Questions about Dave Jones' Video No 626: Capacitance Voltage Dependency

[watchmaker]

I spent time today replicating Dave's class on voltage and ceramic capacitors.

I did things like determine the time constant via calculation (using measured R and C via ST42) and then determining via the scope.  Next, I set the X cursor to 5 times TC and confirmed the expected full charge at the point.  All very cool and solidifying.

I played with frequency and offset V and demonstrated several phenomena to myself.

A plain Jane x7r 104 acted just like in Dave's video. 

THEN came a Vishay 69 pF which raised questions.

When I attempted to determine the TC via scope, I set X1 at the corner of the pulse rise.  Then set Y1 at the base and Y2 at 63.3% (which was 6.33 Vo =10V, pulse with 50% duty cycle).  X2 at the intersection.

I could not get the measured TC to be close to the calculated TC (.994 K and .59pF = approx 58nS).  Got almost twice that.  I ultimately moved X1 to the second base crossing of the capacitor charge trace.  See the jpg.  The trace shows that immediately upon the pulse application, the Vishay C actually has a negative charge before going positive.  When I set X1 to this point, my measured TC came out to 59.7; close enough as Dave sez.

My questions are:  what is causing the initial charge to go negative?

AND:  In this case, is the pulse application the wrong point for X1?

It is enlightening to see how erratic a ceramic capacitor can behave.  I now appreciate Dave's caution to use top tier capacitors and to understand the datasheets of even these most simple of devices.

THANKS!

[RoGeorge]

My questions are:  what is causing the initial charge to go negative?

AND:  In this case, is the pulse application the wrong point for X1?

The wiggling up and down oscillations you see on the yellow trace of the oscilloscope can be from two sources:
- 1. software artifacts from sin(x)/x interpolation (specific to digital oscilloscopes only, but these become visible only when the oscilloscope is pushed close to its limits in terms of acquisition speed - so you also have to tell which model of oscilloscope you are using)
- 2. and this is more probable, the capacitor you measure, together with the probe leads, make a resonant LC circuit.  If you used the alligator GND clip of the probe to connect to ground, that alligator clip has a 2 inch or so piece of wire that is also a coil (any piece of wire has also a small parasitic inductance (L), a small parasitic capacitance (C), and a small parasitic resistance (R))

Please post a picture with the physical setup showing how you connected the probe, to see what else might be the reason.

When measuring high speed signals, you have to remove the parrot-like hook from the tip of the probe, and instead of the GND aligator clip, use the tiny small spring that came with the probes in a separate ziplock plastic bag.  That spring fits the metal ring visible at the top of the probe after the removal of the parrot hook tip.




This video tells the problems of measuring with an oscilloscope probe, in general:

#9: Basic 1X and 10X Oscilloscope Probe tutorial
w2aew
https://youtube.com/watch?v=SX4HGNWBe5M



This one tells about the ground lead alligator, and how to connect the probes when measuring high speed (signals with fast edges) or high frequency:

#68: Oscilloscope Probe Ground lead length effects on signal quality
w2aew
https://youtube.com/watch?v=zodpCuxwn_o



Or, you can make a high speed socket for your probe, socket that you solder on the PCB then insert the probe tip in it:

#111: How to make a high performance oscilloscope probe socket
w2aew
https://youtube.com/watch?v=-4q8geE5ef8

[watchmaker]

Thanks for getting back to me.  I like to understand things on a first principles' basis.

The x7r (NTE 104) worked just as Dave's video showed.  Very nice and regular charge curve.  It was the Vishay 69 pF that did the dip at the pulse application.  The squiggles, according to what I understand from Dave and other sources, can be a result of the cap construction.  (See video 486 "Displacement Current"). 

So my question about the charge trace itself is what causes this Vishay cap to display an initial negative charge.  I was not focused on the squiggles because I think/thought Dave addressed this.

I have a funeral today, but the setup is Sig 824HDX, Sig 1062x, LeCroy 500mhz probe at cap.  Alligator clip to setup, and channel 2. Freq, as shown in the JPG is 1kHz.  Not a stretch for the equip or setup.

FWIW, I have watched those videos and others involving standing waves etc. Been following the MIT classes (including Lewin), Real Analog and the lab books and as the curriculum and labs of LAOE and that by Boylestad.

Since the NTE setup trace is "perfect," I wonder if the issue is not the setup but rather the physics of the cap construction.  (I hold an advanced degree that involved extensive instruction in experimental design; the only independent variable is the different capacitors.  All else is held constant).

I have to check, but I think the Vishay is a different construction.  I will include the part no when I get back.HAZ680#BA###KR

( https://www.digikey.co.th/htmldatasheets/production/1019886/0/0/1/hae103mbabf0kr.html )

   I will also include a screenshot of the NTE cap trace for comparison.

Thanks.

Dewey

[RoGeorge]

Before anything else, try the same measurement using the spring GND instead of using the GND alligator wire.  I expect most of the wiggling will vanish.  The dip and wiggling you noticed in the yellow trace is most probably ringing produced by the relatively long alligator GND wire.  The displacement current video is another can of worms, and I don't think it's related with what you see.

The 1kHz of the square wave doesn't matter, you would see the same if you use 100Hz, or 1Hz.  What matters is how fast the switching occurs (or else said how sharp is the raising edge).  In your red trace, the raising edge takes about 20ns, which is very fast.  (In reality it might have been much faster than that, because you used the probe set to 1x when measuring the red trace - 1x makes any probe much slower than when the same probe is used at 10x - use the probes at 10x whenever possible).

Whatever voltage waveform you measure at the capacitor terminals is because of that raising edge.  The capacitor doesn't "know" if the next raising edge will occur after a milisecond, or after a second.  That is why the 1kHz or 1Hz frequency of the square wave is irrelevant.

The other capacitor was much bigger 104 means 10 followed by 4 zeroes, so 100000pf.  Compared with the current 69pf the previous one was huge.  Therefore the time constant and timebase of your oscilloscope were much slower.  100 thousands compared to 69 means more than a thousand times bigger.  At such slow timebase you wouldn't notice any voltage wiggling, because all the wiggling extinguish very fast.  That is why you didn't notice it in the first capacitor.

If you were to use 50ns/div for the first capacitor, the big one, you might see some wiggling for the big capacitor, too.

[Kleinstein]

The rise time of the square wave looks good. So I don't think the wiggles are an issue of finite sampling rate. It is still a point to look at.
The negative excursion can be seen as the first wiggle, likely from an LC resonance from parasitice inductance from the long ground / aligator clip wire.

With the larger capacitor there the resonance is way lower and the impedance is lower, so that the resonance is not excited that much. There is likely still some oscillation/ resonance, just not visible.

A 69 pF capacitor ( more likely 68 pf as that is a standard values) is most likely not a class 2 ceramic, more like NP0 /C0G. These capacitors are pretty linear and show essentially no effect of the DC bias. With the low capacitance other parasitic capacitance can be significant and thus a high appearent capacitance.

[RoGeorge]

Most probably the red trace looks good only because the probe is not fast enough.  The red trace probe is set to 1x, yellow to 10x, I've noticed that only seconds ago. 

[Kleinstein]

The probe speed may be an issue and effecting the red curve. It is still steep enough to show that the sampling rate and thus the sinx/x interpolation is not causing the ringing. Otherwise the red curve should show similar (or slightly more) ringing.

[watchmaker]

The probe speed may be an issue and effecting the red curve. It is still steep enough to show that the sampling rate and thus the sinx/x interpolation is not causing the ringing. Otherwise the red curve should show similar (or slightly more) ringing.

OK.  I think I figured it out.  Everything starts falling apart below 100pF.  Below that the measured TCs are greater than the calculated.

I reran the demonstration.  First I likely had a loose connection yesterday because the Vishay now gives the same curve as a 69pF NTE.  I selected an NTE and Vishay that both measure 69pF (+.09 to +.39).  They both gave great curves but a measured TC of 100 nS when in series with a 995 ohm resistor.  Consternation.  Reran Dave's video.  Setup was correct.

I was wrong to fudge the measurement by moving the X1 cursor past the initial negative excursion of the charge trace.  The true measurement was at the pulse initiation, which gave a TC of around 95ns if I remember correctly.  Which fits the pattern I found today.

Component values measured with ST42 at .1v and 1kHz.

Then I remembered I had comangled a 100pF NTE and a 59pF Vishay (as measured!) yesterday.

Tested a number of NTE at various values.  Everything above 100pF is close or spot on (the greater the C the better the measured TC.)  Ah ha!

You two were on the trail.  I suspect low values of C get swamped by other sources, such as the well known capacitance for breadboards.  Then there are the device lead lengths.  Etc. The scope is measuring the TC of the SYSTEM.  Got it.

Kind of a useful thing to learn by doing.

Here is a selection of screen shots.  In my setup, the R remained in a fixed position.  The C probe was attached to the return side of the R so it never moved.  The C was connected to the return side of the R and the grounds and gen return were connected to the return lead of the C.

So the only things that moved was the C and the ground/return leads.

All trials were with 1kHz, 5 VP-P pulse at 50% duty cycle.

THANKS!


1000pF


100pF


85pF


NTE 69pF


Vishay 69pF