Zero-ohms Compensation resolution reduction

[joseospina]

Zero Ohms compensation for resistance measurements is nice. However, I wonder if because it makes two voltage measurements then the uncertainty on the result, this is the accuracy, must be multiplied by two. Is it this way?

In fact from tests I've seen conversion time by turning on/off the zero-ohms compensation changes 20 ms. This means that voltage measurement used for compensation should have a resolution close to 6.5 digits from figures provided by NI. However I'm just guessing its resolution from conversion time. Is there some specificaction about this voltage measurement accuracy? Does it always uses 20 ms aperture time?

Thanks

Jose Ospina

Message Edited by joseospina on 08-13-2008 07:08 AM

Message Edited by joseospina on 08-13-2008 07:09 AM

Message Edited by joseospina on 08-13-2008 07:11 AM

[joseospina]

Change "Zero-Ohms compensation" for "Offset Compensated Ohms"

!!! sorry for the lapsus

[Claudia]

Hello Jose,

Offset Compensated Ohms (OCO) allows you to remove offset errors in your Resistance measurements. In order to make resistance measurements, the DMM outputs a DC current that flows through the resistor you want to measure and then the DMM measures the voltage drop. When you enable Offset Compensated Ohms (OCO), the DMM takes two measurements: 1) it turns off the current source and measures the residual voltage in the path; and 2) turns on the current source and measures the voltage again. The resulting measurement is equal to the second measurement minus the first one.

You can fin more information on OCO in the NI DMM Help File under the section:

NI Digital Multimeters Help >> Devices >> NI 4070 >> DMM Measurements >> Resistance >> Offset Compensated Ohms

NI DMM Help File link: http://digital.ni.com/manuals.nsf/websearch/E7307AF8A2CE564F86257364006346A6

If you use the default device configuration for 6.5 digits set by the NI-DMM driver, the uncertainty of your measurement will be the one listed in the specifications. Notice that the specifications for the ranges 100ohms, 1kohm and 10kohm apply when OCO is enabled.

NI 4070 Specifications link: http://digital.ni.com/manuals.nsf/websearch/DC9B978C0ECB851186257442005F5D0A

The default device configuration for 6.5 digits measurements set by the NI-DMM driver can be found in the NI DMM Help File. They are listed under the section:

NI Digital Multimeters Help >> Devices >> NI 4070 >> DMM Measurements >> DMM Measurement Defaults

To configure the DMM using the default settings you only need to call the function niDMM Config Measurement and specify the number of digits you want. See more information in the followign section of the NI DMM Help File:

NI Digital Multimeters Help >> Programming with NI-DMM >> NI-DMM LabVIEW Reference >> VIs >> Configuration >> niDMM Config Measurement

You can find the description of each part of the measurement cycle under the section: 

NI Digital Multimeters Help >> Devices >> NI 4070 >> DMM Measurements >> DMM Measurement Cycle

You can modify the settings of the measurement cycle in your program, except the switching time. The aperture time is the time the DMM is integrating over the signal that you are measuring. Notice that the default aperture time used for 6.5 digits is equal to 6 PLCs for 60Hz and 5 PLCs for 50Hz (PLC = Power Line Cycle). You need to integrate over PLCs to remove any power line coupling to your signal.

Keep in mind that if you decide to modify the measurement cycle manually you need to configure each setting manually. That is, if you decide to set the aperture time manually, you will also need to specify what measurement cycle setting you want to enable, such us OCO, ADC calibration, and DC noise rejection.

More information on this in the following section:

NI Digital Multimeters Help >> Programming with NI-DMM >> Features >> Configure Measurement Timing

And you can measure the effective resolution of your configuration using the following example program that installs with NI-DMM in your computer: "C:\Program Files\National Instruments\LabVIEW 8.2\examples\instr\niDMM\Performance Examples\Maximizing DC Reading Rate.vi"

You mention that you have benchmarked the measurement speed turning OCO on and off, and you got that each measurement took 20ms. It seems that your program tells the DMM that the PLCs are at 50Hz and it sets the aperture time is set for 1PLC. In most cases, the DMM will use the same integration time for your measurement than for the OCO (or Autozero measurement) and that's why you got 20ms for OCO and 20ms for the aperture time.

See more info in the following link: http://zone.ni.com/devzone/cda/tut/p/id/6234

Hope this helps,

- Claudia

[joseospina]

Thank you for your clear answer.

Yor are right, specifications are listed with OCO enabled... thats good for me, since I need to enable OCO and work at the edge of specifications.

About the time needed for the OCO to do the second measurement, what is really happening is that aperture time attribute is configured to be 100ms. However the line frequency is in fact configured to 50Hz, but I thought setting aperture time units to "seconds" will make irrelevant the line frequency attribute.

Two test let me estimate the  OCO additional time:

1) aperture time is set to 500ms, without OCO, conversion time is 510ms, with OCO conversion time is 1010 ms. In this test the ADC calibration and the auto zero options are turned OFF. Number of averages is 1 and settle time is 1ms, DC noise rejection window is rectangular and range is 100 Ohms. Conversion time is manually measured using internal fast interrupting timmers.

2) aperture time is set to 100ms, without OCO conversion time is 267ms, with OCO conversion time is 288ms. In this test the ADC calibration and the auto zero options are turned ON. Number of averages is 1 and settle time is 1ms, DC noise rejection window is gaussian and range is 1000 Ohms. Conversion time is automatically calculated by the DMM drivers using the function niDMM_GetMeasurementPeriod.

So the aperture time of the second measurement (thermal voltage measurement) is the same as that of the first one (with bias current) only in the first test. In the second it seems that it changes from 100ms to 20ms.

In my application high resolution and low conversion speed is useful. That is why i would like to fully understand the measurement cicle times. I've read the documents you kindly suggested. Still no clearification about the OCO additional time is included. Actually the fact that auto-zero adds one aperture time, even if logical, was not clear for me.

In general the three attirbutes: Auto-zero, ADC Cal and OCO consist of an extra DMM measurements each measurement cycle. One characterstic test I made shows this:

-Aperture time 100ms, aperture time units "seconds", PL freq 50Hz, all three options off, settle time 1ms, conversion time 105ms.

-Aperture time 100ms, aperture time units "seconds", PL freq 50Hz, Auto zero ON, the other two options OFF, settle time 1ms, conversion time 186ms. So Auto zero adds 81ms.

-Aperture time 100ms, aperture time units "seconds", PL freq 50Hz, Auto zero ON, ADC Cal ON, OCO OFF, settle time 1ms, conversion time 267ms. So ADC Cal adds another 81ms

-Aperture time 100ms, aperture time units "seconds", PL freq 50Hz, Auto zero ON, ADC Cal ON, OCO ON, settle time 1ms, conversion time 288ms. So ADC Cal adds another 21m

Conversion times ara calculated by DMM driver.

I have not been able to predict these converion time values from information included in the tutorials "adjusting the reading rate of a DMM" or "understanding the NI PXI-4071 7.5 digit FlexDMM" or Digital Multimeter Measurements Fundamentals".

Since as was told I must performed several high precission resistance measurements as fast as possible using a switch to multiplexate the DMM input, i would like to better understand the actual internal bheavior of the DMM for each configuration.

Thank you again

Jose

[Claudia]

Jose,

I am attaching the image of the block diagram of a VI that will help you benchmark more easily the measurement speed. In that VI I used the "Initialize with Options" VI so I could simulate a DMM. I did this because it is convenient, and I would get the same benchmark results if I were opening a real session to a DMM board.

The top and the bottom code as they appear in the image will give you the same results on measurement speed.

Using the default settings for 6.5 digit measurements my measurement period is 286.5ms. By turning on and off different parts of the measurement I find out that the measurement consists of 100ms of aperture time, 60ms of AutoZero, 120ms of ADC Cal, 1ms of Settling Time and the rest is Switching Time, which you cannot change manually.

If I enable OCO I notice my measurement period is 328ms... and here is where you ask yourself why it only added 40ms, right? If you are taking an OCO measurement you no longer need to take an AutoZero measurement. OCO will null out the offset voltages over the whole measurement path, while AutoZero nulls out only the offsets right up to the banana connectors. So the NI-DMM driver sees that you want to make an OCO measurement, and to make your measurement more efficient time wise, it disables AutoZero and only enables OCO. Your measurement period now consists of 100ms of aperture time, 100ms of OCO, 120ms of ADC Cal, 1ms of Settling Time and the rest is Switching Time.

The other questions you may be asking is why AutoZero takes only 60ms and not 100ms. The developers of the DMM perform thorough tests to find out what was the optimum measurement time for an AutoZero and ADC Cal measurement so you could get maximum reading rate and still correct for offset and gain errors, and that's how the maximum time used for AutoZero is 60ms. ADC Cal (corrects gain errors) really consists of two internal measurements: a measurement of an internal reference voltage and a measurement of its own autozero path. That's why ADC Cal takes 120ms.

If you make your Aperture Time less than 60ms you will notice that the times for AutoZero and ADC Cal use the same value than the aperture time you chose. For example, if I make my aperture time 40ms, my settlign time 1ms and I enable AZ and ADC Cal, then my total measurement time is 166ms, where 40ms are aperture time, 40ms are AutoZero, 80ms are ADC Cal, 1ms of Settling Time and the rest is Switching Time. I am sending my benchmark results on a table in the same image.

It seems that you are interested in taking good resistance measurements as fast as possible. I would recommend you to first look at your measurement system. The DMM is a very good measurement device, but if your signal path is noisy you will measure noise and the effective resolution and quality of your measurement will be limited by the level of noise in your system and the quality of the connections. Once you have your system set up (with switches and cables), you could test what is your own settling time. You might need to change the default settling time the NI-DMM driver uses to the one your system actually needs. Then use the default configuration settings for 6.5 digits and look at the level of noise in your system (you can do this by connecting a short at the end of your switch and using the example program I recommended to you in my previous post; that example program measures the standard deviation of your samples to get the noise and calculates the effective number of digits you can get in your setup). Once you have optimized your switch system, your cables, etc to get the quality measurement you want, then you could start twiking the measurement speed.

Hope this helps,

- Claudia