Current measurement with noise.

[paulca]

I want to measure current in mA through my headphone amplifier.  I put a 0.1 Ohm resistor in the low side of the circuit.

The scope GND connected to the "floating" power supply NEG output.  (on the side of the 0R1 connected to the PSU -).  I understand this will earth the PSU it's fine.

The scope probe was clipped to the other side of the resistor.

On the scope however, all I can see is 2 traces.  Not 1, 2.  The smaller low frequency signal (the current/voltage drop I want) is modulated on top of a larger wide noise band in the high kHz and MHz.  It's there even with the 20Mhz bandwidth limit on.

Neverminding the particular measurements in this case, it's a recurring problem I have with the scope.  Seeing a "wall" of noise with your signal modulated on the "top trace" and the "bottom trace".

Are there any trips tricks to collapse these? 

In the particulars of this test:  The circuit under test is a headphone amp.  The signal it is receiving is coming out the back of a supposed "output filterless DAC", which considering it's output has up to 400mV peak to peak digital high MHz I2S and clock noise might .... you know... need a filter on it.  This is almost certainly the main source of noise.  Secondary source would be the radiated noise from the same digital circuit.  I may need a bigger resistor or a few 0.1R in series to get to see a 40mA trace.

So, other than a low pass filter on the DAC output, is there any scope magic I can try?

[paulca]

Knew you might ask.  Here is a basic screenshot.

[mikerj]

The scope GND connected to the "floating" power supply NEG output.  (on the side of the 0R1 connected to the PSU -).  I understand this will earth the PSU it's fine.

Did you physically clip the ground probe to the 0v side of the resistor, or somewhere else on the 0v rail?  You need to probe right on the resistor or as close as possible.  If the current through the resistor is modulated then you can try enabling averaging in your scope, assuming it has this feature.

[EEEnthusiast]

Most likely that the probe is forming a loop and picking up noise. If you are interested in the DC value of the current, then just use a multimeter across the 0.1R . With the averaging and low bandwidth of the DMM, you may get better results.

[jonpaul]

wrong method use a VOM or millimeter,to check,  DC current.

If you use a sensé resistance it should drop 100 mv or so, 0.1 ohm may be way low.

What's the spec of the amp current draw?

j

[paulca]

Just catching back up now.

Sorry.  The issue is the current sense resistor.  The problem isn't too much noise it's too little signal.  The scope is set to 10mV per div.  I was above to post another question about how to make it quieter.   It's only 40-50mVpp noise.

I had been measuring something else which was pulling about 0.4A DC.  I tried a 10R and got burnt, nearly (did momentarily) short the scope earth with 16VDC via the 10R.  Hence burnt finger.

Still 04A across 10R is going to burn anyway.   It prompted me to order some 0.1Ohm resistors.

Then I set about testing the headphone amp.

Dumb.  Basic maths error.  The headphone will be pulling 20-30mA quiescent and 60-80mA flat out.  A 0.1Ohm resistor will be measuring 2 - 8 mV.  My scope can't do that.  Not with "any" noise at all.  The fact I could recognise the signal on top the noise is a credit to it.

I ordered 1Ohm 1/2Watt 1% resistors.

Next time I will do the maths and not get burnt or... go the other way.

Also... wrong tool indeed.  For these kinds of things involving power measurements I'm going to use the cheap chinese 100Mhz / 20Mhz scope which is DC powered and floating.  That way I can't, as easily short something and burn my fingers!  More importantly it won't be the $500 scope taking the school boy error.

EDIT:  I am hoping to measure the AC signal on the current caused by the amplifier "doing work".  I'm hoping the questiant and noise current can be made to go away and it "should" show something like the real waveform.

[paulca]

Is putting the scope probe into x1 mode worth doing as "noise supression"?

[bdunham7]

Knew you might ask.  Here is a basic screenshot.

I have no idea what is going on with that screenshot.  You could try using 1X, changing timebases (faster) and maybe taking a single shot and zooming in on it to see what is going on.  If I understand what you are doing--measuring your DC current supply--you shouldn't be using AC-coupled input.  With your current levels you should be able to see a reasonable signal even with a 0R1 resistor.  I actually use little 0R1 resistors like this all the time.

[Andreas]

is there any scope magic I can try?

Hello,

As 100 kHz or 1 MHz bandwidth should be enough you could further reduce the bandwidth by a math function for the channel.
(sometimes it is also called "high resolution" display)

with best regards

Andreas

[fourtytwo42]

Well it could be your headphone amp is class-d quite common for that application and that is the source of the switching noise or it could be some switching noise from an smps getting into your scope. The two approaches I would try is firstly differential measurement, that is two probes on on each end of the resistor and use the scope maths to invert one and sum the traces, the ground clips can be on the chassis. The other thing I would try is a bypass capacitor across the current shunt to provide a path for the HF switching noise. In either case scopes are often not as sensitive as needed so a good dmm on averaging mode might be better BUT you cannot be certain what it is measuring 

[paulca]

I have the average current provided by the power supply or even an inline current meter (DC jack in and out) I made.  However that's not the part that interests me.  It's the peaks.   When a big sound happens at high volume I want to know what kind of current I am dealing with.

It has occurred to me the headphones themselves are a kind of current sense resistor and if I just put the driver ohms into Ohms law with the signal voltage it should give me an approximate current.  Close enough.  It's how I speced the amp, but I've never tested it.  I could maybe do that with the math function or just do it in my head.  Of course as it's AC signal it might require a bit more mathematics than that.  Also I don't think headphone drivers make good current sense resistors.  Pretty sure the speaker itself will have an impact on the current.