CHAPTER 3 Circuit Design
3.4 S IMULATION RESULTS …
The circuit in Fig. 3. 2 is used to construct a 32x32 array which is simulated by HSPICE. Since the array size is differs from the original definition of the space constants D, all the space constants in the simulation are shrunk with the same factor, 32/180. The simulation results are shown in Fig. 3.
6 ~ Fig. 3. 9. In this simulation, a voltage pulse is applied to the 5x5 cells in the center of the array to imitate the 500Hz flashing light stimulus and the spatiotemporal patterns of the 17th row are observed.
Fig. 3.6 is the monte carlo analysis of OFF bipolar cell with the original V-I converter structure and the modified linear transconductor structure. Monte carlo analysis target on MOSFETs of channel width and threshold voltage of Rm amplifier, band-pass filter, and transconductor. The simulation result of the original structure shows that the variation of output current level is 8mA while the variation of the modified circuit is less than 0.04mA. An OFF bipolar cell block features low DC current variation is proposed to solve the DC level variation problem in the previous work and to increase the stability of the circuit and the chip.
Then we show the transient response of the circuit when a pulse-stimulus is supplied. Fig. 3. 7 shows the output currents of the center cell of the 2-D array. The simulation is performed under the condition that a pulse signal with a turned-on duration of around 1ms is incident on the middle six cells of the array. The simulated input to the middle six cells are periodic stimulus with 100pA transient current added to a 100pA background-induced DC current. All other twenty-six cells are supplied with background currents of 100pA. All above-mentioned currents are supplied to the bases of the photo-BJT of all cells. The output of photoreceptor in Fig 3. 7(a) has overshooting and undershooting at the turn-on and turn-off similar to the CNN model simulation results in the previous section. It is also clear in Fig. 3. 7(b) that the output of the horizontal is analogous to that of the PH1 but has a larger magnitude.
As mentioned in the previous section, the gate bias of MSM11, MSM12, MSM21, MSM22, MSM31, MSM32, MSM41, MSM42, MSM51, MSM52, MSM61 and MSM62, VSM, controls the diffusion range of the horizontal, on bipolar cell, off bipolar cell, amacrine cell and ganglion cell, respectively. Fig 3. 8 shows the steady state output of space domain response with subjected to suitable gate bias voltages of VSM. Fig 3. 8(b) is the results of the horizontal when the middle six cells are incident with stronger light while the other cells are incident with background light. As could be seen in the figure,
a higher gate biasing voltage, VSM, causes lower resistance of the smoothing network, and a wider diffusion range is achieved. The results of photoreceptor with the same stimulus are shown in Fig 3.
8(a). As discussed previously, the edge of the incident pattern would have higher contrast in the output of PH1 but the extent of contrast varies with different diffusion range.
In Fig. 3. 9, the x-axis is time and the y-axis is the pixel location which denotes space. The black and white bars denote the spatial and temporal region of the input stimulus. The stimulus is applied to the 15th to the 20th pixel from 1001 µsec to 2000 µsec. The waveform at the right of each pattern is the spatial domain waveform(s) obtained at the time marked by the vertical arrow(s). The waveform at the bottom of each pattern is the temporal domain waveform obtained at the location marked by the horizontal arrow. Fig. 3. 9(a) and (b) represent the spatiotemporal patterns of the photoreceptor and horizontal cell, respectively. It can be seen from Fig. 3. 9(a) that the photoreceptor‘s signal level drops when there is a stimulus, and it returns to its original level when the stimulus disappears. Slight undershooting and overshooting in temporal domain can be expected in the periphery as it reacts to the stimulus directly. In the spatial domain, there is strong contrast at the edge of the stimulus. At the edge pixels inside the stimulus, the signal level is higher than the other stimulated pixels. Contrarily, at the edge pixels where the stimulus is just absent, the signal level is higher than the other silent pixels. Therefore, the spatial range of the stimulus can be well defined. In the temporal domain, the horizontal cell has a similar reaction to the photoreceptor, as can be seen in Fig. 3. 9(b). However, since the horizontal cell performs lateral diffusion in space, its spatial domain waveform spreads wider than that of the photoreceptor. Fig. 3. 9(c) and (d) represent the spatiotemporal patterns of the ON and OFF bipolar cells, respectively. The OFF bipolar cell performs bandpass-filtering on the signals from the photoreceptor. In the amacrine cell, the rectified signals from both bipolar cells are added with different gain factors, as shown in Fig. 3. 9(e).
Moreover, the response to the appearance of the stimulus is weaker than the response to the disappearance of the stimulus. Thus the amacrine cell provides a strong inhibition to the ganglion cell when the stimulus disappears in time. Therefore, the ganglion cell exhibits clear turned-ON
reaction, as can be seen in Fig. 3. 9(f).
Table II
The transistor’s sizes of circuits in Fig. 3. 2(a)-(d)
Transistor numbers W/L
(µm/µm) Transistor numbers W/L
(µm/µm)
MPH21-MPH24, MO5~MO6 8/0.7 MPH15~MPH18 16/0.35
MPLP1, MPLP3,
MCM9~MCM12 4/1 MLPON1, MLPON3 0.4/1.5
MLPON2, MLPON4 1.2/1.5 MRM1 3/0.4
Fig. 3. 1. The block diagram of a single pixel.
(a)
(b)
(c)
(d)
Fig. 3. 2. (a) The circuit of the photoreceptor and horizontal cell (PH1, PH2, H); (b) The circuit of ON bipolar cell (ONBIP); (c)OFF bipolar cell (OFFBIP); (d) The circuit of amacrine and ganglion cells (Ama, GC).
(a) (b)
(c)
Fig. 3. 3. (a) A linearized transconductor using floating voltages sources; (b) Flipped Voltage Follower implementation; (d) Transconductor using two identical FVFs to implement the floating batteries.
(a)
(b)
Fig. 3. 4. (a) The whole chip architecture of the implemented chip and (b) the implementation of smoothing networks.
RA0 buffer
AND gate MSB
LSB
R0
R1
R2
R31 1 0 1 0 1 0 1 0 1 0
RA1 RA2 RA3 RA4
(a)
(b)
Buffer:
(c)
AND gates:
in1 in2 in4in3 in5
out
(d)
Fig. 3. 5. The schematic of the address decoder used in the chip (a) column address decoder; (b) row address decoder; (c) buffer; (d)AND gate.
(a)
(b)
Fig. 3. 6. The monte carlo analysis of OFF bipolar cell with (a) the original V-I converter structure (b) the modified linear transconductor structure.
Fig. 3. 7. The HSPICE circuit-simulated time domain waveforms of the neuromorphic model of the ON sluggish sustain cell set in a 32x32 array for (a) photoreceptor, (b) horizontal cell, (c) ON bipolar cell, (d) OFF bipolar cell, (e) amacrine cell, and (f) ganglion cell. The light stimulus is applied at 1000th~2000th micro-second and Vsm = 2.2V.
Fig. 3. 8. The HSPICE circuit-simulated space domain waveforms of the neuromorphic model of the ON sluggish sustain cell set in a 32x32 array for (a) photoreceptor, (b) horizontal cell, (c) ON bipolar cell, (d) OFF bipolar cell, (e) amacrine cell, and (f) ganglion cell. The light stimulus is applied at 15th~20th pixel and Vsm = 2.2V.
Fig. 3. 9. The HSPICE circuit-simulated spatiotemporal patterns for (a) photoreceptor, (b) horizontal cell, (c) ON bipolar cell, (d) OFF bipolar cell, (e) amacrine cell, and (f) ganglion cell. These waveforms are recorded from the 17th row of the array. The x-axis is normalized time and the y-axis is the pixel location which denotes space. The stimulus is applied to the 15th to the 20th pixel at time point 1001 to 2000. The waveform at the left of each pattern is the spatial domain waveform(s) obtained at the time marked by the vertical arrow(s). The waveform at the bottom of each pattern is the temporal domain waveform obtained at the location marked by the horizontal arrow. Vsm = 2.2V.