High gain current amplifier has excess noise

[alan]

Hello people,

I'm building a very simple current amplifier consisting of a LT1793 and
a 10G resistor. I took reasonable precautions to reduce stray
capacitance and leakage currents (e.g. the - pin is not inserted into
the op-amp socket). It works ok, except at frequencies above a few tens
of Hz there is excess noise over the expected johnson noise of the
resistor. It seems to level off to about 25ish uV/rtHz by ~100Hz. I
have nothing hooked up to the input of the amp, so the input voltage
noise should not be a problem. Also, the input current noise is only
supposed to be .8fA/rtHz or so, so that shouldn't be a problem either.
Right now, the amp is powered by 7V batteries, so power supply shouldn't
be a problem either. I suspect it might have something to do with
"something is coupling to something else", or that there are some output
pathologies with running the amp at such a high gain. Due to the self
capacitance of the resistor, the amp has a rolloff near 400Hz or so.

Any ideas what the problem is?

 

[Winfield Hill]

alan wrote...

Hello people,

I'm building a very simple current amplifier consisting of a LT1793
and a 10G resistor. I took reasonable precautions to reduce stray
capacitance and leakage currents (e.g. the - pin is not inserted into
the op-amp socket). It works ok, except at frequencies above a few
tens of Hz there is excess noise over the expected johnson noise of
the resistor. It seems to level off to about 25ish uV/rtHz by ~100Hz.
I have nothing hooked up to the input of the amp, so the input voltage
noise should not be a problem. Also, the input current noise is only
supposed to be .8fA/rtHz or so, so that shouldn't be a problem either.
Right now, the amp is powered by 7V batteries, so power supply shouldn't
be a problem either. I suspect it might have something to do with
"something is coupling to something else", or that there are some output
pathologies with running the amp at such a high gain. Due to the self
capacitance of the resistor, the amp has a rolloff near 400Hz or so.

Any ideas what the problem is?

A common source of overlooked noise is e_n-Cin noise. This is given
by i_n = e_n Cin 2pi f, where Cin is the sum of all the capacitances.
In your case we'd have to have Cin > 200pF to get 20nV at 200Hz across
your 10G resistor, but I imagine you'd notice 200pF if it was there...

Since there's no current flowing in your 10G resistor, it shouldn't be
excess noise. By process of elimination, we have rectified RFI pickup.

 

[Joerg]

Hello Alan,

As Winfield suggested, it might also be RFI. Hang a spectrum analyzer
onto the output and look.

I had a similar situation at a client. A new Doppler system showed
excess noise. All sorts of theories wafted through the meeting room.
When they wouldn't believe me that it could be RFI I went to the lab,
played with the mixer frequency a little and after a few minutes you
could hear BBC World Service, crisp and clear. Unfortunately it was
shortly after the hour so I missed Big Ben.

Regards, Joerg

 

[qrk]

Hello people,

I'm building a very simple current amplifier consisting of a LT1793 and
a 10G resistor. I took reasonable precautions to reduce stray
capacitance and leakage currents (e.g. the - pin is not inserted into
the op-amp socket). It works ok, except at frequencies above a few tens
of Hz there is excess noise over the expected johnson noise of the
resistor. It seems to level off to about 25ish uV/rtHz by ~100Hz. I
have nothing hooked up to the input of the amp, so the input voltage
noise should not be a problem. Also, the input current noise is only
supposed to be .8fA/rtHz or so, so that shouldn't be a problem either.
Right now, the amp is powered by 7V batteries, so power supply shouldn't
be a problem either. I suspect it might have something to do with
"something is coupling to something else", or that there are some output
pathologies with running the amp at such a high gain. Due to the self
capacitance of the resistor, the amp has a rolloff near 400Hz or so.

Any ideas what the problem is?

Quickie calculations:

Input current noise:
0.8fA/rtHz * 10G = 8 uV/rtHz

Resistor noise:
sqrt(4*k*T*R) = 13 uV/rtHz

Combined noise = sqrt(8uV^2 + 13uV^2) = 15.3 uV/rtHz

If you use LTspice, that shows the noise settling down at 20 uV/rtHz
above 100 Hz. What your seeing is pretty close to expected.
What are you expecting?

 

[Phil Hobbs]

Quickie calculations:

Input current noise:
0.8fA/rtHz * 10G = 8 uV/rtHz

Resistor noise:
sqrt(4*k*T*R) = 13 uV/rtHz

Combined noise = sqrt(8uV^2 + 13uV^2) = 15.3 uV/rtHz

If you use LTspice, that shows the noise settling down at 20 uV/rtHz
above 100 Hz. What your seeing is pretty close to expected.
What are you expecting?

The input capacitance causes a rising noise gain beginning at
f=1/(2*pi*Rf*Cin). That usually dominates with large feedback resistances.

Cheers,

Phil Hobbs