JP2002071466A - Semiconductor device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow the signal amplification and signal process to be smoothly performed by stably correcting a bias component at a low noise. SOLUTION: There are provided a first bias circuit 120 which is connected to a measurement resister (bolometer 104) and provided with a reference resistance (reference resistance circuit 102) whose resistance temperature coefficient is equal to a measurement resistance (bolometer 104), and a second bias circuit 121 which is connected to a cancel resistance 106 and provided with a reference resistance (reference resistance circuit 112) whose resistance temperature coefficient is equal to the cancel resistance 106. The first bias circuit 120 acts as a means for applying its output voltage to the measurement resistances (bolometer 104) while the second bias circuit 121 acts as a means for applying its output voltage to the cancel resistance.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor device, and more particularly to a semiconductor device provided with an integrating circuit comprising a resistor and a capacitor for storing a current flowing through the resistor. The present invention relates to a semiconductor device such as an infrared sensor, a microwave / millimeter wave detector, a temperature sensor, a magnetic sensor, a pressure sensor, a gas sensor, a flow sensor, and the like for detecting a change in, for example, microwave, temperature, magnetism, and pressure. In the following, the description will focus on a thermal infrared imaging apparatus, but the present invention is applicable to all semiconductor devices that integrate and detect weak signal components.

[0002]

2. Description of the Related Art As a conventional thermal infrared imaging apparatus, for example, there is one described in Japanese Patent Application No. 9-315455.

FIG. 14 is a sectional view showing a conventional thermal infrared imaging device described in Japanese Patent Application No. 9-315455. As shown in the figure, a scanning circuit 1402 is formed on a semiconductor substrate 1401, and a plurality of light receiving units 1400 for converting incident infrared rays into electric signals are provided thereon.

That is, one light receiving section 1400 constitutes one pixel of the image pickup device, and it is necessary to integrate a plurality of pixels two-dimensionally in order to obtain a two-dimensional infrared image. is there. In addition, a scanning circuit 14 for reading out captured image data is provided below each light receiving unit 1400.
02 are similarly two-dimensionally integrated.

The light receiving section 1400 has a film-like structure called a diaphragm.
In 03, a hollow portion 1404 is formed in the lower part. This hollow portion 1404 is formed by removing a sacrificial layer that has been formed in advance by etching.

On the other hand, on the surface of the diaphragm 1403,
An infrared absorption layer 1405 for absorbing infrared light is formed, and a thermoelectric conversion element 1406 for converting heat generated by receiving infrared light into an electric signal is formed at a boundary between the diaphragm 1403 and the hollow portion 1404. As the thermoelectric conversion element 1406, in this example, a bolometer whose electric resistance value changes with temperature is used.
For example, titanium is used as the bolometer.

The operation of the thermal infrared imaging apparatus shown in FIG. 14 is briefly described as follows.
First, the infrared light that has entered the light receiving section 1400 of each pixel is absorbed by the infrared absorption layer 1405 and raises the temperature of the diaphragm 1403 of each pixel. At this time, the temperature rise is converted into an electric signal by a thermoelectric conversion element 1406 (bolometer), and is sequentially read out to the outside through a scanning circuit 1402.

Here, the details of the thermal infrared imaging apparatus shown in FIG. 14 will be described. FIG. 15 shows a readout circuit used in the thermal infrared imaging device of FIG. As shown in the figure, the readout circuit 1500 includes a plurality of bolometers 1501, a pixel switch 1509 connected between each bolometer 1501 and the ground,
PN transistor 1502 and cancel resistor 1503
, A PNP transistor 1504 and an integrating capacitor 1
505, a sample and hold circuit 1506, an FPN correction constant current source 1507, and a reset switch 1508 which is opened and closed by input of a reset signal ΦR.

The sample-and-hold circuit 1506 includes preceding NMOS transistors 1510 and 1511, a switch 1512 that is opened and closed according to a sample-and-hold pulse φS / H input from the outside, a hold capacitor 1513, and a subsequent NMOS transistor 1514. 1515.

Here, the bolometer 1501 detects heat generated by incident infrared rays and converts it into an electric signal as described with reference to FIG. For example, the NPN transistor 150
When a voltage is applied to V _b1 to second base, the base of the NPN transistor 1502 - to-emitter voltage and V _BE, the voltage of V _b1 -V _BE applied to bolometer 1501. Then, assuming that the resistance of the bolometer 1501 is R _b1 , the collector of the NPN transistor 1502 has I _c1 =
A current of (V _b1 −V _BE ) / R _b1 flows.

On the other hand, when the voltage V _b2 is applied to the base of the PNP transistor 1504, the cancel resistor 1503 is connected to the collector of the PNP transistor 1504 in the same manner as described above.
The resistor as _{R b2 I c2 = (V b2} -V BE) / current R _b2 flows.

Then, I _c1 and I _c2 are almost balanced but slightly different, so that ΔI = I _c1 −I _c2, which is a slight difference between them, flows through integration capacitor 1505. That is, the difference ΔI is the sum of the signal component and the bias component that has not been completely removed, and most of the bias component has been removed.

As described above, when infrared rays enter from the outside, the temperature of the thermally separated diaphragm 1403 (FIG. 14) rises, and the resistance of the bolometer 1501 changes.
This change in the resistance value changes I _c1 and the difference ΔI
Is stored in the integration capacitor 1505.

By the way, the bias component that cannot be removed is
This is mainly caused by the variation of a plurality of bolometers 1501 that are sequentially selected. Cancel resistor 1
R _b2 is fixed because a single 503 is used, but since a plurality of bolometers 1501 are provided,
If there is a large variation among these many R _b1 , the difference Δ
I will also vary. Therefore, in order to correct such variations, conventionally, an FPN correction constant current source 1507 shown in FIG. 15 is further provided.

The FPN correction constant current source 1507 is composed of a plurality of stages of constant current sources (not shown), and the current values of the constant current sources include I ₀ , 2 · I ₀ , 4 · I ₀ ,. As shown in the above, weighting of an integer power of 2 is applied. Thus, in accordance with the variation of R _b1, by selecting a desired one of these constant current sources appropriate, than reduce the variation of the difference ΔI due to variations in R _b1.

The signal corrected in this manner is stored in an integration capacitor 1505, and then converted from high impedance to low impedance by a source follower in a sample and hold circuit 1506. The signal sampled in time series is After being temporarily held by the hold capacitor 1513, it is output as an output S / H _out . This output S / H _out is _supplied to the scanning circuit 1 shown in FIG.
402.

As described above, in the conventional thermal infrared imaging apparatus, it has been a problem how to improve the temperature characteristic of the integrating circuit including the integrating capacitor 1505 and the like. The following circuit has already been proposed as a circuit for improving the temperature characteristic of the integrating circuit.

For example, there is one disclosed in Japanese Patent Application Laid-Open No. 2-260914. In an integrating circuit composed of a capacitor and a diffusion resistor, another diffusion resistor for compensating the temperature dependence of the leakage current of the diffusion resistor is used. In addition, the leakage current flowing through it is added to the capacitor.

Japanese Unexamined Patent Publication (Kokai) No. 3-103711 discloses that the current value of the constant current source is changed depending on the temperature so that the output voltage of the magnetic sensor does not change depending on the temperature. Also, JP-A-8-32026
Japanese Patent Application Laid-Open No. 6-64106 describes that a constant current circuit having a constant current characteristic with a temperature coefficient of zero is used, and the current flowing through the piezoresistive element is made constant regardless of the temperature.

Further, Japanese Patent Application Laid-Open No. 8-334413 discloses that a temperature compensating resistor having the same material and structure as the bolometer is provided in addition to the bolometer of each pixel to offset the temperature fluctuation of the output offset voltage. I have.

[0021]

However, in the thermal infrared imaging apparatus described in Japanese Patent Application No. 9-315455, the bolometer itself generates heat by applying a voltage to the bolometer. There is a problem that the integrated waveform is bent and the gain of the integration circuit cannot be increased. That is, even if the amplitude of the integrated waveform is reduced by the cancel operation of I _c1 −I _c2 and the FPN correction circuit, the bending of the integrated waveform occupies the dynamic range of the circuit.

Further, in the thermal infrared imaging device disclosed in Japanese Patent Application No. 9-315455, the setting of the bias voltage, such as V _b1, V _b2 to determine the canceling operation of the bias current is subtle, each imaging device There is a problem that the adjustment for each is complicated. Setting of cancel operation of I _c1 -I _c2 and F
If the PN correction constant current source is not set properly, the amplitude of the integrated waveform will fall outside the dynamic range of the circuit. In addition, conventionally, there has been a problem that when the temperature of the device changes, the DC level of the pixel changes.

Accordingly, it is an object of the present invention to stably correct a bias component with low noise so that signal amplification and signal processing can be performed smoothly.

[0024]

In order to achieve such an object, a semiconductor device according to the present invention according to claim 1 includes one measuring resistor for converting a physical quantity into a resistance value, and one measuring resistor for converting the physical quantity into a resistance value. And a canceling resistor connected to the measuring resistor for applying a bias current to the measuring resistor and integrating and accumulating the current flowing through the measuring resistor; and a canceling resistor connected to the measuring resistor and canceling the bias current flowing through the measuring resistor. A readout circuit for detecting a change in the resistance value of the measurement resistor based on the current accumulated in the integration circuit and indirectly measuring a physical quantity, wherein the semiconductor device is connected to the measurement resistor and A first bias circuit having a reference resistance having a temperature coefficient of resistance equal to the measurement resistance; and a reference resistance connected to the cancellation resistance and having a temperature coefficient of resistance equal to the cancellation resistance. A second bias circuit, wherein the first bias circuit is means for applying the output voltage to a measuring resistor, and the second bias circuit applies the output voltage to a cancel resistor. Means.

According to the present invention, by applying a voltage to a measurement resistor such as a bolometer, it is possible to eliminate the effect of the measurement resistor itself generating heat. As a result, the dynamic range of the signal processing circuit including the integrating circuit has a margin, and the gain of the circuit can be increased. In particular, by increasing the gain of the first-stage integration circuit, input conversion noise can be reduced, and S / N can be greatly improved. Further, a function of digitally setting a bias while maintaining characteristics of low drift and low noise can be provided. The configuration is simple, and a high-performance semiconductor device can be realized at low cost.

In the above invention, the integrating circuit is constituted by the first and second bipolar transistors having collectors connected to each other, and an integrating capacitor connected to the collector, and the first bipolar transistor has a canceller at its emitter. A resistor is connected and a second bias circuit is connected to its base, and the second bipolar transistor has its emitter connected to the measurement resistor and its base connected to the first bias circuit. .

The first bias circuit has a temperature-compensated constant current circuit, a reference resistor connected to the constant current circuit and having a temperature coefficient of resistance equal to the measurement resistance, and a first resistor connected to the reference resistor and connected to the reference resistor. And a filter for removing noise from the voltage generated in the resistor. The second bias circuit includes a temperature-compensated constant current circuit, a reference resistor connected to the constant current circuit and having a temperature coefficient of resistance equal to the cancel resistor, and a second resistor connected to the reference resistor and generated in the reference resistor. And a filter for removing noise from the applied voltage.

In the above invention, the semiconductor device is any one of an infrared sensor, a microwave / millimeter wave detector, a temperature sensor, a magnetic sensor, a pressure sensor, a gas sensor and a flow sensor.

[0029]

Next, an embodiment of the present invention will be described with reference to the drawings. [First Embodiment] FIG. 1 is a block diagram of a thermal infrared imaging apparatus showing a first embodiment of the present invention.

As shown in the figure, the thermal infrared imaging apparatus according to the present embodiment has a plurality of readout circuits 117, a bolometer bias circuit 120, a cancel resistor bias circuit 121, and an FPN correction circuit. , A multiplexer 110, a horizontal shift register 118, and a vertical shift register 119.

The bias circuit 120 includes a constant current circuit 101
And a reference resistance circuit 102 and a filter 103. Similarly, the bias circuit 121 includes the constant current circuit 11
1, a reference resistance circuit 112, and a filter 113. The bias circuit 122 includes a constant current circuit 114, a reference resistance circuit 115, and a filter 116.

The read circuit 117 includes the bias circuit 12
0 and 121, an integrating circuit 105, a cancel resistor 106 connected between the integrating circuit 105 and the power supply, and n (n: an arbitrary natural number) bolometers 104 connected to the integrating circuit 105. , This bolometer 10
N pixel switches 107 composed of NMOS transistors connected between
5 and a FPN correction circuit 108 connected to the output of the integration circuit 105 and connected to the bias circuit 122.

As in the conventional example shown in FIGS. 14 and 15, the bolometer 104 is formed on a diaphragm formed for each pixel and is a means for converting a change in incident infrared light into a change in current. The pixel switch 107 is an NMOS transistor formed below the diaphragm, and is a unit for determining selection / non-selection of the bolometer 104.

The cancel resistor 106 is connected to the integrating circuit 105
Is connected to the bolometer 104 through the
4 is a means for canceling the bias component of the current flowing through 4. The FPN correction circuit 108 is means for reducing the variation in the current flowing through the bolometer 104.

The integration circuit 105 is means for integrating the current flowing through the bolometer 104, the cancel resistor 106, and the FPN correction circuit 108. Sample hold circuit 109
Is means for sampling and holding the output of the integration circuit 105.

Now, the bolometer 104 and the pixel switch 1
07, a plurality of pixels are arranged one-dimensionally or two-dimensionally on the semiconductor substrate 1401 shown in FIG. For example, a plurality of such pixels are formed in the y direction,
Each pixel is connected to one integration circuit 105. The canceling resistor 106, the FPN correction circuit 108, and the sample and hold circuit 109 are connected to the integration circuit 105 to form one readout circuit 117.

At this time, in the case of a one-dimensional thermal infrared imaging apparatus, one read circuit 117 may be provided. On the other hand, in the case of a two-dimensional thermal infrared imaging apparatus, a plurality of readout circuits 117 thus formed are arranged in the x direction. Further, by providing the multiplexer 110 close to the readout circuit 117, the signal of each readout circuit 117 can be appropriately switched.

Therefore, the horizontal shift register 118
By controlling the multiplexer 110, one of the plurality of readout circuits 117 can be selected. On the other hand, the vertical shift register 119 is used in both the one-dimensional and two-dimensional cases, and selects one of the bolometers 104 connected to each integration circuit 105 by controlling the pixel switch 107.

Here, the operation of the present embodiment will be described. First, the bias circuit 120 applies a bias to the bolometer 104 via the integration circuit 105. The temperature-compensated constant current circuit 101 generates a constant current having a small temperature dependency, and the constant current is supplied to the reference resistance circuit 102. Then, the reference resistance circuit 102 is connected to the bolometer 10.
Resistance R having the same temperature coefficient of resistance (hereinafter referred to as TCR)
Has a _1, by receiving the constant current I ₁ which is temperature-compensated, and outputs a voltage V ₁ of the same temperature coefficient as the TCR of the bolometer 104.

[0040] This is from the _{V 1 = I 1 × R 1} , by reducing the temperature dependence of I _1, V ₁ is by having the same temperature coefficient as R _1. And this reference resistance circuit 1
The output voltages V ₁ to 02, applied to the bolometer 104 connected to the integration circuit 105 through the filter 103. Then, the current I ₁₁ having small temperature dependence flows through the bolometer 104. If this is the resistance value of the bolometer 104 and _{R 11, I 11 = V 1} / R 11 = I 1 × R 1 / R 11
Next, it is because it is the same the TCR of R ₁ and R _11.

Next, the bias circuit 121 includes the integrating circuit 1
A bias is applied to the cancel resistor 106 via the line 05. The temperature-compensated constant current circuit 111 supplies a constant current having low temperature dependency to the reference resistance circuit 112. This reference resistor circuit 112 has the same TC as the cancel resistor 106.
It has a resistance of R, receives a temperature-compensated constant current, and outputs a voltage having the same temperature coefficient as the TCR of the cancel resistor 106. Then, the voltage output from the reference resistor circuit 112 is applied to the cancel resistor 106 via the filter 113. As a result, a current having a small temperature dependence flows through the cancel resistor 106.

Next, the bias circuit 122 includes the FP
A bias is applied to the N correction circuit 108. The temperature-compensated constant current circuit 114 outputs a constant current having small temperature dependency. This constant current is supplied to the reference resistance circuit 115. The reference resistance circuit 115 has the same TCR resistance as the correction resistance of the FPN correction circuit 108, and receives a temperature-compensated constant current and outputs a voltage having the same temperature coefficient as the TCR of the correction resistance of the FPN correction circuit 108. And the reference resistance circuit 1
15 to the FPN through the filter 116
The voltage is applied to the correction resistor of the correction circuit 108. by this,
The FPN correction circuit 108 outputs a current having a small temperature dependency.

Although FIG. 1 shows the temperature-compensated constant current circuits 101, 111, and 114 as separate circuits, one temperature-compensated constant current circuit will be described later. Alternatively, three constant currents may be generated through a current mirror circuit.

Next, the bias circuits 120, 1 shown in FIG.
The detailed configurations of the readout circuits 21 and 122 and the readout circuit 117 will be described.

FIGS. 2, 3, and 4 show specific circuit diagrams of the present embodiment shown in FIG. The circuit elements shown in these figures are preferably formed on the same semiconductor substrate with good temperature matching between the elements, but use discrete components as adjustment components such as variable resistors and large-capacity capacitors as appropriate. Is possible. Further, depending on the required specification of the temperature drift, the temperature of the semiconductor substrate may be kept constant by using a Peltier element or the like.

FIG. 2 is a specific circuit diagram of the thermal infrared imaging device shown in FIG. As shown in the drawing, the thermal infrared imaging apparatus according to the present embodiment includes a plurality of readout circuits 201.
, A bolometer bias circuit 212, a cancel resistor bias circuit 218, an FPN correction circuit bias circuit 224, a multiplexer 231, a horizontal shift register 232, and a vertical shift register 233.

The bias circuit 212 includes a constant current circuit 213
, NPN transistors 214 and 215 and resistor 2
16 and a filter 217. Similarly,
The bias circuit 218 includes a constant current circuit 219, PNP transistors 220 and 221, a resistor 222, and a filter 223. Similarly, the bias circuit 224 includes a constant current circuit 225 and an NPN transistor 2
26 and 227, a resistor 228, and a filter 229.

The read circuit 201 comprises a resistor 207 and a P
NP transistor 208 and NPN transistor 204
, Bolometer 202, pixel switch 203, FP
It comprises an N correction circuit 236, a reset switch 206, an integration capacitor 205, and a sample and hold circuit 230.

In the read circuit 201 thus configured, the NPN transistor 204 is
As in the conventional examples shown in FIGS. 4 and 15, a plurality of (here, n (n: an arbitrary natural number)) bolometers 202 are connected to the emitter, and an integrating capacitor 205 is connected to the collector. That is, a change in resistance of the bolometer 202 is converted into a change in current and stored in the integration capacitor 205.

The pixel switch 203 includes a bolometer 202
NPN transistor 2 connected between
One or several of the plurality of bolometers 202 connected to 04 are selected. PNP transistor 208
Is connected to the cancel resistor 207 to the emitter and the integrating capacitor 205 to the collector,
The bias component is removed from the current flowing through the bolometer 202.

The FPN correction circuit 236 includes a plurality of NPN transistors 211 whose gates are connected to each other,
A resistor 209 connected to the emitter of the transistor 211
And the NMOS transistor 2 connected to each resistor 209
10 and FPN data buffers 235 and 234 connected to the bases of all NMOS transistors 210.

As described above, one NPN transistor 2
11 and one FPN correction resistor 209 and one NMOS transistor 210 constitute a set of constant current sources, and a bolometer is provided by arranging a plurality of constant current sources having different current value settings. The variation in the resistance value of the resistor 202 can be reduced. This variation in the resistance value of the bolometer 202 is generally called fixed pattern noise, and is hereinafter referred to as FPN.

The current values of the plurality of constant current sources can be set digitally. In other words, the combined resistance can be varied by selecting and turning on / off a desired one of the plurality of NMOS transistors 210 according to the variation in the resistance of the bolometer 202. Variation can be reduced. As a result, most bias components and inter-pixel variations can be removed from the integrated current.

The components remaining at that time include the signal component, the bias that could not be removed, and the variation component between pixels, and the remaining component is applied to the integrating capacitor 205 for a certain period of time. Is accumulated.

The sample-and-hold circuit 230 stores the current in the integration capacitor 20 after accumulating the current for a certain period of time.
This is for sampling the voltage of No. 5 and temporarily holding it. Therefore, the integration circuit can start the next integration before the signal reading is completed, and the integration time can be extended. The longer the integration time is, the narrower the frequency band of the noise is, so that the noise can be reduced.

The reset switch 206 is connected to the integration capacitor 205 and is provided for resetting the integration capacitor 205 to a constant voltage after sampling. The reset is performed by externally setting a reset pulse ΦR below.

By the way, as shown in FIG. 1, the pixel composed of the bolometer 202 and the NMOS transistor 203 is one-dimensional or two-dimensional on the semiconductor substrate shown in FIG.
A plurality of them are arranged in a dimension. For example, a plurality of these pixels are formed in the y direction and connected to one NPN transistor 204. In the case of a one-dimensional thermal infrared imaging device, only one readout circuit 201 may be used. In the case of a two-dimensional thermal infrared imaging device, a plurality of the readout circuits 201 are formed in the x direction. A multiplexer 231 for switching signals is further arranged.

The horizontal shift register 232 controls the multiplexer 231 so that a plurality of read circuits 201
Of the vertical shift register 23
3 controls the pixel switch 203 and selects one or several bolometers 202 connected to the NPN transistor 204.

[0059] Further, in the bias circuit 212, a constant current circuit 213 which is temperature compensated outputs a constant current I ₁ Temperature dependency is small. NPN transistor 214, the constant current I ₁ is applied to the collector, the emitter resistor 216 having the same TCR as the bolometer 202 is connected. Another NPN transistor 21 is connected to the base and collector of NPN transistor 214.
5 are connected to each other, and the NPN
The collector of the transistor 215 is connected to a power supply.

Note that the NPN transistor 214 is connected to the NPN transistor 20 connected to the bolometer 202.
It must have the same structure and dimensions as 4. Also, NP
The base of the N transistor 214 is connected to the NPN transistor 204 of each readout circuit 201 via the filter 217.
Connected to the base.

Therefore, the NPN transistor 214
The base has a temperature-compensated constant current I ₁Receiving, V₁= I
₁× R₁+ V_BE11Output voltage. Where R₁Is
The resistance value of the anti-216_BE11Is an NPN transistor
Reference numeral 214 denotes a base-emitter voltage. V₁NPN
When applied to the base of the transistor 204,
Data 202₁-V_BE12Voltage is applied. Also, V
_BE12Is the base-emitter of the NPN transistor 204
Is the voltage between them. And the NPN transistor 204 and
And 214 have the same structure and dimensions,
_BE11And V_BE12Are almost equal.

As a result, the bolometer 202 has I ₁
× R ₁ is applied, and the resistance of the bolometer 202 is changed to R
_{As 11} , a current of I ₁₁ = I ₁ × R ₁ / R ₁₁ flows. I ₁
, And the TCR of R ₁ and R ₁₁ is the same, so that the current I ₁₁ having small temperature dependency flows through the bolometer 202.

Incidentally, this circuit operates even if the NPN transistor 215 is omitted from the configuration of FIG. 2 and the base and collector of the NPN transistor 214 are connected. However, the NPN transistor 215 supplies the majority of the base current of the NPN transistor 214 via the power supply, so that the current I of the temperature-compensated constant current circuit 213 is increased.
Most of ₁ flows into the collector of the NPN transistor 214. As a result, N which gives the same base voltage
The collector current of the PN transistor 204 is precisely I ₁
Thus, the temperature dependence of the collector current of the NPN transistor 204 can be improved.

That is, it can be said that the addition of the NPN transistor 215 has the effect of reducing the effect of the 1 / f noise existing in the NPN transistor 214, and this point has already been confirmed by simulation by the inventor. Also, the NPN transistor 215
Has the effect of lowering the impedance of the base of the NPN transistor 214.

The bolometer 202 is connected to the resistor 216.
The same material and the same structure can be used.
However, since the current is always applied unlike the bolometer in the use condition, if the bolometer is formed on the thermally separated diaphragm, the bolometer itself will generate excessive heat, and in the worst case, it will burn out. Therefore, even if the sacrifice layer is formed on the sacrifice layer, the sacrifice layer may not be etched in this portion.

Further, when the resistance 216 and the bolometer 202 have the same resistance value, I ₁ and I ₁₁ become equal, and
The _BE11 and V _BE12 can be equally well precision. However, the bolometer 202, for example, in the case of the two-dimensional sensor, x, because of the in-plane variation in the y-direction, R ₁ and R ₁₁
It is difficult to make Therefore, in such a case, the resistance 21 is set so that the two are equal as much as possible.
By arranging 6 near the center of the plurality of bolometers 202 arranged in the x direction or near the center in the y direction, favorable results can be obtained.

Further, if a resistor having a plurality of connected bolometers 202 is used as the resistor 216 in order to make R ₁ and R ₁₁ equal, the difference between R ₁ and R ₁₁ can be reduced by averaging. For example, if three bolometers 202 are connected in parallel and three bolometers 202 are connected in series, the bolometer 2
02, and the same resistance value as that of
The difference between the ₁ and R ₁₁ is also relaxed.

Further, as described above, the plurality of bolometers 202
The effect of reducing 1 / f noise can also be obtained by using a combination of. This is because the 1 / f noise is generally inversely proportional to the square root of the total number of carriers. For example, when 9 bolometers are connected, 1 / f noise is 1 / f noise.
f noise can be reduced to 1/3.

It should be noted that even if the resistor 216 and the bolometer 202 cannot have the same resistance value, the NPN transistor 2
By properly setting the dimensions of 04 and 214,
The same effect as above can be obtained. For example, the resistor 21
If 6 is the resistance of a multiple of the bolometer 202, the emitter area of the NPN transistor 214 leading to resistor 216 at this time, V _BE11 and V by the 1 / a times the emitter area of the NPN transistor 204 connected to the bolometer 202 _BE12 can be precisely equalized.

[0070] This is due to the following reason, the base current I _B and the base - the relationship between the emitter voltage V _BE, the emitter areas A _E, the proportionality factor I _B0, the elementary charge q, Boltzmann's constant k, T is the absolute temperature,

[0071] _{I B = A E · I B0} · Exp [qV BE / k / T] ···· (1)

Is obtained. By increasing the resistance by a times, _IB becomes 1
At this time, the same V _BE can be obtained by making A _E 1 / a times.

Similarly, in the bias circuit 218, the temperature-compensated constant current circuit 219 outputs a constant current I ₂ having a small temperature dependency. The PNP transistor 220
This constant current output is applied to the collector, and the emitter is connected to the resistor 222 having the same TCR as the cancel resistor 207. There is another P on the base and collector.
The emitter and base of the NP transistor 221 are connected to each other, and the collector of the PNP transistor 221 is connected to the ground.

The PNP transistor 220 is connected to the PNP transistor 208 connected to the cancel resistor 207.
The same structure and dimensions as above. PNP transistor 22
The base of 0 is connected to the base of the PNP transistor 208 of each readout circuit 201 via the filter 223.

The base of the PNP transistor 220 is
Current I compensated for_TwoReceiving, V _Two= I_Two× R_Two+ V
_BE21Output voltage. Where R_TwoIs the resistance of the resistor 222.
Resistance value, V_BE21Is the base of the PNP transistor 220.
This is the voltage between the emitter and the emitter. V_TwoIs a PNP transistor
208, the cancel resistor 20
7 has V_Two-V_BE22Voltage is applied. V_BE22Is PNP
This is a base-emitter voltage of the transistor 208.
PNP transistors 208 and 220 have the same structure and
And dimensions, V_BE21And V_BE22Are almost equal.

Therefore, the cancel resistor 207 has:
A voltage of I ₂ × R ₂ is applied, and a current of I ₂₁ = I ₂ × R ₂ / R ₂₁ flows with the resistance value of the cancel resistor 207 as R ₂₁ . Reduce the temperature dependence of I ₂ and reduce the TCR of R ₂ and R ₂₁
, The current I ₁₂ having small temperature dependence flows through the cancel resistor 207.

The PNP transistor 221 has the same effect as the NPN transistor 215. It is preferable that the resistor 222 and the cancel resistor 207 are made of the same material and have the same structure and have the same resistance. However, the resistor 222 is a times the resistance of the cancel resistor 207 and the emitter area of the PNP transistor 220 is PNP. The emitter area may be 1 / a times the transistor 208.

[0078] More Similarly, in the bias circuit 224, a constant current circuit 225 which is temperature compensated outputs a constant current I ₃ small temperature dependency. NPN transistor 22
6 has a collector connected to the constant current output, and has a resistor 22 of the same TCR as one FPN correction resistor 209 connected to the emitter.
8 is connected. Another N on its base and collector
The emitter and base of the PN transistor 227 are connected to each other, and the collector of the NPN transistor 227 is connected to the power supply.

The NPN transistor 226 is connected to the previous FPN.
The structure and dimensions are the same as those of the NPN transistor 211 connected to the correction resistor 209. NPN transistor 226
Is based on each readout circuit 2 via a filter 229.
01 is connected to the base of the NPN transistor 211.

The base of NPN transistor 226 receives temperature-compensated constant current I _3, and receives V ₃ = I ₃ × R ₃ +
_Outputs the voltage of V _BE31 . Here, R ₃ is the resistance value of the resistor 228, and V _BE31 is the base of the NPN transistor 226 −
This is the voltage between the emitters. The V ₃ NPN transistor 21
The voltage of V ₃ −V _BE32 is applied to the FPN correction resistor 209 by applying the voltage to the base of No. 1. V _BE32 is a base-emitter voltage of the NPN transistor 226. NP
Since N transistors 211 and 226 have the same structure and dimensions, V _BE31 and V _BE32 are substantially equal.

Therefore, the FPN correction resistor 209 includes:
A voltage of I ₃ × R ₃ is applied, and a current of I ₃₁ = I ₃ × R ₃ / R ₃₁ flows with the resistance value of the FPN correction resistor 209 as R ₃₁ . Reduce the temperature dependence of I ₃ and reduce the TCR of R ₁ and R ₃₁
Are the same, a current I ₃₁ having small temperature dependence flows through the FPN correction resistor 209.

The NPN transistor 227 has the same effect as the NPN transistor 215 in the bias circuit 212. Also in the case of the resistor 228 and the FPN correction resistor 209, the resistance 228 is set to a times the resistance value of the FPN correction resistor 209,
By making the emitter area of N transistor 226 1 / a times the emitter area of NPN transistor 211, V
_BE31 and _VBE32 can be made equal.

As a specific example of the temperature-compensated constant current circuit 219, the following may be used.

FIG. 3 is a circuit diagram showing a specific circuit of the constant current circuit. For example, as shown in FIG. 3A, the output of the temperature-compensated constant voltage circuit 301 is connected to the input of the non-inverting amplifier 302, and the output of the non-inverting amplifier 302 is connected to the NPN.
The transistor 303 is connected to the base and the emitter thereof is connected to the diffusion resistor 303a. Non-inverting amplifier 302
a is an operational amplifier 302a and resistors 302a and 302c.
It is composed of

As the temperature-compensated constant voltage circuit 301, a band gap reference circuit shown in FIG. 4 is preferably used.

FIG. 4 is a circuit diagram showing a band gap reference circuit. As shown in the figure, the bandgap reference circuit 400 includes a constant current source 402, an NPN transistor 401 whose base is connected to the constant current source 402, a resistor 403 commonly connected to the emitter of the NPN transistor 401, and 405 and the resistor 40
NPN transistors 404 and 406 connected to NPN transistors 3 and 405, respectively, and NPN transistor 406
And a resistor 405 connected to the base and a constant current source 402 connected to the collector.
Connected to the ground and the ground connected to the emitter
And an N transistor 408.

This band gap reference circuit 40
At 0, the emitter currents I _E1 and I _E2 of the NPN transistors 404 and 406 are:

I _E2 · R ₁ = (kT / q) ln (A ₁ · I _E1 / A ₂ / I _E2 ) (2)

There is the following relationship. Where A ₁ and A ₂ are NPN
A value determined by the emitter area of the transistors 404 and 406, k is Boltzmann's constant, T is absolute temperature, and q is elementary charge.

The output voltage V ₄₁ is determined by the base-emitter voltage V _BE3 of the NPN transistor 408 and the resistance 405
(R ₃ ) With the voltage at both ends of

V ₄₁ = V _BE3 + (R ₂ / R ₁ ) · (kT / q) ln (A ₁ · I _E1 / A ₂ / I _E2 ) (3)

Is represented by The equation (3) first and second terms of the right side of has a temperature coefficient opposite to each other, can be made very small temperature coefficient of V ₄₁ by selecting the circuit constants. Usually, such a bandgap reference circuit can easily obtain a temperature coefficient of about 0.01 [% / ° C.] at an output voltage of about 1.35 [V].

The output of the temperature-compensated constant voltage circuit 301 shown in FIG. 3A is output from the NPN transistor 30 connected to the subsequent stage by the operation of the non-inverting amplifier 302.
3 and the diffusion resistor 303a converts it into a required voltage.
Gain G of the non-inverting amplifier 302 is determined by R ₃₁ and R _32, a _{G = 1 + R 32 / R} 31. Therefore, R ₃₁
By making the temperature coefficients of R and R ₃₂ the same, the temperature coefficient of the gain G can be reduced. Preferably, R ₃₁ and R ₃₂ have the same temperature, and may be formed on the same semiconductor substrate, or may be configured using a collective resistor, a potentiometer, a variable resistor, or the like.

The voltage V ₃₂ output from the non-inverting amplifier 302 is applied to the NPN transistor 303 and the diffusion resistor 30.
The current I ₃₁ is converted into a current I ₃₁ having a small temperature dependence by 3a. Assuming that the base-emitter voltage of the NPN transistor 303 is V _BE and the resistance value of the diffusion resistor 303a is R ₃₃ ,

I ₃₁ = (V ₃₂ −V _BE ) / R ₃₃ (4)

## EQU10 ## In a general silicon IC, the temperature coefficient β of V _BE is about −2 [mV / ° C.], and the temperature coefficient γ of R ₃₃ is
It is, P ^- when using (P-type low concentration) diffusion resistance, 0.
Since it is about 2 [% / ° C.], the temperature coefficient of I ₃₁ can be made substantially zero by appropriately selecting the circuit constant.

Here, the condition for making the temperature coefficient of I ₃₁ almost zero is that the temperature coefficient of V ₃₂ is δ [mV / ° C.]

R ₃₃ = (δ−β) / γ / I ₃₁ (5)

## EQU10 ## For example, 200 [μA] as I ₃₁
If you think that δ is small and negligible,
If R _{33 is set} to about 5 [kΩ], the temperature coefficient of I ₃₁ can be made almost zero. From equation (4), V _{32 at} this time is 1 when V _BE is 0.7 [V]. .7 [V].

Therefore, the temperature-compensated constant voltage circuit 3
When the bandgap reference circuit 400 shown in FIG. 4 is used as 01, the gain of the non-inverting amplifier 302 is 1.7 / 1.35 = 1.26. Depending on the circuit constant, this gain is 1, and the non-inverting amplifier 302 can be omitted. Further, the non-inverting amplifier 302 can be appropriately changed to another means for converting a voltage.

As can be seen from equation (5), even if the temperature coefficient δ of the temperature-compensated constant voltage circuit 301 is not small, the temperature coefficient of I ₃₁ can be reduced by appropriately selecting the circuit constant. . However, when the circuit constant is fixed in a mass-produced product or the like, since the variation of δ leads to the variation of I ₃₁ , the smaller the absolute value of δ, the better.

The configuration in which the diffused resistor 303a is connected to the emitter of the NPN transistor 303 has the effect of reducing the effects of shot noise of the NPN transistor, Johnson noise of the base resistor (r _bb ), and circuit noise connected to the base.

FIG. 5 is a graph showing the current noise flowing through the collector when the diffusion resistance 303a is changed. In the figure, R is Johnson noise of the diffusion resistor 303a, I _C is shot noise of the collector current, and I _B
Represents shot noise of the base current, r _bb represents Johnson noise of the base resistor, and the total represents the sum of these noises. Here, the collector current is 100 [μ
A] and a base resistance of 100 [Ω], which is usually due to the fact that the current flowing through the bolometer is about 100 [μA].

As is clear from the figure, the total noise is reduced by increasing the emitter resistance.
Settles at a certain lower limit of noise current. Emitter resistance is 1
The total noise starts to be reduced by setting it to 00 [Ω] or more. However, by setting it to about 3 [KΩ] or more, the noise current can be suppressed to about twice the lower limit of the noise current. Collector current of 100
In the case of [μA], by setting the emitter resistance to 50 [KΩ] or less, the voltage across the emitter resistance becomes 5 [V] or less, and can be handled by a normal BiCMOS circuit.

By setting the value of the emitter resistance to 30 [KΩ] or less, the voltage between both ends of the emitter resistance becomes 3 [V] or less, and the dynamic range of the circuit has a margin.
Therefore, the value of the emitter resistance is set to 100 [Ω] to 50
[KΩ], preferably 3 [KΩ] to 30 [KΩ].

By using the configuration in which the resistor is connected to the emitter, such an effect of noise reduction can be expected. The range of the resistance slightly changes depending on the current flowing through the collector. As the collector current increases, the preferable resistance value tends to decrease.

Further, as the bias circuit connected to the integration circuit, particularly low noise is required since the integration circuit handles a weak signal. Therefore, it is conceivable to remove high-frequency components of noise by using a low-pass filter. However, using only a low-pass filter can remove high-frequency components, but removes low-frequency noise, particularly 1 / f noise. It is difficult. Therefore, the configuration in which a resistor is connected to the emitter as described above can reduce noise including 1 / f noise present in each circuit element, and is therefore preferable for applications requiring low noise.

Next, when it is desired to appropriately change the current value of the temperature-compensated constant current circuit 219, a configuration as shown in FIG. 3B can be said to be preferable. As shown in the figure, the constant current circuit 324 includes a constant voltage circuit 304 and a non-inverting amplifier 305.
, N NPN transistors 306, and a plurality of resistors 307 and N connected to each NPN transistor 306.
It includes a MOS transistor 308 and a shift register 309 connected to each NMOS transistor 308.

The output of the temperature-compensated constant voltage circuit 304 is connected to the input of a non-inverting amplifier 305, and the output of the non-inverting amplifier 305 is connected to the bases of a plurality of NPN transistors 306. A resistor 307 is connected to the emitter of each NPN transistor 306, and the respective resistance values are set to R, 2
R, 4R,..., 2 ^{n -1} · R.

The emitter area of each NPN transistor 306 is inversely proportional to the resistance value of the resistor 307, and A _E , A _E /
2, A _E / 4,..., A _E / 2 ^{n -1} . An NMOS transistor 308 is connected between each resistor 307 and the ground, and its gate width is
, W / 4, W / 4,...,
W / 2 ^n-1 is set. Shift register 30
Reference numeral 9 is connected to the gate of each NMOS transistor 308 to control on / off of the transistor.

At this time, the current flowing through the collector of each NPN transistor 306 is constant because the base voltage is constant.
It is changed in binary in inverse proportion to the resistance value of the resistor 307.

The reason for changing the emitter area in a binary manner is as follows. When the resistance value of the resistor 307 is changed in binary, each NPN is almost in inversely proportional thereto.
The base current of the transistor also changes. At this time, from equation (1), the emitter area A _{E of} each NPN transistor is obtained.
If V is also changed in inverse proportion to the resistance value, V _BE becomes the same between the transistors. As a result, the voltages applied to the respective resistors 307 become exactly the same, and errors can be reduced.

On the other hand, the reason why the gate width of the NMOS transistor 308 is also changed in inverse proportion to the resistance value of the resistor 307 is as follows. Drain current _ID and drain
The relation between the source-to-source voltage V _{DS is} as follows.
_GS , the gate threshold voltage is V _T , the proportional coefficient is I _D0 , and the gate width is W,

[0114] _{I D = W · I D0 [} (V GS -V T) · V DS -V DS 2/2] ···· (6)

Is obtained. When the resistance value of the resistor 307 is changed in a binary manner, the drain current of each NMOS transistor also changes almost in inverse proportion thereto. At this time, each NMO
If the gate width of the S transistor 308 is also changed in inverse proportion to the resistance value of the resistor 307, the drain-source voltage between the transistors becomes the same.

As a result, the voltages applied to the respective resistors 307 become exactly the same, and errors can be reduced. However, when the voltage between the drain and the source is negligibly smaller than the voltage across the resistor 307, there is no problem even if such a change is not made.

In this circuit, data is sent to the shift register 309 in accordance with the current required for the constant current circuit 219, and the ON / OFF of the NMOS transistor 308 is controlled, whereby the temperature is changed in a binary manner and the temperature is compensated. An electric current can be obtained. That is, the value of the temperature-compensated constant current can be digitally set.

At this time, the resolution of the obtained constant current is substantially the LSB constant current value, and the maximum current of the obtained constant current is about twice the constant current value of the MSB. For example, when the constant current value of the MSB is 200 [μA] and an eight-stage constant current circuit is configured, the LSB is 1.56 [μA], the resolution is about 1.56 [μA], and the maximum current is about 400 [μA]. ]. This resolution indicates an ideal case, and is actually affected by errors in the constant current value of each bit.

FIGS. 3A and 3B described above are for obtaining a current (sink current) flowing into the constant current circuit. On the other hand, FIG. 3C described below is a circuit in which a current (source current) flowing out of the constant current circuit is also obtained.

In FIG. 3C, the constant current circuit 325
Is a PNP transistor 311 having an emitter connected to the resistor 313, a PNP transistor having an emitter connected to the base of the PNP transistor 311 and a base connected to the collector of the PNP transistor 311.
P transistor 312 and constant current circuit 3 connected between the collector of PNP transistor 311 and ground
10, a resistor 317, a PNP transistor 316 having an emitter connected to the resistor 317, a PNP transistor 319 having a base connected to the collector of the PNP transistor 316, and a PNP transistor having a base connected to the emitter of the PNP transistor 319. 318,
A resistor 320 connected between the emitter of the PNP transistor 318 and ground, a resistor 315, and an emitter connected to the resistor 315 and the PNP transistor 3
The PNP transistor 314 includes a PNP transistor 314 whose base is connected to the base, a PNP transistor 321 whose base is connected to the base of the PNP transistor 318, and a resistor 322 connected between the emitter of the PNP transistor 321 and the ground. I have.

In this embodiment, the temperature-compensated constant current circuit shown in FIGS. 3A and 3B can be used as the temperature-compensated constant current circuit 310. The NPN transistor 311 has the collector connected to the constant current output as described above, and has the resistor 31 connected to the emitter.
3 are connected. The emitter and base of another NPN transistor 312 are connected to its base and collector, respectively, and the collector of NPN transistor 312 is connected to ground. NPN transistor 3
The base of the NPN transistor 314 having the same structure and dimensions is connected to the base of the eleventh transistor 11. The resistor 313 is connected to the emitter of the NPN transistor 314.
A resistor 315 having the same structure and dimensions as that of FIG.
Thus, a source current having the same current value as that of the temperature-compensated constant current circuit 310 can be obtained from the collector of the NPN transistor 314.

Also in this case, the resistance of the resistor 313 can be a times the resistance of the resistor 315, and the emitter area of the NPN transistor 311 can be 1 / a times the emitter area of the NPN transistor 314. This source current is supplied to the temperature-compensated constant current circuit 213 and the temperature-compensated constant current circuit 2 shown in FIG.
25 can be used. Further, by adding a circuit composed of PNP transistor 316, NPN transistors 318, 319 and 321 and resistors 317, 320 and 322, a sink current having the same current value can be obtained.

Using this circuit, several source currents and sink currents are generated from one temperature-compensated constant current circuit, and a plurality of constant current circuits such as the above-mentioned constant current circuits 213, 222 and 225 are formed. Can be replaced by

The constant current circuits 213, 222 and 2
In a preferred embodiment of the present invention, a constant current circuit having a temperature compensation of about 6 bits may be formed as the constant current circuit 213 for determining the current flowing through the bolometer. The bolometer current is about 100 to 400 [μA] assuming a bolometer resistance of about 10 kΩ, and if about 6 bits, the bolometer current can be set at about 1/64 of the resolution.

The cancel current is set to the same level as the bolometer current, but the problems here are the self-heating of the bolometer and the variation in resistance. A bolometer having a high TCR often has a negative temperature coefficient, but when self-heating occurs, the bolometer resistance decreases and the bolometer current increases by about 10 to 20%.

Further, the bolometer resistance is usually 10
It varies by about [% pp], and when it is large, it varies by about 30 [% pp]. Therefore, it is necessary to set the cancel current in consideration of these factors.
It may be set to about 1.5 times. That is, if the cancel current is set to about 1 to 1.5 of the bolometer current, it can be said that it is convenient because the cancel current changes together with the change in the bolometer current.

When increasing the gain of the integration circuit, it is necessary to reduce the resolution for setting the cancel current.
The gain of the integration circuit with respect to the integration current is obtained by setting the integration time to T _s ,
Assuming that the capacitance of the integrating capacitor is C, T _s / C.

[0128] for example, about 30 [.mu.s] as T _s, to set the degree 100 [pF] as C, and preferably as an embodiment of the present invention, the gain is 300,000
About. This means that the integrated output becomes 0.3 [V] even at a resolution of about 1 [μA]. The number of bits for setting the cancel current is preferably 8 bits or more so that the resolution for setting the cancel current does not occupy the dynamic range of the integrator circuit (in this case, the cancel current varies from 100 to 150% of the bolometer current. , And the resolution is 1 μA).

Considering the integral gain, the FP shown in FIG.
It is preferable that the N correction circuit 236 has about 6 bits so that the resistance variation can be reduced to about 1/64. The constant current circuit 225 is F
Determine the maximum current of the PN correction circuit. This constant current circuit 2
Also in the case of 25, the circuit is configured based on the bolometer current, so that it changes following the bolometer current, which is convenient. For example, if the resistance variation of the bolometer is 30 [% pp], the FPN correction circuit 23
6 is set so that the maximum current is 30 [% pp] of the bolometer current.

Since the resistance variation of the bolometer changes depending on the sample, it is preferable to configure a binary constant current source of about 4 bits so that the constant current circuit 225 can be set arbitrarily.

FIG. 6 shows an example of a specific circuit of the filters 217, 223 and 229 of FIG.
As shown in the figure, in this example, a low-pass filter using a resistor 601 and a capacitor 602 is configured. The cut-off frequency of this low-pass filter is 1 / (2, where R is the resistance of the resistor 601 and C is the capacitance of the capacitor 602.
πCR). The operational amplifier 603 is a buffer for lowering the output impedance of the filter, and forms a voltage follower.

Next, the conventional integrating circuit and bias circuit will be compared with the integrating circuit and bias circuit of the present invention with reference to the drawings. FIG. 7A shows a circuit diagram of a conventional integration circuit and a bias circuit, and a graph showing a collector current in each pixel. Similarly, FIG.
Is a circuit diagram of the integration circuit and the bias circuit of the present invention,
3 shows a graph indicating a collector current in each pixel.

First, as shown in FIG. 7A, a conventional integration circuit and bias circuit are composed of a power supply 701 for outputting a constant voltage V ₀ , an integration transistor 702 having a base connected to the power supply 701, It includes a plurality of bolometers 704 connected to the emitter of the transistor 702, a pixel switch 703 connected to each bolometer 704, and an integrating capacitor 705 connected to the collector of the integrating transistor 702. As described above, in the conventional example shown in FIG. 7A, a constant bias is applied to the base of the integration transistor 702, and the collector current (that is, the integration current) is

I _C = V ₀ / R / (1 + αΔT) (7)

Is as follows. Here, R is a resistance value of the bolometer 703 at a certain temperature (for example, 25 [° C.]), α is a TCR,
ΔT is the temperature rise of the silicon substrate. As can be seen from this equation, when ΔT changes, the amount of change in I _C increases in a pixel having a small bolometer resistance value, that is, a pixel having a large bias current.

Conversely, the amount of change in I _C is small in a pixel having a large bolometer resistance value. That is, the collector current I
The variation in the amount of change of _C becomes the fixed pattern noise (FPN) itself when the device temperature changes.

On the other hand, the present invention shown in FIG.
A constant current circuit 706, an NPN transistor 707 whose base is connected to the constant current circuit 706, and a constant current circuit 7
06 and the NPN transistor 70
7, an NPN transistor 708 having a base connected to the emitter, a reference resistor 709 connected between the emitter of the NPN transistor 708 and ground, and an NPN having a base connected to the base of the NPN transistor 708.
A transistor 712, a plurality of bolometers 710 connected to the emitter of the NPN transistor 712, and a plurality of pixel switches 711 connected to each bolometer 710
And an integrating capacitor 713 connected between the collector of the NPN transistor 712 and ground.

As described above, in the present invention shown in FIG.
The voltage of the reference resistor 709 is applied to each bolometer 710. The TCR of the reference resistor 709 is the TC of the bolometer 710.
R, the collector current I _C is a current that hardly depends on ΔT, and the fixed pattern noise (F
PN) can be suppressed.

For example, 320 × 240 pixels, pixel pitch 5
In a two-dimensional thermal infrared imaging device of about 0 [μm], the in-plane bolometer resistance varies about 10 [% pp], and the bias current also varies about 10 [% pp]. Therefore, in the conventional example shown in FIG.
When the device temperature changes by x ° C, the bias current becomes x [° C]
× 2 [% / ° C.] × I _0, but the variation is x [° C.] × 2 [% / ° C.] × 10
[% Pp] × I ₀ That is, fixed pattern noise is generated.

On the other hand, considering a case where an attempt is made to obtain 0.1 [° C.] as a noise equivalent temperature difference (NETD) corresponding to the temperature resolution of the infrared imaging apparatus, the temperature change of the subject of 0.1 [° C.] When the temperature of the diaphragm is changed by about 0.1 [m ° C.] and the TCR of the bolometer is set to 2 [% / ° C.], the bias current I ₀ is changed by about 2E−6 × I ₀ . This corresponds to the degree of bias modulation when the minimum resolution temperature difference is observed, and the above-mentioned fixed pattern noise needs to be less than this. That is, the device temperature change x is 1 [m ° C.]
Means that it must be kept within. However, a temperature stabilizing device such as a Peltier element which is usually used has only an accuracy of about 10 [m ° C.], and thus it is difficult to realize the above conditions.

On the other hand, in the present embodiment, such a problem does not occur because the bias current hardly changes due to a change in device temperature. However, if there is a variation in the TCR of the bolometer, a fixed pattern noise is also generated in the present invention. Normally, the variation of the TCR is about 0.1 [%].
It is preferable to perform temperature control with an accuracy of about [m ° C.] or less.

FIG. 8 is a timing chart for explaining the operation of the circuit of FIG. As shown in FIG.
A vertical synchronizing signal of about [Hz] is input to a data terminal of the vertical shift register 233. φH is for example 7
A horizontal synchronizing signal of about [KHz] is input to the clock terminal of the vertical shift register 233. As a result, the vertical shift register 233 outputs the vertical selection signals V ₁ ,
V _2, · · · are output, while selecting a row, the read operation of the integration such as the readout circuit 201 on each column is performed.

V _C is a voltage waveform (integrated waveform) at the integrating capacitor 205 in FIG. Sample hold circuit 2
By applying a sample and hold pulse φS / H to 30, the voltage after integration is sampled and held in a hold capacitor (not shown) in the sample and hold circuit 230. After sampling, reset switch 2
06, a reset pulse φR is applied and the integration capacitor 2
Reset 05.

The horizontal synchronizing signal φH ′ is input to the data terminal of the horizontal shift register 232 and the clock signal φCLK is input to the clock terminal, so that the horizontal selection signal H
₁ , H ₂ ,... Are obtained. Horizontal selection signals H ₁ , H
₂ ... sequentially selects the multiplexer 231 of FIG. 2, the signals held in the hold capacitor of each column is outputted as V _out through the multiplexer 231.
Note that the horizontal synchronization signal φH ′ is
The same signal may be used, or another signal may be used.

Further, the FPN data buffer 23 shown in FIG.
The horizontal selection signals H ₁ , H ₂ ,... Are input to the control terminal 4 (terminal for controlling writing), and the latch enable LE is input to the FPN data buffer 235.
As a result, the FPN data (D _FPN ) is transferred to the FPN data buffer 235 before reading a certain row, and is transferred to the FPN data buffer 235 and held at the timing of switching the row. The FPN data buffer 235
The constant current value connected to the NMOS transistor 210 and output from the FPN correction circuit 236 is determined.

Next, another embodiment of the present invention (second embodiment)
Embodiments 1 to 5) will be described with reference to the drawings.

[Second Embodiment] FIG. 9 is a circuit diagram of a thermal infrared imaging apparatus showing a second embodiment of the present invention. As shown in the figure, in the present embodiment, one OB readout circuit 901, a plurality of readout circuits 910 similar to those described in FIG. 2, an operational amplifier 909, a bolometer bias circuit 913, A bias circuit 914 for an FPN correction circuit, a multiplexer 916, a horizontal shift register 917, and a vertical shift register 915 are provided.

As described above, this embodiment is characterized in that the OB readout circuit 901 is provided and the operational amplifier 909 is provided instead of the bias circuit 218 in FIG. The + input terminal of the operational amplifier 909 is connected to the integration capacitor 9 in the readout circuit 901.
05 is connected, and a bias voltage V _C0 is applied to its − input terminal. The output terminal of the operational amplifier 909 is connected to the base of a PNP transistor 908 in the read circuit 901.

The OB readout circuit 901 has the NPN
A transistor 904, a PNP transistor 908,
A plurality of bolometers 902, a pixel switch 903 connected between each bolometer 902 and the ground,
An N correction circuit 911, a reset switch 906, an integrating capacitor 905, and a sample and hold circuit 912 are provided.

The OB readout circuit 901 is
2 is a special readout circuit different from the one shown in FIG. 2, and the built-in bolometer has a bolometer (OB bolometer 90) having no sensitivity to incident infrared rays.
2) is used. So-called optical black
This is called a bolometer (hereinafter, referred to as an OB bolometer) and is made by optically shielding incident infrared rays.

However, only the readout circuit 901 has the OB bolometer 902, and the other readout circuits 910 are ordinary readout circuits, and ordinary bolometers are used. Therefore, the read circuit 91
0 has sensitivity to infrared rays.

In this embodiment, the bias of the PNP transistor 908 is set by the OB readout circuit 901.
This is performed by keeping the voltage of the middle integration capacitor 905 at a constant voltage V _C0 . That is, even if the bolometer current changes due to a change in the device temperature and a change in the resistance value of the OB bolometer 902, the current on the PNP transistor 908 side changes following the change. And this PNP transistor 90
The bias voltage to the base of the normal read circuit 9
The PNP transistor 10 is also added to the PNP transistor, so that the integrated current of the readout circuit 910 is also compensated for the temperature change of the device.

Further, in this embodiment, the influence of the self-heating of the bolometer can be compensated. Normally, when an electric current is applied to a thermally separated bolometer, the bolometer itself generates heat due to the Joule heat, and the temperature of the bolometer rises. As a result, self-heating of the bolometer causes
Is bent in the integrated waveform V _C, and the bent occupies the dynamic range of the signal. Further, in the circuit of FIG. 9, the OB bolometer 902 also generates self-heating similarly to a normal bolometer.

Therefore, by applying a bias to the PNP transistor 908 so that the voltage of the integration capacitor 905 of the OB readout circuit 901 becomes constant, the current flowing through the PNP transistor 908 also changes following the change in the bolometer current due to self-heating. I am trying to do it.

Therefore, the voltage of the integrating capacitor 905 in the OB read circuit 901 is kept constant. Therefore, the integration capacitor 905 in the OB readout circuit 901
May be omitted.

[Third Embodiment] FIG. 10 is a circuit diagram of a thermal infrared imaging apparatus showing a third embodiment of the present invention. As shown in the figure, the present embodiment has one OB
A read circuit 1001, a plurality of read circuits 1010 similar to those described in FIG. 2, an operational amplifier 1009,
Bias circuit 1013 for canceling resistor, bias circuit 1014 for FPN correction circuit, and multiplexer 1
016, a horizontal shift register 1017, and a vertical shift register 1015.

Also, in this embodiment, an operational amplifier 1009 is used instead of the bias circuit 212 in the circuit of FIG.
The integration capacitor 1005 in the readout circuit 1001 is connected to the + input terminal, and the bias voltage V _C0 is applied to the − input terminal. The output of the operational amplifier 1009 is connected to the base of the NPN transistor 1004.

The bolometer 1002 in the read circuit 1001 also constitutes an OB bolometer. The normal readout circuit 1019 other than the OB readout circuit 1001 having the OB bolometer has the same configuration as that of FIG. 2 and is sensitive to infrared rays.

As described in the second embodiment, also in this embodiment, the OB read circuit 100
The bias of the NPN transistor 1004 is set so as to keep the voltage of the integrating capacitor 1005 in 1 at a constant voltage V _C0 . Thus, similarly to FIG. 9, the influence of the temperature change of the device and the self-heating of the bolometer can be canceled.

[Fourth Embodiment] FIG. 11 is a block diagram of a thermal infrared imaging apparatus showing a fourth embodiment of the present invention. As shown in the figure, in the present embodiment, a read circuit 1101 and a bolometer bias circuit 110
2, a bias circuit 1106 for the cancel resistor, and F
A bias circuit 1107 for a PN correction circuit, a vertical shift register 1113, a multiplexer 1114, and a horizontal shift register 1115 are provided.

That is, in this embodiment, a bolometer bias circuit 1102 is provided for each readout circuit 1101.
The bias circuit 1102 is characterized in that it has a temperature-compensated constant current source 1103, an FPN correction circuit 1104, and a reference resistance circuit 1105. Then, instead of the bias circuit 1102 having the FPN correction circuit 1104, each readout circuit 1101 does not have the FPN correction circuit.

The bias circuit 1107 applies a bias to the FPN correction circuit 1104 in the bias circuit 1102, and the other configuration is the same as that of FIG. That is, the read circuit 1101 includes the cancel resistor 1108, the integration circuit 1109, and the bolometer 11
And a pixel switch 1 composed of an NMOS transistor
111 and a sample and hold circuit 1112.

As described above, the FP is connected to the bias circuit 1102.
By providing the N correction circuit 1104, sensitivity variation can be corrected for each pixel. This is because the FPN correction circuit in the configuration of FIG. 1 works to compensate for the additional current for the variation in the bolometer current, whereas the FPN correction circuit 1104 in FIG. 11 works so that the variation in the bolometer current does not occur. It is due to. Therefore, the sensitivity to incident infrared rays is proportional to the bolometer current. Therefore, the variation in the bolometer current itself can be reduced, so that the variation in sensitivity of each pixel can be reduced.

Here, the detailed configuration of FIG. 11 will be described. FIG. 12 shows a specific circuit diagram of the fourth embodiment according to FIG. As shown in the drawing, in this embodiment, a plurality of read circuits 1201, a bias circuit 1202 provided in each read circuit 1201,
A bias circuit 1207 for an FPN correction circuit, a bias circuit 1206 for a cancel resistor, a vertical shift register 1228, a multiplexer 1229, and a horizontal shift register 1230 are provided.

Each reading circuit 1201 includes a resistor 1220, a PNP transistor 1221, an NPN transistor 1222, a bolometer 1223, a pixel switch 1224, a reset switch 1225, an integrating capacitor 1227, and a sample and hold circuit 1226.
And

Each of the bias circuits 1202 is connected to the FPN
Correction circuit 1203 and temperature-compensated constant current circuit 120
4 and a reference resistance circuit 1205. The FPN correction circuit 1203 includes a resistor 1215a, PNP transistors 1208 and 1209, and an NPN transistor 1
210, a resistor 1211 and an NMOS transistor 12
12, a data buffer 1213 and 1214, a resistor 1215, and a PNP transistor 1216. The base of the NPN transistor 1210 in each FPN correction circuit 1203 is connected to the output of the bias circuit 1207.

The reference resistance circuit 1205 is connected to the bolometer 12
23, a resistor 1219 having the same TCR and NPN transistors 1217 and 1218.
The base of the fifth NPN transistor 1217 is connected to the collector of the NPN transistor 1218. Further, the resistance of the resistor 1219 is set to a times the resistance of the bolometer 1223, and the emitter area of the NPN transistor 1222 is set to N.
It may be 1 / a times the emitter area of the PN transistor 1222.

The temperature-compensated constant current circuit 1204 outputs a current I ₀ having a small temperature dependency as in the example of FIG.

Now, the NPN transistor 211 of FIG.
As with the FPN correction resistor 209 and the NMOS transistor 210, a constant current digitally set according to the variation in the resistance value of the bolometer is output. This constant current also has a small temperature dependency due to the bias circuit 1207. The FPN correction circuit 1203 is configured to obtain a source current, similarly to the configuration of FIG.

Temperature Compensated Constant Current Circuit 1204 and FP
By adding the currents output from the N correction circuit 1203 and flowing the sum to the reference resistor circuit 1205 and applying the bias voltage output from the reference resistor circuit 1205 to the base of the NPN transistor 1222, each bolometer current has almost no temperature dependency and , Variations are reduced, and variations in sensitivity are also reduced.

As described above, in a two-dimensional thermal infrared imaging device having 320 × 240 pixels and a pixel pitch of about 50 μm, the in-plane bolometer resistance variation is 10% p.
−p]. At this time, in the conventional circuit, the same sensitivity variation as the resistance, that is, about 10 [% pp] sensitivity variation occurs. On the other hand, in the present embodiment, even if there is a variation in the resistance of the bolometer, the bias current is corrected so as to be substantially the same, so that the sensitivity variation can be significantly reduced. The remaining sensitivity variation is due to the in-plane variation of the TCR.
It is about 1 [% pp].

[Fifth Embodiment] FIG. 13 is a block diagram of an entire image pickup apparatus according to a fifth embodiment of the present invention. As shown in the figure, the imaging apparatus according to the present embodiment includes an imaging element 1301, an amplifier 1302, a sample and hold circuit 1303, and an A / D converter 1304.
, VRAM 1305 and FPN memory controller 1
306, FPN memory 2, and FPN memory 1307
, Digital subtractor 1308 and D / A converter 13
09 and NTSC (National Television System Commi
ttee) signal generator 1310 and comparator 1311
, An FPN memory controller 1312, an FPN memory 1313, a temperature stabilizing element 1314 such as a Peltier element, a Peltier control circuit 1315 for controlling the driving of the temperature adjusting element, and an optical system 1316.

The image pickup device 1301 is formed by, for example, forming the structure shown in FIG. 2 on a single silicon substrate. The optical system 1316 faces the image sensor 1301.
Is provided, and light incident from the outside is transmitted through the optical system 1.
The light is condensed on the image sensor 1301 by 316. Then, the image sensor 1301 converts light into an electric signal, amplifies the electric signal by an integrating circuit or the like in the image sensor 1301, and outputs the amplified signal to the outside.

The amplifier 1302 amplifies the output signal of the image sensor 1301 and inputs the amplified signal to the sample hold circuit 1303. The sample hold circuit 1303 temporarily holds the received signal. Note that the amplifier 1302 need not be used if the output signal of the image sensor 1301 is sufficiently large.

A / D converter 1304 converts the signal held by sample / hold circuit 1303 into a digital signal. The number of bits of the A / D converter 1304 is as follows when an infrared imaging device is taken as an example. For example, if the temperature resolution of the subject is 0.1 ° C. and the dynamic range of the temperature of the subject is 100 ° C., a bit number of 10 bits (about 1000 gradations) is required at this time. Then, in order to further reduce the quantization error, when 2 bits (4 gradations) are allocated per minimum temperature resolution, the data width of 12 bits in total becomes the A / D converter 130.
4 needed.

The VRAM 1305 is a memory for holding a 12-bit digital signal of each pixel. For example, when the image sensor 1301 has 320 × 240 pixels, if the capacity is about 320 × 240 × 12 bits, Good. Since data is managed in byte units, it is possible to prepare a large capacity (for example, 320 × 240 × 16 bits) as needed.

[0177] The FPN memory 1307 is
This is storage means for correcting variations that could not be removed by the FPN correction performed in 1, and stores variation data of each pixel for correction. The FPN memory controller 1306 is a circuit for controlling the FPN memory 1307, and the digital subtractor 1308 is for subtracting the variation amount of each pixel from the signal of each pixel coming in real time. The acquisition of the variation data may be performed in the following sequence after acquiring the FPN correction data in the image sensor 1301.

First, the data of each pixel output from the A / D converter 1304 in a state where the incident light is blocked by a shutter or the like has variations that cannot be removed by the FPN correction in the image sensor 1301. This data is stored in the FPN memory 1307.

This operation is performed when the power is turned on or when the previous correction is deviated, and in a normal imaging state, the stored variation data of the FPN memory 1307 is subtracted from the subtracter 1.
The signal 308 is subtracted from the signal of each pixel coming in real time to obtain a signal having no variation.

It should be noted that an adder may be used instead of the subtractor 1308. That is, by using the adder after taking the complement of the data in the FPN memory 1307, an operation equivalent to the subtractor can be performed. It can be carried out. Further, the position of the subtractor 1308 does not necessarily have to be arranged at the position shown in FIG.
/ A converter 1309.

After that, the D / A converter 1309 converts the digital signal thus processed into an analog signal, and outputs the analog signal to the NTSC signal generator 1310. N
The TSC signal generator 1310 outputs an NTSC composite signal by synthesizing the analog signal and a predetermined synchronization signal. Also, the NTSC signal generator 131
Instead of 0, PAL (Phase Alternatio
Other types of signal generators such as “n by Line” and RGB output may be used.

The acquisition of correction data to be supplied to the FPN correction circuit in the image sensor 1301 is performed as follows. Comparator 13 used in the present embodiment
Reference numeral 11 denotes a digital comparator which determines the magnitude relationship between the signal level of each pixel and a predetermined reference level.

This reference level is set to the upper or lower limit of the dynamic range of the signal processing circuit such as the integrating circuit, the amplifier, the A / D converter, etc. in the image pickup device 1301, and the upper or lower limit of the dynamic range is a predetermined level. Can be set to the value added. In order to determine the magnitude relation, any one of a good one that is equal to or higher than a predetermined reference level, a good one that is equal to or less than a predetermined reference level, and a good one that is within a range of two predetermined reference levels is used. May be used.

The FPN memory controller 1312 creates FPN correction data according to the comparison result. The created FPN correction data is stored in the FPN memory 1313. Therefore, the FPN memory 1313 only needs to have a capacity obtained by multiplying the total number of pixels by the number of bits of the FPN correction data. For example, if the number of pixels is 320 × 240, FPN
320 × 240 when the number of bits of the correction data is 6 bits
What is necessary is just to have a capacity of × 6 bits. Further, in order to control data in byte units, it is possible to increase the capacity as needed.

[Sixth Embodiment] In the above-described five embodiments, each read circuit is provided with a plurality of bolometers. However, the present invention is not limited to this. For example, when the imaging device of the present invention is applied to a human body detector, a fire detector, or the like, it is only necessary to detect the presence or absence of infrared rays, and thus it is not necessary to provide an image sensor with a plurality of pixels. Therefore, at least one pixel is required, and in that case, at least one bolometer in the readout circuit is required. Further, in the case of a single pixel, there is no variation between pixels as in the case of a plurality of pixels.
A bias circuit for the PN correction circuit and the FPN correction circuit becomes unnecessary.

By applying the first to fifth embodiments based on the above facts, the following is obtained. For example, in the first and fourth embodiments, only one readout circuit having one bolometer is used, or two bolometers are prepared in one readout circuit and one of them is used.
It is good to make an individual into an OB bolometer. When two bolometers are provided in one readout circuit,
An FPN correction circuit and a bias circuit for the FPN correction circuit may be provided.

In the second and third embodiments, 1
It is preferable to combine one readout circuit with one bolometer and one readout circuit with one OB bolometer. Further, in the fifth embodiment, it is clear that the image sensor 1301 may be one pixel.

[0188]

As described above, according to the present invention, by applying a voltage to a measuring resistor such as a bolometer, the effect of the measuring resistor itself causing self-heating can be eliminated. As a result, the dynamic range of the signal processing circuit including the integrating circuit has a margin, and the gain of the circuit can be increased. In particular, by increasing the gain of the first-stage integration circuit, input conversion noise can be reduced, and S / N
Can be greatly improved. Furthermore, a function capable of digitally setting a bias can be provided while maintaining the characteristics of low drift and low noise. The configuration is simple, and a high-performance semiconductor device can be realized at low cost. As described above, according to the present invention, an excellent effect is obtained that the bias component can be corrected stably with low noise, and the signal amplification and signal processing can be performed smoothly.

[Brief description of the drawings]

FIG. 1 is a circuit diagram showing a first embodiment of the present invention.

FIG. 2 is a specific circuit diagram of the thermal infrared imaging device according to FIG. 1;

FIGS. 3A, 3B and 3C are circuit diagrams each showing a temperature-compensated constant current circuit; FIG.

FIG. 4 is a circuit diagram showing a band gap reference circuit.

FIG. 5 is a graph showing a noise current of a transistor in this embodiment.

FIG. 6 is a circuit diagram showing a filter according to FIG. 1;

7A is a circuit diagram showing a conventional integration circuit and a bias circuit and a graph showing a collector current in each pixel, and FIG. 7B is a circuit diagram showing an integration circuit and a bias circuit according to the present embodiment and FIG. 4 is a graph showing a collector current in a pixel.

FIG. 8 is a timing chart showing the operation of the present embodiment.

FIG. 9 is a circuit diagram showing a second embodiment of the present invention.

FIG. 10 is a circuit diagram showing a third embodiment of the present invention.

FIG. 11 is a block diagram showing a fourth embodiment of the present invention.

FIG. 12 is a circuit diagram showing details of FIG. 11;

FIG. 13 is a block diagram showing a fifth embodiment of the present invention.

FIG. 14 is a sectional view showing a conventional thermal infrared imaging device.

15 is a circuit diagram showing a readout circuit of the thermal infrared imaging device according to FIG.

[Explanation of symbols]

101, 111, 114 ... constant current circuit, 102, 11
2, 115 ... reference resistance circuit, 103, 113, 116 ...
Filter, 104 bolometer, 105 integration circuit, 1
06: cancel resistor, 107: pixel switch, 108
... FPN correction circuit, 109 ... Sample hold circuit, 1
10 Multiplexer 117 Reading circuit 118
... a horizontal shift register, 119 ... a vertical shift register,
120, 121, 122 ... bias circuit, ΦV ... vertical synchronization signal, ΦH ... horizontal synchronization signal, ΦH '... horizontal synchronization signal,
.PHI.CLK ... clock signal, .PHI.S / H ... sample-and-hold pulse, Vout ... output _{voltage, V 1, V 2, ...} , V n ... vertical selection _{signal, H 1, H 2, ...} , H n ... horizontal selection signal.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H01L 29/84 H04N 5/33 H04N 5/33 H01L 27/14 K F term (Reference) 2G065 AB02 BA12 BA34 BC08 BC15 BC16 BC28 BC33 BC35 BE08 CA13 CA21 DA18 2G066 BA09 BB09 BB11 BC02 BC04 BC07 BC11 BC15 CA02 CB03 4M112 AA01 BA01 CA08 CA10 EA02 FA01 FA02 4M118 AA05 AB01 BA30 CA14 CB14 GA10 GB09 5C024 AX06 CX43 HXX HXX HXX

Claims (5)

[Claims] 1. A measuring resistor for converting a physical quantity into a resistance value, a bias current connected to the measuring resistor and applied to the measuring resistor, and a current flowing through the measuring resistor integrated. A readout circuit comprising an integrating circuit for storing and a canceling resistor connected to the measuring resistor for canceling a bias current flowing through the measuring resistor; and a reading circuit for the measuring resistor based on the current stored in the integrating circuit. In a semiconductor device configured to detect a change in resistance value and indirectly measure the physical quantity, a first bias circuit having a reference resistor connected to the measurement resistor and having a temperature coefficient of resistance equal to the measurement resistance And a second bias circuit having a reference resistor connected to the cancel resistor and having a temperature coefficient of resistance equal to the cancel resistor. The semiconductor device according to claim 1, wherein the bias circuit is means for applying the output voltage to the measurement resistor, and the second bias circuit is means for applying the output voltage to the cancel resistor. 2. The first bipolar transistor according to claim 1, wherein the integration circuit includes first and second bipolar transistors having collectors connected to each other, and an integration capacitor connected to the collector. Has the emitter connected to the canceling resistor and the base connected to the second bias circuit, and the second bipolar transistor has the emitter connected to the measuring resistor and the base connected to the first resistor. A bias circuit connected to the semiconductor device. 3. The constant current circuit according to claim 1, wherein the first bias circuit includes a temperature-compensated constant current circuit, a reference resistor connected to the constant current circuit and having a temperature coefficient of resistance equal to the measurement resistance. A semiconductor device comprising: a filter connected to a resistor and configured to remove noise from a voltage generated at the reference resistor. 4. The constant current circuit according to claim 1, wherein the second bias circuit includes a temperature-compensated constant current circuit, a reference resistor connected to the constant current circuit, and having a temperature coefficient of resistance equal to the cancel resistance. A semiconductor device comprising: a filter connected to a resistor and configured to remove noise from a voltage generated at the reference resistor. 5. The semiconductor device according to claim 1, wherein the semiconductor device is an infrared sensor, a microwave / millimeter wave detector, a temperature sensor, a magnetic sensor, a pressure sensor, a gas sensor, or a flow sensor. A semiconductor device, comprising: