Measuring Power Output

[harry]

Is there an ideal way to do this? I'm modifying a 50 watt PA head, it uses 2xEL34s, yet it seems a little bit quieter than 50 watts. I'd like to check what it's actually putting out. Do I need a dummy load?

Many thanks for any suggestions!
Harry

[Enzo]

Yes you absolutely need a load - dummy or otherwise. Tube amps love to destroy their output transformers without a load.

COnsider that the whole rest of the amp has to drive those EL34s to full power.

The cheap and dirty way to approximate the output power is to apply a signal - I use 100Hz myself - and using a dummy load, measure the AC voltage across the load with a meter. VOltage squared over resistance calculates power.

[harry]

Thank you for the response, Enzo. Would there be any problems with using the speaker though? Or is it just less risky to use a dummy load (one less thing to break!)?

When measuring the output, do you just increase the amplitude of the tone going in until the output stops going up (and it can be assumed that the signal is clipping)?

[Earl]

I've never needed to do this, but have had a couple of thoughts about it -

One is - that if you use a speaker for the load, will it's impedance cause a phase shift between voltage and current...making the actual power off by some "power factor" percent?

The other is - if you use a AC voltmeter, how can you tell if the signal at the output terminals is undistorted? The RMS/peak/peak-peak value of a sine wave is easy...but when the waveform becomes distorted, the standard I^2*R, E^2/R doesn't work correctly.

[Enzo]

It would be a very very rare guitar amp whose output was not distorted at ANY level.

If an amp puts out 50 watts nominally, but due to distortion, it puts out a true 50.2 - or maybe 49.8 - watts, what is the difference to the hobbyist? That is close enough to 50 watts. The goal here is to findout what the amp puts out, but this is not a laboratury test.

Same thing with the speaker versus resistor load. There may well be a power factor difference, but it will be a very small percentage difference. I think Harry is concerned his 50 watt amp is putting out 20 watts, as opposed to worrying about one watt here or another watt there.

As you know, the impedance curves of a speaker are far from flat, so power output of real world signals are nominal at best.

[Earl]

Enzo - just for the sake of arguement, and I'm not disagreeing with you -

Say you had a 20V P to P signal at your output terminals. With a meter that reads RMS (in it's frequency range), you'd see 7.07V. With a nominal 8 ohm speaker, you get a RMS power of ~6.25 Watts.

If that signal on the output terminals were a square wave, and since the RMS value of a square wave is sqrt*duty cycle* peak to peak, the power value would be 25 Watts.

Somewhere in between those two numbers would be the closer answer.

However, I'm not sure how a RMS voltmeter reads square wave signals....

[Enzo]

And i am not sure how many Fender Twins - or dual EL34 PA heads - will crank that square wave into a speaker.

My Fluke 75 here just measured a square wave at 6v p-p as 3.34vAC. It is not a true RMS reading meter, but it will more or less accurately measure sine waves up to about 1kHz. DOn't know how different a true reading meter would show. I have some AC meters and other fancier test gear, but I don't want to drag it out at the moment. In the course of guitar amp repairs, I don't use them. And yes, the same size sine measured about 2.1 as we expect.

COntext is everything. Looking to find a rough estimate of output power in a tube final stage is an approximation at best. If we were power matching transmission line feeds or something, then we might need better measures.

[J M Fahey]

Load your amp with a >50W power resistor of appropriate value (4/8/16r) or fashion one from a piece of nichrome wire heater resistance element. You won´t be able to solder it so twist tightly its ends with the connection cable and put it in a porcelain or glass container filled with water. Get a clean sinewave, either from a generator (build a very simple 1 kHz one) or a Test Frequencies CD or generate it with some suitable software and your PC´s soundcard. If you have an oscilloscope to check your distortion, good; otherwise connect any speaker to the same output, with a 470r or 1K resistor in series, *just to hear the grittiness that indicates distortion*.
Measurement: with your load and "distortion detector" connected rise slowly the volume until the sound quality suddenly worsens, gets grittier. You can rise and lower the volume to pinpoint the exact onset of distortion. Measure : the AC voltage ; that value squared and divided by the impedance gives the true RMS power.

[MWJB]

Why not leave the speaker connected so you can hear the level of distortion?

I think Enzo summed it up best in his last post, re context.

the down & dirty method (speaker, even using your instrument as a signal) will be enough to tell you if there is a fundamental problem.

What's the use to a guitarist anyway if his amp makes 47.5W with a 100mV signal into a dummy load, if he can't get near the amp on a tight stage with a hollow body guitar with P90's and a lively speaker? He's never going to subject an audience to 1KHz and make them stare at a sine wave (I'm not knocking the signal generator & scope method as a standard, just suggesting a practical approach for the less technically minded).

[Earl]

To be frank, I've actually never had the need to measure the output power of a guitar amp. They all seem to be pretty loud, and that's usually all it takes .. :-)

OTOH, this board typically seems to take things to a higher level of analysis....that's the main reason I regesterd at AmPage...there's some great minds here!

[Enzo]

To the player, the wattage is not the issue, sound is, as yuo say. To someone working on the amp, it can be a meaningful measure. If your Fender twin puts out about 38 watts, something is wrong.

[Steve Conner]

The way I've always done this is to hook the input of the amp to a signal generator, and the output to a dummy load and oscilloscope. I feed in a sine wave at say 100mV, on the clean channel, and crank up the master volume until I see the output just start to clip. Then I look at the peak voltage on the scope, divide by 1.4 (since I know it's a sine wave) then work out V squared over R.

Scoping the output can tell you a lot of other useful things too.

One thing that bit me once was stray inductance of the dummy load. I was using a big old rheostat made of iron wire wrapped round a ceramic tube. The inductance was enough to screw up the measurement at 400Hz and make me think I was getting 110w, when it was really only 65.

BTW, for a bipolar square wave that goes from +V to -V, such as a clipped amplifier gives out, the RMS, average, and peak values are all the same. The sqrt(duty cycle) thing is only for a unipolar square wave that goes between +V and zero.

[Earl]

"BTW, for a bipolar square wave that goes from +V to -V, such as a clipped amplifier gives out, the RMS, average, and peak values are all the same. The sqrt(duty cycle) thing is only for a unipolar square wave that goes between +V and zero."

Steve, you're absolutely right..thanks for the clarification.

[Old Tele man]

...here's how Fender specifies OUTPUT TEST for '65 Deluxe Reverb:

"22 Watts minimum RMS into an 8-ohm RESISTIVE load at 5% THD with 1KHz input into the NORMAL channel at J1. All controls full CW, all VIBRATO channel controls full CCW, REVERB and VIBRATO off."

...use 400Hz input for bass amps.

[harry]

Thank you for the responses, I will let you know how the measurements go!