The NI PXI-4070 FlexDMM -- A Benchmark Comparison of 6 1/2-Digit DMMs

A test system is only as fast as its slowest component is a simple conclusion that can be drawn from the popular idiom – a chain is as strong as its weakest link. Taken in context, the slowest component of most automated measurement systems is a digital multimeter (DMM). For years, test engineers have employed numerous strategies to extract more speed from these devices in R&D laboratories and on the manufacturing floor. Optimization techniques have included cutting down the number of tests, reducing the accuracy requirements, purchasing multiple DMMs, or even purchasing a more expensive
8 1/2-digit DMM and running it at much lower resolution. These techniques can improve the overall test throughput of a DMM, but typically at the expense of accuracy, system cost, size, or
all. The National Instruments PXI-4070 6 1/2-digit FlexDMM eliminates the need to use such techniques by providing the following advantages over traditional DMMs:

• It has a fast A/D converter with a sample rate of 1.8 MS/s, so it can take measurements more frequently
• It uses the 33 MHz PCI bus, which provides transfer rates at least 100 times faster than GPIB interface
• It takes advantage of high-speed host PC processing, which improves system throughput every time the CPU is upgraded
• The streamlined NI-DMM driver software offers tight integration with LabVIEW
• It executes rms-to-DC conversion, traditionally an analog problem, in the digital domain, decreasing overall measurement time
• It dramatically improves speed and accuracy of function and range changes by using solid-state relays
• Seamlessly integrates with National Instruments PXI and SCXI switches by using thehigh-speed PXI trigger bus

This paper quantifies these advantages by benchmarking the NI PXI-4070 FlexDMM with two of the electronic industry’s most popular traditional 6 1/2-digit DMMs. In this paper, these DMMs will be designated DMM1 and DMM2. The FlexDMM will be tested against DMM1 and DMM2 in three typical measurement scenarios – single-point, multipoint, and multifunction. These benchmarks demonstrate that the FlexDMM is ideal for use in automated tests on both the production floor and in R&D environment.

Test Setup

Measurement speed and system speed are good indicators of the performance of a DMM. Measurement speed is how fast a DMM can take a measurement. System speed is how fast
a DMM can scan with an external switch, change ranges (1 V to 10 V) and functions (VDC to VAC), in addition to taking measurements. A typical automated test system includes both a DMM and an external switch, because users can expand the channel count of their DMM with a switch, without prohibitive expense.

Each system-speed test time includes the time it takes the DMM to hardware handshake with an external switch. Each test was run with a National Instruments SCXI-1128 24-channel solid-state multiplexer. This switch was chosen because of its fast switching
capability.

Various tests were run on each of the DMMs to benchmark the speed of single-point measurements, multipoint measurements, and multifunction measurements. To limit the number of tests, only the three most popular functions were used . VDC, VAC, and 4-wire resistance. Other functions produced similar results. The test system included the following hardware and software:

• NI PXI-1000B Chassis
• NI PXI-8176 embedded controller with:
  • 1.26 GHz Pentium III processor
  • 256 MB of RAM
  • Windows 2000
• NI LabVIEW 6.1
• IVI-compliant DMM-specific instrument drivers

The PXI-4070 FlexDMM was housed in the PXI-1000B chassis and controlled by the PXI-8176 controller through the PXI backplane. All tests were automated using LabVIEW. DMM1 and DMM2 communicate to the PXI-8176 controller through GPIB. Figure 1 shows how the following tests were connected.


Each function was tested to 6 1/2, 5 1/2, and 4 1/2-digits of resolution, except VAC, which was tested at 6 1/2 digit resolution. This was done to get the maximum performance out of all AC
measurements taken and to limit the number of tests performed. The front panel displays on DMM1 and DMM2 were disabled for all tests as was autozero for all three DMMs.

All measurement speed tests are in readings per second, which only includes the data acquisition time. The system speed tests are in "effective readings per second," which describes the total number of measurement points returned from the DMM divided by the total measurement time. Total measurement time includes the initialization, configuration, and data acquisition time for all the functions (VDC, VAC, etc.) tested. For example, in a test where 450 points of each type were measured:
This automated test measures both measurement and system speed. The system speed test measures the time for the DMM to initialize, configure, switch, take one measurement and
transfer the data. The DMMs were configured to perform two-way hardware handshaking with the SCXI-1128 32-channel multiplexer.

RESULTS

Summary
As shown in the graphs above, the FlexDMM is at least 2.5 times faster than the traditional GPIB-controlled DMMs. They also demonstrate that for a single-point measurement, DMM configuration time plays a big factor in the overall time. A DMM’s requirements are different from most measurement devices because they transfer very small amounts of data requiring fast configuration and efficient communication between the DMM and its controller.

Much of the FlexDMM configuration time efficiency can be attributed to its register-based configuration setup versus the traditionally slow GPIB message-based setup. The FlexDMM is configured through a single call in the driver, which directly writes to the measurement setup registers. Traditional DMMs need multiple message-based calls that then have to be parsed internally, making the processor do extra work and slowing setup time. The small latency of the FlexDMM is primarily due to the streamlined NI-DMM driver software and the 1.26 GHz embedded PXI controller.

Multipoint Measurements

This automated test measures both measurement speed and system speed. The measurement-speed test measures how long it takes each DMM to initialize, configure, and acquire 100 points. The system-speed test measures how long it takes each DMM to initialize, configure, scan the entire scan list, taking one measurement on each channel of the switch, and then transferring the data. Each channel scanned was a DC voltage in the 10 V range, so no function or range changes took place. The total number of measurements returned from the DMM corresponds to the number of channels that were scanned through on the switch (channel count). A resolution of 4 1/2 digits was used and two-way handshaking was the method of triggering used.

RESULTS

Summary
The DMM1 and DMM2 transfer speeds are primarily dependent on their communication bus speed and internal data buffer sizes. Hence, to achieve maximum throughput, the channel count size should equal the DMM buffer length. For example, if DMM1 has a buffer size of 512 points, then for the full buffer length scan, one measurement point would be acquired from 512 channels, creating a full buffer and maximizing throughput. The DMMs were tested for their ability to transfer partial buffers and a full buffer.

DMM1 and DMM2 were connected to the PC via GPIB, which has a maximum throughput of 1 Mbytes/s. The PXI-4070 FlexDMM uses the 132 Mbytes/s PCI bus, so it can transfer data more than 100 times faster than GPIB. The ability of the FlexDMM to transfer at these high rates decreases its dependency on its internal buffer. Another notable point is that the FlexDMM uses only a fraction of the PXI bus throughput, which is important when running multiple devices on the same bus. It is important that one device not use the entire bandwidth, and therefore prevent other devices from communicating on the bus. The FlexDMM also benefits from its extremely fast A/D converter, which can take more frequent measurements at the rate of 1.8 MS/s.

Multifunction Speed

The ability of a DMM to take multiple measurements and efficiently change functions and ranges is the key to maximizing test system throughput. For this automated test, the total measurement time involves the DMM taking measurements in the following three functions – VDC, VAC, and 4-wire resistance. For each particular function, multipoint measurements were taken in three different fixed ranges.

Traditional DMMs have three, fixed AC filter settings which depend on the frequency of the input signal. This test included three particular cases:

• DC only (only VDC and 4-wire resistance were measured)
• AC medium filter (60 Hz input signal)
• AC fast filter (5 kHz input signal)

RESULTS

Summary
Traditionally, AC measurements slow down the total measurement time considerably, so we have performed both a DC only test and a mixed AC and DC test. The results clearly indicate that the speed limiting function is VAC. Most traditional DMMs use an analog rms-to-DC converter for AC measurements, which adds long settling delays. The PXI-4070 FlexDMM uses a digital algorithm that only requires a few cycles of a waveform to compute rms values, which dramatically increases AC reading rates. This digital algorithm automatically rejects the DC component of the signal, making it possible to offer a DC-coupled VAC mode that completely avoids the slow-settling input coupling capacitor required with traditional methods.

The digital approach to rms computation offers accuracy benefits as well. The algorithm is completely insensitive to crest factor, and can deliver exceptionally quiet and stable readings. Crest factor is the ratio of the peak value of a signal to its rms value. For a sine wave the crest factor is 1.414, and for a square wave the crest factor is 1. This specification is important because it indicates the maximum peak value of an arbitrary waveform that a DMM can handle without overloading. The crest factor also affects the accuracy of AC measurements. For example, given a DMM with an AC accuracy of 0.03% (this is always specified for sine waves), and an additional error of 0.2% for crest factors between 1.414 and 5, the total accuracy for measuring a triangular wave (crest factor = 1.73) is
0.03% + 0.2% = 0.23%.

Furthermore, the FlexDMM guarantees AC accuracy to 1% of full-scale, rather than the
10% of full-scale offered by traditional DMMs. To further improve system speed the
FlexDMM uses solid-state relays instead of electromechanical relays, thereby improving
speed and accuracy of function and range changes.

Conclusions

These benchmarks have shown that National Instruments has implemented a cost-effective 6 1/2-digit DMM that uses PC technology to achieve extremely fast measurement speeds as well as system speeds in a single-slot 3U PXI module. The NI PXI-4070 FlexDMM can be used in any PXI system; the user can implement numerous other measurement devices and switches in one compact chassis. PXI not only provides fast and controlled triggering across the PXI chassis backplane, but also it affords users the flexibility of upgrading to a faster measurement system by using the latest PC processors. National Instruments LabVIEW can also contribute to performing faster, more powerful postprocessing and data presentation of the acquired data when it is used in conjunction with the measurement system.

Reader Comments | Submit a comment »