The Transient Frequency Behavior Test

William H. Graff

For domestic manufacturers of narrowband equipment, understanding this test is now essential.

Originally conceived as voice-only services using what is now called frequency domain multiple access, narrowband transmitters continue to evolve into complex systems allowing voice and data to share smaller and smaller slices of permitted spectrum. The current land-mobile frequency infrastructure has shown remarkable flexibility in meeting this demand for more-sophisticated services.

During the past decade, the changes in telecommunications testing and standards have also been rapid and dramatic. Major milestones during this period were the release of TIA/EIA 603, Land Mobile FM or PM Communications Equipment Measurement and Performance Standards, in February 1992; the refarming of the available VHF and UHF spectrum by the Federal Communications Commission (FCC) in February 1997; and the release of a supplement to TIA/EIA 603-1 in March 1998.

One major testing change that took place over the past few years for domestic manufacturers of narrowband devices was the introduction of the transient frequency behavior test. Manufacturers need a thorough understanding of this test to successfully meet today's standards.

FCC Changes: A Brief History Lesson

When the FCC refarmed the land-mobile bands in 1997, they not only narrowed the channels, they also added important changes to required testing. These changes reflected the desire of the commission to satisfy demand for additional channel capacity and recognition that digital-serial communications would increasingly compete for bandwidth with traditional telephony.

One change was a requirement for greater attenuation of transmitter spurious emissions. The FCC changed the 43 + 10 log (P) requirement, which probably dated back to the tube days, to a more-severe 50 + 10 log (P), and the conducted and radiated ERP limits from –13 dBm to –20 dBm, which caught a number of manufacturers by surprise despite more than enough advance notification.

These changes were overdue. The European Union, with its adoption of many of the European Telecommunications Standards Institute (ETSI) specifications, had been using far stricter standards. The EU standards for conducted and radiated emissions were already below the –30 dBm or even –36 dBm level.

When the final stage of the FCC's spectrum reorganization goes into effect in 2005, spurious levels below 55 + 10 log (P) or –25 dBm will be mandatory.

To put these radiated limits in perspective, if a test uses the most severe ETSI limit for radiated spurious emissions of –36 dBm, a measuring instrument would detect approximately –46.6 dBm at 3 m. By extrapolating the FCC Class A device limits above 216 MHz from 10 to 3 m and converting the result to dBm, the limit appears very close to –51 dBm, which is almost within the permitted 4-dB variance allowed for open area test sites (OATS).

Another change was a requirement for spectral transmission efficiency, defined as providing a data-transmission rate of at least 4800 bps for each 6.25 KHz of spectral bandwidth. Driving this change was the expectation that speeds over transmitter telephony channels should parallel landline modems. This requirement would now exclude all simple data modems derived from voice equipment occupying 25-KHz channels using direct frequency-modulation techniques providing data rates no better than 9600 bps. It would also force adoption of Gaussian filter techniques for manufacturers wishing to send F1D.

A third change was the addition of the transient frequency behavior test, which had been published in the 1993 edition of TIA/EIA 603. It was not new; the test had existed for years in the ETSI standards. For domestic manufacturers, however, it was a new requirement.

In part, the test requirement was a response to interference complaints from users of data modems operating in simplex mode. These modems were subject to turn-on transients when the data transceiver would switch from receiver to transmitter. This transient effect occurs when any transmitter reaches full power many milliseconds before the synthesizer locks on the frequency. An emission is produced as the modem powers up many megahertz before the frequency-determining circuits lock on to the assigned frequency, quickly sweeping the carrier across the band.

For voice-only products, this transient effect was not considered a big problem. Turn-on transients from push-to-talk radios were tolerable, producing only an easily ignored pop, if anything at all, on the afflicted channels.

With the advent of data radios, however, transient effects became more critical. Disturbances to the spectrum caused increased bit error rates and error-correcting retransmissions. The problem worsened as designers pushed for faster and faster bit rates and the error corrections became more frequent. Clearly an intolerable situation was approaching.

The transition to using the transient frequency behavior test was not always trouble-free. Manufacturers were sometimes surprised to find that products that behaved according to requirements in steady-state mode did not meet transient behavior requirements.

In the early 1990s, for example, a device came to M. Flom Associates (Chandler, AZ) test laboratory for type examination. The device, based on a popular amateur handheld transceiver board tuned for the 450 to 470 MHz range, had a power output of 4 W, a data transmission rate of 9600 bps, and swept the frequency band from 7 MHz below its assigned frequency several times per second. It exhibited good behavior in steady-state mode and, according to the existing rules, could have passed type approval.

The laboratory, however, pointed out the transient behavior problems and eventually convinced the applicant that the radio did not fit into the principles of good engineering practices, and the device was reworked.

Defining the Test Procedure:
One Lab's Experiences

The experiences of test laboratory M. Flom Associates in developing an inexpensive and efficient system for the transient frequency behavior test offer a helpful guide to manufacturers who need to implement this test.

As M. Flom engineers first modeled it, the transient frequency behavior test was simple in concept. A reference RF generator was modulated with a 1000-Hz tone. Deviation was set equal to the appropriate channel bandwidth, and the reference signal was set to the exact frequency of the test transmitter carrier through a suitable combining network. The output of this combining network fed a test receiver whose output was in turn connected to an oscilloscope. When the test sample was keyed on, the oscilloscope would show the low-level modulated signal being extinguished, and the higher level carrier predominating. Any variation in carrier frequency relative to the reference signal would show up on the oscilloscope trace as a spike, indicating the difference from the assigned frequency to the actual transmitter frequency versus time.

There were two test conditions that needed to be satisfied: carrier-on time and carrier-off time. Trigger levels were to be set for –50 dBc, meaning that for radios of 5 W or less, the system would trigger on signal levels of –13 dBm or less.

M. Flom Associates began working on implementation of the test procedure as soon as the test was announced in the Code of Federal Regulations. There were several problems associated with the new test. First, there was no off-the-shelf solution for performing this test inexpensively. The only domestic manufacturer of instruments for the test offered a modulation domain analyzer, but it could not trigger on levels less than –30 dBc. Single-box solutions from Europe were very expensive. Directional couplers were bandwidth limited, and several of the couplers would be required to completely cover both the UHF and VHF bands.

One crucial element of the test was selection of the basic test receiver. The laboratory already owned an HP 8920A communications test set and its big brother, an HP 8901A modulation analyzer. So when the testers set out to adapt the existing instrumentation to the new FCC requirements, they had to choose a receiver only. After much consultation with the laboratory's programming group, the testers chose the older, but better understood, HP 8901A modulation analyzer.

Next the testers had to define the test problem. Test requirements were well specified, even though difficult to extract from a first reading of the test. The transmit signal was to be attenuated 40 dB below the test receiver's maximum input level. A reference signal generator was then to be introduced 20 dB below this carrier-input level. This translated into a dynamic range of 60 dB. The 8901 had a tuned RF sensitivity of at least –25 dBm with a maximum RF input of 1 W, which gave the testers a 65-dB range.

The directional coupler was potentially a problem. Because M. Flom Associates was a contract test laboratory, the testers did not want to employ multiple couplers covering multiple frequency ranges. Instead they chose a 4:1 resistive divider with 25- legs. This model offered a 10-dB attenuation from one port to any other port and no frequency-selective attenuation. The testers built the divider using common axial lead 1/4-W resistors inside a small enclosure. The divider worked well. Using a network analyzer, the testers were able to show good frequency response with predictable attenuation well past 550 MHz.

Triggering methodology was determined by trial and error. The testers expected to need a trigger to start measurements. Some of the first experiments used the voltage differential produced by total radio current through a resistor hooked in series with the power supply. The idea was that when the power amplifier turned on, an increase in current above receiver idle or quiescent would logically occur. A 1- resistor could produce more than enough voltage for triggering.

This triggering idea was at first accepted by the FCC for several of the laboratory's filings, but eventually rejected. The testers then turned to a common crystal detector. Their concerns were that a detector with satisfactory sensitivity would not produce sufficient trigger voltage. They were also concerned that excessive transmitter power would destroy the detector, but they were wrong. Output with the test setup typically gave more than 0.55-V-dc output when the oscilloscope was set to trigger on levels less than 2 mV, maintaining trigger levels down to the 50-dBc levels specified. The resistive combiner's 10-dB attenuation proved sufficient to protect the detector over several years and hundreds of test programs.

The Final Test Solution

The final test setup M. Flom Associates developed included a 30-dB attenuator, a fixed attenuator with attenuation sufficient to limit the carrier to a safe level, a 4:1 resistive combiner, a reference RF-signal generator, a crystal detector, a modulation analyzer, and an oscilloscope (see Figure 1).

For the test setup to operate optimally, two important issues must be considered. First, the FM modulation range and input attenuation must be set to fixed internal values. The autorange functions are capable of fouling the measurement. Second, the reference RF-signal generator should be exactly on frequency; even a difference of 50 Hz is very noticeable. It will be necessary to manually tune the instrument so it will recognize the low-level reference signal (see Figure 1).

Figure 1. Transient behavior test setup.

The oscilloscope should be set up in dual-trace mode, with one trace to monitor the demodulated output of the test receiver and the other to look at the trigger from the crystal detector. Time scale should be set to 10 µs/division per TIA/EIA 603. Vertical scale should be equal to channel spacing: 25 kHz, 12.5 kHz, or 6.25 kHz. The modulated 1000-Hz tone should just touch the upper and lower bounds of the display. In the second window, the trigger level should be monitored. The trigger level should be placed as close as possible to the beginning of carrier rise for the carrier-on condition and as close as possible to the beginning of carrier fall for the carrier-off condition. For equipment capable of multimode operation, the test should be repeated for all applicable channel bandwidths.

The most common source of error will be setup errors on the modulation analyzer. It is absolutely necessary to disable the internal FM and attenuation autorange functions. Sometimes it is necessary to freeze the IF bandwidth and to disable the internal error reporting.

Also, the oscilloscope should be set to record dc input levels from the demodulated output of the test receiver.

If all of the above settings are used, clear, repeatable test results should be achieved. For accurate diagnostics, designers should also check equipment across low, middle, and high frequencies within the band.

A Not-So-Transient Future

The 1998 addendum to TIA/EIA 603 updated the standard to reflect FCC, refarming of the VHF and UHF bands. Several of the items were expected, such as spectrum-analyzer bandwidths, updated and clarified frequency tables, frequency tolerance, and emission masks. The changes to transient frequency behavior, however, were unexpected.

The original standard required a trigger level at –50 dBc, and the revised standard set a less-severe threshold at –30 dBc. This new requirement brought the domestic land-mobile standard into conformity with its European counterpart and allowed use of modulation-domain analyzers for single-box test solutions. This revised methodology was adopted by the FCC and continues unchanged to this day.

For developers of narrowband transmitters, understanding and using the transient behavior test is essential now and will be for many years to come. Despite the current interest in broadband radio transmissions, narrowband spectrum transmitters in a form similar to those available now will continue to be in use for decades.

William H. Graff is director of engineering at M. Flom Associates (Chandler, AZ). He can be contacted at whgraff@worldnet.att.net.

Back to January/February Table of Contents