Chapter 4. Discussion

This doctoral thesis reports on the development of electronic and optoelectronic circuits and devices which would be useful for increasing the level of integration of pulsed TOF laser rangefinders and reducing their physical size and complexity. The performance of the innovative features was verified with prototypes of a laser radar module in a real distance measurement environment. The work started with the development of electronics for the receiver channel. Timing discrimination is of great interest in pulsed TOF laser rangefinding, because the measurement is based on the time required for an optical pulse to travel to the target and back and the dynamic range of the input signal level is large. Two types of timing discriminator were developed, a leading edge discriminator in which the signal can saturate in the amplifier channel and a high-pass timing discriminator in which the signal has to be processed linearly and gain control structures are usually used. In addition, a new idea is presented in which the high-pass timing discriminator is combined with analogue time interval measurement.

The leading edge discriminator has a large dynamic range (1:4000), as the signal is allowed to saturate in the receiver, whereas the signal for the high-pass timing discriminator has to be processed linearly. The transimpedance preamplifier usually limits the dynamic range of the signal to about 1:100–1:200, and in this work a current-mode attenuator between the photodetector and the preamplifier and a voltage-mode attenuator between the preamplifier and the postamplifier were used successfully to enlarge the dynamic range up to 1:620. The walk error of the leading edge discriminator after compensation is ±35 mm and that of the high-pass discriminator is ±4 mm. Analysis of the walk error in the leading edge receiver shows that this can be reduced by increasing the bandwidth of the receiver and shortening the rise time of the optical pulse.

Three prototypes for laser rangefinding devices were constructed in order to verify the usability of the receiver channel. The prototypes consumed about a fifth of the power required by earlier realisations which contained discrete components in their receiver channels and achieved similar or even better performance. The prototypes can be used as such for measurement applications involving mm-level accuracy, but serve in particular to show the potential of the new technique with respect to the construction of laser radar modules with a high level of system integration.

Integration of the photodetector into the receiver chip was also investigated by designing and testing several photodetector structures using standard silicon processes. The responsivity of the integrated photodiode (0.3 A/W) is about a half of that of discrete PIN photodiodes at 850–900 nm wavelengths, but the noise of the receiver is lower, because of absence of the parasitic capacitances and inductances caused by packages, PCB wiring, bond wires, the I/O cell and ESD protection structures, and partly compensates for the reduction in the signal. Integration of the photodetector into the receiver chip may be useful in short-range applications, but due to the high internal gain, APD is normally the most attractive alternative, possibly using hybrid realisation, although the latter would increase the number of components, as a high voltage has to be generated for the APD.

Finally, the operation of the integrated imaging TOF laser radar was demonstrated with a chip which can be used to measure distances in three directions with a single laser pulse. The chip includes four photodetectors and receiver channels to produce logic-level timing signals and a three-channel time-to-digital converter. The optics and the laser pulse transmitter are external components in the system.

In some applications a visible measurement beam helps to set up the measurement, and an additional visible pointer beam or an IR viewer is essential when using infrared lasers. This envoked the idea of using visible lasers as pulse sources. CW lasers developed for DVD-RAM purposes with a wavelength of about 650 nm are currently available that achieve an optical power of 50 mW. This power is very low relative to the high-power infrared pulse lasers used in this work (~100 W), but it is enough for short-range applications (up to 10 metres) or for applications in which reflectors can be used as targets (Banner). The driving electronics of the CW laser diode can be constructed with a supply of a few volts, which is advantageous, as the drivers of high-power pulsed lasers needs hundreds of volts in order to achieve pulse widths below 10 nanoseconds.

There are only a few papers in the literature concerned with transimpedance preamplifiers in which measurements are performed with a photodiode capacitance comparable to that available here. Phang et al. (1999) reported a preamplifier implemented in a 0.35 µm CMOS process which also employed fully differential structure in connection with the external photodiode generating the input to it. The circuit achieved a transresistance of 19 kΩ , a bandwidth of 70 MHz and an input-referred noise of 6.7 pA/√Hz with a photodiode capacitance of 5 pF. The same group described in Zand et al. (2001) an improved version of the preamplifier in which a common gate input was used to isolate the capacitance of the external photodiode. This preamplifier was also implemented in a 0.35 µm CMOS process and achieved a transresistance of 33 kΩ , a bandwidth of 255 MHz and an input-referred noise of 6.8 pA/√Hz with a photodiode capacitance of 2 pF. Pennala et al. (2000) reported a receiver channel implemented using a MAXIM GST-2 semi-custom bipolar (f_T = 27 GHz) array that achieved a bandwidth of 1 GHz, a transresistance of 11 kΩ and an input-referred noise of 10 pA/√Hz with a photodiode capacitance of 1.5 pF. The performance of the transimpedance preamplifiers described to date is summarised in Table 6.

In the field of timing discriminators, Ruotsalainen et al. (1995b) reported a ±60 ps walk error in a dynamic range of 1:4.4 in a 1.2 µm CMOS process and a ±30 ps walk error in a range of 1:70 in a 1.2 µm BiCMOS process. Both circuits employed a high-pass timing discriminator and walk error compensation with an offset voltage, and the pulse had a rise time of ~1 ns, a half-value width of ~7 ns and a fall time of ~2.5 ns. Jackson et al. (1997) reported a ±150 ps walk error in a range of 1:100 using a lumped-element R-C for shaping the pulse in a CFD. The chip was implemented in a 1.2 µm CMOS process and the rise and fall times of the pulse were ~5 ns and ~15 ns. Simpson et al. (1997) reported a ±250 ps walk error in a range of 1:100 using a high-pass timing discriminator implemented in a 1.2 µm CMOS process. The rise and fall times of the pulse were 10 ns. Pennala et al. (2000) reported a ±20 ps walk error in a range of 1:17. The chip was implemented in a MAXIM GST-2 technology semi-custom bipolar array and the rise time and width of the optical pulse were ~40 ps and ~80 ps respectively. The detection method was high-pass timing discrimination together with walk compensation by means of an offset voltage. The combination of the high-pass timing discriminator and time-to-amplitude conversion presented in paper III achieves a ±3.5 walk error in a range of 1:21. The performance of the timing discriminators is summarised in Table 7.

To the best of the author’s knowledge, no papers have yet been presented in the field of imaging devices utilising a focal plane array and the pulsed time-of-flight measurement principle. Lange & Seitz (2001) described a 3-D time-of-flight camera with 64 x 25 pixels that employs a focal plane array, but the measurement is based on continuous wave (CW) modulation. The unambiguous distance range is 7.5 metres and the chip can measure 10 frames per second with an accuracy of less than 5 cm.

In conclusion, this thesis describes high performance circuits for pulsed time-of-flight laser rangefinding and high performance laser radar modules developed especially for industrial applications. The results pave the way to a component-like laser radar construction which would certainly increase the number of applications for such a technology. Moreover, the laser radar chip, including customised photodetectors with receiver channels and multi-channel time interval measurement circuitry, has been shown to be successfully implementable using current semiconductor processes.