Amplifier Circuit for low frequency signals

Thread Starter

Marvin Chewchut

Joined Nov 9, 2018
4
Hi everyone,
I would like to ask for guidance regarding with my project. My project needs a digitizer circuit with a capability to amplify small signals (minimum of around microvolt) with minimum frequency response of 0.01 hz and must be up to 100 hz or above. I would like to ask what amplifier/s best to be used for this given setup. It is best that the amplifier has a adjustable gain feature so that i could maybe create a programmable gain feature for my project.
 

wayneh

Joined Sep 9, 2010
16,765
What do you mean by “digitizer”? Do you want to convert a 1 microvolt signal to 1 (5V or whatever) and less than 1 microvolt to 0?

Isolating the signal from the noise will be a big challenge.
 

Thread Starter

Marvin Chewchut

Joined Nov 9, 2018
4
What do you mean by “digitizer”? Do you want to convert a 1 microvolt signal to 1 (5V or whatever) and less than 1 microvolt to 0?

Isolating the signal from the noise will be a big challenge.
Oh sorry for the wrong inputs. Actually thats in terms of nanovolts. Im really sorry for the wrong inputs.

Our project uses a 24 bit ADC and since (correct me if im wrong) it is 24 bit, its sensitivity is only around microvolt. I would like to use an amplifier to shift the signal to a range that the ADC can sense. Im still new on this topic and I hope i will be enlighten so that im following the right path hehe
 

MrChips

Joined Oct 2, 2009
22,870
We see this too often, someone asking for help but fails to give the full picture up front. This leads to a game of 101 Q & A.

What are you trying to measure?
What is the source of your signal?
 

ebp

Joined Feb 8, 2018
2,332
Nanovolt signals are very difficult to deal with. Even the very best auto-zero amplifiers will have offset voltages of hundreds of nanovolts. Thermoelectric potentials could easily be three orders of magnitude greater than your signal with just a few degrees of temperature difference. Circuity would need to be built with extreme care to be sure all the unwanted thermocouples in the circuit compensated for each other and the circuit would need to be built to allow insulating it and protecting it from any air current or temperature gradient. Power supplies would have to be extremely stable and amplifiers would have to have high power supply rejection ratio. Thermal noise from a 1 ohm resistor at 25 °C would be about 1.3 nV RMS for a 100 Hz bandwidth - and that is for a resistor which would produce no noise other than thermal noise. A leakage resistance of 1 gigohm from 1 volt into a 1 ohm resistor will produce a voltage of 1 nV.

It is absolutely impossible to get remotely close to low nanovolt performance out of any circuit built on a plug-in breadboard.

I know the late Jim Williams wrote about ultra low level circuits in at least one of the applications notes he authored for Linear Technology. There are probably better amplifiers around now than when he wrote the ap notes, but many of the basic issues are still the same. Linear Tech is now part of Analog Devices. You may find other useful info at AD. Maxim is worth a look, since they have both amplifiers that might be usable and high resolution ADCs.
 

Thread Starter

Marvin Chewchut

Joined Nov 9, 2018
4
We see this too often, someone asking for help but fails to give the full picture up front. This leads to a game of 101 Q & A.

What are you trying to measure?
What is the source of your signal?
We were trying to build a seismometer using only an affordable +-1g accelerometer. We were trying to test if its possible to use an accelerometer as alternative sensor for a seismometer. We will amplify and digitized the accelerometer to the extend that it can read very very small signals from the accelerometer. Though we haven't started it yet, thats why im trying to search in this forum for related problem.
I would also like to ask if the project sounds plausible?

Nanovolt signals are very difficult to deal with. Even the very best auto-zero amplifiers will have offset voltages of hundreds of nanovolts. Thermoelectric potentials could easily be three orders of magnitude greater than your signal with just a few degrees of temperature difference. Circuity would need to be built with extreme care to be sure all the unwanted thermocouples in the circuit compensated for each other and the circuit would need to be built to allow insulating it and protecting it from any air current or temperature gradient. Power supplies would have to be extremely stable and amplifiers would have to have high power supply rejection ratio. Thermal noise from a 1 ohm resistor at 25 °C would be about 1.3 nV RMS for a 100 Hz bandwidth - and that is for a resistor which would produce no noise other than thermal noise. A leakage resistance of 1 gigohm from 1 volt into a 1 ohm resistor will produce a voltage of 1 nV.

It is absolutely impossible to get remotely close to low nanovolt performance out of any circuit built on a plug-in breadboard.

I know the late Jim Williams wrote about ultra low level circuits in at least one of the applications notes he authored for Linear Technology. There are probably better amplifiers around now than when he wrote the ap notes, but many of the basic issues are still the same. Linear Tech is now part of Analog Devices. You may find other useful info at AD. Maxim is worth a look, since they have both amplifiers that might be usable and high resolution ADCs.
Thank you sir for your guidance. I will take note of this and I will try to look for this devices
 
Last edited by a moderator:

crutschow

Joined Mar 14, 2008
26,329
I would also like to ask if the project sounds plausible?
It's iffy.
The noise level of your accelerometer is likely in the microvolt region making it very difficult to see nV signals. :(

What is the frequency range of the seismometer signals you want to observe?
If the bandwidth is sufficiently small, you may be able to us low-noise AC amplifiers and high-order filters to get the signal to the level you can detect
That would avoid the problem of DC offsets in the circuit.
But it's still a very teeny signal to try to detect.
 

ebp

Joined Feb 8, 2018
2,332
If you are considering MEMS accelerometers, check the specifications for hysteresis. I've never worked with such accelerometers, but I have a vague and possibly completely wrong memory of seeing something about hysteresis issues.

You would probably have to place the amplifier right at the accelerometer. Normal seismometers transducers get strung out on long cables that would be completely unworkable with nanovolt signals. Slight movement of the cable would probably produce nanovolts due to triboelectric effects.

It sounds like an interesting project, but with some serious technical issues.

The last time I had any involvement with geophones it was trying to figure out methods to keep the instrumentation from being blown by lightning strikes near the the 'phones. Apparently this can be quite a problem.
 

Thread Starter

Marvin Chewchut

Joined Nov 9, 2018
4
What is the frequency range of the seismometer signals you want to observe?
It's iffy.
The noise level of your accelerometer is likely in the microvolt region making it very difficult to see nV signals. :(

What is the frequency range of the seismometer signals you want to observe?
If the bandwidth is sufficiently small, you may be able to us low-noise AC amplifiers and high-order filters to get the signal to the level you can detect
That would avoid the problem of DC offsets in the circuit.
But it's still a very teeny signal to try to detect.
I would like to amplify signals even just around 0.01 hz to 20hz

If you are considering MEMS accelerometers, check the specifications for hysteresis. I've never worked with such accelerometers, but I have a vague and possibly completely wrong memory of seeing something about hysteresis issues.

You would probably have to place the amplifier right at the accelerometer. Normal seismometers transducers get strung out on long cables that would be completely unworkable with nanovolt signals. Slight movement of the cable would probably produce nanovolts due to triboelectric effects.

It sounds like an interesting project, but with some serious technical issues.

The last time I had any involvement with geophones it was trying to figure out methods to keep the instrumentation from being blown by lightning strikes near the the 'phones. Apparently this can be quite a problem.
can i have some pointers regarding with lightning strikes? i might run into this problem

i mean what components best to used in order to prevent this kind of problem
 
Last edited by a moderator:

MisterBill2

Joined Jan 23, 2018
7,964
Definitely an optical approach will give both better sensitivity, a higher output and much less noise. Also, check out the geophones used in searching for oil underground. They have excellent low frequency sensitivity. And they could also be home built
 

ebp

Joined Feb 8, 2018
2,332
I would take it from the TS's statements that the objective is to investigate the possibility of using accelerometers, not implement something that has already been done:
"We were trying to build a seismometer using only an affordable +-1g accelerometer. We were trying to test if its possible to use an accelerometer as alternative sensor for a seismometer."
 

MisterBill2

Joined Jan 23, 2018
7,964
I would take it from the TS's statements that the objective is to investigate the possibility of using accelerometers, not implement something that has already been done:
"We were trying to build a seismometer using only an affordable +-1g accelerometer. We were trying to test if its possible to use an accelerometer as alternative sensor for a seismometer."
What we have already presented is that given the output of the accelerometer is 1 volt per "G", I think, and that the anticipated signal is nanovolts, or microvolts, it is quite possible that the inherent noise from the device is greater than the anticipated signal. But since the need is for a limited frequency response amplifier, it may be that one exists. Analog Devices would be a likely company to search for such, if it is to be found. But not all experiments result in successes.

But then I recall the following poem from some unknown author:

They said it was an impossible task,
and to look at the task, who wouldn't.
So we tackled the task
that could not be done,
And you know what!
It couldn't!!
 

MisterBill2

Joined Jan 23, 2018
7,964
Hello,

For those very low frequecies, you likely are forced to use a DC coupled amplifier.
With the nanovolt range, the offset of the amplifier may become a problem.

For an other idea, read this article:
https://www.analog.com/en/analog-dialogue/articles/low-noise-inamp-nanovolt-sensitivity.html

Bertus
For certain it will need to be DC coupled. But right now I am imagining using the voltage to frequency modulate an oscillator, since very small changes in frequency are easy to detect. But probably I am ignoring something there. That was just a quick thought.
Of course, a means of mechanicaly amplifying the vibration shaking the accelerometer, such as a resonant suspension, could be a big benefit.in fact, that may be the very best option, since it would not add any electrical noise.
 

MrChips

Joined Oct 2, 2009
22,870
Here is a thought. Use four accelerometers wired in a bridge configuration with two sensors out of phase and the other two in phase. Then use a low-noise IN-amp such as AD8429.
 

MisterBill2

Joined Jan 23, 2018
7,964
Here is a thought. Use four accelerometers wired in a bridge configuration with two sensors out of phase and the other two in phase. Then use a low-noise IN-amp such as AD8429.
Accelerometers are single-ended output devices and putting them in a bridge configuration would not work, both because of the physical arrangement and because of the electrical arrangement. That would only give 4 times the output, not a worthwhile improvement, for 5 times the cost.
 

wayneh

Joined Sep 9, 2010
16,765
But, using multiple sensors would allow cleaning away a lot of the noise. Only a signal appearing on all sensors would pass.
 
Top