Overcoming the Problems With Precision Full-Wave Rectifiers

Even though we publish new blogs daily from your intrepid editor and a number of well-known writers, sometimes it's good to have a look back at our well-received articles from the recent past. Here is one that discusses the precision full-wave rectifier.

A precision full-wave rectifier is used in signal processing, typically with audio signals or as part of a data acquisition system. Complex bipolar waveforms are rectified so that only positive voltages are present.

There are many appnotes that cover precision full-wave rectifier circuitry, but they show versions powered from both positive and negative supplies. That makes sense, since the circuit is processing bipolar signals. However, in this blog, we see a special version of the circuit. This one uses RRIO (rail-to-rail input and output) op-amps and is powered from a single (or unipolar) supply.

The author discusses previous methods used in similar circuits. These rely on the op-amp's output swinging all the way to the supply rails and then going into saturation. This works as long as you are processing fairly low frequency signals and don't care about the time it takes for the op-amp to come out of saturation. And as long as you don't have stringent accuracy requirements.

The author then shows how to use an active clamp circuit to effectively change an op-amp from an amplifier with a gain of +1.00V/V to an amplifier with a gain of -1.00V/V. Accuracy is a function of resistor matching, op-amp offset voltage, and op-amp bias current. You must also consider the frequency content of the signals being processed and compare that to the gain-bandwidth product of the op-amp.

The article, part of our SIGNAL CHAIN BASICS series, was written by Rick Downs, at the time an Applications Engineering Manager for Texas Instruments.

6 comments on “Overcoming the Problems With Precision Full-Wave Rectifiers

    March 12, 2013

    I ran the circuit through LTSpice using the LT1491 and a BAT54 diode. Works pretty well. Definitely need the rail-to-rail capability for both input and output. The one thing I noticed that are not shown in the article is that the output will dip below the reference, especially when at 2.5V, for a diode drop's worth when the input is on the negative slope transistion through the mid-value.

    Otherwise, a nice little full-wave rectifier.

  2. Brad Albing
    March 12, 2013

    I can see where it might swing one diode drop below the 2.5V reference. Is that just very short term during the “zero-crossing” (actually the 2.5V reference)?

  3. RedDerek
    March 12, 2013

    I did the simlation at 1kHz and the dip was momentary and only at the middle of the negative slope of the input (the zero crossing). I can offer the simulation file if desired, but loading up LTSpice and using the LT1491 and 1N4148 diodes in the library work well. It is a quick and easy simulation to put together. It did not seem to matter as to where the reference was, it was always at the reference value.

  4. brhans
    March 13, 2013

    Doesn't this circuit just move the saturation problem to a different area, but still suffer from it to the same extent?

    For any input voltage greater than your reference, the output of U2 will swiftly swing all the way down to its negative saturation and stay there until the input returns to a value below the reference.

    The 'dip' in your observed output voltage during 'negative slope transistion through the mid-value' is very neatly explained by this as U2 takes its time to recover from negative saturaion.



  5. eafpres
    March 13, 2013

    Is it a SPICE thing or is the phase of Vin- incorrect in the figures?  It is a nifty little circuit.

  6. mtripoli
    March 13, 2013

    Interesting. I don't have the LT1461 you mention in my standard library from LT. I do have the LT1462… Nonetheless, I got the SPICE model for the OPA364 (the device shown in the original article) from TI and ran it through LTSpice; the waveforms produced are identical to what is shown in the first diagram. However, when changing the bias to 2.5V everything went to hell; the output was half-wave and not full-wave as shown. I'm not really interested enough to stop and figure out why…

    You mention using the BAT54; this is a schottky. The “original” schematic shows a 1N4148; not sure it's fair to compare between the two.

    I did some work a little while ago on these type of rectifiers. I remember looking at this (as well as about 10 others of the same ilk). I remember seeing a note about using the specific device shown as others have different CMRR, phase inversion characteristics, etc. This may account for why you see what you do. 

    EDIT: I transposed the numbers in the LT1461/91 – my mistake. I ran it again with the LT1491  and indeed do see the “dip” you see. It appears that this is one of those cases where the device itself is important. 

    Here's the LTSpice with the TI model if anyone cares to mess with it: 


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.