Peak Detector

1. Field of the Invention

The present invention relates to peak voltage detectors, and more particularly, to peak detectors wherein high accuracy and minimum susceptibility to noise are desired.

2. Description of the Related Art

FIG. 1 shows a typical prior art peak voltage detector 10 which consists of a comparator 11, a charge pump 12 and a hold capacitor C.sub.h. When V.sub.in is greater than V.sub.out, the noninverting output of the comparator will be high and the inverting output will be low. This turns transistor Q1 off and transistor Q2 on. Current I1 thus flows through transistors M1 and Q2, with the current in transistor M1 being mirrored by transistor M2 as current I.sub.ch. While transistor Q2 is on, I.sub.ch flows into capacitor C.sub.h to charge C.sub.h. When V.sub.in is less than V.sub.out, the inverting output of the comparator will be high and the noninverting output will be low. Transistor Q1 is now on and transistor Q2 is off. Now current I1 will flow directly from V.sub.cc through transistor Q1. The current in transistor Q2 and thus the current I.sub.ch will be zero, and capacitor C.sub.h will hold the previous voltage.

FIG. 2 shows the voltage and current waveforms for the circuit of FIG. 1. One of the disadvantages of the prior art described above is that the output voltage V.sub.out is almost always higher than the actual peak value of V.sub.in. This is because comparator 11 cannot respond instantaneously to its inputs. As a result, even after V.sub.out becomes greater than V.sub.in, I.sub.ch stays on for a finite time delay (d), as shown in FIG. 2. During the finite time delay (d), capacitor C.sub.h continues to be charged. This results in an error voltage V.sub.e1. The amount of overshoot represented by V.sub.e1 depends on the magnitude of I1, the delay d and the size of capacitor C.sub.h.

Another disadvantage of the prior art is that the topology shown in FIG. 1 is highly susceptible to noise spikes occurring near the peaks of V.sub.in after the true peak has been captured by capacitor C.sub.h. FIG. 3 presents the voltage waveform of V.sub.in with a voltage spike V.sub.spike occurring near the peak of V.sub.in and the voltage waveform of V.sub.out. FIG. 3 also shows the resulting current pulse 26 developed in I.sub.ch due to the voltage spike V.sub.spike. Current pulse 26 further charges capacitor C.sub.h and creates an error voltage V.sub.e2 in the output voltage V.sub.out.

To overcome the drawbacks of the prior art described above, the present invention provides a current reduction circuit and a small leakage current to reduce the amount of error in the output voltage occurring due to the noise spikes in the input voltage and the characteristics of a non-ideal comparator.

The present invention provides methods and apparatus for detecting a peak voltage with high accuracy and minimum susceptibility to noise.

The peak voltage detector according to the present invention includes an input terminal and an output terminal, a comparator for comparing the input terminal voltage and the output terminal voltage, the comparator having the input terminal voltage and the output terminal voltage as inputs and first and second complementary comparator output voltages as outputs, a hold capacitor coupled to the output terminal, and a charge pump coupled between the comparator and the hold capacitor.

The charge pump of the present invention has a differential input pair for receiving the first and second comparator output voltages as inputs, a current mirror for mirroring a current in the differential input pair to the charge current, a current reduction circuit for reducing the charge current that charges the hold capacitor responsive to inputs from the input and output terminals, and a current source coupled between the current reduction circuit and ground.

To detect the peak of the input terminal voltage, the peak voltage detector of the present invention compares the input terminal voltage to the output terminal voltage, and charges the output terminal when the input terminal voltage is greater than the output terminal voltage until the output voltage is substantially equal to the input terminal voltage. As the output terminal voltage approaches the input terminal voltage, the level of the charge current that charges the hold capacitor is substantially and smoothly reduced for increasing output terminal voltages, thus minimizing the error in the acquired output voltage due to the non-ideal comparator and noise at the input terminal.

To reduce the error in the output voltage further, the peak voltage detector of the present invention may also have a leakage current coupled between the output terminal and ground.

FIG. 1 is a circuit diagram of a prior art peak voltage detector.

FIG. 2 presents the waveforms of the input and output voltages and the respective charge current of the peak voltage detector shown in FIG. 1.

FIG. 3 presents the waveforms of the input voltage with a voltage spike and the corresponding output voltage and the current induced by the voltage spike of the peak voltage detector shown in FIG. 1.

FIG. 4 is a circuit diagram of a peak voltage detector according the present invention, the

FIG. 5 is the waveform of the current through the current reduction circuit as a function of the difference between the input and output voltages of the circuit of FIG. 4.

FIG. 6 presents the waveforms of the input and output voltages and associated charge current of the peak voltage detector shown in FIG. 4.

FIG. 7A presents the waveform of the input voltage with a voltage spike and the comparison of the corresponding output voltage waveforms according to the prior art and the present invention.

FIG. 7B presents the waveforms of the charge current due to the voltage spike in FIG. 7A according to the prior art and the present invention.

FIG. 8 is an alternate circuit implementation for the circuit of FIG. 4 which accommodates low supply voltage operation.

FIG. 4 shows a peak voltage detector 40 according to one embodiment of the present invention, having a current reduction circuit 48 for reducing the charge current I.sub.ch ' as the difference voltage V.sub.in '-V.sub.out ' becomes smaller. Peak voltage detector 40 includes a comparator 41, a charge pump 42 and a charge capacitor C.sub.h '.

Comparator 41 compares V.sub.in ' and V.sub.out ', with each of the complementary outputs thereof provided to the respective transistor of differential pair 49 in charge pump 42. The inputs V.sub.in ' and V.sub.out ' to comparator 41 are also provided to current reduction circuit 48 within charge pump 42.

Charge pump 42 includes current reduction circuit 48, differential input pair 49 consisting of transistors Q11 and Q12, and a current mirror 50. If V.sub.in ' is greater than V.sub.out ', then transistor Q11 will be off and transistor Q12 will be on. The fraction of the current I20 which will flow through transistors Q12 and M21 to be mirrored to transistor M22 as the current I.sub.ch ' will depend, however, on the current reduction circuit 48. If, on the other hand, V.sub.in ' is smaller than V.sub.out ', then transistor Q11 will be on and transistor Q12 will be off, turning off I.sub.ch ' in any event. In this case capacitor C.sub.h ' will retain its previous charge (voltage).

Current reduction circuit 48 includes identical transistors Q13A, Q13B and Q14. When V.sub.in ' is substantially greater than V.sub.out ' during the peaks of V.sub.in ', the base of transistor Q14 will be at a substantially higher voltage than the bases of transistors 13A and 13B, turning on transistor Q14 and turning off transistors Q13A and Q13B. Also transistor Q12 will be on and transistor Q11 will be off. Consequently the current 120 will flow through transistors Q12 and M21, and be mirrored to transistor M22 as charging current I.sub.ch '.

As V.sub.out ' increases to approach the peaks of V.sub.in ', the voltage on the bases of transistors Q13A and Q13B during the peaks will approach the voltage on the base of transistor 14, starting to turn on transistors Q13A and Q13B. Now the voltage of the common emitter connection of transistors Q13A, Q13B and Q14 will adjust to reduce the current in transistor Q14 below I20, so that the current I20 divides between transistors Q13A, Q13B and Q14. The part of I20 that flows through transistors Q13A and Q13B is not mirrored to the charging current I.sub.ch ', with the net result that the charging current I.sub.ch 'will be less than I20 by an amount dependent on the relative values of V.sub.in ' and V.sub.out ' and the number of transistors Q13, as more or fewer can be used.

When V.sub.out ' increases to nearly equal the peaks of V.sub.in ', the voltage on the bases of transistors Q13A and Q13B during the peaks will nearly equal the voltage on the base of transistor Q14. Now the current I20 divides nearly equally among transistors Q13A, Q13B and Q14, each conducting a third of I20. Again the part of I20 that flows through transistors Q13A and Q13B is not mirrored to the charging current I.sub.ch ', with the net result that the charging current I.sub.ch ' will be approximately one third of I20. (or I20 divided by N+1, where N is the number of transistors Q13 if more or fewer are used). Thus as shown in FIG. 5, the current I.sub.Q14 in transistor Q14 becomes smaller as the difference voltage V.sub.in '-V.sub.out ' becomes smaller in region 50. As a result, as V.sub.in '-V.sub.out ' gets smaller, I.sub.ch ' becomes smaller, charging capacitor C.sub.h ' at a lower rate. Hence, the present invention allows V.sub.out ' to "soft land" to its final value.

If V.sub.out ' exceeds the peaks of V.sub.in ' (off peak or due to prior noise at a peak of V.sub.in '), the comparator will begin to shut off transistor Q12 and turn on transistor Q11, so that the still further reduced current in transistor Q14 (region 51 of FIG. 5) will flow through transistor Q11, not transistor Q12, thereby terminating the charging current I.sub.ch '.

FIG. 6 presents the voltage waveforms of V.sub.in ' and V.sub.out ', and the associated current waveform of I.sub.ch '. Because of current reduction circuit 48, Ich' becomes smaller as the value of V.sub.out ' becomes closer to the value of V.sub.in '. Since comparator 41 is not ideal, there is still a certain amount of delay d', but during the delay period d', because the charging current I.sub.ch ' is substantially smaller than the charging current I.sub.ch in the prior art of FIG. 2, the error voltage V.sub.e3 created in the present invention is much smaller than the error voltage V.sub.e1 of the prior art.

FIGS. 7A and 7B present a comparison between the error voltage of the present invention and the error voltage of the prior art created by noise in the input voltage. FIG. 7A presents the voltage waveform of an input voltage V.sub.in ' having a V.sub.spike ' occurring near the peak of V.sub.in ', and the corresponding voltage wave forms of V.sub. out of the prior art and V.sub.out ' of the present invention. FIG. 7B presents the resulting prior art current pulse 54 and the present invention's current pulse 56 created by V.sub.spike '. In the prior art, after the true peak of the input voltage is captured, current pulse 54 due to V.sub.spike ' is at the full amplitude of current source I1. However, in the present invention, after the true peak of the input voltage is captured, since the value of V.sub.out ' is close to the value of V.sub.in ', current pulse 56 due to V.sub.spike ' is small. Current pulse 56 is substantially less than the level of the current source (I1 or I20), as shown in FIG. 7B. Thus, the resulting error voltage V.sub.e4 of the present invention is much less than error voltage V.sub.e2 of the prior art for a given noise level in the input wave V.sub.in '.

A straight-forward implementation of the circuit of FIG. 4 does not allow operation of the circuit at less than 3 volts. In particular, current I20 in FIG. 4 flows through the differential amplifier comprised of bipolar transistors Q13A, Q13B and Q14. To achieve stable operation, the current source I20 will require at least 0.5 v across it, and the Vce of Q14 will also require 0.5 V or more. Also the base voltages of Q11 and Q12 will dictate the voltage drops allowed across Q14 and the current source I20. Consequently the design of the peak voltage detector in the form shown in FIG. 4 for low supply voltage operation becomes very difficult. However, FIG. 8 shows an alternate circuit implementation for the circuit of FIG. 4 which accommodates low supply voltage operation.

In the embodiment of FIG. 8, transistors Q25, Q26A, Q26B, Q27, Q30, Q31A and Q31B are identical, and the resistors R21, R22, R23A, R23B and R24 are the same size. The current of current source I30 flows through transistor Q30 and resistor R24, setting the base voltage of transistor Q30 and thus the base voltage of transistors Q31A, Q31B and Q25. This in turn sets the current through transistors Q31A and Q31B, though the division between transistors Q28 and Q29 of the current through transistors Q31A and Q31B depends upon the difference between the output voltage VOUT" and the input voltage VIN". In particular, in the region of peaks, and when the output voltage is initially considerably lower than the input voltage, transistor Q28 will be substantially on and transistor Q29 will be substantially off. With transistor Q29 off, transistor M23 does not mirror any current to transistor M24 and accordingly transistors Q26A and Q26B will be off. Thus, the current IQ25 through transistor Q25 and resistor R22 will equal the current through each of transistors Q31 and the associated resistor R23. Since the positive input to the comparator 50 is higher than the negative input, the output of the comparator holds transistor Q22 on and holds transistor Q21 off. Consequently, the current IQ25 through transistor Q25 will also flow through transistors Q22 and M21, to be mirrored to transistor M22-to provide the charging current I.sub.ch " to charge the capacitor CH" to drive the output VOUT" toward the intended peak value.

As VOUT" approaches the voltage of the peak of the input voltage VIN", Q29 will begin to turn on, with the voltage of the common emitter connection of transistors Q28 and Q29 readjusting to reduce the current through transistor Q28. The current through Q29 is mirrored by transistors M23 and M24 to transistor Q27 which in turn is mirrored to transistors Q26A and Q26B. This current, which flows through resistor R22 as one of the current components therein, provides part of the voltage drop across resistor R22. Since the base voltage of transistor Q25 is set by the current source I30 to be equal to the VBE of transistor Q30 plus the voltage drop across resistor R24, and the emitter voltage of transistor Q25 is 1 VBE below the base voltage of transistor Q25, the voltage across R22 and thus the total current through resistor R22 is set. Consequently the larger the current component through transistors Q26A and Q26B, the smaller the current component IQ25 through transistor Q25, which of course so long as VOUT" is less than VIN" (Q22 on and Q21 off), will flow through Q22. This current will be mirrored by transistors M21 and M22 to provide the output charging current I.sub.ch ", thereby providing the same soft landing at the peak voltage as hereinbefore described with respect to FIGS. 4 and 5. Of course when VOUT" exceeds VIN", even at the peaks of VIN", Q22 will remain off even during the peaks in the input voltage, and transistor Q22 will be off and transistor Q21 will be on, so that any current flow IQ25 through transistor Q25 will be directly from VCC through transistor Q21 and not through transistor Q22. Thus no current component in transistor Q22 will be mirrored by transistors M21 and M22 to the output. In the circuit of FIG. 8, since the current through transistor Q25 is equivalent to the current IQ14 of FIG. 4, the need for the standoff voltage for either Q14 or I20 of FIG. 4 is eliminated. The voltage across transistor Q25 still needs to be 0.5 volt.

To avoid the accumulation of the effects of noise at and near the peak of the input signal, a small leakage current is provided to capacitor C.sub.h. In particular, as shown in FIGS. 4 and 8, the output of the peak detector provided to the holding capacitor C.sub. is also connected through a current source to ground so as to provide the small leakage current I.sub.1 constantly discharging the capacitor. This leakage current may be provided by a resistor to ground, or by way of an active current source, the design of which are well known. In any event, the leakage current avoids the accumulation of the effects of noise spikes at or near the peaks which could create inaccurately high peak detection values. While the leakage current could, depending on its magnitude, decrease the detected peak below that actually present, such a decrease at most would be only a small percentage of the total signal and would be uniform servo burst to servo burst, and therefore would not affect the relative amplitude measurements.

While the present invention has been particularly described with reference to the various figures, it should be understood that the figures are for illustration only and should not be taken as limiting the scope of the invention. Many changes and modifications may be made to the invention, by one having ordinary skill in the art, without departing from the spirit and scope of the invention as disclosed herein.