How To Measure Watts Per Channel

Hi fellow forum members

Perhaps a "follow up" to the DC Offset thread would be, "How to measure Watts Per Channel."
1- The best way to measure.
2- What we are measuring.
3- How to read (measure) rated distortion. (proceedure/observation)
4- Compare channels.

Ron

1. 8ohm hi power resistive load. scope. audio osc. ac rms vtvm.
2. power: P=E x I ...... (I=Erms/R[8ohms])..... so P=wattsrms
3. distortion analyzer
4. do measurment on each ch. and record results.

I'm sure others will elaborate on this basic info.

oldhifiguy pretty much sums it up.

Need a very low distortion signal source ($$), a dummy load capable of taking the power of the DUT, and a scope to look at the signal to see when it starts to clip.

Also need a very good meter that reads true RMS, and is reasonably wide bandwidth (most distortion analysers also can function as a very accurate AC voltmeter).

Simple formula for power

Here is the method i use. Use a scope and ac volt meter and sine wave generator. drive the amp till the wave form just clips into your load resistors. Then back off the drive a little. read the voltage off the volt meter. use voltage squared divided by the resistance of the loads. for example 30 volts into 8 ohms is 30 squared = 900 / 8 = 112.5 watts

You can read the voltage right off the scope. Drive the amp until it just clips, then back off just a bit. P= peak to peak voltage squared / 8 X load resistance.

Peak to peak NOT

Peak to peak voltage will not give you RMS power. You need to take one half of peak to peak and multiply by .707 to get RMS voltage. Then you can calculate power by squaring the RMS voltage and dividing by the load resistance

peak to peak

If I remember correctly that method will give readings that are high. I think you have to convert the peak to peak voltage to rms by multiplying the ptp value by .707. then plug that value into the formula. Most meters read directly in rms volts.

grateful said:

If I remember correctly that method will give readings that are high. I think you have to convert the peak to peak voltage to rms by multiplying the ptp value by .707. then plug that value into the formula. Most meters read directly in rms volts.

Click to expand...

The Vpeak squared divided by 8 x load resistance takes care of that. This is because power equals effective voltage squared divided by load resistance. Since effective voltage for a sine is peak divided by square root of two, and peak is peak-to-peak divided by 2, we get effective voltage = peak-to-peak divided by 2xsqrt(2). You square that and get peak-to-peak/8. Then divide by load resistance, and voila, you get exactly what was written two posts up.

Output power is measured at a set level of distortion, best observed with a distortion meter or spectral analyser, usually this is either the rated distortion, 1% or 10%. Usually, wideband measurements are made at 1% distortion, 1kHz either at 1% or rated distortion.
That being said, distortion increases very rapidly at the onset of clipping, so even just verifying the waveform visually on a scope is enough. No visible clipping on the scope will put you at about 1-2% distortion max - backing off just a little will drastically reduce distortion but only slightly (few %) reduce output power. Once you got to that point, measure the voltage.

Output voltage is best measured with a true RMS meter, or at least a wideband VTVM or regular voltmeter. Most RatShack digital meters are quite bad over some 300Hz unless you go for an upscale model, and most of those don't do too well over a few kHz, but even that is usually enough (see below). An aside: non-RMS voltmeters measure peak values but are internally calibrated to show 0.707 of it in AC volt mode, assuming you are measuring sinewaves and want to know the effective value. This assumption is fine for our application, but may be totally inapropriate for other applications.

Some words of warning:

1) Avoid measuring the full power bandwidth of an amp or indeed measuring power output at very high frequencies (say over 10kHz). If you do it, you need to do it very quickly. The reason for this is that the power at such frequencies in an audio signal tends to be very small even if the amp is driven to clip parts of an audio waveform. Assuming a non-inductive and powerful enough dummy load and a good power amp, the amp and load will have no problem with it, but the output amp compensation network (sometimes referred to as a Zobel network, the series cap and resitor in parallel with the output) will not take this for very long as it will be taking over the role of a load to a signifficant degree - and will eventually (sometimes this is measured on the order of 10-20 seconds) burn out the resistor. Full power bandwidth is best measured with the zobel taken off, and the dummy load connected via twisted pair directly to amp output on the board. All of this assumes the simple generator-amp-dummy load-scope+meter hookup. There are other much more complicated methods of measuring that do not require these precautions.

2) The maximum power dissipation of a class (A)B amp occurs at around 1/3 of the power output. Very few amps are designed to provide this power level on a continious basis. Some are not designed to provide the full power output on a continious basis. This is because the average power of an audio signal is only s small fraction od the peak power. Even so, an amp which cannot withstand these conditions for at least 10 minutes should be considered a bad design - or, put in a place that does not provide sufficient ventilation! In other words - keep the measuring short and/or ventilate the amp well. Keep in mind that testing for output power at low impedance loads stresses the amp considerably more than normally. Especially avoid asking for it by 'testing' the protection circuitry - it does not protect against stupid users! For low impedance loads, heat generated in the power supply and transformer may become an issue as well.

3) Keep your load well connected. Losing the load at full power can be disaster for some amps, even if they can be used without load assuming they were powered up without one. It is the removal of a load that can kill an amp because of inductive backlash - even connecting a load can cause this since it is near impossible to connect a load without contact bounce, hence a series of connects and disconnects. In other words: do not operate speaker switches at high power levels, keep cables sohort and braided for minimum inductance, and avoid inductive loads (blowing a fuse in an amp with a heavily inductive load can also lead to havoc). Also important to remember with bipolar amps: a hot amp will be considerably less resilient than a cold one.

I find that measuring at the specified output rating (watts per channel) is the most accurate. Read the distortion level at that piont.
Going even slightly over the point where the amp clips on the 'oscope results in distortion rising very fast.

Why do some use the 1% distortion level?

Ron

1% was considered the limit of audible (harmonic) distortion in the days of tube amplifiers. Given the typical harmonic distortion spectrum of (especially a single-ended) tube amp, that rule of thumb is probably pretty reasonable.

You might not like the sonic consequences of 1% THD on a typical push-pull solid state amplifier.

I use a Sound Technology ST1701 Oscillator and Distortion analyzer. This unit combines the wattmeter, oscillator, and auto null distortion bridge in one unit. I interface the Sound Tech and stereo with a ST1200 switch and filter panel. I can switch select 4,8,or 16 ohm 200 watt loads, switch input sources driving the Amp Under Test, and interface the oscilloscope to view either channel, input level, or distortion output, and switch to monitor speakers all by pushbutton. To correctly measure power, the rated Total Harmonic Distortion of the AUT must be known. Both channels are driven equally. Input level is increased while watching the THD. Power is read when the rated THD is reached. This measurement is valid within the AUT's rated band width. Popular number freaks like to measure power at 1 kHz, which typically gives 10 to 20% higher power.
An AC variac is needed for extremetly high power testing to compensate for AC line sag.
Tube amplifiers will show gradual increases in THD as the input level is raised, while transistor amps have an abrrupt knee, called the clipping point, where distortion rises dramatically.
Monitoring the distortion filter output is a neat trick. On tube amps the THD is sine wave like (dominant 2nd harmonic) On transistor stuff it typically is fast spiky stuff with high amplitude but little RMS value (RMS is considered a trick to make transistor "test" better than tube). Transistor amp usually THD is highest at 20kHz, while tube is highest at 20Hz.

Would it be more meaningful to measure watts first than switch to distortion?
Most work I do is on 25+ year old gear. I doubt very much that the distortion level is on spec at rated output.

Ron

My experience with 25+ years classic gear is on the contrary, if the unit got its necessary tune up, power usually exceeds rated output specs, and THD is lower at rated output.
Most amps only have a single transformer, it is necessary to measure both channel simultaneously, so you need two load resistors!

It would be cool if we could find a copy of the FTC wattage ratings rules that were used in the vintage era to see specifically how they defined power ratings and the procedures required. I do recall a pre-warm up period required. It may well be that most vintage amps and receivers exceed their ratings because of these FTC rules that made them have to be very conservitive in their ratings.

Lefty

FTC specified line voltage, both channels driven, 8 ohm load, 30 minute warm up at 1/3 rated power, full bandwidth and distortion specified, power measured RMS.
While it is true you can set to rated power, and get lower distortion figures, there is more fun to be had by finding the point where distortion meets spec, and then measure power. Typically we get a higher power than is specified...a little audio bonus!

Thanks, guys -- interesting post. :thmbsp:

Holy Cow

I have a degee in Computer Science...but feel like a 1st grader....

Seems to me that you should be able to get a 'black box'...? or something, hook it to the speaker outputs, turn on the amp and read it.

I know, too simple...but one can dream. :yes:

Sound Technology and a few other companies make this 'black box', but it isn't cheap.

Think multiple house payments for a decent used one, and the equivalent of a decent down payment on a home for a new one.

It is easy to measure the output voltage (which is what "power meters" do), but with a complex load (i.e., a real speaker) converting that to watts is nontrivial. A signal generator will help, but you still have the challenge of measuring THD, which is again nontrivial (especially in highly-fed-back amp designs with distortion levels that are "in the grass" of even some fine analytical equipment).

The pros have already explained how it is done. "You gotta break eggs to make real mayonnaise".

Easy Method

Just recalled something I read years ago about measuring RMS power that doesn't take much in the way of expensive equipment. Put the dummy load resistors in a oil bath and run balls out for a given time and measure the temperature increase, put the numbers into a calculation and out comes true RMS power :thmbsp: I guess it wouldn't tell you much about the distortion value at the power but sure is a simple idea using basic physics :scratch2:


Lefty