempeature, it does not define what it means. It is
not as obvious as you think.

Don, I get a power ratio of 1e+17 and voltage ratio of 3e+8 for 170 dB.
What exactly does it mean to have a 170 dB above ktb noise?

P = k T B

is the noise power from a conductor (be it 1 or 1,000,000 Ohms). P is in
Watts, k is Boltzmann's constant, T is the *source* temperature (Kelvin)
and B bandwidth (Hz).

Often the source is at about 298k (room tempeature), but sometimes it is
a lot lower. Point a highly directional microwave antenna at the sky and
you get an average tempeature around a few K (not sure of the exact
number). At some parts it will be lower.

Point it towards the moon or Sun, and you will get a lot more. Point it
at the earth, and it will be about 290 K (average tempeature of the earth).


For a good understanding of noise figure, noise tempeature, noise factor
and how to measure them, take a look at an Agient application note or
two. AN 57-1 is the most useful.

http://www.home.agilent.com/cgi-bin/pub/agilent/expandedresults/cp_ExpandedResults.jsp?NAV_ID=-14355.0.03&LANGUAGE_CODE=eng&contentType=Editorial&entityType=ED30&COUNTRY_CODE=GB

but -2 and -3 are worth reading too.

I
understand that ktb noise is receiver input noise. I see the noise
graph in above link also uses that terminology; e.g. over ktb. So 170
dB over ktb means the terrestrial noise is 3e+8 times greater than
receiver input noise? The noise graph shows dB increasing as frequency
decreases. Shouldn't it be the opposite? Perhaps those numbers are
negative; e.g., -120dB. According to the graph your 170 dB seems
reasonable. Have you seen any graphs that come close to 37Hz?

I would not put too much faith in them. Man made noise depens very much
on location and will chage over time. Bluetooth, mobile phones etc will
put it up very much as these technologies develope.

I've no idea what it would be like at 37Hz. But at that, you wil not be
able to make a high gain antenna.

If you can cool your detector, and all what it sees (which is basically
everthing) you can reduce the kTB bit. Liquid nitrogen is commooly used.

Last, how do they figure the terrestrial noise values? Is it based on
a dipole antenna at a fixed length or what? For example, wouldn't a
200-meter dipole pick up more noise than say a 10 cm radius 100-turn
coil?

Yes, but it might not scale the way you think. A 20 dB (relative to a
dipole) gain Yagi will probaby pick up no more/less than dipole, which
by definition (of dBd) has 0 dB of gain. The reasona being the high
gain anteann will pick up more from one directiona than another. So you
gain 100x more power from one directiona, but make the direction ony a
few degrees, rather than your 360 degree.

Thanks,
Paul

 

[Winfield Hill]

Apostrophe Police wrote...

On Thu, 11 Aug 2005 19:02:05 -0700, Winfield Hill wrote:
pmlonline@gmail.com wrote...
...
... a transformer should also be fairly noise free-- ...

Correct, it's low-frequency noise is given by its dc resistance.
But that won't abrogate the reality of ambient E-M signal noise.

One Demerit to Winfield Hill, for one extraneous apostrophe in an
otherwise possessive "its", and FIVE Punitive Demerits, because
Win Hill is one of our Gurun.

Right. I know well the correct usage, and an improper usage grates
on me when I see it. Yet somehow my typing mode sticks an extraneous
apostrophe in every now and then, which I don't see it until it's too
late, if at all.


--
Thanks,
- Win

 

[Don Lancaster]

pmlonline@gmail.com wrote:

Don Lancaster wrote:

The fact that the signal phase is known suggests that you have
access to the generator of the signal, which would make your
problem a piece of cake for a commercial lock-in amplifier.

Or, to use a much more modern term, synchronous demodulation.

CMOS 4016 ferinstance.



Don, that would be great, but are you sure that's the correct part #?
I looked it up and found a switch.

Paul

The switch changes the gain of the opamp from +1 to -1 and does so in
synchronization with the positive and negative signal peaks, thus
performing full wave rectification.

This used to be called a lockin amplifier.
A more modern term is synchronous demodulation.

Signals other than the synchronized one decorrelate with time.

Why don't you try to find out who wrote that book on the 4016 and then
read it?

--
Many thanks,

Don Lancaster
Synergetics 3860 West First Street Box 809 Thatcher, AZ 85552
voice: (928)428-4073 email: don@tinaja.com

Please visit my GURU's LAIR web site at http://www.tinaja.com

 

[Don Lancaster]

Bret Ludwig wrote:

Don Lancaster wrote:

Huh?

One quarter of a ten cent CMOS 4016 hung on any opamp to make +1 -1 gain.

There's even an ugly and unsubstantiated rumor that someone once wrote a
book on this.



But it was poorly distributed and severely in need of editing.

Only went seven languages and one and a quarter million copies so far.

--
Many thanks,

Don Lancaster
Synergetics 3860 West First Street Box 809 Thatcher, AZ 85552
voice: (928)428-4073 email: don@tinaja.com

Please visit my GURU's LAIR web site at http://www.tinaja.com

 

[Don Lancaster]

pmlonline@gmail.com wrote:

Winfield Hill wrote:

pmlonline@gmail.com wrote...

Winfield Hill wrote:

That's correct, it's a switch, as Don intended.

A cd4066 is better than a 4016, and a 74hc4053 is even better yet.

Well, I wonder how you can make a single 4016 switch chip into a
lock-in amplifier or synchronous demodulation.

Easy as pie, piece of cake.


You're going to make me think? I guess you make a mixer, but not sure
how you'd make a lock-in amp. For mixer I would put switch directly
across signal output and an oscillator would control the switch. So
the 37Hz signal with short the output half the cycle. LOL, that's a
really cheesy mixer though, no?

Paul

Sigh.
You use the 4016 to configure an opamp as a +1 -1 gain.
Then synchronously switch in phase with the positive and negative signal
peaks.

One of the CMOS cameos.


--
Many thanks,

Don Lancaster
Synergetics 3860 West First Street Box 809 Thatcher, AZ 85552
voice: (928)428-4073 email: don@tinaja.com

Please visit my GURU's LAIR web site at http://www.tinaja.com

 

[Dave]

pmlonline@gmail.com wrote:

Chips have a temperature range. I can understand the upper limit but
often wonder why the lower. Perhaps the material properties change too
much. Too bad, I could pack some dry ice around the components to
lower the noise. This chip is good for -50 C. Dry Ice is around -79
C.

Can you get a mil-spce version? It might operate to lower temperatures.
The package is made of different things, and they will all expand at
different rates. The lower temperature limit might be set by that.

It is not uncommon to cool to FET preamplifers to liquid helium
temperatures or below.

Does anyone understand the cause of 1/f (flicker noise)?

I think it is actually f^a, where as is about -1. Despite being called
1/f a lot of the time, I think the constant a is a bit more or less than
-1. (Can't recall which).

I have never seen a satisfactory explanation.
I read about

it but still haven't seen any explanation for the cause. Is there any
way to lower 1/f noise?

No idea, other than raising the frequency. There may be other ways.

I thought about lowering thermal noise by
packing dry ice around the RF coils and R's. I wonder if it would be
worth it.

Probably, but liquid nitrogen would be a lot better. That will produce
only 25% of the thermal noise as room temperature, but dry ice will
produce 65% as much noise as at room temp.

Liquid helium is more difficult to get, more expensive, so is probably out.

I've stuck things in liquid nitrogen and never had problems with
operation. Resistors change by a small amount, but it is pretty small
for modern devices.

Too bad components such as op-amps can't take dry ice
temperatures. Maybe if there's a device, perhaps some optical
amplifier that could take such temperatures then one could take
advantage of the dry ice. Hey, whatever works.

Paul