|
Description | |
Electrical schematics | |
Prototype measurements |
 |
Ritorna alla pagina iniziale |  |
Return to English homepage |
RF peak detector
I admit it from the beginning: I've not designed the circuit discussed here, but I've found it in a paper printed in QST magazine, january 19871 (seei references); but I've found it gives exceptional performance, despite to its simplicity that makes it exceptionally reliable too.
Everyone knows the simple peak detector that uses a diode and a capacitor, shown in figure 1; the main drawback in the performance of this very simple detector is that, due to the non linear diode V-I relationship, the output voltage level of this circuit is close to the input peak voltage value only for relatively high input signals: 300 - 500 mV for germanium or hot-carrier diodes, up to 700 mV - 1 V for silicon detector diodes; for lower signals at the input, the output voltage level inesorabily falls down to a mere percentage of peak input voltage.
In fact, in the circuit shown in figure 1 the output is related to the input by the following relationship:
Vout = max(Vin - Vd).
where Vd is the voltage drop across the diode that, unfortunately, "eats" most of the input signal, when the latter is a low level one, leaving almost nothing across the output load resistor R2.
Despite its limits, this detector fulfills requirements in many applications, but in some istances a better sensitivity is needed, for example in field strenght meters, RF millivoltmeters, low power wattmeters, etc.; usually the approach in such cases consists in preceeding the detector by a suitable RF amplifier, but this leads to design challenges in satisfying requirements of bandwidth, dinamic response, stability and many others.
To solve the low sensitivity problem, the output signal should be picked-up before the diode itself, but obviously this could not work, but...
The idea consist in using another diode (D2), same as the detector (D1), where a current flows that is the same magnitude of the peak current flowing in the detector diode (D1), such that the diode voltage drop across D2 matches as close as possible the drop across D1.
With reference to figure 2, we can see the same figure 1 detector but with U1, D2 ed R3 added to.
Let us consider for a while what happens if the input Vin is a DC voltage: in the resistor R3 flows all the current (ID2) flowing in the comparing diode D2, while in R2 flows all the current (ID1) coming through the detector diode D1 (we can ignore capacitor C1 since Vin is a DC level); the voltage drops on R1 and R2 are given by:
Vdet = ID1 x R2 (voltage drop on R2) = V+ (voltage at U1 non inverting input),
Vfb = ID2 x R3 (voltage drop on R3) = V- (voltage at U1 inverting input).
The operational amplifier (U1) compares these voltages and gives its output voltage:
Vo = AV x (V+ - V-), (where AV is the voltage differential gain of the op-amp).
Vo that is applied to D2, controls ID2 and consequentely Vfb = V- ; the negative feedback tends to keep V- = V+ = Vdet .
Therefore Vfb = Vdet and, provided that the diodes are the same and their currents matches (ID1 = ID2), then their voltage drops are the same (VD1 = VD2); therefore:
Vin = VD1 + Vdet = VD2 + Vfb = Vout .
As final result, thanks to U1 that equalizes the currents flowing into the two diodes, the output voltage equals the input voltage.
If this is the case, as it should be normally, then must be considered that in the resistor R2 flows an average current that is lower than the peak current passing through D1; therefore the ratio R2:R3 has to be choosen such that the DC current flowing through D2 equals the peak current passing through D1.
Grebenkemper in his original paper written in 19871, suggests an optimal ratio of 5:1 (R2 = 1 Mohm, R3 = 180 kohm), explaining that a lower ratio reduces circuit sensitivity at low input levels, while an higher one introduces a gain peaking that gives the circuit a greater than unity gain for some input levels range. Later in 1993 the author gives some corrections3 to his original design, among them the change of R3 value from 180 to 100 kohm, leading to a R2:R3 ratio of 10:1 .
In my application I opted for a ratio R2:R3 = 10:1, getting the results reported in the gain measurements.
In the complete electrical schematics, the input load resistor (R1) value is not specified, since it is dependent by the detector load impedance wanted for a specific application; in a RF low power wattmeter a non inductive 50 ohm resistor with adequate power rating will probably be choosen: it will be the dummy load for the device under test; for applications such a detector in a RF millivoltmeter, this resistor will be 1 Mohm or it will be eliminated at all.
In this application and others similar high impedance application, it may be desiderable to have the input DC decoupled; this can be done by putting a capacitor before the detector diode (D1); take into account the input signal reduction due to the capacitive divider formed by C1 and this added capacitor.
About C1, its value depends on the lowest frequency that must be detected: my application uses only 1 MHz signals.
The complete detector comprises circuits not yet described; they are:
A low-pass filter (R3, C2) between the diode detector and the op-amp non inverting input: also in this case component values may be tailored to the frequency range expected for the input signals.
Polarization circuit (D3, D4, R6, C5, C6): it puts signal COMMON line at 0,7 V above supply ground; this is necessary to make possible the offset zeroing of the op-amp: in practice this is equivalent to a three voltages supply circuit: +14,3 V , +0,7 V e -0,7 V.
Op-amp offset compensation circuit (RV1, R4, R5); trimmer RV1, ranging between +0,7 V e -0,7 V with respect to the COMMON line, controls the amount of voltage across R4, that sums to the voltage across R7 (corresponding to R3 in figure 2) allowing for both positive and negative offset correction at the op-amp input: the multi turn trimmer RV1 has to be trimmed for a zero output voltage when no input voltage is applied.
Finally capacitor C3 improves op-amp stability, by reducing its gain at high frequencies, while resistor R8 is the circuit output load.
It should be noted that, starting from R2 on, all the rest of the circuit - included the operational amplifier - works at DC: therefore it is not necessary to use high speed components and, if well assembled, this circuit can be used to detect signals at very high frequencies: 500 MHz and beyond.
About the construction, I suggest a double face PCB with a ground plane connected to the COMMON line on the component side, while all the wirings are carried on the solder side with paths as short as possible; use non inductive resistor for R1 and R2 and good quality ceramic capacitors for C1 and C2.
All diodes I used are 1N4148 silicon detectors: I found in fact that is not necessary to use hot-carrier diodes as D1 and D2, furthermore it is not necessary to have these diodes matched, as pointed also by the author in his 1993 corrections3 .
As op-amps you can use other types too, like TL071/TL072/TL074, LF353, LMC272 etc. The LM358 is not a great choice for this application.
This peak detector can be used anywhere a RF level measure is needed, application limits are as follows:
Input signal frequency range: from DC to 500 MHz and beyond.
Minimum level of input signal: less than 30 mV rms for a gain error better or equal to -3 dB; this level corresponds to a 18 uW rms power on a 50 ohm load.
Maximum level of input signal: around 10 V rms, equivalent to 2 W rms on a 50 ohm load; (this maximum level depends on op-amp features and the power supply voltage: the given figure is relative to the use of TLC274 op-amp with a 15 V supply).
Input dinamic range: better than 50 dB.
Possible applications include:
directional couplers to measure forward and reflected power on a transmission line (it is the original application of this circuit);
low power wattmeters;
RF millivoltmeters;
field strenght meters;
radio receivers;
etcetera...
I used it as a detector in a circuit for RF currents measurement in an electrosurgery unit working at1 MHz.
1John Grebenkemper, KA3BLO, "The Tandem Match - An Accurate Directional Wattmeter", QST, january 1987, pp. 18-26;
2John Grebenkemper, K6WX (ex KA3BLO), "TANDEM MATCH CORRECTIONS", Feedback QST, january 1988, p.49;
3John Grebenkemper, K6WX, "AN UPDATED TANDEM MATCH", Technical Correspondence, QST, july 1993, p.50;
4Chapter 22 - Station Setup and Accessory Projects: The Tandem Match - An Accurate Directional Wattmeter - The ARRL Handbook CD version 2.0 - ©1997 by The American Radio Relay League, Inc.
5Frank Van Zant, KL7IBA, "HIGH-POWER OPERATION WITH THE TANDEM MATCH DIRECTIONAL COUPLER", Technical Correspondence, QST, july 1989, pp.42-43;
6Chuck Hutchinson, K8CH and Zack Lau, KH6CP, "Improving the HW-9 Transceiver", QST, april 1988, pp.26-29;