SIMULATION RESULTS

CHAPTER 2 CIRCUIT DESIGN OF THE RETINAL CHIP

2.4 SIMULATION RESULTS

Thirty-two basic cells describe in the preceding section are connected in series to form a 1-D array to perform the HSPICE simulation. The following simulation results are all from the simulation using HSPICE model for 0.35µm standard CMOS process. A standard BJT device is used to model the photo-BJT used in real chip. The BJT’s base is each cell’s input to which a base current could be supplied. This base current simulates the photo-stimulus illuminated on the real chip because the base current induces the flow of emitter current proportional to this base current, and furthermore the photo-stimulus similarly induces the emitter current proportional to the intensity of incident light.

Therefore the input of each cell is a current in the following simulation.

First we show the transient response of the circuit when a pulse-stimulus is supplied.

Fig. 18 shows the output currents of the center cell of the 1-D array. The simulation is performed under the condition that a pulse signal with a turned-on duration of around 1.2ms is incident on the middle six cells of the array. In the simulation a virtual

background illumination current of 1.2nA is added by each cell’s offset transistor MN13.

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 the cells. The output of PH1 in Fig. 18(b) 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. 18(c) that the output of PH2 appears to be the low-passed output of a pulse signal and the output of the horizontal is analogous to that of the PH1 but has a larger magnitude.

Fig. 19 shows transient response of the output currents when the same pulse signal mentioned above is incident on the middle of the array while the transistor MN13 that supplies additional virtual background illumination is turned off. Notice that a

background-induced photocurrent of 100pA still exists in this case. Comparing this result with the one in Fig. 18, which is simulated when offset current of 1.2nA is added by MN13; it is clear that the undershooting phenomenon vanished without the aid of MN13 in spite of the background-induced photocurrent exists. In fact, the offset current added mimics the background illumination illuminated on the whole chip. If the intensity of the background illumination is strong enough that the background-induced photocurrent reaches an adequate level to make all the transistors work normally, that is, not being turned off, the offset current provided by MN13 will be no more needed. The additional offset current is only required under weak background illumination.

Fig.20 demonstrates the result of simulation under the condition that an overlarge offset current of 11.6nA is supplied. On suchlike condition, the overshooting and

undershooting of PH1 no more exist. As discussed in previous section, the time constant of PH2 varies with different bias current of the transistor MP7. The higher level of bias current, I_D_,MP₇, makes lower time constant τ . The disappearance of overshooting and

undershooting under an overlarge offset current arises from the variation of time constant of PH2. An overlarge virtual background illumination or real ambient illumination results in the increase of the photo-input current which also makes the drain current of MN4 to increase. Therefore, drain current of MN8 which is mirrored from that of MN4 turns out to be larger, and consequently I_D_,MP₇ increases and the time constant of PH2 decreases with the larger offset current. As explained in the simulation of the CNN model, the smaller the time constant, the less noticeable the overshooting and undershooting would be. When the time constant is too small that the delayed signal, or low-passed signal in the same meaning, of PH2 could catch up with that of PH1, the overshooting or

undershooting then no more appears. From what has discussed above, a conclusion is drawn that a proper level of offset current added to the photocurrent ensures the correct temporal response of the retina cell.

The following simulation is performed to show the influence of bias current of MP7,

7 D,MP

I , on the PH2’s time constant. The transfer function of PH2 is assumed to be a single pole low-pass form as described in equation (4). Several ac current signals with different frequencies are supplied to the input of PH2, that is, MN8 for the calculation of current gain of PH2 for different-frequency input. The input of PH2 is the drain of MN8 and the output is the drain current of MN11. The time constant is thus the inverse of the 3-dB frequency of the simulated transfer curve according to the single pole assumption.

Vary the DC level, also the bias current of MP7, of the input of PH2, and repeat the simulation above. Fig. 21 shows the time constant variation when supplying PH2 with different level of DC current. It is clear in this figure that the time constant becomes smaller when a larger I_D_,MP₇ is supplied. According to equation (9) and (10), transconductance of MP7 is proportional to I_D_,MP₇ or the square root of I_D_,MP₇. Therefore, the time constant is then inversely proportional to I_D_,MP₇ or the square root of I_D_,MP₇ because of the fact that the time constant is inversely proportional to the transconductance of MP7.

As mentioned in the previous section, the gate bias of MNL and MNR, Vsm, controls the diffusion range of the horizontal elements. Fig. 22 shows the steady state output of the horizontal and PH1 with subjected to different gate bias voltages of V_sm.Fig. 22(a) is

the result of the horizontal when the middle six cells are incident with stronger light while the others 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 PH1 with the same stimulus are shown in Fig. 22(b). 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.

As for the space-time pattern of the 1-D retinal array, the six center cells are turned on for some time and turned off suddenly during simulation, and the figure of their outputs versus time and position is color-coded as demonstrated in Fig. 23. In the figure, the output patterns of the horizontal, PH1, and PH2 are shown. The time axis is located on the bottom and the space axis is on the left. The upper color bar in red indicates the duration the light is incident, and the one in the right side shows the range of incident cells. The lower most plot sketches the transient response of the most middle cell and the steady-state spatial response when the six cells are turned on is plotted in the right side of the space-time pattern. The color bar in the right most is the color mapping from the normalized value to corresponding color assigned.

As shown in the figure, the output of PH1 spreads out laterally when the light is just incident. After some time, the output begins to contract in space gradually and reaches a steady-state value. The contraction is caused by feedback of both the horizontal and PH2 to PH1. During the turn off transient, output current suddenly drops and returns to the steady state value. The pattern of horizontal appears similar .characteristic of spreading and contraction but the extent of spreading is wider than that of PH1 since horizontal produces its output by spatially averaging the input from PH1. PH2 which is the

low-passed output of PH1 in temporal domain is shown in Fig. 23(c).This spatiotemporal pattern is analogous to the measured patterns of the input of bipolar cells sent by cones in mammalian retinas [5].

Table I

The designed device dimensions in the retinal basic cell.

Device label Width/length

MN1=MN2=MN12(W/L) 2µm/1µm

MN3(W/L) 1.5µm/1µm

MN4=MN5=MN6=MN7=MN8=MN10=MN11(W/L) 1µm/2µm

MN9(W/L) 0.4µm/2µm MN13(W/L) 1µm/10µm MP1=MP2(W/L) 1.5µm/1µm MP3=MP4(W/L) 2µm/1.5µm MP5=MP7=MP9(W/L) 2µm/1µm

MP6(W/L) 8µm/1µm MP8(W/L) 2µm/0.4µm MP10(W/L) 35µm/12µm MSL=MSU(W/L) 1µm/1µm

G=4 τ=τ

₁

τ=τ

₂

τ=τ

_Η

D=D

PH2

Output

PH1

Horizontal

Photo-input

Fig. 6. The actual architecture of our basic cell. τ is the time constant and D is the diffusion constant.

Fig. 7. The transient simulated result of the CNN model. (a) The input signal, (b) the output of PH1, (c) the output of PH2.

Fig. 8. The simulated space-domain response of the CNN model. (a) The input signal, (b) output of the horizontal, (c) the output of PH1.

Fig. 9. Simulated transient responses of the model with different diffusibility of the horizontal applied. The time constant of PH2 is kept to be 0.25ms in the simulation.

Fig. 10. Simulated transient responses of the model with different time constant of PH2 applied. The space constant of the horizontal σ is kept to be 50 in the simulation.

(a)

(b)

Fig. 11. Simulated space-domain response of the model with different diffusibility of the horizontal applied.(a)Output of the horizontal, (b) output of PH1.

Horizontal PH2 Photo-Input PH1

V_bias1

V_bias2 V_bias4 V_bias3

VDD

V_sm

…..

VDD VDD VDD VDD VDD VDD VDD VDD

MN1 MN2

to left cell to upper

cell MSL

MSU

I

_PH1

I

_PH2

I

_In

Fig. 12. The circuit of a basic cell constructed from the architecture in Fig. 6.The arrows represent the four current outputs. IIn, IPH1, IH, and IPH2

are for the output of photo-input, PH1, horizontal, PH2 respectively.

Current Conveyor

i X

i Y

v Y

v X v ^Z

i Z

X Y

Z

(a)

Ｚ

ＹＸ

i

_ｚ

＝ i

V

_Ｙ

V

=V

(b)

Fig. 13. The current conveyor. (a) The block diagram of a current conveyor, (b) the schematic of an example of current conveyor.

I

_in

C

I

_{o u t}

M1 M2

VDD VDD

(a)

1/g

_{m 1}

C

v

_gs2

g

_{m 2}

v

_gs2

i

_in

i

_out

g

₁

/d

₁

/g

₂

s

₁

/s

₂

d

₂

(b)

Fig. 14. Current delay element. (a) The circuit of a current delay element, (b) the small signal equivalent circuit of the current delay element.

CELL CELL CELL CELL …….

CELL CELL

CELL

CELL CELL CELL CELL ……. CELL

Fig. 15. The architecture of the 2-D retinal array and other peripheral circuits.

………

Fig. 16. The schematic of the address decoder used in the chip.

Fig. 17. The output buffers of the chip. (a) Output buffer for PH1, PH2, and the horizontal, (b) output buffer for the photo-input.

Fig. 18. HSPICE simulated transient response in the condition that Vbias1=1.5V, Vbias2=0.6V, Vbias3=1.1V, Vbias4=430mV, Vsm=1.1V. The stimuli is a pulse with 100pA transient current added to an 100pA background induced DC current. A virtual background current of 1.2nA is supplied. The simulated results of (a) photo-input, (b) PH1, (c) PH2, (d) the horizontal.

Fig. 19. HSPICE simulated transient response in the condition that Vbias1=1.5V, Vbias2=0.6V, Vbias3=1.1V, Vbias4=0, Vsm=1.1V. The stimuli is a pulse with 100pA transient current added to an 100pA background induced DC current. No virtual background current is supplied. The simulated results of (a) photo-input, (b) PH1, (c) PH2, (d) the horizontal.

Fig.20. HSPICE simulated transient response in the condition that Vbias1=1.5V,

Vbias2=0.6V, Vbias3=1.1V, Vbias4=530mV, Vsm=1.1V. The stimuli is a pulse with 100pA transient current added to an 100pA background induced DC current. A virtual background current of 11.6nA is supplied. The simulated results of (a) photo-input, (b) PH1, (c) PH2, (d) the horizontal.

Fig. 21. Simulated time constant with different I_D_,MP₇ supplied.

Fig. 22. The steady-state space response when the middle six cells are turned on. (a)

Photo-input, (b) output of the horizontal, (b) output of PH1. The simulation is performed under the condition that Vbias1=1.5V, Vbias2=0.6V, Vbias3=1.1V, Vbias4=430mV. A virtual background current of 1.2nA is supplied.

(a) (b)

(c)

Fig. 23. HSPICE-simulated spatiotemporal patterns of the circuit. The time axis is

located on the bottom and the space axis is on the left. The right most color bar in (c) shows the mapping from normalized to assigned color. Space-time pattern of (a)PH1, (b) the horizontal, (c) PH2.

CHAPTER 3 LAYOUT DESCRIPTIONS AND EXPERIMENTAL RESULTS

In this chapter, the layouts and chip photographs of the retinal chip is described in section 3.1. The experimental environment is illustrated in section 3.2. Finally, the experimental results are shown in the last section to verify the functions of the chip.

3.1 LAYOUT DESCRIPTIONS

The retinal chip is designed and fabricated in 0.35µm double-poly triple-metal CMOS process. The architecture and circuit are described in detail in the previous chapter. The layout of the chip is shown in Fig. 24 and Fig. 25.

The retinal chip contains a sensory array of 32x32, address decoders, readout buffers, and other circuits. Fig. 24 shows the layout of whole retinal chip. The sensory array is in the center and the input and output pads are arranged in its peripheral. The row address decoder and the column address decoder are on the left side and the upper side of the sensory array respectively as labeled in the figure. Total area of the chip is around 2.6mm x 2.6mm. ESD (Electrostatic Discharge) protection circuits for input and output pads are included in the chip.

The layout of a basic cell in the chip is shown in Fig. 25. In each basic cell, there

are twenty-six NMOS devices, eleven PMOS devices, and a parasitic PNP

phototransistor, including eight NMOS switches. Each basic cell occupies an area of 73.3 µm x 73.3µm with a fill factor of 0.25. The region of the photo-BJT is labeled in the figure. It is surrounded by dashed line in the lower-left of the figure. Except the phototransistor region, the other region of the basic cell is covered by metal layers to prevent the incident light from affecting the other part of the circuit. The

phototransistor region which is not covered by metal layers is transparent and light could pass through easily. The region right adjacent to phototransistor is for the large capacitance MP10 as labeled in the lower right corner of the cell. The locations of other parts of the cell are shown in the figure as labeled.

A photograph of the whole chip and the basic cell are shown in Fig. 26 and Fig.

27 respectively. In Fig. 27, the region of four main part of a basic cell, photo-input, PH1, PH2, and the horizontal are labeled. Only the area of photo-BJT is not covered by metal layers and is transparent.

3.2 EXPERIMENTAL ENVIRONMENT

The diagram of the measurement setup is demonstrated in Fig. 28. The projector and the LED are used to project patterns on the chip. By controlling the function generator that drives the LED, we could control the flash frequency and luminance of the light. The projector is controlled by a computer, so as to project different patterns onto the chip. When measuring the transient response, the row and column addresses are kept the same to observe the response of a selected cell. Additional counter is needed to measure the spatial response or space-time patterns. The outputs of the chip are connected to the terminals of the oscilloscope for observation and recording.

3.3 EXPERIMENTAL RESULTS

Since most of the transistors in our chip operate in subthreshold region and the SPICE model for subthreshold circuit may not be accurate, the simulation results in the previous chapter might be different from the measurement results of the chip.

Therefore, it is necessary to compare the measurement results with SPICE simulation results. The measurement results are shown in the following.

First the measured transient response is shown. Fig. 29 shows the result of measurement when a periodic light source is incident on the chip. In the figure, overshooting and undershooting of PH1 could be observed clearly and so could that of the horizontal as simulated previously. To verify that the measured output of the chip is consistent with the HSPICE simulation results and the original CNN model, a normalized output of each part of the cell and simulation curves are plotted in the same drawing for comparison. The comparison is shown in Fig. 30. It could be found from the figure that the measured curves match the simulation curves qualitatively.

Fig. 31 shows the measurement of PH2 with different Vbias3. The measured time constant calculated from the rise and fall time of PH2 with different background illumination and Vbias3 is shown in Fig. 32. Since it is impossible to apply test signal directly to the input of PH2, the accurate time constant of PH2 is unavailable via chip measurement. The rise and fall time of the output of PH2 with supplying periodic signal are used to calculate the approximate time constant. The calculated results are shown in Fig. 32. As discuss in previous chapter, the time constant decreases with larger background illumination which could be evaluated roughly by IPH2. IPH2

represents the DC level of output of PH2. In the measurement, different background levels are produced by the aim of MN13 which adds an additional offset current to each cells input.

The spatial response is shown in Fig. 33. As mentioned in the previous

discussion, the smoothing or diffusion range could be tuned by controlling bias Vsm. A larger Vsm makes a wider smoothing range. The comparison of the measured spatial response and simulated response is shown in Fig. 34. The high contrast at the image edge is not as clear as that of simulation results, but the smoothing measurement result of horizontal closely match the simulation result.

Finally, the spatiotemporal patterns of the chip with subjected to periodic signal are shown in Fig. 35. The output of PH1 and horizontal are shown. The measured pattern is similar to HSPICE simulation results shown in the previous chapter. The output of PH1 spreads out laterally when the light is just incident. After some time, the output begins to contract in space gradually and reaches a steady-state value.

During the turn off transient, output current suddenly drops and returns to the steady state value. The pattern of horizontal appears similar .characteristic of spreading and contraction but the extent of spreading is wider. This spatiotemporal pattern is also similar to the biological measurement of real retinas shown in Fig. 36. The method of the biological measurement and recording are explained in [5].

Table II

Summary of characteristics of the proposed 32x32 retinal chip.

Process 0.35µm DPTM CMOS

Power Supply 3V

Resolution 32x32

Basic Cell Area 73.3µm x 73.3µm

Photo Sensor Area 40µm x 34µm

Number of Transistors in Basic Cell 39

Fill Factor 0.25

Total Area 2.6mm x 2.6 mm

Total Transistors of the Chip 41k

Power dissipation <30mW

VDDD CA1 VSSh VDDh CA2 CA2 CA2 CA2 VSSA

VSSD

VSSA BIAS1 BIAS2 BIAS3 BIAS4VDDh VSSh sm VP VB

VDDA

Fig. 24. Layout of the whole retinal chip. Main parts of the chip are labeled. Sensory array is in the center, the row and column address decoders are in the upper and left of the array respectively. Output buffer and address buffer are also labeled.

Photo-BJT

PH2 Horizontal

PH1

MP10

Fig. 25. Layout and floorplan of a basic cell of the sensory array. The four regions of main parts of the basic cell are surrounded by solid lines. The lower two sub region surrounded by dashed lines are phototransistor and PMOS capacitor MP10.

Column Decoder

Ro w De co de r

Output Buffer

Row Address Buffer

Column Address Buffer

Sensory Array

Fig. 26. Photograph of the whole retinal chip. Main parts of the chip are labeled.

在文檔中由哺乳類動物視網膜生物模型所設計之金氧半仿視網膜晶片 (頁 27-0)