A 700 MHz laser radar receiver realized in 0.18 μm HV-
CMOS
Mikko Hintikka; Juha Kostamovaara
Department of Information and Electrical Engineering, Circuits and
Systems Research Group, University of Oulu, Linnanmaa, Finland
This is a post-peer-review, pre-copyedit version of an article published in
Analog Integr Circ Sig Process.
The final authenticated version is available online at:
http://dx.doi.org/10.1007/s10470-017-1041-0.
1
ABSTRACT
This study presents a CMOS receiver chip realized in 0.18 µm High-Voltage CMOS (HV-CMOS) technology and
intended for high precision pulsed time-of-flight laser range finding utilizing high-energy sub-ns laser pulses. The IC chip
includes a trans-impedance preamplifier, a post-amplifier and a timing comparator. Timing discrimination is based on
leading edge detection and the trailing edge is also discriminated for measuring the width of the pulse. The transimpedance
of the channel is 25 kΩ, the uncompensated walk error is 470 ps in the dynamic range of 1:21,000 and the input referred
equivalent noise current 450 nA (rms).
Keywords:- laser radar receiver; laser ranging; optical sensors; timing discrimination
1 INTRODUCTION
The pulsed time-of-flight laser range finding principle is based on the measurement of a transit time of a
short laser pulse travelling from the laser transmitter to the target and back to the pulse detector electronics. The
measured flight time of the pulse can be converted to a distance of the target based on the known velocity of
light. Fig. 1 shows a typical construction of a pulsed TOF laser range finder. The achievable precision is typically
at the level centimeters. [1]
Fig. 1 Block diagram of a TOF based laser radar
Since the use of optical signals for measuring distances removes the need for physical contact with the
target, laser range finding has been applied in many fields, including the automotive, military or robotics for
target identification and range determination [2–4]. The unique advantages of the pulsed TOF range finding
technique, are high precision and short measurement time. It also gives a high spatial resolution due to the fact
that electromagnetic radiation at optical frequencies can easily be focused with optical lenses.
The main factors limiting the measurement accuracy of a TOF laser radar are noise limiting the single
shot-precision and amplitude variation of the received echo causing systematic timing error (known as timing
walk). Typical TOF based laser radars are using laser pulses with a pulse width of 3 – 5 ns limited by the
limitations of high-speed laser diode drivers [5]. However, in principle, the single-shot accuracy, timing walk as
well as eye-safety could be improved if shorter, even sub-ns pulses, were used. This would also help detecting
multiple targets within the measurement range. Recently, techniques to produce high-energy sub-ns pulses with
semiconductor laser diodes have been presented, and thus there is an interest to study what kind of performance
would be available in laser ranging using these transmitter techniques [6–9].
In this work, a laser radar receiver that can be used to detect high-energy sub-ns pulses has been designsed
and chracterized. The proposed integrated CMOS laser radar receiver is designed for a fast and accurate pulsed
TOF laser range finder especially targeted to measure small distance variations of a distant target e.g. for
measuring the vibration of a surface at some distance.
The paper is organized as follows. Section II describes the sub-ns pulse detection from the accuracy and
precision points of view. The construction and an implementation of the receiver channel are presented in section
III. Section IV presents the most essential verification results from the chip. Conclusions are given in section V.
2 SUB-NS LASER PULSE DETECTION
2
Typical pulsed TOF laser range finders nowadays use 3 – 5 ns laser pulse in the transmitter. Even though
the usage of a shorter pulse width would be beneficial from the accuracy point of view, the problem is that
the peak power level available from commercial laser diode sources giving a single sub-ns pulse is limited
to sub-W range [10, 11]. However, recently, a number of special gain-switched laser constructions have
been suggested to increase the available energy level [6, 9, 12, 13]. A particularly interesting semiconductor
laser approach is based on the “enhanced gain switching” principle, which is capable of producing laser
pulses with an energy level and pulse width of ~ 1 nJ and ~ 100 ps, respectively [7, 9]. A notable feature of
this technology is that the driver requirements are quite moderate and can be straightforwardly realized with
standard MOS circuit technology. The pulse transmitter used in the receiver verification measurements of
this work is based on a MOSFET based driver presented in [8]. The transmitter utilizes a custom designed
bulk laser diode working in the “enhanced gain switching” mode and it generates a 100 ps (FWHM) pulse
with a peak power of 10 W at a wavelength of 860 nm. The properties of the sub-ns pulse detection are
analyzed in the following subsections to motivate the receiver design.
2.2 Timing walk error
The amplitude of the received pulse in a pulsed TOF may vary a lot due to changes in the reflectivity,
orientation and distance of the target. When using the leading edge timing discrimination principle, where
the timing comparator generates the timing signal as the received echo pulse crosses a certain threshold at
the input of the comparator, the amplitude variation causes timing error (typically known as timing walk
error) [14]. The amount of walk error depends largely on the slew rate of the pulse. The shorter pulse rise
time gives a lower walk error, as shown in Fig. 2 where three pulses with different pulse widths and rise
times are shown with the corresponding timing walk error.
Fig. 2 Walk error
2.3 Noise and jitter
The input referred noise current is one of the most critical parameters in a trans-impedance amplifier (TIA),
which is normally used as the preamplifier in a pulsed mode laser radar receiver. The noise of the TIA
usually dominates over all the other noise sources in the amplifier channel and therefore determines the
sensitivity of the receiver [15–19]. Fig 3 illustrates the definition of TIA noise as the equivalent input noise
current I
n, TIA
at the input of the noiseless TIA. The equivalent input noise current source together with the
noiseless TIA reproduces the output noise of the actual noisy TIA.
3
Fig. 3 TIA definition
The input referred RMS noise current relates directly the sensitivity of the laser radar receiver and can be
expressed with a single number. It is determined by dividing the total RMS output noise voltage by the
TIA’s mid-band transimpedance value [15].
() ()
dffIfZ
R
i
n
BW
T
T
rmsn
2
2
2
0
,
1
⋅=
∫
>
, (1)
where
|
()
|
is the frequency response of the transimpedance amplifier and R
T
is its mid-band value.
The bandwidth of the TIA must be optimized so as to reduce the total integral noise. However, the
bandwidth needs to be sufficient for detecting the pulses. The time bandwidth product optimizing the SNR
for Gaussian shape pulses is defined to be 0.44, which means that the optical bandwidth is 0.44
⁄
-, where
T
P
is the pulse width (FWHM) [20]. When increasing the bandwidth of the TIA while decreasing the pulse
width, the noise is increasing under square root as in (1). That means that if the pulse width decreases by
the factor of ten, for example, and thus the bandwidth of the TIA increases by the factor of ten
correspondingly, the total noise increases roughly by a factor of three only. If the laser pulse energy can be
kept constant, that will improve the signal-to-noise ratio and also the timing walk error. Fig. 4 shows three
pulses with the same pulse energy but different pulse widths and the corresponding noise levels of the
receiver. In reality this tendency is softer however since the spectral noise of the preamplifier also tends to
become lower at lower bandwidths. [15]
Fig. 4 Noise level vs. pulse width
The jitter in the pulse detection is proportional to the ratio of the receiver noise and the slew rate of the
timing signal.
( )
SNR
t
tv
rnoise
jitter
≈
∂∂
≈
σ
σ
(2)
As already explained above, a shorter pulse with a certain energy improves the SNR and also has a shorter
rise time, and thus the jitter in the detection is improved as well. [21] The above brief analysis shows thus
that from the point of view of measurement accuracy it is in principle advantageous to shorten the laser
pulse in laser ranging, especially if this can be done without decreasing the total energy of the pulse.
4
3 RECEIVER CHANNEL
Typically the receiver channel electronics used in a TOF based laser range finder consists of a trans-impedance
pre-amplifier, a voltage type-post amplifier and a timing discrimination generating the timing mark from the
received echo pulse. The distance measurement is based on the measurement of time interval between the
generation of the laser pulse and the detection of the laser echo with the receiver. The bandwidth requirement
for the receiver arises from the characteristics of the laser pulse used in the ranging. The receiver realized in this
work includes the amplifier chain as mentioned above and also two timing comparators, which are used to detect
the leading and the trailing edges of the analogue timing pulse at the output of the receiver channel. Both edges
are discriminated for measuring the width of the received pulse. This information can be used for walk error
compensation, for example. The optical pulse is converted to a current pulse in an external Avalanche Photo
Detector (APD). Fig. 5 shows a simplified block diagram of the receiver chip realized in this work.
An APD with the diameter of 100 µm and internal capacitance of 0.5 pF was used as the photo detector
of the receiver channel. The diameter selection is determined by the stripe width of the laser diode, since they
should match [22]. Further minimizing the input capacitance the input pads with smallest ESD clamp diodes
were used and unnecessary metal was removed from the pads. In [23] it was shown that the discrete high-speed
APD outputs a ~ 300 ps pulse if a ~ 100 ps optical input pulse was used. As pointed out above, the bandwidth
requirement for the channel becomes from the pulse width and ~ 300 ps means that the bandwidth of the receiver
channel should be in the range of ~ 1 GHz. The maximum input signal from the APD can be however very high
and thus the input current for the receiver channel was limited with external cross-coupled Schottky diodes. The
transimpedance of the channel needs to be high enough to receive a small signal (slightly above the noise level)
at the level detectable by the comparators. The input-referred noise current was estimated to be ~ 450 nA and
thus the transimpedance of 25 kΩ gives ~ 11 mV noise voltage to the output of the channel. With this trans-
impedance the signal level with a SNR of 10 is ~ 110 mV.
Fig. 5 Block diagram of the receiver
3.1 Receiver Stages
3.1.1 Preamplifier
The transimpedance preamplifier which is used to convert the APD current to a voltage signal includes the
core amplifier with the active nested feedback topology and the overall shunt feedback resistors. Fig. 6
illustrates the topology as a single-ended version. The inner feedback loop makes a voltage amplifier
compose of two transconductance gain stages (g
m2
) and the active feedback (g
mf
). Basically the inner
amplifier follows the Cherry-Hooper topology, [24] yet with an active feedback. The outer loop with the
passive shunt feedback resistor performs the current to voltage conversion. Basically three gain stages could