Latest RF power sensor technology - interview with Rohde & Schwarz

Our editor, Ian Poole interviews Thomas Reichel, Head of Power Meter Development at Rohde and Schwarz about RF power sensor technology and looks at key elements in their design.

One of the key parameters for any radio frequency system is the level of the RF power that is generated. These days tighter controls are required for many RF-based equipment from cellular handsets to base stations as well as satellite systems, and all forms of RF systems where RF power is present.

With the recent announcement by Rohde and Schwarz of their 67GHz coaxial power sensor, we thought it would be interesting to discuss power sensor technology with Thomas Reichel, the head of the Power Meter Development Group to get a real insight behind what makes a good RF power sensor and what the design challenges are.

It is clear from the way that Thomas Reichel talks that he knows both the marketplace and the technology behind power sensors in real depth. He also has an infectious enthusiasm for the subject.

Basic RF power sensor requirements

I asked Thomas about RF power sensors and what should people look for in an RF power sensor when they choose one? He replied: "Customers should look at what they really want the power sensor for - is it for universal use where a more general power sensor is needed, or is there a specific requirement for which a more specialised sensor may be required?"

He also notes that care should be taken in choosing the manufacturer of the RF power meter because they are normally used for very accurate measurements - full calibration, traceability and also support are essential. It is important to find a company that can provide all of these.

Right sensor for the right job

There are a number of different types of RF power sensor - what are the main types and what is each one particularly suited to?

"There are two main types of RF power sensor - thermal and diode sensors. The thermal types have been around for many years - twenty or thirty years ago they were the main type. However since the beginning of the 1990s the industry needed capabilities the thermal sensors could not provide because of the new types of modulation being used. Now eighty to ninety percent of all power sensors are diode based."

Looking at the different types of sensor, there must be good reasons for using the different types - what are the limitations of each type?

"Well, there is a lot to say about the way in which each type works. Although thermal sensors are slow over a relatively large portion of their dynamic range and they only can measure the average power of a signal, they still deliver the most accurate measurement results. One reason is the effect of harmonics and non-harmonics can be predicted very well. Diode power sensors are, in general, much more critical when it comes to modulation or superimposed spurious, but modern types like wideband power sensors or multi-path power sensors out-perform their simple predecessors very well in these aspects. So, today, diode based RF power sensors offer sufficient accuracy for the majority of applications. The diode based types need very careful design, but they offer many advantages in terms of functionality, speed, and dynamic range - they can actually measure down to levels of around -60 or -70 dBm whereas the thermal power sensors typically have a lower range limit of between -30 to -35 dBm. Reading the peak power or the rise time of a pulse and making statistical evaluations of the envelope within some milliseconds are typical tasks for the new designs.

Reichel added: "For about eight years, so-called USB power sensors have been available. These are in fact, miniaturized power meters that are operated directly from PC's or other measuring instruments. If properly designed, these sensors outperform their traditional counterparts in terms of accuracy, speed and versatility. Rohde & Schwarz was first in this field and feels dedicated to go this way further."

RF power sensor design challenges

I asked Reichel about the key challenges in the design of an RF power sensor?

He replied by saying: "There are many challenges in power sensor design, but probably the main ones in the RF area revolve around the impedance matching required to provide the very high accuracy levels that need to be achieved. If there is a poor match, then there will be multiple reflections of considerable amplitude between the power sensor and the DUT, resulting in the DUT emitting more or less power than in the ideal case of a perfectly matched load. Therefore the measurement result of the sensor will be different from the expected one, even if the sensor itself performs very well. The power loss at the sensor input due to sensor mismatch doesn't cause mismatch errors at all, since this loss is included in the calibration factor of the sensor. In contrast to the effect of multiple reflections, reflection loss is independent from the DUT and can therefore be determined very well"

"There are other challenges as well. For thermal sensors, enabling the sensor to remain accurate when the environment is changing presents real problems. As the sensor basically operates by detecting the temperature change resulting from power being dissipated in a load, external temperature changes need to be removed from the equation."

"Another challenge is in achieving the dynamic range needed. It took a lot of development work to be able to achieve a dynamic range of 55dB with a thermal power sensor. In previous times, linearity was a problem at the top end, which has been solved meanwhile by accurate numeric compensation. The low power end of the scale is still a challenge due to the influence of noise, zero offset and zero drift."

Accuracy

I asked Reichel about power meter measurement accuracy, and wondered what some of the major pitfalls were when making measurements.

He answered by saying that: "One of the major problems we find is that people use the wrong type of sensor. Besides a selection by frequency and power, it is necessary to look at the required measurement and its uncertainty as well as at the DUT and the waveform being measured to select the correct sensor - thermal, multiple-path, wideband. CW power sensors shouldn't be used any more, since they don't like any modulation of the envelope or even spurious, delivering unpredictable results in these cases."

"Another major problem we find is that the RF connector on the power sensor becomes damaged with use. As the RF connection directly to the sensor is very important any damage to the connector can result in a poor connection being made and this will have a significant effect on the measurement."

"There are other problems that can occur - the frequency input may not match the signal frequency. Power sensors make a lot of frequency-dependent numeric compensations. Therefore they need a numeric frequency input - that is until they can make their own frequency measurements."

"There may also be a large zero offset if the zeroing has been forgotten - it is surprising how often this happens. The zero offset is an additive power contribution independent from the power level, leading to large relative (dB) errors at low levels. Zeroing reduces this effect by considering the outcome of a test measurement without signal, executed by the user. Zero offsets mainly stem from un-symmetries in the signal path between the RF transducer and the first amplifier stage, generated by the sensor and the base unit as well in traditional designs. Since integrated power meters like all the sensors of the Rohde & Schwarz NRP family contain all the analogue circuitry, these offsets can be measured in the factory and stored in the data memory of the sensor. They are automatically considered from the first measurement, so that an additional zeroing by the user can be omitted in many cases."

"Another important consideration is the mismatch uncertainty and whether this has been considered. At least 'guessing' the amount of mismatch uncertainty should be a must for any power measurement, since this contribution to total uncertainty often exceeds the uncertainty specified for the sensor alone. This is especially true, when frequency rises and the matching of the DUTs and the power sensors as well get worse. However, if the impedance mismatch of the DUT and the power sensor are known in a complex notation, a proper numerical compensation is possible. The R&S NRP power sensors support this by a measurement function called 'gamma correction', which only needs the complex reflection coefficient of the DUT to be input to the sensor."

"Gamma correction is key to renew the concept of so-called offset tables. They normally account for nominal loss or gain ahead of the sensor without considering mismatch at the sensor side or the DUT side as well. Storing all four S parameters instead of nominal loss only enables more accurate results. Rohde & Schwarz was first with having implemented this feature in its power sensors of the NRP series."

"While there are a number of potential errors that need to be considered, when these are properly taken into account, the use of an RF power meter is by far the most accurate way of measuring RF power levels."

Spectrum analyser power measurements

Many spectrum analyzers have a good power measurement capability these days - why does anyone need a separate power meter?

"Yes spectrum analyzers are very good, but they cannot provide the accuracy that RF power meters are able to provide. Spectrum analyzers have several drawbacks if very accurate power measurements are to be made. The first is that they need a cable to connect the instrument to the item under test - this has an unknown loss, and even if it is measured, it is difficult to characterise it exactly, so there is a large uncertainty here. Power meters connect directly to the source and do not need a connecting lead so they are inherently more accurate."

"Impedance matching is another key issue. An RF power meter will be able to provide a VSWR of better than 1.2:1, whereas a spectrum analyzer may have a VSWR of anywhere up to 2:1. Although they are better at lower frequencies, the uncertainties increase as the frequency rises. The impedance will change when ranges are changed this also creates inaccuracies."

"However it should be said that spectrum analyzers are often quite good when it comes to relative measurements, although some precautions must be taken to avoid the pitfalls."

New Product

With the launch of their new 67 GHz power sensor, I asked what was so special about it?

"I think the frequency limits and the accuracies we have achieved are very impressive. It is not easy to create an accurate power sensor at these frequencies for a variety of reasons."

Wanting to explore this a little further I asked what formed the major challenges of a 67 GHz sensor?

The answer from Thomas Reichel was quite definitive: "Mechanical problems," he said: "At these frequencies the dimensions are very small, and there are issues with mechanically matching the coaxial feeder to the transducer while maintaining the impedance match are quite exacting."

"To achieve this we developed a new method for making the transition - previously this consisted of several different parts. We took a radical look at the design and came up with what we call a soft transition which is manufactured using photo-lithography to give the required accuracy."

One things that has stuck me - the development of this power sensor cannot have been cheap - are there many users for them?

Reichel replied: "There are more than you think. There are many signal generators, network analysers, spectrum analysers and the like that extend to these frequencies and calibration facilities for these will need RF power meters that are capable of operating at these frequencies. In addition to this, there is growing number of 60GHz short range communications systems that are being introduced - again these need power meters during the design, manufacture and for servicing and repair."

As our conversation came to an end, I thanked Thomas Reichel for his open and interesting answers about RF power sensors and their design.

Thomas Reichel is Head of Power Meter Development at Rohde & Schwarz. He received a Dipl.-Ing. degree from Technical University Braunschweig in 1978 and started his career as an analogue hardware designer for different electronic instruments. Since 1991 he has been in charge of the power meter and voltmeter activities of Rohde & Schwarz. His professional interests include network analysis and general metrology. He is owner of several patents in the field of RF measurement technology.

Rohde & Schwarz is an independent group of companies specializing in electronics. It is a leading supplier of solutions in the fields of test and measurement, broadcasting, radio monitoring and radiolocation, as well as secure communications. Established more than 75 years ago, Rohde & Schwarz has a global presence and a dedicated service network in over 70 countries. It has approx. 7400 employees and achieved a net revenue of € 1.2 billion (US$ 1.7 billion) in fiscal year 2008/2009 (July 2008 to June 2009). Company headquarters are in Munich, Germany.