Implementing time-to-digital converters in size- and power-constrained sensor applications

2Sensors need to convert physical values such as pressure, weight, distance, and length into a digital value. In the first step of such a conversion, the physical is converted into a temporary electrical analog value such as voltage or time.

Analog-to-Digital Converters (ADC), which currently dominate in market presence, use electrical voltage as a temporary value. For example, a weight is converted with a Wheatstone bridge and strain gages into a differential voltage. An ADC converts the differential voltage to a digital value.

A Time-to-Digital Converter (TDC) can do the same type of analog-to-digital operation, but the difference with TDCs is that time is used as a temporary value (“time”) instead of volt-age. With a Resistor-Capacitor (RC) circuit, a weight is converted with strain gages into a time difference. A TDC converts the time difference to a digital value. The basic principle is the same. Only the temporary value during the process of conversion is different.

Though historically ADCs have been primarily used, TDCs can offer technical advantages in size, ease of development, and power consumption.

Process technology challenges alleviated

TDCs are undemanding in matters of process technologies. A TDC is basically a purely digital circuit. No special process technologies in the wafer fabrication are required for them to become more powerful. Everything can be integrated and manufactured on standard CMOS processes.

Today’s sensor applications often require intelligent front-end systems. Such digital intelligence (such as microprocessors and non-volatile memory) must operate with the converters in the smallest spaces. A TDC is capable of being manufactured in all kinds of wafer process technologies. The integration of a high-resolution ADC into a System-on-Chip (SoC) can be challenging and can potentially cause problems at migration to smaller process technologies such as at the deep-submicron level, and at migration to altered wafer process technologies. With a standard microprocessor today, for example, a 10 bit to 12 bit ADC is integrated. However, even at a 14-bit resolution, measurements are very imprecise. High-end ADCs can achieve 24-bit resolution, which is a 4,000 times higher resolution when used in standard microprocessors.

In contrast, a TDC is easy to migrate from one technology to another because of its digital nature, and automatically improves performance at smaller process technologies due to taking up less die space, which facilitates faster gate switching and less switching delay. A migration requires minimal R&D effort with little risk of failure due to a TDC’s ease of porting digital values. A TDC in general can achieve a resolution of 30 bits because time can be measured and resolved much more precisely than analog voltage. Such resolution is achieved regardless of having an on-chip microprocessor or not. TDCs are undemanding because they are built with digital gates that are the same as the ones in a microprocessor.

Simplified front-end design

TDCs also require little effort for front-end design. In contrast, an ADC needs a very good pre-amplifier with a stable and precise amplification of 50 to 200 times to measure strain gages. Such a pre-amplifier is more difficult to develop and port to a different process technology.

A TDC needs an analog front-end circuit for the conversion into time; however, in most cases the required circuits are very simple and easy to integrate on different wafer process technologies. For the same set-up as the ADC example above, a TDC does not need a pre-amplifier. The sensor signals can be compared directly with simple comparators. The reason for this is the inherently higher 30-bit resolution of a TDC. All the problems caused by pre-amplifiers and portability do not exist with TDCs.

Low power consumption

More and more battery-powered applications drive the demand for less and less power consumption. With a TDC, such power-optimized applications can be realized without modification. TDCs are quickly and easily turned on and off; the power-consuming circuit for signal processing at high-precision measurements can be switched on only when it is really needed.

For example, designers are facing battery-powered applications that require a 10-year life cycle in permanent operation. As such, little power is available to run the system. Ultrasonic heat meters with a TDC-based IC inside are capable of processing everything from temperature and ultrasonic measurements within as little as 2 µA power consumption. This gives enough room for the power needed to operate the microprocessor and covers the remaining tasks needed in applications such as heat meters. The requirements of an ultrasonic water meter are between 5 and 10 times higher.

Case study: Ultrasonic flow metering

To achieve an ultrasonic flow measurement with the required accuracy, the time difference must be resolved into the lower picosecond (ps) range, such as 50 ps (Figure 1). With a “normal” counter, developers would need a clock frequency of 20 GHz for this task; that cannot be achieved without a lot of time and effort.

Figure 1: Ultrasonic water and heat metering with acam TDC-GP22.
(Click graphic to zoom by 1.9x)

Acam uses the gate delay of digital circuits (such as those from an inverter) to resolve in the picosecond range. It measures how many gate delays have passed in a certain time difference. Acam developed digital circuits to measure almost any time difference stable over temperature, voltage changes, and process variations. Also, for increased resolution requirements, these TDCs address the complete spectrum of measurement applications. They have a stable resolution starting at 10 ps (TDC-GPX, Figure 2), and 98 percent of all available applications can be covered with that. The high resolution demanded today is no longer an obstacle. Easily a resolution below 5 ps can be achieved and covers demand. Converter integration in a complex product on a deep-submicron process can lead to a smart solution with very little silicon area consumption.

Figure 2: The acam TDC-GPX multifunctional Time-to-Digital Converter (TDC) has four operation modes with 81 ps to 10 ps resolution.
(Click graphic to zoom by 1.9x)

The future of TDCs

Demands for integrating TDCs will rise as do the demands for smaller ICs, higher integration, more integrated intelligence, smart sensors, little or almost no power consumption, faster product development, and flexibility to quickly generate new products in alternative process technologies. Semiconductor manufacturers have yet to invest and promote TDC technology as a parallel technology to ADCs, but niche applications are utilizing TDC circuits and have come to dominate metering segments with big market shares. TDCs already cover more than 90 percent of the ultrasonic flow metering market share, more than 70 percent of magnetostrictive ultrasonic position-ing, more than 50 percent of positron emission tomographs for cancer diagnosis tools, and double digit percentages of industrial and science segments.

Additionally, new sensor segments continue to adopt TDC technology. In the field of capacitive sensor elements, TDCs fit almost perfectly to measurement requirements. Capacitive MEMS sensors and complex sensor systems need powerful converters and built-in intelligence to cover the sensors of the future. TDCs can address the challenges in these applications, and embedded developers will recognize their advantages to the demanding applications of the future.

Augustin Braun, together with Andreas Larsch, is CEO and owner of acam messelectronic gmbh, a company located in Stutensee near Karlsruhe, Germany, that is specialized in solutions and concepts for highly precise measurement technology.

acam messelectronic gmbh