# Peak Detector: A Classic Analog Circuit Still in Wide Use

There are two groups of classic circuits: those which have served us well in the past, but have been superseded by better designs, and those which are still in use, perhaps with updated components.

In the former group, for example, goes the superregenerative receiver invented by E.H. Armstrong. While it is still used in specialized, niche designs, it has pretty much been made obsolete by the superheterodyne architecture (also developed by Armstrong — who also invented FM radio!). And who knows? The superhet itself may be made obsolete by software defined radio (SDR) and zero-IF designs.

But some simple, basic circuits are still going strong after many decades. Consider the analog peak detector shown in Figure 1 in simplified form. It's sometimes called a peak-hold circuit (and sometimes called a full-wave rectifier, although I think designation is misleading, for reasons too long to explain here). The peak detector does one thing: it monitors a voltage of interest and retains its peak value as its output.

Figure 1

The basic peak-detector circuit requires just a few high-quality analog components to capture and hold the signal's maximum value.

How does the circuit operate? It's simple: the input signal charges the hold capacitor, and the diode prevents the capacitor from discharging. The input op amp, in conjunction with the capacitor, presents that held value as the output via the driver op amp. As the input voltage increases further, the capacitor is charged to the higher voltage; if the input voltage decreases below the previous value, the voltage on the capacitor stays at the previous peak value.

By adding a simple comparator to the output to compares the present input value to the already held value, it can also indicate that a peak has been reached when the present input value is less than the held peak by some desired amount, Figure 2. This transforms the circuit from providing a peak-hold function to implementing the peak-detect function, with comparator hysteresis to establish a valid-peak threshold.

Figure 2

By adjusting the hysteresis of an output comparator, the designer can set the threshold for the “valid peak detected” value.

It may seem that such a circuit itself is nearly obsolete. After all, wouldn’t it make more sense to use an ADC on the signal of interest, with a basic software loop checking the newest converted value against the previous ones?

Reality is that it would not make more sense, for both hardware and software reasons, except for fairly slow to moderate-speed analog signals. On the hardware side, the cost of the ADC and its input signal conditioning could be significant (the ADC has to sample at much higher than the Nyquist rate, in practice); and the power consumption and footprint would be higher. On the software side, the checking loop would need some filtering algorithm to make sure a little noise on the signal did not cause a false output (some sort of moving average, perhaps?), and soon the “basic” software loop could easily grow to consume a inordinate amount of the CPU resource.

In contrast, the analog circuit requires no software, little power or real estate, and is inherently self-filtering (up to a point). The main component constraints are that the designer's bill of materials (BOM) calls out low-leakage, high-quality components, which are a little more costly than commodity parts.

More important, this peak detector can operate at frequencies much higher than those for which there are ADCs, since its only semiconductors are the op amps and diode, both of which are available into the tens of GHz. This can be for a relative assessment where the voltage alone can stand as a proxy for actual power levels, or for absolute measurement when used in conjunction with the known impedance across which the voltage is measured. Due to its virtues, this design is still used as the core of the RF peak-power measurement function, either using discrete components, as a single-purpose IC/module, or as part of a larger IC.

The peak detector was such a common configuration that some analog vendors offered it as a standard application design. For example, the data sheet for the OPA111 op amp from Burr-Brown (now part of Texas Instruments) showed a possible circuit (see Figure 18), and Analog Devices offered the PDK01 IC, which I believe (I'm not sure on this) was a monolithic version of an older, multichip Burr-Brown device.

A relative of the peak detector is the crystal (diode) radio, which uses a diode to demodulate (detect) a basic amplitude-modulated broadcast signal; see Figure 3.

Figure 3

The venerable crystal radio circuit is a self-powered envelope detector, and a precursor of the peak detector.

This circuit is an envelope detector that tracks the overall modulated signal shape (a.k.a. envelope) — the information-bearing part of the signal — while ignoring the much higher-frequency carrier waveform. (If you don't understand how the crystal radio works, please review your analog basics.)

Note that the early crystal radios were entirely self-powered, as enough energy could be captured via a long-wire antenna to drive high-impedance headphones (usually around 20kΩ). To anyone who insists that energy harvesting is a new phenomenon, or that RF energy harvesting is a recent development, just point to the classic crystal radio of the early 20th century as a counterexample. [See Nothing new about energy harvesting and Remember: voltage is not power (but we still need it).]

Are there any similar long-lived “classic” circuits that are still in use that you know of? Are there any that you have used?

Related posts:

## 22 comments on “Peak Detector: A Classic Analog Circuit Still in Wide Use”

1. Davidled
December 5, 2013

This Figures reminded me that I and my colleague friend used Napkin to design the circuit in Cafeteria. Based on the analog circuit textbook, we redesigned circuit on Napkin. No simulation tool, Napkin was a part of simulator tool with textbook. Then this schematic was rebuilt in breadboard. Through this process, we all built deeply the knowledge of analog circuit with trial and error.

2. eafpres
December 6, 2013

@Bill–I think I would put a full wave rectifier circuit onto your list of still there but maybe with new parts.  As an example, there are some LED lamps called “driverless” which use a rectifier in front of a series of LED elements to power a luminaire directly from AC line current.  Old meets new.

3. antedeluvian
December 6, 2013

Bill

Are there any similar long-lived “classic” circuits that are still in use that you know of? Are there any that you have used?

In aswer to both your questions- the 4-20mA source driver. No matter how many different variations there are with the functionalty integrated onto a chip- they all seem to use the same circuit internally.

Oh and I have also used the self same peak detect circuit, but as in your description, more recently I have worked with ADCs for the same purpose.

4. Davidled
December 6, 2013

I wonder LED driver chip would include a full wave rectifier circuit instead of the separation in the PCB board. This decision is depending on how to design the LED circuit in order to transform AC power Line to DC.

5. eafpres
December 6, 2013

@DaeJ–in most designs the so-called driver circuit contains the rectifier as well.  I was referring to designs that don't convert to DC, but do rectify the AC.  In theory they are simpler but so far there are other issues like variable output of the LEDs as the current goes up and down.  There are some commercial products on the market that are actually driving the LEDs with rectified AC but in most cases they are DC and the driver has the rectiifer and all the other components to generate DC into the LEDs.  The driver turns out to be the least reliable part of many LED lighting products.

6. Vishal Prajapati
December 7, 2013

@eafpres, Another circuit I can recall that is being used as an LED driver is NE555. The most morden non isolated buck converter ICs use same internal structure with two opamps and one Flipflop to drive the external or internal MOSFET. It also use capacitor to determine the frequency of switching.

7. Davidled
December 7, 2013

There is other drive such as HV9801A: Vin max is 450, and output current is controlled by External FET with PWN in order to dim LED. This LED is made by Supertex INC. There is no rectifier circuit except a few diodes outside chips. Inside chips consist of VDD Regulator, average current regulator, and OR, SR and AND gates.

8. Peter1010
December 7, 2013

The PKD-01 was designed by Precision Monolithics (PM), which was later acquired by Analog Devices. Inside PM there was a lot of discussion whether the PKD-01 should be offered as a device. The PKD-01 was a modified version of the Precisions Monolithic GAP-01 which, among other things, could be configured as a synchronous demodulator for instrumentation

I along with Wes Feeman wrote articles, seminars and application notes for the PKD-01 and GAP-01.

Burr Brown did have peak detectors as well as other specialized analog circuits.

9. Bill_Jaffa
December 7, 2013

Thanks, that's good historical information–often lost “forever”!

10. samicksha
December 7, 2013

any views on full-wave rectifier design wherein we have the necessity of a transformer with a center-tapped secondary winding.

11. Victor Lorenzo
December 8, 2013

One more classic circuit is the push-pull power output stage with the center tapped transformer that was used in the audio tube amplifiers, also used in audio amplifier for the firsts transistor radios.

It was amazing! With just a few components you could make a 200W, extremely clean at high rates, audio power amplifier. Ouuuppppps! But dangerous too with those more than 350V in the output tubes plates.

12. Victor Lorenzo
December 8, 2013

I agree with you, it is becoming rare with so many switching regulators and PFC circuits around.

But we can still find some specialized lab equipments that still use the 'old fashioned' power transformer+rectifier+linear regulator to avoid having noises comming from the switching regulator to interfere with the high gain input stages.

13. eafpres
December 8, 2013

Some other examples using traditional full wave rectifier are AC-AC power converters used in motor drives.  Some of these include a transformer in the output stage as isolation, even if the voltage isn't transformed.

14. Victor Lorenzo
December 8, 2013

@Bill, “After all, wouldn't it make more sense to use an ADC on the signal of interest, with a basic software loop checking the newest converted value against the previous ones? ” There're several algorithms for implementing the peak detector in the digital domain for real-time applications but it some times gets a little bit tricky, not so obvious I would say, as things are not processed 'in a loop'.

I fully agree with you at “the ADC has to sample at much higher than the Nyquist rate ” and, if you let me, I would add that the 5xNyquist criteria that usually serves for being able to record/reproduce analog signals with acceptable quality is still short for this application. You may lose important events present in the analog signal.

15. Victor Lorenzo
December 8, 2013

@Bill: “A relative of the peak detector is the crystal (diode) radio “, I should admit it, I played a lot with them and tried lots of different diode types trying to find 'the best' diode for the circuit. At first I only used the Antenna+LC+Diode+Headphones, and soon started putting some germanium type transistors here-and-there. I was only 11 or 12 year old, but my father had everything I needed (tons of passive components, tubes transistors, old radios for salvaging) for those little projects (including lots of books!)

16. vvc
December 11, 2013

A couple of very important analog circuits still widely used today in most anything requiring a frequency source are the Colpitts and Hartley oscillator topologies. Another circuit which is ubiquitous in most analog circuits is the differential amplifier found in the majority of operational amplifier IC's (as illustrated in the gain elements of your peak detector).

17. pmoyle111
December 11, 2013

For completeness, the circuit is actually missing the reset which would be a fet or analog switch across the cap. Otherwise you can no longer detect any other peaks except those greater than the last one.

To make an enveope follower, a resistor is just placed across the cap where the rc time constant is relative to what's inside the envelope so that the cap is naturally discharged as the signal falls.

In the case of the crystal radio, there should be a small cap across the output. As I remember something like an .02 uF mica, this along with the resistance of the headphones made the envelpe follower.

The envelope follower was very commonly used in AGC circuits back in the day.

Back about 20 yr ago I used the PDK01 in a couple of custom instrument designs. The “good old days”.

18. Bill_Jaffa
December 11, 2013

Good points about reset and other aspects–I was just trying to capture the “essence” in that basic schematic. As with all real circuits, there are always more things you need to add in the final design, of course.

19. mediatechnology
December 11, 2013

I have a forum thread “Level Detectors Absolute Value, Peak and RMS” that many of you may find interesting. It contains a number of diode-based Design Ideas collected over the years as well as the classic Blackmer RMS detector and comparator-based peak detectors. I do remember the PMI GAP-01 and PKD-01 and one of the posts contains some applications.

20. RedDerek
December 31, 2013

Hmm,

1. Depletion MOSFET configured as a self-limiting circuit.

2. Multipliers – reason that should be soon blogged about.

Others mentioned thus far are all good old basics as well.

21. Kev1
June 12, 2014

Hi Dave, just to say I used the PKD-01 in the distance past, a great little chip and at the time beat the competition on it's versatility.

We were very sad at the time it was announced that production was to end, we were kindly supplied with some last production run chips for spares, but they were never required. Probably still in my 'misc' drawer…

This is one of the chips that I will always remember, in fact that's how I found your post, just surfing to see if they still get a mention!

22. Peter1010
June 12, 2014

Yes the PKD-01 was a quite nice chip. Internal to Preciscion Monolithics there was quite a political battle about whether the PKD-01 or GAP-01 was going to be the better revenue generator.

Marketing favored the PKD-01 and engineering the GAP-01.