Output voltage waveform of a PP3 battery

[onemilimeter]

Heard before that battery will give "clean" voltage. Figure below shows output voltage of a PP3 battery. The peak-to-peak voltage is approximately 2mV. The measurement was done with several capacitors (electrolytic 220uF x 1, ceramic 0.01uF x 1, ceramic 0.1uF x 5, ceramic 0.022uF x 1)  connected in parallel with the output. It looks to me that adding capacitors will not help to reduce to the output noise. The scope is Tektronix TMD2024B. Any suggestion to reduce the noise below than 1mV? Cheers.

[scrat]

I think that only a low-dropout regulator can improve this.

[mikeselectricstuff]

Heard before that battery will give "clean" voltage. Figure below shows output voltage of a PP3 battery. The peak-to-peak voltage is approximately 2mV. The measurement was done with several capacitors (electrolytic 220uF x 1, ceramic 0.01uF x 1, ceramic 0.1uF x 5, ceramic 0.022uF x 1)  connected in parallel with the output. It looks to me that adding capacitors will not help to reduce to the output noise. The scope is Tektronix TMD2024B. Any suggestion to reduce the noise below than 1mV? Cheers.

Are you sure that's not just scope noise? 2mv p-p seems in the right sort of zone for input amp noise. How does it change when you replace the battery with a short?

[ejeffrey]

I am pretty sure you are just seeing the input noise of your scope.  My TDS2024 displays a very similar trace with a 50 ohm terminator over the input.  The main cause is johnson noise in the input attenuator resistors.  Also, 2 mV p-p is a hand-waving 0.4 mV RMS, which is equivalent to 28 nanovolts / sqrt(Hz) spectral density over a 200 MHz bandwidth -- not a bad number overall.

If you want to see the real noise of the battery, you could try building an AC-coupled amplifier with a gain of 100 or so.  You can get op-amps with ~3 nV / sqrt(Hz) input noise, and discrete transistors to 1 nV.  Then compare the noise with the battery attached vs. shorted inputs.  I still don't think you will be able to see the difference.

[onemilimeter]

Are you sure that's not just scope noise? 2mv p-p seems in the right sort of zone for input amp noise. How does it change when you replace the battery with a short?

mike, I think you're right. I removed the battery and placed a 10kohm resistor and a 0.1uF capacitor (both connected in parallel) across the probe and the noise is approximately 2mV peak-to-peak too.

Does it mean that the TDS2024B is not suitable for voltage measurement lower than 2mV?

[onemilimeter]

I am pretty sure you are just seeing the input noise of your scope.  My TDS2024 displays a very similar trace with a 50 ohm terminator over the input.  The main cause is johnson noise in the input attenuator resistors.  Also, 2 mV p-p is a hand-waving 0.4 mV RMS, which is equivalent to 28 nanovolts / sqrt(Hz) spectral density over a 200 MHz bandwidth -- not a bad number overall.

If you want to see the real noise of the battery, you could try building an AC-coupled amplifier with a gain of 100 or so.  You can get op-amps with ~3 nV / sqrt(Hz) input noise, and discrete transistors to 1 nV.  Then compare the noise with the battery attached vs. shorted inputs.  I still don't think you will be able to see the difference.

jeff, you're right. I think it's the noise from my scope. Do you mind to share how did you get the 28 nV/sqrt(Hz)? Cheers.

[ejeffrey]

jeff, you're right. I think it's the noise from my scope. Do you mind to share how did you get the 28 nV/sqrt(Hz)? Cheers.

There is a rule of thumb (based on an assumption of gaussian white noise and some sleazy approximations) that you divide peak-to-peak voltage seen on an oscilloscope by 5 to get the RMS. That gives 0.4 millivolts.  Then you just divide by the square-root of the noise bandwidth of your scope.  Unfortunately this isn't known unless you have a calibrated noise source, but the 3dB bandwidth of the scope is a reasonable guess.  0.4 millivolt / sqrt(2*10^8 Hz) gives a 1 Hz noise density of 28 nanovolt.

To do a proper measurement you would want to get a real RMS measurement of the noise voltage (the scope's RMS measurement function might be good enough) and either a measurement of the effective noise bandwidth with a calibrated noise source or a spectrum of the measured noise.  However, the rough calculation assuming white noise an a brick wall filter gives something right in the ballpark of what you expect, so I think there isn't a lot of reason to go through all that.

[tekfan]

If your scope has an averaging function you may try that.
You can also measure  the noise with a multimeter set to the AC range. Use the shortest leads possible when doing such a measurment and preferably a shielded can.
I tried it with my multimeter just out of curiosity and I got less than 10?V for 2 AA batteries. The noise is probably much less than this in reality.

[onemilimeter]

... less than 10?V for 2 AA batteries. The noise is probably much less than this in reality.

Did you mean 10uV?

[tekfan]

... less than 10?V for 2 AA batteries. The noise is probably much less than this in reality.

Did you mean 10uV?

Yes I meant 10uV. Looks like you can't write the micro sign.

[onemilimeter]

Did you mean 10uV?

Yes I meant 10uV. Looks like you can't write the micro sign.

True... we can't insert the "micro" symbol.

I followed your method by measuring the battery voltage using HP 34401A. The measured AC voltage was 8uV.

[Zero999]

Yes 10µV. It's possible to get the micro symbol but it's a pain. I have common symbols on Opera notes. Firefox can do a similar thing with the clippings extension.

[Mechatrommer]

i keep a special characters in a file for easier copy paste whenever needed, if somebody interested you may copy as well.
±¢£©® °¹²³ªº × µß ¼½¾ ðØ?€ƒ™œ
last time (older smf) ohm symbol didnt show up, but it seems it working in smf2 edit post box
edit: ekkk! wrong! ohm symbol still show up as "?" after posting.

[Zero999]

It displays on the preview page.

EDIT: Square root and pi don't work either.

[ndictu]

Well, the page is in UTF-8, the preview does display it OK but that doesn't go through database, once you save it gets messed up so I'm assuming the character encoding in database is set up to something wrong. I did a test install of SMF and it's all in latin1_swedish_ci which is mysql default so the installer doesn't specify anything.

It could be probably fixed by just changing the encoding of the post table (or all tables, to avoid conflicts) to some unicode, like utf8_general_ci. UTF8 is a superset of latin1 so it shouldn't break anything, but you never know ...

[onemilimeter]

Adding the following features in this forum?

[Ronnie]

There is a rule of thumb (based on an assumption of gaussian white noise and some sleazy approximations) that you divide peak-to-peak voltage seen on an oscilloscope by 5 to get the RMS. That gives 0.4 millivolts.

A good reference material from Analog Devices regarding RMS to peak-to-peak ratios read figure 1-73 page 85 (1.83) at http://www.analog.com/library/analogDialogue/archives/39-05/Web_Ch1_final_R.pdf