NI 408x DMM Specifications Questions

[JLAnderson]

The NI 408x DMM specification format has changed from the NI 407x series. This leaves some questions and lack of information on how to calculate the accuracies for ranges less than the maximum (e.g. 5.5 digits). There is a general lack of documentation for the new meters with few hits on the NI website.

1. The 2 year 10 to 28 degree specification for DCV is gone and the new specification is presumably for 0 to 55 degrees C. How would you use Tempco with this new specification since anything outside of 0 to 55 is not used? The temperature range should be specified as a footnote like it is on page 5.
2. The additional noise error figure is now called DC Voltage Noise (typical). The chart axis have changed RMS Noise (ppm) to Measurement Noise, RMS (Volts) and Integration Time (NPLC) to Aperture Time (seconds). The multipliers have been added to the graph instead of being in a table. The Aperture time only goes up to 100 mS (6 PLC) where the old chart went to 100 PLC (1.666 S). The low end of the chart now goes to about 0.6 uS while the old one is 1.666 mS. So the chart is shifted significantly toward the lower end. An explanation on why this was done would be helpful.
3. The note for figure 1 states to add five times the typical RMS noise to the accuracy specification. The NI 4071 specification stated multiplying by 6 to get Vpp which implies 3 for Vp. The 6 represents 6 standard deviations of RMS noise is peak to peak noise with coverage of 99.7%. Keysight DMM documentation states to multiply by 3 to get Vp so it can be added to the accuracy specification. Please explain where the 5 came from because 5 standard deviations are not used anywhere else that I know of. It makes no sense to not include 100 mS in multiplication factor converting RMS to Vp. It should read something like: “Multiply the RMS noise by 3 (Vpeak ) and add to the accuracy specification”. 
4. In digitizer mode, the graph showing digits or bits vs. rate is gone. Presumably the new PXIe interface is fast enough to read all the data into the computer at the maximum rate of 1.8 MS/s. The NI 4071 specifications stated the accuracy was the same as DC voltage specifications.  The NI 4081 specifications are completely different and much worst in digitizer mode. There is a RMS noise table but no explanation on how to use it. Accuracy is a function of sample rate (digits) and range. Please explain how to determine the accuracy for any rate and range. 
5. Data for default aperture times cannot be found anywhere. A table for resolution, reading rate, aperture time and RMS noise like shown in the NI 4065 specifications would be very useful. Putting the RMS noise into a spreadsheet makes it much easier to add more points to the curves and getting more accurate data for the additional noise calculation. The RMS noise charts also use linear grid lines when the data is logarithmic. This makes it harder to interpolate values.

Without complete specifications, I can’t determine the suitability of these meters for future use, especially if the NI 4071 is discontinued.

John Anderson

[WillyWad]

Hey John,

1. Regardless of the ambient temperature range, tempco is implemented based on the temperature of the device during it's last external calibration (Textcal), as mentioned in the footnote you referred to. Tempco should be used if you are more than +/- 10 degrees Celsius from Textcal even if you are within the 0 to 55 degrees Celsius operating temperature range.
2. I do not know precisely why the changes you mentioned on the DC Voltage Noise graph were made. This decision would have been made based on the experience of the engineers and technical writers who work with the 408x DMMs. My initial guess is that it seems the graph begins to level out around 100ms. Or, in other words, further increasing the aperture time would have minimal effect on the DC voltage noise that would be seen on the measurement. 
3. I've looked into why 5 is the chosen factor in this scenario, but could not find anything that specified. If you still want to know, or it has bearing on an application you are working on, I would suggest contacting our Applications Engineering team for support. They will likely need to escalate to our Product Support Engineers for the 408x DMMs to get this information, but that would be the most effective way for us to sort that out for you.
4. If you refer to Tables 7 and 8 in the specifications sheet for the PXIe-4081, this shows the accuracy you can expect based on your reading and range. This is done using the method of Accuracy = ±(ppm of reading + ppm of range). From there, you can determine how sampling rate affects the accuracy by referring to Figures 4 and 5. Base on the rate and range, you will see what noise to expect with your reading and apply that on top of the accuracy you previously determined.
5. In regards to your final comment, I can submit your comment to our team that manages this documentation. They will review it and determine the best course of action for providing revised and updated documentation.

I hope this information is helpful for you. I can see that there are still some gaps in the information we've reviewed so far, and I understand how that can make it difficult to determine if a specific device is right for your application. I briefly suggested this above, but considering the depth of your questions, I would suggest reaching out to our support team to open a ticket and review this information directly with an Applications Engineer. The AE you will work with can do some deeper research and consult our PSEs and developers as necessary. (You can reach the support team at 866-275-6964 or by emailing support@ni.com).

Take care,

Will

[charlesy]

Hello Mr. Anderson,

Thank you for the detailed questions. Here is some clarification and rationale for the 408x specs.

In the past, we received feedback that the 407x specs were difficult to interpret and compare with other dmms. With the 408x, as you pointed out, we updated several aspects of the specs, considering both industry norms and aiming for increased consistency with our related SMU products.

You are correct -- for the 4081, DCV is now specified over the full 0-55C environmental range, though explicit specs are given assuming device temperature is within 5C of the last self cal. The footnote on page 5 (resistance accuracy over full temperature range) was included to differentiate that column from the other accuracy columns that constrained temperature to +/- 10C from External Calibration. Because the voltage reference is so stable, there wasn't much value in adding those additional columns to the DCV accuracy table.

For DCV, tempco is provided for guidance on device drift with or without self cal, but it is not necessary to add that back in if you stay within 5C of the last self cal.

We moved away from using absolute environmental temperatures (0-55C, 18-28C, etc.) for accuracy criteria, instead using device temperature (Tselfcal +/- 5C, Textcal +/- 10C, etc.). This is because DMM temperature can vary significantly depending on module load and activity in various PXI chassis. The onboard temperature is what matters - not that of the outside air.

In many cases, the 4071's primary absolute accuracy specs required +/- 1C, but we assumed most customers would have a hard time holding that, so we loosened the 4081's specs to allow for +/- 5 C. Relative accuracy specs, as is customary, remain at +/- 1C.

We documented aperture specs in absolute time rather than power line cycles because the it provides consistency among systems deployed in 50 Hz and 60 Hz regions… 100 ms aperture for a "6.5 digit" measurement has the same timing in the U.S. and Europe. The difference between 5 and 6 periods isn't very significant when windowed with the default High-Order Noise Rejection.

As a PXI instrument, we have the benefit of two isolation layers from AC MAINS to the analog front-end (an AC-DC conversion in the chassis and an isolated DC-DC in the DMM). Power line noise within the instrument is not dominant, though it obviously could be depending on system cabling.

The noise curves sweep aperture time from 555 ns because the 408x is now capable of acquiring that narrow a sample. The curve stops at 100 ms because you get diminishing returns beyond that. For very long apertures, you may see slightly increased drift because there hasn't been recent auto zero. Rather than using a long aperture to reduce noise, it is better to increase Number of Averages, which will repeat the Auto Zero stage.

For the noise multiplier, we used 5x not only to convert from RMS to peak, but also margin up from typical since you would be applying this scaling on top of a warranted accuracy. A noise parameter has already been included in the absolute accuracy specs, assuming a 100 ms aperture. You only need to account for additional noise if you use a shorter aperture.

For Digitizer mode on the 407x, DC accuracy was listed as "typical". For the 408x, we opted to provide warranted DC accuracy specifications. Because you can't interrupt a continuous waveform with correcting functions (like auto zero), it's not possible to achieve the same DC accuracy as the DCV DMM mode. DC Accuracy is not dependent on sample rate, but noise is. We provided the typical noise curves in RMS so the user could consider it in conjunction with other system noises (rss-ing them, for example). We did not attempt to convert to Peak (or Peak-to-Peak) because Digitizer mode is inherently a multi-point acquisition where DC accuracy applies to the average over many points. Compare to DMM mode which is often single-point, where the noise contribution needs to be considered as a peak to add to the accuracy.  

Keep in mind that Digitizer mode's AC performance specs (flatness and bandwidth) are still "Typical", which is not unusual for this type of device. 

The default aperture times and other configuration parameters associated with a given "number of digits" setting are defined in the driver. As such, they are documented in the NI-DMM Help. You can find it under:
NI Digital Multimeters Help -> Devices -> NI 4081 -> DMM Measurements -> DMM Measurement Defaults

I apologize for the long post, but wanted to give some context for spec changes between the 407x and 408x.

If you are at NIWeek, I hope you've found it to be interesting and educational. If you have further questions, please don't hesitate to ask.

Charles Yarbrough
National Instruments

[JLAnderson]

Thank you for your responses and I have some follow up.

1. DoD portable equipment can operate in extreme temperatures (hot and cold). This is why I have been asking about temperature coefficients. The Navy only uses a 1 or 2 year calibration interval. My question has been answer satisfactorily.
2. The new chart saves a number of extra steps to get the DC RMS noise without having convert PPM and multiply by scaling factors. I don’t know where you got the 555 nS value for the low end. I could only estimate it to be a little less than 600 nS from the graph. So does that mean the lowest number of 555 nS applies to 3 digits @ 1.8 MS/s?
3. Attached is an excerpt from on another on-line forum (with some editing) shows the relationship between RMS and peak-to-peak voltages. Peak is ½ of peak-to-peak that is added to the base accuracy. If you are saying 5 represents peak, then peak-to-peak is 10 standard deviations. This is off the chart at better than 99.999% coverage. In general “typical” measurements are not warranted. I will check with our experts to see how we will handle this. The “PXI Express DMMs Form Foundation for More Accurate, Smarter Test Systems” document shows a 12 hour noise graph. I would like to see that plotted as a normal curve to see what the standard deviations are.  Perhaps I should have mentioned the NI article “How Do I calculate Accuracy for DMMs?” that was created after I asked questions on how to determine the accuracy for a 4071 when less than 7.5 digits is selected. For example the aperture for 4.5 digits is 50 uS. This is significantly less than the 7.5 digit aperture of 100 mS. Using the RMS noise chart to find the noise at 50 uS is adds a significant amount of noise thus reducing accuracy at 4.5 digits compared to 7.5 digits as expected.  Using few digits will reduce test time as long as the accuracy is sufficient. This is why I need to know the apertures for the different number of digits. See number 5 below.

4. It sounds like digitizer accuracy depends on the actual measurement being taken. I will check with our experts to see how we will handle this.

5. The NI 408x devices do not show up under DMM help but all the other meters are there.

The meters look good but the spec sheets need some work. They don’t even specify how many digits can be used for each function. Example accuracy calculations are appreciated even if it is just a white paper.

John Anderson

[kirsch]

1.  The NI 408x devices do not show up under DMM help but all the other meters are there.

The NI 408x was just released and is first supported in NI-DMM 15.2. Is this the version you have installed?

[JLAnderson]

After a 6 hour ordeal to download and install NI-DMM 16.0 (upgrading from 3.1) I found the default documentation that looks the same as the NI 4071.  But my complaint was the website help doesn't get updated to the latest information in a timely manor.  In the past our IT security was blocking access to all help files so I had to go on line since pdf files have not been available for some time.

John Anderson

[ChristophersonJ]

Hey John, 

I'm sorry that you had a rough time with the newest driver, but hopefully now you are able to access either the online or offline help for NI-DMM 16.0, which should both be available to you. 

When you mention our online information are you referring to the help documentation or to other articles such as KBs, white papers, etc.?  If so I can tell you that our content teams are actively seeking to improve our process for keeping our resources up to date, so that when we release new products, our information can be updated as quickly as possible (it's a pain point for us too).

But certainly as always if there is ever a question you have, or clarification you need, you can either leave feedback on the resources themselves, or give our support line a call and our AEs can help you track down what you need. 

So just to clarify have you now been able to find the clarifications you needed from our 408X documentation, or do you still have some outstanding questions?

Best Regards, 

Chris J