Test Cells Monitor Immunity or Emissions

A new breed of compact and portable test cells allow you to perform EMC tests at the bench.

David Marsh, Contributing Technical Editor -- Test & Measurement World, 4/1/1997

Most engineers associate EMC immunity-testing with room-sized anechoic chambers and prohibitive cost of ownership. But now several companies offer compact and much lower cost EMC test cells that perform these tests on equipment from pagers and mobile phones to small rack-sized instruments (see the Product Survey Table on page 16). With a base price around £20k, many R&D departments can afford one of these cells. As well as instant availability during product development, payback also comes when you’re ready to submit finished equipment to an independent test house. Testing equipment in-house identifies early problems, and gives you confidence that test house results will meet acceptance limits for final product certification.

Major measures of the success of any process are the degree of repeatability that you can achieve, and how easily you can achieve it. For EMC test cells to be really useful, they must make it easy for you to make acceptably reproducible measurements without needing expert assistance.

You normally use EMC test cells to test your equipment’s immunity to an external RF field, to meet specs such as IEC1000-4-3 (see the test set-up in Figure 1). Notice that you can also use many EMC test cells "in reverse," to test radiated emissions from your equipment (see Figure 2). To meet the RF immunity specs, the EMC test cells in the Product Survey Table generate a 3 V/m or 10 V/m field from 27 MHz or 80 MHz to 1 GHz. The specs demand that the strength of the RF field is uniform within 6 dB over at least 75% of the test area. Many EMC test engineers disregard the 3 V/m spec, arguing that passing a transmitter or being underneath a striplight subjects your equipment to larger fields. The 10 V/m spec is the minimum acceptable test for real world conditions, and in practice you’ll want some comfortable margin (say +6 dB, or 20 V/m).

The ideal RF test site is an open space in an electrically quiet vacuum. Earthbound RF sites that approximate to ideal conditions are open area test sites, which EMC test engineers use for making reference measurements. Test results from open area test sites show good reproducibility because of the lack of external influences, particularly reflections. EMC test cells try to simulate free space, or far field radiation, and to reproduce open area test site results. Notice that if you use most cells for radiated emission testing, you’ll most likely be making near field measurements. Near field measurements--measurements from a distance comparable to wavelength--contain a lot of energy other than the far field RF "footprint" that your equipment produces. Near field measurements are notoriously difficult to correlate with open area measurements, but are very useful for making comparative measurements during product development.

While you’re busy worrying about reproducible measurements and correlation with open area test results, consider the challenge for test cell designers. You expect them to deliver a cell that reproduces neutral conditions in an enclosure that fits in your lab. EMC test cells enclose the RF field, so there are inevitably reflections and resonances that affect field strength uniformity, even when the cell is empty. You also expect cells to work uniformly over at least three decades in frequency for test objects of arbitrary size or shape. The bane of RF designers is voltage standing wave ratio (VSWR). VSWR refers to the ratio of peak to mean voltage on a transmission line, caused by reflections due to imperfect impedance matching. Your equipment under test forms part of the load matching. Typical TEM cells have VSWR figures around 1.5:1 or better when empty. Your equipment interrupts the test cell’s geometry, affecting VSWR. Poor VSWR performance renders RF measurements meaningless and can destroy RF power amplifiers, which is why test cell power amps have comprehensive VSWR overload protection.

Figure 2. The configuration for EMC emissions testing.

How Test Cells Work
Most small EMC test cells employ variations of stripline designs, basically parallel plates to which you apply an RF voltage. Ideally, if the plates are one metre apart and you apply a 10 V signal, the field strength between the plates is 10 V/m. In practice, as frequency rises, wavelength relative to the plate dimensions causes non-linear effects. You will also want the RF field to be contained (shielded). To address these problems, Crawford designed the Transverse Electromagnetic Mode (TEM) cell in the 1970s. TEM refers to the RF propagation mechanism within the cell. You can visualise a TEM cell as a length of coax with your equipment in between the centre conductor and the shield. Well designed TEM cells are useful over five decades of frequency, with an upper frequency limit of around 1 GHz. For higher ranges, GHz-TEM or GTEM cell designs extend the upper frequency limit to around 20 GHz, but they are much more expensive (GTEM cells of similar size to cells in this survey cost upwards of £80k).

Geometry and septum arrangement play a great part in EMC test cell design. The basic TEM cell is an RF absorber lined box structure with a septum (RF emitter) that runs beneath the test chamber and terminates in a load resistor. Your equipment sits above the septum. The RF field is normally square or rectangular in the plane immediately adjacent to the septum, decreasing in strength as you move vertically away. Wavelength is short at high frequencies (~330 mm at 1 GHz) and wavelength reacts with your equipment, causing resonances and complex multi-modal wave patterns from quarter wavelength upwards. Practical limits for reproducible results may only be 350 MHz to 500 MHz, rather than the 1 GHz that you expect. A 300-mm cubic volume might only be useful for a 75- to 100-mm high object for the frequency range that you need to qualify. As a rule of thumb, don’t expect to reliably test items that are larger than one-third the distance between the septum and the outer conductor.

Using Test Cells
Chase EMC’s emCell is a representative evolution of classic TEM cell construction. The emCell is a free-standing box structure that accommodates a 5U 19-inch rack instrument. It uses a combination of ferrite and carbon-loaded polyure-thane foam for the RF absorber lining, and has an offset septum that extends the uniformity of the RF field. Useful frequency response lies between 100 kHz and 2 GHz without needing software flatness corrections. Current instrumentation limits high frequency operation to 1 GHz. Field uniformity compares with large test chambers for equipment up to 0.025 cubic metres in volume, but you can test larger equipment with a reduction in performance. You can also use the emCell to monitor near field radiated emissions from your equipment by connecting a spectrum analyser to its input. Spectrum analysers are favourite for development use because they quickly identify peak emissions.

Like the other cells in the Product Survey Table, the emCell comes as a system. Instrumentation comprises the signal generator, RF power amplifier and attenuators, and includes cabling. You also get Chase’s IM PAK software (you have to supply a Windows PC). The software includes pre-calibrated transducer factors. Combined with optional transducers, the software supports full compliance immunity tests to IEC1000-4-6. Chase calibrate each emCell using an isotropic field sensor and in-house calibration software. Each corner point of the test area volume is individually measured at 1% steps over the frequency range, and the software combines these measurements to report field uniformity within the test volume. Some EMC test cells include an isotropic sensor in the feedback loop to linearise cell response, using a combination of calibration factors and real-time measurements.

Rohde & Schwarz and Wayne Kerr use dual septum designs to extend the size and uniformity of the RF field. Rohde & Schwarz’s S-LINE design consists of symmetrical two-wire TEM lines in a shielded enclosure (see Figure 3). Symmetrical TEM lines improve volumetric efficiency compared with a conventional single TEM line by as much as 3:1. Where a single TEM line creates a field that decays by 20 dB as you move vertically away from the septum, the equivalent test in an S-LINE enclosure shows a variation of around 4 dB. The relative field strength remains within the standard 6 dB tolerance within the whole cubic area, over almost all of the frequency range. In Wayne Kerr’s design, each septum is driven in anti-phase, effectively bisecting the chamber. In theory, a null exists halfway between each septum, creating a virtual ground plane. The RF field in Wayne Kerr’s dual septum cell is mirrored both sides of the virtual ground plane. You can minimise the contribution from equipment cabling by running the cables through this virtual ground plane area, which makes reproducing development test results easier.

Figure 3. A dual septum EMC test chamber increases volumetric efficiency by up to 3:1.

MA Instruments use a different design approach in their MAC micro-anechoic chambers. The MAC cell has a lengthened cuboid shape with a flattened horn at one end. Your equipment sits at the far (flat) end, basically in a waveguide. As a 300-mm cube, the MAC 1 cell is the smallest of four cells that offer volumes up to one cubic metre. MAC 1 and MAC 2 (500-mm cubes) are hybrid designs between the larger MAC cells and a TEM cell. The larger MACs work as waveguides across their frequency band, while the smallest two MACs change between TEM and waveguide mode depending on frequency. With adequate instrumentation, MAC cells are useful to 2.5 GHz. One of the MAC cell’s key properties is that emissions test results correlate far better with open area test sites. Typical TEM cells offer far field emission measurements above about 950 MHz, while MAC 1 resembles a far field from about 95 MHz (40 MHz for a MAC 2).

Monitoring a Unit Under Test
If your equipment has a display, you can simply watch for problems while you sweep the cell through its frequency and power level spectrum. Many cells have shielded mesh windows to provide visual feedback (check if this feature is optional because custom doors are expensive). What if your equipment is a sub-assembly, or lacks a display? And what do you regard as a "failure"?

Most cells have two sets of filtered electrical access ports -- line power cabling and equipment monitoring channels. A line impedance stabilising network (LISN) decouples conducted RF emissions picked up by your equipment from the main supply line. A LISN provides an "isolated" point to measure how much RF energy is being piped back into the line (make sure yours is uninterruptibly grounded!). Equipment monitoring channels are often just 25-way filtered D connectors. You probe nodes in your equipment and bring signals out of the cell for external monitoring. If you have a specific need, consult the cell manufacturer because most will configure turnkey systems to your specs.

Unit under test failures can be difficult to quantify. Most often, equipment fails at one or few precise frequencies, which is why RF immunity tests step through the spectrum in small increments (typically 1%). At some frequency and power level, a structure, a length of wire or even a printed circuit board trace resonates and causes the equipment to malfunction. Digital systems tend to crash, while analogue systems become noisy or unstable. If you can monitor key nodes in your equipment, you can automate your test by reporting failures versus test cell frequency and power level. Automated tests are especially useful for development work, comparing results from various "tweaks". Comtest’s G-Strip and the MAC systems in Table 1 usefully include data acquisition systems and software that record test data and pass/fail threshold values alongside cell settings.

Getting Started
Before you begin in-house EMC testing, consider your measurement expectations. Are you most interested in immunity or emission tests? Will you use the cell for compliance or pre-compliance tests? What do you consider "acceptably reproducible measurements"? Under-specifying a test cell buys disappointment, while over-specifying can jeopardise your project because cell cost rises very steeply against performance. Most buyers should not expect to make perfect measurements, but ones that are good enough to demonstrably advance your project. Balance performance criteria with your real needs.

Consider the size and shape of the items that you want to test. Follow the test cell manufacturer’s recommendations for your type of equipment and fully evaluate the results, including repeatability. If you have a test item with EMC test house results, look for correlation with their qualified results. Now try an item that is too large for the uniform test volume and check how well test runs compare. If the item is too large, you should expect less accurate results, but can you compensate by testing at a higher V/m level. Remember the spirit of what you’re trying to achieve. If your equipment survives a 10 V/m field that contains 20 V/m peaks, it’s likely to meet 10 V/m legislative limits.

When you start testing, observe careful layout. For equipment testing to IEC1000-4-3, include at least one metre of cabling in the test field. To get reproducible results you must place your equipment accurately in predetermined positions (radiating each face) and take great care with cable runs. Cabling forms an excellent antenna, and causes more measurement problems in EMC test cells than anything else. One metre of cable becomes quarter wave resonant at only 75 MHz. To help you re-create test conditions accurately, document your initial test runs with measurements, drawings, and photographs. To assess repeatability, run a series of tests with the same equipment under test, removing it, and replacing it in the test cell between test runs. You will see what to expect as systematic error, and adjust your expectations accordingly.

Useful Books on EMC Testing
Electromagnetic Compatibility, Jasper Goedbloed, ISBN: 0-13-249293-8.
EMC for Product Designers, Tim Williams, ISBN: 0-7506-1264-9.
The Engineer’s EMC Workbook, John Middleton, ISBN: 0-9504941-3-5.

You can reach David Marsh, Contributing Technical Editor for Test & Measurement Europe, at 101453.2302@compurserve.com.