measure mimimun voltaje using USB-6289 (avoid noise)

[Jonathan.espinosa]

Hi.

I need some help with my application.

The principal idea is in a future measure a magnetic field using a magnet-resistor circuit (internaly: use a bridge weastone).

(see the diagram please...)
- ok before to do this idea, i need to measure on mico-volts o nano volts (if exist some posibility) with this card
- i need a reference voltaje of 2.5v. i suppose to use a voltaje divisor with resistors of 1K.(Letter R)
- before to use the circuit i'm using another voltaje divisor with resistors of 1K. (Letter S)
- Using a very-good cable (twisted-pair, low-impedance, shielded) aprox 6 ft of long. i send both signals to compare.
- actualy i used both metods Diferencial (Ai0-Ai8) or NRSE(Ai0-AiSense). to compare the best metod.
- i using resistors to Ground-Reference, (in accordance with M Series User Manual page 4-12) to reduce the noise
- i'm using the channel 1 to check the maximum measure that i can hope
- i'm program Ao0 with 5V
- my range to measure is 0.1v to -0.1v

my results are:
using standard deviation.
on Ai0 - 2.0e-5 v
on Ai1 - 1.04e-5 v

ok after this story,
-i need to reduce the noise
-have a better measure

some idea?
why i can't use 16 bit of resolution as minimum?. to have 3e-6 V
which is the better technique to measure microvolts

thanks so much if you can help me!!
i have a headache with this problem, hahaha.....

[Henrik_Volkers]

When it comes to low level measurements and noise (say you can't 'mean-out' the noise) look for good low noise preamplifier.

Bridge measurements are close to the physical level of what is possible. (HBM, to name one). Your limit is the thermal noise of the bridge resistors.

[Jonathan.espinosa]

Hi Henrik 

Thanks for your suggetion, actualy the idea is reduce the noise on conexions cable and power supply, someone said use a very low distorsion power supply. like as LM399h.

and try other options like this.

Greetings.

[ChrisR.]

Hi, Jonathan.  The 6289 specs (http://www.ni.com/pdf/manuals/371292g.pdf) show you should expect 9uV of noise on the ±0.1V range with the built-in filter disabled.  This is close to the 10.4uV you're measuring on channel 1.  The noise drops to 2uV with the filter enabled, so you should do that.  To get lower noise than that, you need to do some form of digital filtering.  That could be as simple as averaging a number of samples.  Depending on the measurement bandwidth you need (i.e., how fast you need your measurements to respond to a change in the bridge), you should be able to get the noise quite low.

Also, if I understand your picture correctly, you don't need the bias resistors because the bridge itself is providing a path for bias currents and is setting a good common-mode voltage for the input channels.

Hope this helps.

[OmarIsi]

Hi Jonathan.espinosa,

In the specs file you can see that the device has a 16bit Analog Input resolution in ADC, so this is the maximum resolution you can have in measurements. With a range of 0.1 to -0.1V you will have a resolution of 0.8uV, you will be able to detect maximum changes in voltage by this value multiples. we need to consider that if we had a voltage above 0.1V limit the resolution will change automatically. 

I see you have a grounded referenced signal, so you don need to use the bias resistors. you can connect signal in a NRSE way or you can measure the difference in both singals directly using the differential measure method.

I think the difference in measure is because the variation in the resistor values, the supplier says that we have a 1KOhm resistor, but this value is not exactly the same in all resistors, we can have an important variation betweem them. To reduce this problem you can change the resistor for bigger values, like 10KOhm or 1MOhm, so we can reuse the variation. You can test this configuration...

I hope information helps you

Regards, 

[johnsold]

I agree with Henrik that you should consider a good, low-noise, high CMRR differential amplifer. The other suggestion you had about clean power supplies is good also. It seems to me that I have run into problems obtaining LM399s, so that particular part may not be a good choice for new designs.

Making measurements with resolution in the microvolt or smaller range is not trivial. You should consider the entire measurement system as a whole. Perform an error analysis and a noise analysis. This will help you determine where to focus your efforts on inprovement.

Lynn

[Henrik_Volkers]

Spoiler

LM399 can make a stable source (IF you can get your hand on them ;), need a burn in and carefull treatment)  BUT to get a low noise source use lead acid cells.

First: No multiplexing in the input stage! 

Second: Check your background noise: Using a Dummybridge with short cutted sense wires AT the bridge and the hole cable setup while all systems (of your test setup) are running.

[Jonathan.espinosa]

Hi Everybody
I'm sorry, on morning after almost 3 weeks i have a chance to check the card and all suggestions that all give me.

-i chance the configuration using the diferential mode. (check please the diagram2.jpg file.)

-maybe the mode of the connexion of, the card, power source, cable and my bridge, it's the problem with the grounds loops

- actualy my better measure is around 6.2e-5v - using 5v to -5v volts range.

Hi ChrisR.

Thanks for the tip, i need to enable the digital filter and check what happned. In the picture "table 4.3", it show two ways to conect the ground reference. why i have a best measure if i connect the second schematic.

For example, when i measure a floating signal, of a reference voltaje circuit.
i'm using ADR441BRZ is a reference of 2.5v.
-on the first case, i'm using 100k ohm resistors, that on the second case is directly the connection.

Hi OmarIsi

The card specifications indicates that it have a 18 bit of resolution

I checked the resistors and changed the configuracion for 100 kohm but i have a bad results. why?   😮

Hi johnsold.

I changed the reference, i'm using ADR441BRZ. it's better than lm399. thanks

Hi Henrik_Volkers

I dont have a multiplexor on the input

my background noise is around of  1e-5v

i hope more ideas and what is the correct mode to connect.

thanks so much ....

[Henrik_Volkers]

Get your hands on a low noise bridge amplifier.

Gain 10 to 100

Needed bandwidth ?