Half wave rectifier waveform

[stitch]

I have set up a half wave rectifier circuit as shown in the link below.  The upper oscilloscope screen shot shows several cycles.  The lower screen shot is a close-up of the bottom of the waveform.  I expected to see a "rectilinear" shape at the bottom of the waveform, but I was surprised to see a curved shape there instead.  Can anyone help explain this curve?  Thank you.
http://www.late-date.com/halfwave.html

[IanB]

Nature abhors sharp corners 

I think the curve happens when the forward voltage across the diode decreases close to zero and the diode tries to switch off. However, there is at the same time a collapsing magnetic field with stored energy in the transformer that is trying to push current through the diode while it is trying to switch off. These effects fight each other and prevent the voltage dropping to zero as fast as it otherwise would.

[stitch]

I completely forgot about the transformer.

[qno]

Part of the effect is the reverse recovery time of the diode.

[stitch]

Oh wow!
Both of your explanations (combined) really illuminated the solution.
I sketched out an explanation to myself.
http://www.late-date.com/halfwave-explain.html

[w2aew]

Of course, another affect could be that the Rigol scope may not like that much overdrive when you turn the vertical scale to 1V/div, and you're seeing the overdrive recovery of the scope.  I don't know the Rigol scope, so I can't say for sure if this is what's happening, but it's something to look into.  Is there a way on the Rigol to "zoom" the waveform without changing the vertical scale setting (so that you can avoid the overdrive situation)?

[stitch]

I have not found a way to zoom without changing the vertical scale settings.  But I did change the V/div to 2V/div just to see if the waveform looked the same. It did.  But then I noticed something...
As I re-positioned the waveform on the screen (by using the vertical positioning knob only) the waveform changed. When I positioned the wave higher on the screen, the bottom of the wave became even more curvilinear.  When I positioned the wave lower on the screen, the bottom of the wave became more rectilinear - the curving flattened out.

Here are the screen shots:

http://www.late-date.com/trace-position.html

[tekfan]

That definitely looks like you're overloading the vertical preamps. It's never a good idea to have a large part of the waveform off the screen.

[hitachi8]

is every digital scope act like that?

[ejeffrey]

Most amplifiers of any tipe are like that, whether in a DSO, an analog scope, or anything else.  They become much slower when driven into saturation.   You can easily demonstrate this with op-amps: make a simple 10x amplifier with an op amp, and drive it with a function generator.  As soon as the peak of the waveform reaches the maximum output voltage you will start to see 'rail sticking' where the amplifier takes a long time to recover.

There are things you can do to mitigate this, but it is better to avoid overdriving amplifiers.

[tekfan]

Every analog and digital scope acts like that. Even if you've got a digital scope you're actually overdriving analog preamplifiers.
Some scopes have very fast overload recovery for just these situations but they are very expensive.

[stitch]

Highly informative.  Thank you.

So now my question is:  Should I attribute the curving at the bottom of the waveform entirely to overdriving the scope, or does capacitance and inductance in the circuit contribute (in some part) to the explanation?

Also, I checked my probes for compensation against the oscilloscope's square wave.  The probes were fine.  But I noticed that the square wave showed no "distortion" even at 20mV/div.

[onewatt]

I think something else might be causing the distorted waveform that you see.  Anytime the ac sinewave fundamental is distorted, harmonic frequencies are created. And, in this case, diode rectifiers always create even-order harmonics, like the 2nd and 4th harmonic frequencies.  The total superposition of a halfwave rectified sinewave and even order harmonic frequencies creates the distorted waveform that interests you.  I see this type of diode distortion all the time; even order harmonics are easily seen using the scope's FFT function. I think what's happening is the halfwave rectifier is doing what it normally does, that is, grossly distorting the ac sinewave fundamental by chopping half the waveform off -- leaving only a halfwave DC pulsating waveform.

Also, the rectifier diode is not turning-on exactly at the ac zero-crossing point; that is because the diode has a 0.7volt barrier potential and has a non-linear curve at the knee voltage.  When the diode finally conducts, the ac fundamental frequency is distorted, however slightly, from the zero crossing point.

Very often a rectifier diode is used as a 1/2 power switch in hair blowers and other consumer motor driven appliances.  If you check the ac sinewave across those devices, you will also see that at half voltage, which is usually the low power setting, when the rectifier diode is operating as a halfwave rectifier, the pulsating DC waveform is slightly distorted because of generated even order harmonics. 

Other switching semiconductors like triac lamp dimmers really distort the ac fundamental by clipping the sinewave at different degrees depending upon the amount of dimming.  The more the sinewave is chopped, the more distortion and, consequently, in a triac, the more odd and even order harmonics that are created.  Triacs really chop the ac fundamental in a sharp leading edge. It's the leading edges, like in a digital pulse, that contain the high-frequency harmonics.

Triac high-frequency harmonics, and by the way -- rectifier diodes, do not contribute to the operation of lamps or motors, which is why such devices have a low power factor that's caused by the consumer device drawing more current than it uses.  Any non-linear device will cause a lower power factor by generating harmonic frequency currents that do not contribute to true power; harmonic frequencies merely heat components and cause electrical noise and crosstalk.

[IanB]

Every analog and digital scope acts like that. Even if you've got a digital scope you're actually overdriving analog preamplifiers.
Some scopes have very fast overload recovery for just these situations but they are very expensive.

Coming back to an old thread now I have played with a DSO a bit, I have a question. If I adjust the vertical sensitivity while the scope is sampling, I can understand what I am doing to the analog front end. But suppose a hold a sampled waveform in the digital memory, and then use the vertical controls to zoom in on that? My intuition tells me that I am now purely working with the digitally sampled signal in memory and the analog section is out of the picture. Of course, the vertical resolution of the sampled signal is also limited to the (8-bit) ADC depth so I can't zoom in too much, but presumably I can magnify whatever detail is actually there?

[onewatt]

I'd like to offer an alternative theory why the waveform changes shape when you vary the vertical position.  It has to do with the number of pixels on the screen.  The spatial frequency of the distance between pixels is undersampling the data when you vary the vertical positioning.  There are just so many pixels and, yes, sized does matter.  When you change the waveform’s vertical positioning, the number of screen pixels are inadequate to represent the data smoothly.  In effect, the Nyquest spatial frequency is exceeded. So the scope display algorithm begins anti-aliasing the jagged edges by moving the pixels around and maybe adding a Gaussian filter to blur the image lines because the last thing Rigol wants you to see is an ugly jagged line. 

The spacial frequency of the screen's pixels are undersampling the data; much like temporal undersampling when wagon wheels going backward in a western movie; or video morie patterns when a news anchor wears a closely spaced pin-stripped shirt; or when a gamer complains about jagged diagonal lines.  Same with the scope screen. A horizontal line is easy to display as a straight line of pixels, but a diagonal line shows a jagged line because of the nature of square pixels.  The scope display algorithm is merely moving the pixels around to anti-alias the waveform, to show a smooth continuous line to the eye…  Rigol does not want you to see alias artifacts, so they fudge the data… surprised?

[IanB]

So the scope display algorithm begins anti-aliasing the jagged edges by moving the pixels around and maybe adding a Gaussian filter to blur the image lines because the last thing Rigol wants you to see is an ugly jagged line.

The Rigol has two display modes. One is points, where the sampled data points are plotted as dots wherever they land, and the other is sin(x)/x interpolation, where the dots are joined with a smooth line using an interpolation function. No funny business is mentioned beyond this.

[onewatt]

I guess the question is... Is the change in waveform from the vertical position control due to input ADC signal sampling error, display aliasing artifacts, or actual electrical changes in the waveform caused by overloading the input signal opamps?

I see what you're saying about sin(x)/x interpolation though.  However, with regard to the rectified waveform, I was thinking more about equal-time-sampling vs real-time-sampling.  In the scope pics, dots were not used.

And whatever sampling method was used (and I'm assuming equal-time-sampling), the dots have to be connected.  And since I see a fairly smooth curve, I'm guessing that some type of display anti-aliasing technique is being used and this anti-aliasing technique is what's causing the change in waveform when the vertical position is changed. When I talk about display anti-aliasing, I'm talking about something different than input sampling aliasing caused by an under-sampling analog-to digital converter; where lower frequency alias waveforms are created. However, I'm not a scope expert, but I have noticed these rectified waveforms since school and everyone always asks, "what's this distortion about?"

The display font I'm using to type this is not anti-aliased on the screen.  Use windows magnifier and you will see the individual pixels that make up the font.  However, the menu font on my Firefox browser is all blurry around the edges.  I would be interested in holding a magnifying glass up against the scope's screen and tell me if the individual pixels are visible or is there any attempt to join the pixels with various shads of pixels, which would indicate some type of display anti-aliasing?  Maybe the pixels are arranged in a way to imitate a curve where there is none. I had a Rigol DS1102CA 100 MHz scope a few years ago.  Unfortunately, I am presently scope-less.

I've seen this type of rectifier waveform without an isolation transformer, too.  I will have to think more about the idea of overloading the vertical amps that might cause this waveform artifact.  I'm not too familiar with observing this effect, possibly because I didn't know enough to look for it.

I think I've said my piece... maybe you can resolve this conundrum... I figured I had figured it out years ago, but apparently, rectified waveform distortion is more complicated than I first realized.