DE4243481C1 - Wideband rectifier with lower cut=off frequency of several hundred Megahertz - has feedback circuit consisting of two series diodes between output and inverting input of operational amplifier, and adjustment resistor between diodes and non-inverting input - Google Patents

Abstract

The rectifier input (E) is connected via a coupling capacitor (C0) and a first resistance (R0) to the anode and cathode respectively of first (D1) and second (D2) diodes. The cathode of the first diode is connected to the positive input of an operational amplifier (OP12). The second diode's anode is earthed and a further capacitor (C12) couples the first diode's cathode and second diode's anode. A fifth diode (D5) is connected between the second diode's anode and a sixth diode's (D6) anode whose cathode is connected to the amplifier's feedback path. The series connected fifth and sixth diodes are forward biased by a dc voltage (Ub12). A parallel resistance (R12) connects the positive input of the operational amplifier and the feedback path and is in parallel with the capacitor and series diodes. The feedback path contains a third and fourth diode (D3,D4) in series and a feedback resistance (R34). The fourth diode's cathode is connected to the amplifier's negative input. The feedback resistance lies between the negative input and the parallel resistance. ADVANTAGE - Wideband rectifier enables optimisation of linearity and temp. characteristics using simple, cheap circuit. Improves low frequency characteristics of rectifier, and shifts lower effective break frequency by factor of two. Low power requirements.

Description

The invention relates to a broadband rectifier.

Broadband and energy-saving rectifiers, that work up to frequencies of a few hundred MHz currently only possible with passive diode circuits. It are often from the Villard circuit or Voltage cascade derived circuits used. Depending on Choice of the resistance ratio α = R1 / R0 can be good Approximations to real mean, RMS or Peak value rectifiers can be reached. Value often becomes on an approximation of behavior to a real one RMS rectifier. The well-known circuits are therefore also used as a rms effective rectifier designated.

The values for α given in the literature apply only assuming ideal diodes or very high ones Input levels. With real diodes and realistic Input levels must be α by simulation or experimental be optimized. The known circuits indicate when the circuitry complexity is kept low at real diodes and realistic input levels, independently of the chosen α, a strong temperature response as well as high Linearity error.  

From DE-PS 28 42 749 is a circuit arrangement for Temperature stable high frequency peak value rectification known, which in the preamble of claim 1 listed features, except for the use of only one Diode in the compensation circuit and the missing connection this diode with feedback branch of the Operational amplifier has.

From DE-AS 28 23 819 is a rectifier circuit for the measurement technology is known with a rectifier diode and one of the rectifier diode DC amplifier, that of the rectifier diode is connected downstream and its negative feedback branch two has negative feedback diodes. In addition, the Rectifier circuit in parallel to that on the Rectifier diodes following capacitor have a resistor. The circuit according to this citation shows none Connection of the compensation diodes with the Feedback branch of the operational amplifier.

In the electronics magazine 2 / 29.01.1982, pages 57 to 59 is a highly linear, broadband precision RF Demodulator shown which is a full-wave rectification using two operational amplifiers.

The invention has for its object a to create broadband rectifier, the one Optimization of linearity and temperature behavior made possible by a simple and cheap circuit.

The solution to this problem can be found in the Claims 1 or 2 specified features.

The basic idea of the invention is to use a as a reference signal To use DC voltage. The reference rectifier then only consists of a voltage divider from two diodes and a resistor. The output voltage of the reference rectifier is then the DC voltage minus two diode voltages. By varying the Resistance values can have different characteristics and Temperature responses can be set. It is expected that for a certain resistance value and a certain one Curve shape of the reference rectifier Rectifier similar characteristic and temperature response having. It has been found that this expectation with the accuracy requirements made in practice is fulfilled. The input voltage of the Reference rectifier therefore corresponds very well to that desired rectifier output voltage if both Show rectifier the same value. This equality can be forced that the Reference rectifier as a negative feedback network in one Operational amplifier is used.

The functioning of the invention in the above sense was through extensive simulation and measurements proven.

The invention also has the technical object based on the low frequency characteristics and that Decay behavior of the rectifier circuit after Claim 1 and similar circuits to improve.  

The solution to this problem is obtained by the in claim 2 specified characteristics. Measurements showed that the lower one effective cutoff frequency approximately by a factor of two is moved below.

The invention is described below with reference to the drawing illustrated embodiments explained in more detail.

It shows

Fig. 1 shows a simplified circuit of the quasi RMS value rectifier, with only one operational amplifier

Fig. 2 shows a quasi-rms rectifier with operational amplifiers arranged symmetrically to the input path.

The circuit arrangement according to FIG. 1 shows a broadband rectifiers whose input E is connected via a coupling capacitor C0 and a first resistor R0 connected to the anode of a first diode D1 and the cathode of a second diode D2 connected to the cathode of the first diode D1 to the positive Input of a single operational amplifier OP12 is connected, the output of which forms the measurement output. The second diode D2 is connected to ground with the anode and a further capacitor C12 is active between the cathode of the first diode D1 and the anode of the second diode D2. The anode of a fifth diode D5 is additionally connected to the anode of the second diode D2, the cathode of which is connected to the anode of a sixth diode D6, the cathode of which is connected to the feedback branch of the single operational amplifier OP12. The latter two diodes D5, D6, connected in series, are forward biased by a single DC voltage Ub12. In addition, a parallel resistor R12 arranged in parallel with said capacitor C12 and the series-connected diodes D5, D6 acts between the positive input of the single operational amplifier OP12 and the feedback branch. The feedback branch consists of a third diode D3, a fourth diode D4 and a feedback resistor R34, the latter two diodes D3, D4 being connected in series and with their anodes facing the output of the single operational amplifier OP12, while the cathode of the fourth diode D4 is connected to the negative input of said operational amplifier OP12 and the feedback resistor R34 lies between the negative input of the operational amplifier OP12 and the parallel resistor R12.

The circuit in FIG. 1 is the simplified version of the broadband rectifier of FIG. 2. If R0, R56 and Ub12 are set to zero and the amplifier OP12 with its negative feedback network is omitted, the circuit is identical to the known voltage cascade. If one neglects the alternating components and the frequency response at low frequencies and is only interested in the steady state, the version of FIG. 1 is identical to the version shown in FIG. 2. The following dimensions are assumed: Ub12 = 2 _* Ub1, R56 = 2 _* R5, C12 = C1 / 2, R12 = 2 _* R1 and R34 = 2 _* R3. The capacitor C12 can only be charged with positive input voltages. The result of this is that a greater alternating component of the output voltage and a greater drop in the frequency response can be expected at low frequencies than in the version of FIG. 2. If the discharge behavior is not to be deteriorated, C0 must be selected to be somewhat smaller than C12. The cut-off frequency of the high pass, formed from C0 and the effective input resistance of the rectifier, is then so high that the frequency response is additionally deteriorated at low frequencies. If one chose C0 larger than C12, the discharge behavior of the circuit would be significantly deteriorated. If the voltage at C12 drops below half the output value, C0 is also discharged via R12. In FIG. 2, C0 can be omitted if the input signal can be provided with sufficiently low blocking DC voltage. If DC voltages have to be blocked, the connection of C0, which is connected to the diodes, remains free of DC voltages as long as the AC component of the input signal has a symmetrical amplitude distribution. This means that C0 can be chosen to be very large in the version of FIG. 2 without impairing the properties of the circuit. The additional component complexity of the version of FIG. 2 is essentially limited to the two operational amplifiers and some resistors. Very slow amplifiers with good DC characteristics can be used. Such operational amplifiers are cheap and available with low power consumption. The version of Fig. 1 should therefore only be used when there is a lack of space.

Because the capacitor C12 only turns on during positive half-waves can be loaded is the frequency response at low frequencies zen uand the suppression of alternating components at low Frequencies not optimal.

The capacitor C0 is essential for the function of the circuit necessary. If it is chosen too small, it deteriorates the frequency response at low frequencies. If it gets too big chosen, the swing-out behavior deteriorates. To Discharge from C12 to about half the initial value becomes D1 conductive. Now C0 is also discharged via R12. The end So the oscillation process is strongly non-linear and slow.

The circuit arrangement shown in FIG. 2 shows a broadband rectifier, the input E of which is connected via a coupling capacitor C0 and a resistor R0 on the one hand to the anode of a first diode D1 and on the other hand to the cathode of a second diode D2. The cathode of the first diode D1 is connected to the positive input of a first operational amplifier OP1 and the anode of the second diode D2 to the positive input of a second operational amplifier OP2. The positive input of the first operational amplifier OP1 is connected to ground via a first capacitor C1 and is connected via a first resistor R1 to the associated feedback network, consisting of a third diode D3 and a third resistor R3. A fifth diode D5 is connected to ground with its anode and is connected with its cathode to the feedback network of the first operational amplifier OP1. The diode D5 is biased in the forward direction by means of a first DC voltage source Ub1. The positive input of the second operational amplifier OP2 is connected to ground via a second capacitor C2 and is connected via a second resistor R2 to the feedback network consisting of a fourth diode D4 and a fourth resistor R4 of the second operational amplifier OP2. The anode of a sixth diode D6 is also connected to the feedback network of the second operational amplifier OP2 and is connected to ground with its cathode and is forward-biased by means of a second DC voltage Ub2. The output of the first operational amplifier OP1 is connected via a seventh resistor R7 to the positive input of a third operational amplifier OP3 and the output of the second operational amplifier OP2 is connected via an eighth resistor R8 to the negative input of the third operational amplifier OP3. The output of the third operational amplifier OP3 represents the measuring output, which is fed back via a tenth resistor R10 to its negative input, while its positive input is connected to ground via a ninth resistor R9.

The following dimensioning is assumed for the broadband rectifier in FIG. 2: Ub1 = Ub2, R5 = R6, D5 = D6, C1 = C2, R1 = R2, D1 = D2 = D3 = D4, R3 = R4 and R7 = R8 = R9 = R10. If R0 is set to zero, the circuit behaves similarly to an ideal peak value rectifier at high input levels. If the magnitude of the input voltage is greater than the voltage across the capacitors C1 or C2, one of the two diodes D1 or D2 becomes very low-resistance. The capacitors C1 and C2 are therefore charged to the peak values very quickly. The capacitors are discharged relatively slowly via R1 and R2. If R0 is greater than R1, the charge time constant has approximately the same value as the discharge time constant. The circuit behaves similar to an average rectifier. By changing the resistance value of R0 it is possible to choose between the extreme peak value and mean value rectifiers. If R0 is between zero and R1, a continuum of intermediate solutions is possible. The output level of the RMS rectifier is between the output levels of the other two rectifiers for all waveforms. It is therefore to be expected that an R0 can be found in which the circuit is a good approximation for an RMS rectifier.

The operational amplifiers OP1 and OP2 with their Negative feedback networks R3, D3, R4 and D4 are on their Outputs the voltages at the capacitors C1 and C2 around a diode voltage offset is available. This will the voltage drops across diodes D1 and D2 in the first Approximation compensated. By changing the Resistance ratio R3 / R1 can change the temperature response and the characteristic curve of the circuit can be optimized. The Provide bias generator Ub1, R5, D5, Ub2, R6 and D6 at very low input levels, that the diodes D1 to A small direct current flows through D4. If one were to connect R1 to R4 directly to ground, would be at small input levels, the diodes D2 and D4 have a very high resistance and thus the circuit gain very high. Offset voltages of the operational amplifiers would therefore be small input levels cause much larger errors than at high input levels. Another advantage of Bias generation is that the diode voltages of D1 and D2 cannot be built up by the input signal have to. This causes a rough equalization of the characteristic curve  (Approach to ideal diodes) reached. The resistors R7 to R11 and the operational amplifier OP3 form one Differential amplifier that the results of the two One-way rectifier summarized.

The peak-to-peak voltage of the disturbing alternating components decreases by a factor of three. These results apply to the same swing-out time constant t _a for both circuits (t _a = R1 _* C1 = R2 _* C2 = R12 _* C12).

The capacitor C0 can be omitted (short-circuited) or be chosen very large, without the decay behavior the circuit deteriorate since it is free of DC voltages remain. The swing-out behavior is now linear and faster.

The term symmetrical square wave signal is used below. A symmetrical square wave signal is to be understood as a signal that has the voltage rate c _{f *} u _eff (c _f : crest factor) and - c _{f *} u _eff over the same length of time. The rest of the time the signal has the voltage value 0 volt.

A rough estimate for the ratio α = R1 / R0, at which a good approximation to the real RMS rectifier is achieved, will now be given. The symmetrical rectangle with variable crest factor c _{f is} used as the test signal. All diodes are ideal, which means that with negative diode voltages no diode current flows and when a positive diode current flows the diode voltage is zero. The output voltage u _{a of} the quasi-effective value rectifier is described under these conditions by equation 1.

With α = ∞ u _a indicates twice the peak voltage of the input signal. The mean value _* α is displayed with α «2 _* c _f . If one chooses α = 2 _* c _max , the dependency of the output voltage u _a on c _{f is} minimal in the range 1 <c _f <c _max . With c _f = c _max ^1/2 , the maximum value of u _a = u _max results. With c _f = 1 and c _f = c _max , the minimum values of u _a = u _{min result} .

With the setting of α = 8 the following applies: c _max = 4; u _min = 1.6 _* u _eff and u _max = 2 _* u _eff . For the ratio v = u _max / u _min to be minimized, v = 1.25 results. The larger c _{max is} chosen, the greater the ratio v.

In the case of the mean value and peak value rectifier, the symmetrical square wave signal v = c _max . The output voltage of the quasi-effective value rectifier is therefore somewhat more independent of the crest factor than for the other two rectifiers. At c _max = 4, there are maximum level changes of 1.94 dB and 12.04 dB due to a change in the crest factor.

With cosine-shaped modulation of the idealized Equation 2 effective RMS rectifier applies equation 3 is used.

Equation 3 cannot be solved according to u _a , but it can be solved numerically by the Newton-Raphson zeroing method. For α = 8, u _a = 1.7143 _* u _eff .

So far we have limited ourselves to signals with symmetrical amplitude distribution. At this point, we want to use the example of the asymmetrical and DC-free square wave signal to investigate how the idealized effective rms rectifier behaves with asymmetrical signals. Our test signal has the amplitude of c _{f *} u _eff for the duration of T _p / (1 + cf²) and the amplitude of u _eff / c _f for the rest of the period Tp. Equation 4 describes the relationships in both versions of the idealized rms rectifier. Equation 5 applies if the capacitor C0 is short-circuited in the version of FIG. 2.

Tables 1 and 2 summarize the results of Equations 1 through 5. The idealized RMS rectifier can be optimized for a crest factor range of 1 to c _fmax using the resistance ratio α = R1 / R0. The parameter α is to be set to twice the value of c _fmax . For c _fmax = 4, a level error between -1.16 dB and +1.34 dB depends on the curve shape. For c _fmax = 10 ^1/2 , the errors are between -0.96 dB and +0.97 dB.

Table 1

Curve shape dependence of the idealized quasi-effective rectifier with α = 8.

The character "*" indicates the maximum of the respective function via c _f .

Table 2

Curve shape dependency of the idealized effective effective rectifier with α = 2 * 10 ^1/2 .

The character "*" indicates the maximum of the respective function via c _f .

In the circuit of Fig. 1, C12 can only be charged during the positive signal peaks. The frequency response and the suppression of alternating components at low frequencies is correspondingly poor. The basic idea of the invention is to make charging of C12 possible even in the case of negative signal peaks. The goal is achieved by dividing the rectifier into two equivalent rectifiers of different polarity and by subtracting the two results. The two errors mentioned are roughly halved if t _{a is} kept constant.

Since the decoupling capacitor C0 is no longer required or remains DC-free in the circuit according to FIG. 2, the following negative effect of FIG. 1 is omitted . If a value of half the steady-state voltage is reached during decay, C0 discharges together with C12 via R12. If the frequency response error at low frequencies caused by the high pass, consisting of C0 and the effective input resistance of the rectifier, should be negligibly small compared to the frequency response error of the actual rectifier, C0 must be selected greater than C12. The decay behavior is therefore non-linear and slower than in FIG. 2.

Using an example, the dimensioning of the circuit using the QEFF program is to be demonstrated. The Qeff program is a simulation program for the circuits of FIGS. 1 and Fig. 2. We want to develop a quasi RMS value rectifier, which is used with a measuring instrument that can change the input gain in 5 dB steps. To avoid frequent switching at the range limits, the automatic divider should have a hysteresis of ± 2.5 dB. The level range of interest of the rectifier is therefore 10 dB. The amplifier that drives the rectifier should be able to deliver a peak voltage of ± 8 V. The maximum permissible crest factor is 3.16, so that the rectifier's working range is from 0 dB to +10 dB. The -3 dB bandwidth should be 320 MHz. If the HP-5082-2835 diodes, which have a maximum capacitance of 1 pF, are used, R0 can be estimated at 249 ohms. According to Section 3, this results in an R1 of 1575 ohms for the idealized rms effective rectifier. Since the differential resistance and the series resistance RS of the diodes increase the effective value of R0, R1 must be chosen somewhat larger than 6.32 _* R0. The resistor R3 serves to optimize the linearity and the temperature dependence of the circuit. Experience has shown that it must be selected larger than R0 and smaller than R1. The resistor R5 and the auxiliary voltage Ub must be selected so that the diodes D5 and D6 are still safely operated in the pass band even at the highest levels.

The optimization takes place in three steps. after the Components have been preset as described above, R2 is set so that at normal temperature the maximum positive and maximum negative level errors den have the same amount. The program is on "Curve shape dependency only at normal temperature" set.

The optimization of R3 can be done without observing the Dependence on the curve shape. A program run takes less than 30 s here, since only the sinusoidal Signal is used.

The last stage of optimization should only be one Be in control at all temperatures and waveforms. If necessary you have to change R2 slightly and R3 to adjust.

Table 3 shows the result for the example. The maximum level errors are almost symmetrical with +1.02 and -0.98 dB. The ratio of R1 / R0 is therefore almost optimal. The linearity and temperature dependence for sine tones is also optimized and negligible. The power consumption of the rectifier circuit according to FIG. 2 is approximately 130 mW if power-saving operational amplifiers with an operating voltage of ± 5 V are used. If the bias voltage is taken from an operating voltage of 1 V instead of 5 V (for example by DC-DC conversion from the operating voltage for the operational amplifiers), the power consumption can be reduced to approximately 70 mW.

This example shows that the Characteristic equalization and temperature compensation through D3 and R3 works very well. An alternative would be following, occasionally used method. A medium-frequency sine wave whose amplitude is linear is dependent on a control voltage, is on one given identically constructed rectifier. The Control voltage is a measure of the rms value when both Rectifiers have the same output voltage, what can be achieved by a simple control loop. This procedure can only achieve that Linearity and temperature dependence for sinusoidal signals is optimal, the linearity and temperature dependence the other waveforms are the same as the one here used method that is much easier and cheaper is to be realized.

The circuit of this example was constructed in accordance with FIG. 2 with short-circuited C0 and checked for agreement with the simulation with regard to the characteristic with sinusoidal modulation and curve shape dependence with two-tone and noise. The simulation and measurement results agree very well in terms of measurement accuracy and component tolerances. The temperature response has already been successfully checked for several other dimensions.

The frequency response drop of the example circuit up to 100 MHz is less than 0.2 dB and corresponds to the frequency response of a first order low-pass filter with a cut-off frequency of approximately 460 MHz. The frequency response at low frequencies corresponds approximately to that of a first-order high-pass filter with the cut-off frequency f _gu = 2.5 (2 · π · t _a ). The alternating components are approximately 1.5 times larger than in a true RMS rectifier with the same decay time constant t _a .

The low-frequency properties were also determined in the circuit according to FIG. 1. The f _gu is slightly less than twice as high as in the circuit according to FIG. 2 if C0 = 5 C1 is selected in the circuit according to FIG. 1. Note that the circuit of Fig. 1 has very poor timing with this value of C0; If C0 is made significantly smaller, the frequency response deteriorates, since then the high-pass filter from C0 becomes relevant to the effective input resistance R _{eff of} the rectifier (R _eff approximately R1 / 2.5). The peak-to-peak voltage of the alternating components is approximately three times greater in the circuit according to FIG. 1 than in the circuit according to FIG. 2

Table 3

Printout of the QEFF program

Claims (3)

1. Broadband rectifier, the input (E) of which is connected via a coupler capacitor (C0) and a first resistor (R0) to the anode of a first diode (D1) and the cathode of a second diode (D2), with the following features:

• the cathode of the first diode (D1) is connected to the positive input of a single operational amplifier (OP12), the output of which forms the measurement output,
• the second diode (D2) is connected to ground with the anode, and a further capacitor (C12) is active between the cathode of the first diode (D1) and the anode of the second diode (D2),
• - At the anode of the second diode (D2) the anode of a fifth diode (D5) is additionally connected, the cathode of which is connected to the anode of a sixth diode (D6), the cathode of which is connected to the feedback branch of the single operational amplifier (OP12) stands, and
• the two latter diodes (D5, D6) connected in series are biased in the forward direction by means of a single direct voltage (Ub12),

characterized by

• that between the positive input of the single operational amplifier (OP12) and the feedback branch a parallel resistor (R12) arranged in parallel with the named capacitor (C12) and the series-connected diodes (D5, D6) is effective,
• - The feedback branch consists of a third diode (D3), a fourth diode (D4) and a feedback resistor (R34),
• - That the two latter diodes (D3, D4) are connected in series and with their anodes facing the output of the single operational amplifier (OP12),
• - That the cathode of the fourth diode (D4) is connected to the negative input of said operational amplifier (OP12), and
• - That the feedback resistor (R34) is between the negative input of the operational amplifier (OP12) and the parallel resistor (R12).

2. Broadband rectifier, the input (E) of which is connected via a coupling capacitor (C0) and a resistor (R0) on the one hand to the anode of a first diode (D1) and on the other hand to the cathode of a second diode (D2) with the following features :

• the cathode of the first diode (D1) is connected to the positive input of a first operational amplifier (OP1) and the anode of the second diode (D2) is connected to the positive input of a second operational amplifier (OP2),
• - The positive input of the first operational amplifier (OP1) is connected to ground via a first capacitor (C1) and is connected via a first resistor (R1) to a feedback network belonging to the first operational amplifier, consisting of a third diode (D3) and a third resistor (R3), connected,
• a fifth diode (D5), the anode of which is connected to ground and the cathode of which is connected to the feedback network of the first operational amplifier (OP1), is biased in the forward direction by means of a first DC voltage source (Ub1),
• - The positive input of the second operational amplifier (OP2) is connected to ground via a second capacitor (C2) and is connected via a second resistor (R2) to a feedback network belonging to the second operational amplifier, consisting of a fourth diode (D4) and a fourth resistor ( R4) of the second operational amplifier (OP2); connected,
• a sixth diode (D6), the cathode of which is connected to ground and the anode of which is connected to the feedback network of the second operational amplifier (OP2), is forward-biased in the forward direction by means of a second direct voltage (Ub2),
• the output of the first operational amplifier (OP1) is connected via a seventh resistor (R7) to the positive input of a third operational amplifier (OP3),
• the output of the second operational amplifier (OP2) is connected to the negative input of the third operational amplifier (OP3) via an eighth resistor (R8),
• - The output of the third operational amplifier (OP3) represents the measurement output and is fed back via a tenth resistor (R10) to its negative input, while its positive input is connected to ground via a ninth resistor (R9).