This disclosure relates generally to electronics and more particularly to reference voltage circuits.
A reference voltage circuit is a circuit that produces a fixed voltage to a device. The fixed voltage is substantially constant despite variations in temperature. Conventional bandgap reference voltage circuits use a combination of a bipolar (or diode) base-emitter junction voltage (Vbe) and a proportional to absolute temperature (PTAT) voltage. Vbe is roughly 650 mV at room temperature and has a negative temperature coefficient (TC). The PTAT voltage has a positive TC which, when added to the negative TC of the Vbe, creates a low temperature coefficient reference voltage of about 1.24 volts.
When fabricating voltage reference circuits in integrated circuits, pressure from the package on the integrated circuit die can alter the fixed voltage produced by a voltage reference circuit. One way to avoid this problem is to use a ceramic package that can be hermetically sealed and does not induce pressure on the die. Another way to avoid this problem is to use a die coat that displaces pressure normally placed on the die. These methods can increase the production cost.
A reference voltage circuit corrects for bandgap voltage shifts induced during fabrication. The reference voltage circuit generates a reference voltage using first and second base-emitter pairs. The reference voltage circuit sums the voltage across the first base-emitter pair with a difference voltage multiplied by a factor of K. During a first time period, the difference voltage is the voltage across the first base-emitter pair minus the voltage across the second base-emitter pair, and during a second time period, the difference voltage is the voltage across the second base-emitter pair minus the voltage across the first base-emitter pair.
Particular implementations can provide one or more of the following advantages: 1) the reference voltage circuit can correct for shifts in the bandgap voltages induced during fabrication; 2) the reference voltage circuit can correct for offset due to an operational amplifier; 3) the reference voltage circuit can be fabricated at a reduced cost compared to conventional reference voltage circuits that are insensitive to the fabrication process; and 4) post-fabrication testing of the reference voltage circuit can be reduced or eliminated in some cases, saving time and cost of fabrication.
The details of one or more disclosed implementations are set forth in the accompanying drawings and the description below. Other features, aspects, and advantages will become apparent from the description, the drawings and the claims.
The reference voltage circuit includes two bipolar junction transistors (BJTs) Q1 and Q2. Transistors Q1 and Q2 each comprise a base-emitter pair, and the transistors each have approximately equal emitter areas.
Transistors Q1 and Q2 are coupled to a network of resistors R1-R7, a network of switches 102, 104, 106, 108, 110, and a feedback loop 112. The feedback loop includes an operational amplifier that includes a first stage op1a and a second stage op1b. The output of stage op1b is the bandgap reference voltage, Vbg. The output voltage is fed back into the network of resistors.
In operation, the operational amplifier serves to drive current into transistors Q1 and Q2. A control circuit provides control signals p1-p4 to the network of switches. In general, the control signals oscillate at a same frequency (e.g., 500 kHz) but at different respective phases and duty cycles.
The reference voltage Vbg is the sum of 1) the voltage across one of transistors Q1 and Q2 (Vbe), and 2) a difference between the voltages across both transistors Q1 and Q2 (ΔVbe). For example, Vbe can be the voltage across Q1 when control signals p3 or p4 are high and the voltage across Q2 when control signals p1 and p2 are high. A network of resistors amplifies ΔVbe. Vbe has a negative temperaturee coefficient and ΔVbe has a positive temperaturee coefficient. By adding Vbe to ΔVbe in a proper ratio, Vbg is substantially constant despite temperature changes.
The control circuit is configured to generate the switching signals so that, during a first time period, ΔVbe is equal to the voltage across Q1 minus the voltage across Q2, and during a second time period, ΔVbe is equal to the voltage across Q2 minus the voltage across Q1. By continuously toggling the switches, the resulting average ΔVbe over time cancels out any voltage shift from stress on the package.
Pressure from the package on the integrated circuit die can induce a shift in Vbe. In addition to the Vbe shift, the pressure also causes a ΔVbe shift. The effect on Vbg from the Vbe shift is 1:1, so that a lmV shift in Vbe also shifts Vbg by 1 mV. However, ΔVbe is typically amplified, e.g., by a factor of 5, 10, or 20, so that a 1 mV shift in ΔVbe shifts Vbg by 5, 10, or 20 mV. Thus, most of the resulting voltage shift in Vbg is due to the shift in ΔVbe.
In the example reference voltage circuit of
When signals p1 and p2 are high, resistor R5 is connected to transistor Q2, thus increasing its Vbe. The ΔVbe is applied across R2. When signal p1 is high, the differential input to op1a is negative and the differential input to op1b is the positive output of op1a. When signal p2 is high, the polarities of both amplifiers op1a and op1b is reversed, so the feedback loop maintains a negative feedback through both phases. When signal p1 is high, the output Vbg includes the ΔVbe of Q2-Q1 and the offset of the opamp (op1a and op1b). When signal p2 is high, the output Vbg includes the negative offset of the opamp and the ΔVbe from Q2-Q1. Similarly the opamp offset is inverted between when signal p3 is high and when signal p4 is high, and the ΔVbe is from Q1-Q2. The reference voltage circuit output multiplies the ΔVbe and opamp offset by a factor K. In this example, K is approximately R3/R4.
At time t1, the clock signal rises. At time t2, signal p1 rises. The difference in time between t2 and t1 is generally some time shorter than the period of the clock signal or half of the period of the clock signal. At time t3, the clock signal falls and signal p1 falls. At time t4, signal p2 rises. The difference between time t4 and t3 can be the same as the difference between times t2 and t1.
At time t5, the clock signal rises and signal p2 falls. At time t6, signal p3 rises. The difference between time t6 and t5 can be the same as the difference between times t2 and t1. At time t7, the clock signal falls and signal p3 falls. At time t8, signal p4 rises. The difference between time t8 and t7 can be the same as the difference between times t2 and t1. At time t9, the clock signal rises and signal p4 falls, and the control circuit begins to repeat the sequence between t1-t9.
In this example, the resistance of resistor R5 is as (m+1)/(n−1), but the resistor can have other values, e.g., to achieve different gains in the system. The currents labeled in the system can be expressed as follows:
The difference in the Vbe of Q1 and Q2 depends on the ratio of currents through the collectors. For purposes of illustration, emitter current replaces the collector current in this analysis, which is a valid simplification for large β. Assuming the transistors Q1 and Q2 are operating in the region of relatively constant β, ΔVbe can be expressed as follows:
For purposes of illustration, the value of the adjustable trim resistor R7 can be assumed to be zero. Then Vbg can be expressed as follows:
Vbg=Vbe+I3·(m+1)+k·(I1+I2+I3),
Where Vbe is the voltage from the emitter to the base of transistor Q2. The currents can then be expressed in terms of ΔVbe because Vx=m*ΔVbe, as follows:
Hence, Vbg can be expressed as:
This expression of Vbg can be written as:
As a result, n, m, and k can be selected to provide varying levels of gain for ΔVbe. The trim range is set by resistor R7. Once the trim range is determined, the resistance of resistor R6, k, can be reduced by half the trim range. This sets the nominal trim range center value to the nominal bandgap voltage and allows the trim to go positive or negative as required. Hence the trim range need not be included in nominal calculations for purposes of illustration.
Alternatively, the circuit can include a current source 302 to set the current flowing through transistors Q1 and Q2, as shown in
As shown in
A control circuit or other circuit generates two control signals, p1 and p2. In general, the control signals oscillate at a same frequency but at different respective phases. The offset of the two phase circuit goes through the gain (m+1) in each phase. The gain is set by (R3+R4)/R4 in one phase, and (R1+R2)/R2 in the other. Because the resistors can have some matching error, the offset cancellation can depend on the matching of the resistors. In some cases the matching can be made better than 1%, cancelling 99% of the offset. For an operational amplifier with 5 mV of offset, the net result can be 50 uV.
At time t1, the clock signal rises. At time t2, signal p1 rises. The difference in time between t2 and t1 is generally some time shorter than the period of the clock signal or half of the period of the clock signal. At time t3, the clock signal falls and signal p1 falls. At time t4, signal p2 rises. The difference between time t4 and t3 can be the same as the difference between times t2 and t1. At time t5, the clock signal rises and signal p2 falls, and the control circuit begins to repeat the sequence between t1-t5.
The reference voltage circuit drives current through first and second base-emitter pairs (602). During a first time period, the reference voltage circuit generates an output Vbg by summing 1) the voltage across the first base-emitter pair and 2) the voltage across the first base-emitter pair minus the voltage across the second base-emitter pair, multiplied by a factor K (604). During a second time period, the reference voltage circuit generates the output Vbg by summing 1) the voltage across the second base-emitter pair and 2) the voltage across the second base-emitter pair minus the voltage across the first base-emitter pair, multiplied by K (606). The second time period is substantially the same length of time as the first time period. By continuously alternating between the first time period and the second time period, the reference voltage circuit can cancel offset voltages induced in the base-emitter pairs during fabrication.
The reference voltage circuit can additionally, or alternatively, operate as follows. During the first time period, the reference voltage circuit generates the output Vbg by summing 1) the voltage across the first base-emitter pair and 2) the voltage across the second base-emitter pair minus the voltage across the first base-emitter pair, multiplied by a factor K. During the second time period, the reference voltage circuit generates the output Vbg by summing 1) the voltage across the second base-emitter pair and 2) the voltage across the first base-emitter pair minus the voltage across the second base-emitter pair, multiplied by K.
While this document contains many specific implementation details, these should not be construed as limitations on the scope what may be claimed, but rather as descriptions of features that may be specific to particular embodiments. Certain features that are described in this specification in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable sub combination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can, in some cases, be excised from the combination, and the claimed combination may be directed to a sub combination or variation of a sub combination.