1
\$\begingroup\$

I am following this circuit:

Circuit to measure the capacitance. The opamp is OPA604 with supply voltage +-5V, Cx and Cf is 5pF, Rf is 10Mohm, the diode is 1N4007 and the output capacitor is 1nF. I also have 3 CD4051AN analog multiplexer/demultiplexer at 2 sides of Cx to change the capacitors and between the diode and the output capacitor to discharge the output capacitor. Then I connect Vi to about 100kHz sine wave signal.

The signal at the output of op-amp seem to have a low frequency noise, which makes the whole output seemed to be moving up and down:

And the signal after the diode still have the sinewave there, which I cannot understand.

Then I tried changing the 10M resistor by a 636k resistor, and the low frequency noise at the op-amp output just disappeared. The signal between the diode and the output capacitor also seemed to be much more constant.

So could anyone please explain to me where is the low frequency noise come from? And also how can there is a sine wave between the diode and the capacitor (since the diode forbid any discharge to the capacitor). Any pointing is also greatly appreciated. Thanks in advance.

\$\endgroup\$
  • \$\begingroup\$ Worth asking what kind of 10M resistor (link to datasheet if need be) \$\endgroup\$ – Brian Drummond Dec 5 '14 at 14:07
  • \$\begingroup\$ I suppose that at a frequency of 100kHz the diffusion capacitance comes into play ("how can there is a sine wave between the diode and the capacitor ?"). \$\endgroup\$ – LvW Dec 5 '14 at 14:18
  • \$\begingroup\$ @Brian it is a brown resistor, So I think it is carbon film resistor. I took it from the rack in my lab so I am not sure about the datasheet. the resistor value measured by my multimeter is 10.00 Mohm \$\endgroup\$ – mihota Dec 5 '14 at 15:44
  • \$\begingroup\$ @lvW the sinewave amplitude of 400mV is reduced to about 20mV when I changed the 10Mohm resistor to the 616kohm resistor, with the same input frequency, so I am not sure if it can be caused be the capacitor \$\endgroup\$ – mihota Dec 5 '14 at 15:49
  • \$\begingroup\$ Carbon composition can be noisy, I'd expect carbon film to be better, but maybe worth looking for metal film here. \$\endgroup\$ – Brian Drummond Dec 5 '14 at 16:27
1
\$\begingroup\$

How have you assembled it? On a PCB, plug-in breadboard or overboard.

The gain at 50/60Hz will be very large and it is easy to get coupling from the AC power lines. I would suggest placing a grounded metallic plate under the circuit - make sure you insulate it appropriately. Attach that to the circuit ground rail.

To isolate the low frequency interference disable the 100kHz signal.

As others have said the 1n4007 is not at all appropriate, use a 1n4148 or similar signal diode. You will not get good accuracy with just a simple rectifier, it will be very non-linear. You will probably need to use a synchronous rectifier, you can use an CD4053 or similar to create that.

The ripple at the output is just because you have a large amount of low frequency interference - the time constant of the 1000pF cap and the scope probe input resistance is such that it follows the interference - get rid of that and it should be much better.

\$\endgroup\$
  • \$\begingroup\$ Thank you for your response. Unfortunately the project is finished and I have no way to try your suggestion now. But I will make sure to keep them in mind for future development. Regards, \$\endgroup\$ – mihota Aug 24 '15 at 4:17
0
\$\begingroup\$

I can't explain the output from the op-amp being 103.8kHz but I can possibly explain the ac appearing on the output capacitor - two possible reasons - the first is that your scope probe is loading the 1nF capacitor causing it to lose a little charge between cycles of the 103.8kHz - try using a x10 probe and set your scope to avoid aliasing - in the 2nd picture the time base is way too low to avoid aliasing of any 103.8kHz signals.

The 2nd reason is the crappy 1N4007 and its crappy reverse recovery time - this will kind-of act like a pure perfect diode with a capacitor in parallel - try something like a 1N4148 or BAS16.

If the signal on the 1nF is truly 90 odd hz (and not aliasing) then it still could be the scope probe discharging the capacitor. Also, your first picture does show signs of an underlying low frequency and this could be related to rectified AC hum getting onto the power rails.

Regarding the 103.8kHz, what is connected to the input and what are the power rails like on the op-amp?

\$\endgroup\$
  • \$\begingroup\$ It is kind of confused me because when I change the 10Mohm resistor by an 616kohm resistor, the sine wave at the output capacitor significantly reduced. But I will sure try your suggestions (it is Saturday here already, and my lab doesn't open in weekend, so I will have to wait to next Monday) About the underlying low frequency noise, it cause the signal in the first photo to be kind of unclear (a big blur line instead of a small clear line). I also realized that its frequency is about the frequency of the sine wave in the second photo. \$\endgroup\$ – mihota Dec 5 '14 at 16:42
0
\$\begingroup\$

First what the value (range) of Cx?
As Andy said get rid of the 1N4007, and how about an (opamp) active rectifier circuit on the back end.
As far as you low frequency pick up. Is the whole thing in a metal box or is it open to the air? There could be all sorts of AC fields getting into your input. Reduce or disconnect the 100 kHz sig gen and find out the frequency. Then see where it's coming from..(move the circuit around and see if the pick-up changes) your scope, florescent lights, power supply, something else??

\$\endgroup\$
  • \$\begingroup\$ The value of Cx and Cf is 5pF. I will definitely try other diode when my lab open next Monday I am a noob about op-amp so could you please give an example of active rectifier circuit? The whole circuit is in open air, so I will sure try to cover it with some copper plate next Monday. \$\endgroup\$ – mihota Dec 5 '14 at 16:48
  • \$\begingroup\$ Ok I think the circuit on the wiki page ( en.wikipedia.org/wiki/Precision_rectifier) look fine. Now my question is can I used OPA604 and my 1N4007 for this circuit? It is a bit hard to order things here. \$\endgroup\$ – mihota Dec 5 '14 at 17:11
  • \$\begingroup\$ The circuit you found is fine. (there are more advanced versions, but start simple.) The OPA604 is fine also. The problem with the 1N400x series of diodes is that they are mostly made for high current rectifying circuits and they will be slower than a small signal diode. Here, mccsemi.com/up_pdf/1N4001-1N4007(DO-41).pdf Reverse recovery time is 2us... 100kHz is perhaps fastest you'd want to go. \$\endgroup\$ – George Herold Dec 5 '14 at 18:01
0
\$\begingroup\$

The capacitance of the diode is unbalancing your feedback network and being modulated by the output sine wave from 19pF to 4.5pf from 0V to 10V of reverse voltage (see the picture) below. This is way higher than the feedback capacitor when the voltage is near zero, and that is why the problem went away when you lowered the feedback resistor because the reactance of the feedback capacitor became less significant in the feedback network and stops the oscillation. you can also stop the low frequency noise by adding a series resistor to the output of the OPAMP. enter image description here

1N4001-1N4007 capacitance vs reverse voltage.

The reason you get the sign wave in the output is that because the diode has internal capacitance there is a reactance that permits the sine wave to pass and acts a capacitive divider in series with the capacitor being measured.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.