Author Topic: low Signal amplification ?  (Read 1842 times)

0 Members and 1 Guest are viewing this topic.

Offline namsterTopic starter

  • Regular Contributor
  • *
  • Posts: 58
  • Country: dz
low Signal amplification ?
« on: October 25, 2019, 06:46:13 pm »
Hi,
I want to amplify a signal that has a full scale of 20mV, the major problem is the bottom of the signal (bellow the offset of Amplifier), I use a 4 wire Kelvin measurements, the two methods I think about is:
1 - a non inverting Amplifier with Jfet input like 34420a and a DAC For offset compensation
2- an Instrumentation Amplifier  ( Like AD8552 with 1 µV Offset )
so how can I amplify a voltage of 10µV with a gain of 350!
« Last Edit: October 25, 2019, 07:09:13 pm by namster »
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: low Signal amplification ?
« Reply #1 on: October 25, 2019, 07:12:57 pm »
How about just using a MAX4209 and manual zero'ing to get rid of the final offset?
 

Offline dom0

  • Super Contributor
  • ***
  • Posts: 1483
  • Country: 00
Re: low Signal amplification ?
« Reply #2 on: October 25, 2019, 07:20:49 pm »
Bandwidth? Acceptable offset error?

Also

Quote
a full scale of 20mV

vs

Quote
mplify a voltage of 10µV with a gain of 350!

Those numbers are four orders of magnitude apart.
« Last Edit: October 25, 2019, 07:23:11 pm by dom0 »
,
 

Offline _Wim_

  • Super Contributor
  • ***
  • Posts: 1568
  • Country: be
Re: low Signal amplification ?
« Reply #3 on: October 25, 2019, 07:21:15 pm »
If bandwidth is low, typically a chopper-stabilzed (CMOS) opamp is recommended for this application
 

Offline _Wim_

  • Super Contributor
  • ***
  • Posts: 1568
  • Country: be
Re: low Signal amplification ?
« Reply #4 on: October 25, 2019, 07:56:43 pm »
Those numbers are four orders of magnitude apart.

I think he means the smallest expected input signal for the amplifier is 10µV, the largest is 20mV, both will be amplified with a gain of 350. The small signal has a high risk to be swamped by the offset voltage due to temperature variations
 

Offline dom0

  • Super Contributor
  • ***
  • Posts: 1483
  • Country: 00
Re: low Signal amplification ?
« Reply #5 on: October 25, 2019, 08:06:57 pm »
20 mV * 350 = 7 V, so that will require antics with some of the suggested parts (with supply voltage limits of 5 V).
,
 

Offline _Wim_

  • Super Contributor
  • ***
  • Posts: 1568
  • Country: be
Re: low Signal amplification ?
« Reply #6 on: October 25, 2019, 08:11:51 pm »
20 mV * 350 = 7 V, so that will require antics with some of the suggested parts (with supply voltage limits of 5 V).

Gain could be split up in 2 stages, with the second stage a gain of only 5 and a standard +-15V opamp.
 

Offline moffy

  • Super Contributor
  • ***
  • Posts: 2216
  • Country: au
Re: low Signal amplification ?
« Reply #7 on: October 25, 2019, 11:43:53 pm »
On: https://www.analog.com/en/products/ltc1052.html#product-overview
you will see a design for a low noise, low drift amplifier that should suit your needs.
 

Offline namsterTopic starter

  • Regular Contributor
  • *
  • Posts: 58
  • Country: dz
Re: low Signal amplification ?
« Reply #8 on: October 26, 2019, 04:35:43 pm »
sorry for my english , i will add some clarification , the full scale of voltage is 20mV and it correspand at 20mOhm , the resolution of the instrument is 10µOhm so the lowest voltage to be amplified is 10µV , the gain of amplification have to be 350 , the BW is about 20Hz or bellow , i tested a non inverting amplifier with DAC to compensate Offset like HP34420a but that work correctly the gain have a exponontial form , also a standard instrumentation Amplfier don't give correctly at low voltage input ,
 

Offline dom0

  • Super Contributor
  • ***
  • Posts: 1483
  • Country: 00
Re: low Signal amplification ?
« Reply #9 on: October 26, 2019, 04:39:41 pm »
Oh, a micro-Ohm meter. In a nutshell: You're doing it wrong. Don't try to measure it at DC, use an AC signal and a synchronous demodulator (as simple as two analog switches). Every commercial milli/micro-ohm meter has done something to that tune since the late 60s.

Here is one of many articles on the subject: http://cappels.org/dproj/dlmom/dlmom.html

The AC approach largely eliminates drift, thermal offsets and precision (except gain) requirements from the frontend and also allows you to reject mains noise etc. pretty well.
,
 

Offline namsterTopic starter

  • Regular Contributor
  • *
  • Posts: 58
  • Country: dz
Re: low Signal amplification ?
« Reply #10 on: October 26, 2019, 04:54:39 pm »
and What about 4 kelvin measurement used by HP 3458A 34420a 34401a , can you achieve a 10µOhm resolution ? , Also AC excitation has disadvantage of Skin Effect which can affect the measurement
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: low Signal amplification ?
« Reply #11 on: October 26, 2019, 08:58:13 pm »
also a standard instrumentation Amplfier don't give correctly at low voltage input

Minimum common mode for MAX4209 is Vss - 0.1 (ie. below the rail). Even if it wasn't though, simply adding an extra negative power rail is hardly a big deal.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 15153
  • Country: de
Re: low Signal amplification ?
« Reply #12 on: October 27, 2019, 07:57:01 am »
Most of the modern high resolution DMMs use multiplexer and do a kind of pseudo differential measurement in 4 wire ohms mode. So they measure one end first and than the other. So the actual amplifier sees kind of modulated DC and thus low frequency AC. In addition they usually offer an more where the excitation current is turn on an off to compensate for possible thermal EMF at the DUT.  In addition possible offsets from the multiplexer and protection circuit are also corrected. AFAIK they usually don't reverse the current flow, but still use a kind of modulated excitation.

With the modulation slow (e.g. some 1-20 Hz) enough skin effect should not be such a big problem. In the extreme one could test at 2 frequencies and warn the user if the DUT shown a significant difference.

To really get  very high resolution an AC current (could be a kind of square wave) is the way to go, because of possible thermal EMF at the DUT. DC wise some 1 µV resolution is feasible, better gets tricky.  AC can go down to the 1-10 nV range with some patience.
 

Offline namsterTopic starter

  • Regular Contributor
  • *
  • Posts: 58
  • Country: dz
Re: low Signal amplification ?
« Reply #13 on: October 27, 2019, 06:20:34 pm »
@Marco
the MAX4209 is not verry stable with a typical Gain error Temperature of 50ppm and 200ppm for non linearity that is not verry good specs for my application .
@Kleinstein
for reducing Thermal EMF it exists a methode called current reversal , or delta method , for my milliohmeter project i want a instrument that give a good performance for a not expensive cost , in the first version i used a H bridge wich reverse current but there was a high leakage current , in second version I would like to try to keep the same concept but by improving all the circuit , for know i try to amplifiy the signal wich has a full scale of 20mV
i tried a circuit wich have two Opamp with two gain of 10 and 35 (same as 34220a ) but the gain is not linear what can be the problem ?
the offset can be componsate  by a DAC

« Last Edit: October 27, 2019, 06:39:09 pm by namster »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf