1
\$\begingroup\$

So I grabbed my Fluke True RMS multimeter yesterday and plugged it in to my Samsung Galaxy Note's audio jack, and started to play some music. I saw AC voltage frequency around 0 – 15 Hz (it's way too slow and unhearable). Is that the frequency of the audio signal? Why is it so slow? I thought that audio signals like MP3 music playing comprises of sums of sines and cosines blended perfectly that look like below

enter image description here

Also, why is it that the audio signal doesn't reach 20 Hz – 20kHz which is what I expect to see in the voltmeter.

\$\endgroup\$
  • 1
    \$\begingroup\$ Can you confirm the accuracy of your readings? you ought to know most of the audio which can be reproduced in a smart phone is in the range of 60~18KHz or 20~20KHz in a high quality speakers. Also confirm your Fluke model # and that is has 100KHz BW on the DSO scope \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 2 '12 at 1:40
  • \$\begingroup\$ Fluke 112 True RMS \$\endgroup\$ – WantIt Dec 2 '12 at 2:08
  • 1
    \$\begingroup\$ Your question is inconsisent with a Fluke DMM 112 true RMS meter. It has no waveform display. However you did not mention the frequency of the DMM in the HZ mode which has a 50KHz BW and would count the zero crossings. Your question is inconsistent with the test modes available.. tequipment.net/FlukeDigital112MultimeterSP.html#SC30390 \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 2 '12 at 2:18
  • 1
    \$\begingroup\$ @vvavepacket - David probably deserves the accept for this one, I think he put things pretty clearly in the first place. \$\endgroup\$ – Oli Glaser Dec 2 '12 at 3:59
  • 1
    \$\begingroup\$ People seem to be mistaking the stock photo used as an example for a measurement that has been taken by the poster. It isn't. \$\endgroup\$ – Chris Stratton Jan 9 '13 at 17:07
8
\$\begingroup\$

You are feeding hundreds or thousands of different frequencies into a device that is designed to display a single frequency. What do you expect to happen? At best the device will display just one of the many frequencies that are present in the music (and that number will be meaningless in the larger context). At worse, you will confuse the thing and it will give you a completely bogus number.

If you used an o-scope and used the auto-measure feature to tell you what the frequency is then you would likely get an equally bogus measurement. The only difference is that with an o-scope you could actually see that the number was bogus.

How do you know that the music does not contain frequencies in the 0-15 Hz range? It is entirely possible! Drums will often have some low-level noise in that range that is often not audible but is also not often filtered out of the final recording. (Note for the pedantic: frequencies under 10 Hz are often filtered out, but there could easily be stuff in the 10-15 Hz range). I also recently worked with a recording studio that had equipment putting noise in the 0-5 Hz range into their recordings.

Get an MP3 of a sine wave and redo your measurements using that, not music. If the multimeter reports the correct frequency then my answer is correct. If it reports some other frequency then the problem is something else.

\$\endgroup\$
  • \$\begingroup\$ David your answer is correct. \$\endgroup\$ – WantIt Dec 1 '12 at 22:34
  • 1
    \$\begingroup\$ I'm thinking if no other test gear becomes available then some sound card scope would at least give a better idea of the issue here. Visual Analyser is one of the best I am aware of - it has a basic scope, SA with (vaguely trustworthy) THD+n measurements, Sig Gen, and a few other useful features. \$\endgroup\$ – Oli Glaser Dec 2 '12 at 1:34
  • 1
    \$\begingroup\$ Why the downvote here too? Upvote to compensate. \$\endgroup\$ – Oli Glaser Dec 2 '12 at 2:02
  • 1
    \$\begingroup\$ @Richman - I see your point, but We don't know whether it has true DSO capabilties yet. I'm suspecting that is a picture of the actual MP3 waveform (from some audio software) and the fluke is just trying to display the frequency in counter mode (which will probably just pick up the largest harmonic or just get totally confused) \$\endgroup\$ – Oli Glaser Dec 2 '12 at 2:34
  • 1
    \$\begingroup\$ +1 for you @OliGlaser i dont understand why people downvote your answer. i second you regarding the fluke simply get confused. \$\endgroup\$ – WantIt Dec 2 '12 at 2:38
2
\$\begingroup\$

You're trying to measure the audio waveform using a multimeter?

Well, there's your problem. A multimeter can only respond to changes in the input voltage or frequency up to a few hertz (the display update rate).

You need to use the proper tool - An Oscilloscope.


If you had the multimeter set to measure frequency, you would have a different issue, namely the fact that music is not just one frequency. As such, the multimeter's frequency counter would not be able to find one frequency to "lock" on to, and you will likely get noisy, fairly random results.

\$\endgroup\$
  • \$\begingroup\$ oh i see, then what is the frequency to consider when designing a pre amplifier? is it 20 hz, or 20kHz, minimum maximum, average? \$\endgroup\$ – WantIt Dec 1 '12 at 4:39
  • \$\begingroup\$ @vvavepacket - Human hearing is sensitive to frequencies in the range of ~20 Hz to ~20 Khz, though as you age, the highest frequency you can hear decreases. \$\endgroup\$ – Connor Wolf Dec 1 '12 at 5:30
  • \$\begingroup\$ 0~15Hz would feel interesting on a flight simulator shaker synchronized to music ... real music.. I see another Disneyland Invention coming \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 1 '12 at 15:13
  • 1
    \$\begingroup\$ Since when does a DMM with a waveform display only have a bandwidth of a few Hz. Show me an DMM meter with these specs. Otherwise correct your statement. Not possible. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 2 '12 at 1:48
  • \$\begingroup\$ @Richman - What? What is a DMM with a waveform display? Do you mean a portable oscilloscope? That's not a DMM, you know. \$\endgroup\$ – Connor Wolf Dec 2 '12 at 6:26
1
\$\begingroup\$

Case of the Erroneous Frequency Counter on Audio

It would have been a Fluke if we assumed all DMM's are standard. They're not exactly the same, but pretty standard. One big difference these days is the use of digitizing the inputs and computing true RMS results from instantaneous readings from an ADC. Fluke RMS meters also can measure frequency and the 100 Series Specs have a bandwidth of 50 Hz ~ 50 KHz for voltage and the lowest voltage Full scale is 300mV.

Some even have DSO scope results with limited settings. But not the 100 Series. enter image description here But 300mV voltage scale does not mean you can accurately measure frequency at low voltages. Most often counters with analog inputs have hysteresis to avoid displaying random readings from noise for an unconnected input.

As far as I can tell from the Fluke 120 Calibration Manual, it is only calibrated with a standard 6V signal, so the hysteresis may in fact be much larger than expected. But at least I verified it will measure frequency from 50 to 50KHz.

So if the amplitude of the audio changes the frequency of your measurement, chances are good that you have a noisy signal and it is not measuring all the "zero crossings" We normally assuming it is an AC coupled signal or at least no DC offset as this will affect which zero crossings are counted. Now when one looks at music on a Spectrum Analyzer like AC/DC's Hell's Bells, the spectrum may be fairly broad and flat but I saw the peak well below 100 Hz during the chorus. THis means the average frequency of all the voltages would only be below 100Hz as the higher frequencies do not reach the zero crossing threshold to be counted. They are riding on the crests of each beat and wave. This means EQ and Bass boost will affect the readings as well as amplitude for hysteresis.

Rule of Thumbs

  • If changing the threshold or doubling the amplitude has no effect on the frequency count, the result should be triggered properly.
  • Do not expect accurate readings from low signals with lots of noise.
  • Otherwise, make sure the signal is as strong as the manual's calibration signal.
  • remove DC if any present on signal unless input is known to be AC coupled.

Hypothesis To get the OP's 0 to 15 Hz result from a Samsung Galaxy audio jack, I would hypothesize it was counting a percentage of the number of amplitude peaks above and below both +/- hysteresis thresholds that result in a zero crossing. So if the dominant signal was in 100Hz range and broadband noise just below that ( excuse me .. music from 20 ~20KHz) and the signal was a few volts, it might have only counted 0~15% of the zero crossings that were loud enough.

( apologies for goofing assumptions on previous answer. Not all true RMS DMM's are the same and I saw graph and mis-read)

enter image description here

So if you want to count the root of your squared sums, make sure your signals are big enough to overcome the hysteresis for frequency measurements. ( the Voltage and current readings are instantaneous without hystereis ).. I think that covers it.

\$\endgroup\$
  • \$\begingroup\$ This answer is several times better than your previous answer. I just have one minor point (which does NOT invalidate your conclusion). I doubt that the DMM is using an ADC on the AC input, and calculating the RMS from that. More likely it is using an RMS to DC conversion chip similar to the LTC1966, and then runs that output into their normal low-speed ADC. This would be much lower power than an 200 KHz ADC plus signal processing, and thus have a chance of being powered from a battery for months or years like a typical DMM. \$\endgroup\$ – user3624 Dec 2 '12 at 16:03
  • \$\begingroup\$ Thanks for sharing that. I was not familiar that chip. It appears to use the differential S&H method with a fast sampling rate so 800KHz BW is possible to -3dB while the Fluke has 100Khz for voltage and 50KHz for current. I wonder if the Samsung phone needs to filter the Class D output and how much is needed. LTC chip Note 15 also says "At higher input frequencies, increasingly large portions of this noise are aliased down to DC. Because the noise is shifted in frequency, it becomes a low frequency rumble and is only filtered at the expense of increasingly long settling times. " \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 2 '12 at 16:43
  • \$\begingroup\$ I doubt that the phone output needs to be filtered. That output needs to be filtered inside the phone for EMI and ESD reasons anyway, and doing a proper filter on something that low power is trivial and inexpensive. I have looked at the output of my Samsung Galaxy S2 and it does not have any noticeable Class-D switching artifacts on its output-- but there is a chance that the OP's phone is different. \$\endgroup\$ – user3624 Dec 2 '12 at 16:52
0
\$\begingroup\$

Fake Name is spot on, you need an oscilloscope to work with AC signals, which means it's not an optional piece of equipment if you want to do anything vaguely interesting with electronics. It's the main tool that gives electrical engineers "an eye" into the electronic world (there are others - spectrum analysers (shortened to SA), time domain reflectometers (TDRs), LCR meters, and many other weird and wonderful test instruments)
However, in any standard electronics lab, you are pretty much guaranteed to see a scope commanding centre stage on the test bench.

To answer the comment, when designing a preamplifier there is a range of frequencies to consider.
Most decent audio amplifiers have a range fo approximately 20Hz, to 20kHz (i.e 20 cycles per second to 20,000 cycles per second), so you need to beable to see how your amp performs over the entire range. This bring up another useful tool, a signal generator. As you can guess from the name, these generate various types of waveform at different frequencies (all adjustable of course)

Then there are tools like Audio Precision gear that will monitor the output and tell you how flat your response is and how much THD+n (Total Harmonic Distortion + noise) your amp is adding to the input signal (a very clean source signal and an SA can come in handy here too if you don't have lots of money for AP gear)

\$\endgroup\$
  • \$\begingroup\$ Is the OPs photo a fake or does it really have a digiztized display representing the signal. Isn't that the def'n of a DSO? I know there Quality factors regarding the aliasing filter and sampling rates. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 2 '12 at 1:51
  • \$\begingroup\$ @Richman - not sure, it looks strange to me - wouldn't mind knowing the model number. I know Fluke make handheld scopes but I haven't seen a display like that on one. \$\endgroup\$ – Oli Glaser Dec 2 '12 at 2:00
  • \$\begingroup\$ Whoever downvoted - why? \$\endgroup\$ – Oli Glaser Dec 2 '12 at 2:01
  • \$\begingroup\$ I should point out that your answer does not address the main questions in the OP. It does have useful and correct info, just not directly addressing why he was getting the low frequency readings. I didn't do the downvote. In fact, I upvoted just 'cuz I don't want you to take this comment too personally or critically. \$\endgroup\$ – user3624 Dec 2 '12 at 3:42
  • \$\begingroup\$ @DavidKessner - no problem - I think your answer is better worded actually. I initially though the image was acually from the Fluke (some handheld DMM/scope I'm unaware of), which was what confused me - my comment to your answer is what I'm thinking is happening, which is essentially what you appear to have been saying from the beginning. Certainly if it's just a counter it's not going to display anything sensible for anything harmonically complex and time variant. Thanks for the upvote anyway :-) \$\endgroup\$ – Oli Glaser Dec 2 '12 at 3:55
0
\$\begingroup\$

A frequency counter is a device to count how many times per second some kind of event happens. In most cases, the event would be an electrical signal going from a voltage below some threshold to a voltage above it, and then back below. One should not think of them as "frequency detectors" in the linear-waveform sense, but rather as "event rate" meters.

If one feeds into a frequency counter a waveform where each period contains a "high" portion, a monotonically decreasing portion, one "low" portion, and a monotonically increasing portion, then each cycle of that waveform may be detected as an "event", and the rate at which those events occur will match the fundamental frequency of the waveform. On the other hand, if one feeds in a more complex waveform that does not meet the above criterion (implying that it contains substantial amounts of several frequencies, none of which dominates) then a count of how many times the voltage goes between high and low levels will be relatively meaningless.

The frequency measurement feature on Fluke meters is designed for measuring the rate at which clearly-defined and regularly-spaced events occur. It is unsuitable for measuring events which are not clearly defined, and is of only marginal utility with events that are not regularly spaced (the more irregular the events, the less useful the readings).

\$\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.