Testing amps for output wattage

[blindboybenton]

Hello
I have a question about testing amps for output wattage

I do all this guitar amp repairs as a hobby really. I don't have the best test equipment
Over the last few months ive had a few people ask me too test their amp because they think its too quite etc. A old retired engineer too me this way of testing

Warm up the amp inject a signal in from a signal generator , wire up the dummy load and connect a DMM across the load set to AC volts. Use the oscilloscope to bring up the gain/vol until you see the signal start too clip. Back it off so the signal is perfect

Then take the reading from the DMM ie let say its 14.5v ......... 14.5 x 14.5 divide by 8 or 4 ( ohms ) = 26 watts

Is this correct ????
The reason why i think its not accurate is sometimes the wattage does not seem right, when i see what the amp is pushing out at max its way more than the amp should be able too ie 2x el34 kicking out 70-80watts.

Im i doing anything wrong ?

My oscilloscope 15mhz is old and not been calibrated , It does the job for tracing the signal. The sig gen is an old gould. If any one could help and keep to simple terms
im not a trained and experienced as you guys

Bassman

[Mick Bailey]

This method only works if your DMM reads true RMS. If it reads peak AC voltage then your output calculation will be too high. If this is the case multiply the reading by 0.707 before squaring it and dividing by the resistance.

Your scope is OK uncalibrated and just using it to view the onset of clipping.

[TomCarlos]

Unless your scope is way out of calibration, you can compare what you see on the scope against the DMM (assuming it is a True RMS meter).

Start with the scope, use the GND option to adjust the horizontal line to sit on the "0" axis. Then enable the AC option. Half the wave should be above the line, half the wave below. The top part of the wave is called the Peek. Take that reading and multiply it by .707. That will be your RMS voltage (and hopefully the same as your True RMS DMM. Same math...

Make sure your dummy load is 4 or 8 ohms and make sure it can dissipate the heat generated for that load. Here is a photo of my dummy load... a row of metal resistors... 1 row is 16 ohms, another 8 ohms, another 4 ohms.

[Chuck H]

+1 to the above advice. I'll just warn that once you start calculating output "accurately", you may be disappointed by the truth. Example... There's no way a Marshall 50W head, with 2xel34's at 420Vp, is putting out 50W. Very often closer to 35W or 40W But then, we don't use such an amp completely clean all the time. The closer the wave form gets to a square wave, the closer the true RMS voltage matches the peak voltage. In this case it can be surprising how many watts an amp is making. In either the true, clean watts or the clipping wattage you can pretty much toss any numbers attached to an amp by the manufacturer out the window.

Though I still call any guitar amp with two big bottles a "50 watt" and one with four a "100 watt". Likewise, my personal little amp is a sort of "18W" thing that only puts out 14 clean watts

[Malcolm Irving]

To check on your 'scope - how about setting it to DC and looking at the voltage reading it gives for a 9V battery and compare this with the reading on the DMM?
It won't be 9V exactly but the DMM should be pretty reasonable for DC volts. If the 'scope checks out OK, you can then be more confident about reading the peak-to-peak voltage for a good sine wave on the scope. Then Vp-p/(2 x sqrt(2)) = Vrms and (Vrms x Vrms)/R is the power.

[Leo_Gnardo]

I'll just warn that once you start calculating output "accurately", you may be disappointed by the truth. Example... There's no way a Marshall 50W head, with 2xel34's at 420Vp, is putting out 50W. Very often closer to 35W or 40W But then, we don't use such an amp completely clean all the time. The closer the wave form gets to a square wave, the closer the true RMS voltage matches the peak voltage. In this case it can be surprising how many watts an amp is making. In either the true, clean watts or the clipping wattage you can pretty much toss any numbers attached to an amp by the manufacturer out the window.

Though I still call any guitar amp with two big bottles a "50 watt" and one with four a "100 watt". Likewise, my personal little amp is a sort of "18W" thing that only puts out 14 clean watts

I've been doing power measurements much like Bassman1965 ever since dinosaurs stalked the earth... works OK for me. And as you say Chuck it almost always falls short of the manufacturer's claimed power. By 40% in the case of those wonderful old Marshall 50W heads but they sound so good I forgive 'em. Something to think about, the power claimed is often at some 5 or 10 or 20% distortion figure, not some fractional per cent the way hi fi amps are measured. That way MI manufacturers can play the specsmanship game, try to impress the potential buyer that they're going to get more more watts therefore a better amp, for those who think along those lines which includes about 95% of amp buyers.

Running against the trend, nice old Marshall 100W often turn out more than 100W, some I've seen in the 120W range. Not that it's a whole lot louder, just proves it's a very healthy amp in the power department. Some guitarists around here like running theirs with KT88 and that's a treat to hear and the power figure almost always in excess of what's claimed. Good all around.

[R.G.]

I had to sit through a semester's worth of lectures and homework on measurement and instrumentation. Seemed ridiculous at the time.

Analog meters are fundamentally current meters, with resistors and other things arranged around them to make them read out in applied voltage, taking into account the resistance of the coil wires inside the meters. They read only average DC, with a bit part of the averaging being the meter's mechanical inertia. The AC ranges are done by inserting a capacitor to take out any real DC, then rectification by one or another means, and reading the average DC current from that voltage. What changes on the AC scales other than DC blocking and rectification is that the scale is then marked in "AC volts". The scale markings are what the RMS of a sine wave of that size would be.

On to DMMs. Unless you know the insides of your DMM, including the programming that is probably inside the uC that does the work, you can't tell much. What you can do is feed it various known signals and waveforms and find out how it acts. The early, cheap DMMs were 0-2V or 0-200mV A-D readers and were scaled and the decimal point put in place to match the scale switch. AC was read in much the same way as with analog meters, by DC-block/rectify/filter/scale-the-answer.

Fancier DMMs and later DMMs may do anything. I don't actually know the insides of any of them, but I suspect that fancier true-RMS meters actually do a raw sampling of the input, be it AC or DC, then go compute the RMS.

The issue you're facing is calibration. Your scope is showing you oddities. We had several lectures on calibration, too. The standard for calibration used to be some master reference kept by the National Bureau of Standards, and all secondary masters were calibrated to that, and other tertiary standards to those and so on. The string of references was kept written down, and when an instrument was formally calibrated, the sticker on it would indicate the calibration date and have the note "traceable to NBS" on it, indicating that the chain of references was documented.

It's expensive and time consuming to toddle off to a calibration lab to get things calibrated, so all us home hacker engineers used reference cells to be sure we were not too far off. These "reference cells" are called "batteries". The output voltage of a battery is a reading of the electrochemical potential caused by the atoms in the reacting chemicals in the battery as long as it's unloaded. So a meter directly across a fresh battery is a good, if not perfect, way to roughly calibrate a meter, crudely at least. The variance from ideal is that there are different electrochemical voltages for the same battery at different points in its life and different temperatures.

If I were worried about answers that were grossly off, more than a few percent, I would go buy a fresh alkaline battery and use that to "calibrate" the meter or scope. Fresh alkaline batteries have an open circuit voltage of 1.60V. This drops precipitiously to 1.5n volts under any load at all, but a meter or scope lead isn't much of a load. So read the battery with the meter, then the scope, and note any differences from 1.60V. That will tell you pretty quickly if you have gross (i.e. > a few percent) errors.

[J M Fahey]

^^^^^^^^^ that
Beyond instrument quality, precision, etc, which is fine but may have been lost along the way, I love the "back to basics" approach.

The fresh alkaline battery is as good as it gets, and in fact not far from the original standards , which are often based on fancy chemistry "reference batteries" which can be bought at Lab supply shops.

The standard experience is finding amps deliver less than advertised, specially Tube amps, mainly because tubes are current limited and inefficient by definition.

If the amp measures more, like in the OP post, the most common explanation is that load is higher impedance than expected, which happens with real speakers all the time, because they are nominal impedance only at a narrow range of frequencies, usually between 250 and 350Hz or so ... a frequency nobody uses for testing

At 1 kHz which is a typical test one, it's noticeably higher, and of course measurements are inflated by the same amount.

Or the AC meter used may be uncalibrated.

In general the DC side is incredibly accurate, as R.G. states above, basic "meter" is a very good DC to Digital converter and display, 0-200mV or 0-2V, with adequate attenuation at the input to measure higher voltage, very easy, but then for AC you must convert that to DC and there is where you open the can of worms.

I suggest you repeat the test, rechecking the load resistance just in case, and if younused actual speakers out of convenience, repeat using plain resistors.

A side problem may be that many power resistors, specially if based on Nichrome wire ( the most common resistive wire, such as found in stoves, toasters, hair dryers, etc.) rise their resistance a lot when hot, again twisting measurements.

[blindboybenton]

Many thanks for all this info. I will try the 9v battery and see how accurate the scope is and see what i get

Bassman

[tedmich]

a ternany scale: "off" "loud" and "broken"
has always worked well enough for me.

[blindboybenton]

im sorry too say that my ears don't have a built in RMS meter

[g1]

It should also be mentioned that it is possible to distort the signal without actually clipping the power amp, and this can affect the reading.
The amps controls need to be set so what you see at the output is the same shape as the sinewave at the input.
If the amp has an effects return or power amp in jack, you should use them for the power measurement, it takes the other controls out of the equation.

[nickb]

Can I just add that if you don't happen to have a standard battery lying around, a $1 7805A linear regulator hooked up to your 9V battery gives 5V at 2% accuracy. 2% is good enough for the vast majority of measurements.

[Enzo]

And guitar amp preamps rarely do clean at all volumes, so if you are looking for max clean from a powr amp, you should feed the test signal directly into the power amp.

If your amp was making a true 50 watts, and now is only making 25 watts, the result will only be 3 decibels less loud. That is not enough, just enough to notice. So when someone says their amp is not loud, it often is not power output. Power is not loudness.

[R.G.]

Power is not loudness.

Dead right! Knowledge is power, and loudness is definitely not the same thing.

In fact, loudness may be the opposite of knowledge, so where does that leave us?