Single Supply Precision Rectifier Design and Optimization 

The only exception to this being over-the-top op-amps which are designed to transition between input topologies and prevent reverse breakdown of the input transistors at the expense of drastically increased input bias.

+TecKonstantin Thanks, and thank you for sharing that circuit!

+Michael Hawthorne The explanation perhaps wasn't the clearest but it should work as drawn. When the input is signal positive the comparator output is high, the transistor is on, pulling the input to the HC4066 low, so the analog switch is open and the opamp runs as a non-inverting amplifier. I don't see why there is a need for the transistor inverter - just swap the inputs to the LM393 and connect the output directly to the HC4066 input retaining the 3.9K pullup resistor.

+Michael Hawthorne TMM is correct, what I meant by "OFF" was that the logic output of the transistor inverter would be off (i.e., a voltage low state); for this to be true, the NPN transistor must of course be on, as you pointed out. +TMM In theory, the transistor inverter is not necessary as you could use the comparator in an inverting configuration. In practice though, I've found the LM393 to be very prone to oscillation in an inverting configuration with a high source impedance (10k, in this case). However, the only other comparators I had on hand had longer propagation delay times (which would further limit frequency response), and the usual methods for preventing oscillation would have similar effects (hysteresis would cause a discrepancy in the zero-crossing detection, and capacitor compensation results in additional comparator delay). The inverter was a quick and easy solution that added minimal additional delay.

+TMM Actually, the inverting comparator oscillation issue can probably be fixed by connecting the comparator's positive input to ground via a 10k resistor, rather than connecting it directly to ground. Then you can get rid of the inverting transistor completely.I probably should have tried that. :\ Also, an LM211 comparator would probably be a better choice for this application, but I didn't have any on hand.

I didn't try it but I suspect that if you just use a BJT to pull the non-inverting input to ground you will get some distortion there. A FET switch will probably be better, and the 4066 is simple and cheap so I just went with that.

+Smudger Dave Any op amp can be powered from either a single or a dual supply. A "single supply op amp" has had certain characteristics that are typically advantageous for single supply operation - such as rail to rail outputs - emphasized when its internals were designed by microchip/TI/Linear/Analog/whoever. But when using the op amp in an application, the op amp itself doesn't know or care if you run it from a single 5v supply or a split +-2.5v supply, which is exactly what I did for the dual supply rectifier circuit in this video.

For single supply half-wave rectification, no diode is used. Unfortunately, it typically relies on driving the op amp to ground, which puts the op amp into saturation, and as demonstrated in the video, this is the cause of the glitching that you see at higher frequencies.

So there's no way of reducing the glitch on the half wave circuit?

There is, but to fix a problem you have to understand its source. As shown in the video, the glitch observed in the half wave rectifier circuit is due to the op amp's output saturating, delaying its response. You can get an op amp with a higher slew rate (and higher price tag), and that will help some, but doesn't address the core problem. A switching circuit is probably the most obvious approach, as demonstrated in the full wave rectifier circuit at the end of the video. That circuit should be easily adapted to a half-wave rectifier by using the analog switch to ground the input signal during the unwanted half of the signal. Of course, you would also have to ensure that your signal source is OK with that. If you can provide more specific details on your desired application and performance, then perhaps I can give a more specific answer.

I expect the diode would likely introduce some distortion. A p-channel JFET would probably be better suited to replace the 4066 (I just had a bunch of 4066's on hand), and the BJT is not required if you use the comparator in an inverting configuration.

Thanks, You're right!! Use of the diode, as I suggested, sucks. Becomes reverse biased when input goes neg. Won't clamp the op amp "+". Would the P-ch JFET be a problem when the source-gate bias reverses? Perhaps an N-ch and reverse the comparator? Is it ok to exceed the 4066 input pin limit of -0.5v? Seems to work well with your -1v demo. I appreciate your videos, keep up the good work.

Have you solved your frequency-limited problem? I am curious to know what you used to do so.

+Dennis Lubert Yes, fair point. I need to use something longer/thinner anyway, the camera tends to auto-focus on my hand or the screwdriver handle if they get in the frame. Glad you like the video editing! :)

There's still the problem of op amp saturation recovery. For example, when the input goes negative, the non-inverting op amp will saturate, and it will take some time to come out of saturation when the input goes positive again. The same problem will occur with the inverting op amp when the input goes positive. These saturation recovery delays will result in "glitches" in the rectified waveform and limits the circuit's high frequency response. You can certainly use the comparator in an inverting configuration and save yourself a transistor. However, many comparators (such as the LM393 used here) have their inverting input pin right next to the output pin, and especially with high source impedance (such as the input protection resistor in this circuit) capacitive coupling can cause unwanted feedback, resulting in "chatter" on the comparator's output. Careful circuit layout prevents this, but using the comparator in a non-inverting configuration is more forgiving of sloppy layout (i.e., testing the circuit on a breadboard).

I honestly never encountered any 'feedback' issues as the way you described. Even not in 300-400kHz Class-D switching circuits, where these kind of problems should be much more significant. The reason why I am talking about alternatives, is because in practice, I now need to buy at least two different packages. Which takes up a lot of space. There are some comparator-opamp-in-one packages on the market, but the choices are very slim and limited (+ pricey). In the end it all comes down to practical choices. Nowadays it's not so difficult to find affordable fast high-speed opamps (BW around 200MHz) for less than a dollar. It would be interesting to compare these with this circuit.

With reasonable circuit layout and/or low source impedance, you probably wouldn't encounter the feedback I described. But larger source impedances and stray parasitics can cause nasty oscillation on the rising/falling comparator edges (it's pretty easy to replicate on a solderless breadboard, for example). ADI did an "ask the engineer" piece which talks a bit about it: www.analog.com/library/analogDialogue/Anniversary/5.html. I wouldn't expect the same problem on a printed circuit board or a dead bug solution with careful layout. I didn't try any high speed op amps, but its not really the speed of the op amp that is the limiting factor in most cases, it's how fast the op amp can come out of a saturated state. The MCP601 op amp I used certainly isn't high speed (2.8MHz bandwidth), but even frequencies in the audio range get distorted when used in a typical rectifier configuration. Note in the 'scope waveforms that once the op amp output starts tracking the input again, it catches up quite quickly; it's the delay in tracking the input, caused by saturation, that poses the problem. Unfortunately, saturation recovery time is not specified by manufacturers, so you'd need to get a reasonable sample of op amps and test them to see how they perform. If you have an op amp that doesn't distort your waveform at the frequencies you need, you can just use one of the single op amp circuits presented in the beginning of the video. You're absolutely right about making practical choices, and it all depends on your application and design. Do you really care about that little bit of distortion in the bottom of the waveform? If you're implementing a peak detector circuit, probably not. Are you digitizing the rectified waveform? If so, many PIC microcontrollers (even the really tiny ones!) have built-in comparators that you can use, no need to buy another IC. Are you optimizing for cost, size, power consumption, etc? You have to take it all into consideration of course and make the trade-offs that best suit your needs. This is just another circuit that may be useful for some situations, but certainly will not be optimal for others.

devttys0 fg

+Евгений Жук (Nacky) No, not just you apparently; not sure how that happened, I definitely shot the video in 1080p!

+devttys0 Thanks for quick response, I thought there was a problem on my side. The vidio is great anyways.

+devttys0 when uploading youtube converts to all other resolutions, starting with the lowest. As long as it is doing that, only those low ones it has converted to are being displayed. Finally it will reconvert your resolution into its own format and only then it will be displayed. You can somewhat "fix" this by uploading it hidden, and then when its rendering is finished, unhide it. Oh btw. what do you think of enabling the feature of subtitles for others to add (in different languages)?

+Dennis Lubert I heard about that but thought that its long time gonsky, as everybody seems to use the hide-unhide trick.

+Dennis Lubert Interesting...I uploaded the video as private, then only *after* youtube said it was finished rendering did I add the video annotations and make the video public (which is what I always do). Odd that youtube said rendering was finished, but clearly had only finished for lower resolutions; it did seem that the rendering finished really quickly for this video! I'll have to watch out for that in the future... I've turned on community subtitle submissions for this video (ru-vid.com_video?v=1TLZBWx6BCc), I'll be sure to add it to my other videos as well!