Measuring output power?

[tompagan123]

Greetings,

How do you all measure output power?  I've seen a YT video where Gerard Weber measures the AC volts across the speaker and applies ohms law to calcutate the power.  Easy peasy.  Anyone else do it like that?

Thanks again!

Tom

[sluckey]

I do.

[shooter]

Just finished calculating gains for each stage n overall Amp pwr.  Pre = 59, Pa=16.5  pwr =21.4 which shouldn't be since it's PSE el84's

[HotBluePlates]

Greetings,

How do you all measure output power?  ... measures the AC volts across the speaker and applies ohms law to calcutate the power.

General Radio made the 1840-A Output Power Meter. After you strip away all the bells & whistles, GR puts a resistance from hot to ground at the output of a device and measures voltage. The switching and metering is then calibrated to represent that voltage in terms of decibels and power (because the ultimate load resistance is a fixed constant; therefore, measured volts converts directly to measured power).

If you don't have a load resistor, you can just measure at the speakers terminals.

[jjasilli]

Just finished calculating gains for each stage n overall Amp pwr.  Pre = 59, Pa=16.5  pwr =21.4 which shouldn't be since it's PSE el84's

I think power ratings are nominal by necessity.  Without measuring, output power can be calculated using the Ohm's Law power formula, and plugging in the plate voltage, and primary impedance of the OT (assuming a correct speaker load).


If you measure per a real speaker:  Does it's nominal impedance match the OT secondary?  If NO, then you get a different power reading.  If YES, still the speaker shows Impedance -- i.e., a different fixed resistance for every frequency.  The industry standard is 1000 Hz with a clean (undistorted) output wave; but different speakers will have a different resistance at any given frequency.  So the amp's output power reading may change with different signal frequencies, and with different speakers at the same frequency.


If you measure per a fixed resistor, similar complications persist.

[tompagan123]

Ah.. So then as far as signal quality at the measurement point, that would have to be subjective.  perhaps the highest voltage achievable for a clean sounding open E chord?    "clean for  a guitar amp could actually be a really high THD.   Maybe using Maj7 chord searching for the  point where it becomes dissonant would be useful?    I dunno, just wondering..

[jjasilli]

Power is measured at one specific frequency, from a signal generator.  If there's a range of frequencies -- like from a music source; multi-note guitar chord; white or pink noise generator --  ea. frequency will be at different voltage, to the chagrin of your meter or 'scope, which are not designed to measure more than one voltage at a time (from the same signal).  There are multi-frequency analyzers.  They will show different wave heights for each frequency.  The wave height is the voltage at that frequency.  The wattage (power), is the Area Under the Curve of the Waveform.  That will vary with each frequency.


A similar issue exists with regard to the horsepower of a gasoline engine. It varies with RPM and other factors.  So, the stated horsepower is nominal (and might even be complete hogwash).

[PRR]

> calculating gains for each stage n overall Amp pwr.  Pre = 59, Pa=16.5  pwr =21.4

I do not follow that math.

And gain is not power. What if I had a guitar that put out 10 Volts? Looks like 50,000 Volts out, WOW! Yet as we know, two EL84 can only make 12V or 13V across 8 Ohms.

Speaker is dubious. An "8 Ohm" speaker is above 16 Ohms in much of the guitar power spectrum, and can touch 50 Ohms on the lowest notes. You can easily get an "impossible" reading.

Power tests are quite simple and standardized.

Since there is always distortion, figure how much distortion you will call "clean". In low-fi work this is often 5% THD.

Load with a *resistor*. Because of the many different impedances of real speakers, loudspeaker tests are always run on a resistor.

Feed a sine-wave input. Start mid-band, perhaps 400Hz. For extra fun, later repeat the test across the audio spectrum.

You will generally want any Master Volume FULL up... you are measuring final stage output, not what you get from preamp strain.

Increase level until you hit the THD number.

Without a THD meter, use a 'scope. Increase until the wave is clearly bent. Decrease until not bent. Find a point in the middle where wave is slightly bent. Call that the power.

While you are set up, turn the level up to very-bent and leave it that way for a long time. Then turn down to half of just-bent and leave it that way. This covers the maximum-abuse conditions. If an amp is over-worked, these tests may cause a failure (better on the bench than at the gig). (Max-abuse for class-A is idle, but you've probably run that test already.)

There is also the early-Traynor technique. Again, a mid-band sine into a resistor, but you use an Averaging meter (old passive needle-meter) and find the maximum you can get. This will typically be rounded squares out, 50+% THD, and 1.5X-2X the "clean" power output. In some sense, this is "cheating", but in guitar-land it is appropriate for highly overdriven tones. It also tells directly the thermal power that the speaker may have to handle.

You can instead use a Peak-to-Peak voltmeter (some not-all VTVMs). For clean-clipping amps, the P-P reading will rise to a point and then plateau, even as you crank more signal through. Back off a bit from that plateau and call it Power. This will be good for push-pull 6L6/6V6 with NFB, dubious for the EL types, and often way-wrong for single-ended amplifiers (which clip asymmetrically).

It is always useful to have a low-level monitor speaker across the load resistor. The meter will not tell you if the amp is making weird sounds as it strains. Often the progression of harmonic overtones is more musically interesting (or distressing) than the actual power. Put 100 ohms in series with a speaker and bridge that across the load resistor.

[Jim Coash]

Greetings:  Of all the questions I had to deal with during my 40 years in the A/V business, those dealing with power were the most frustrating.  So many times I lost a sale to another store because the salesperson there managed to convince the customer that his package offered more "watts" for the money.  We all are familiar with Ohm's law, but its relationship to power is so difficult to explain.  I always had to give a detailed explanation about speaker impedance, efficiency, headroom, bandwidth, IM, THD, RMS and Peak power.  Almost always to someone who could not really understand the science.  The most common statement I made with regards to power was that watts were to speakers as horsepower is to tires.  Speakers do not have watts, tires do not have horsepower.  The amount of power a speaker can handle has much more to do with clipping than actual measured wattage.  Distortion, not power, damages speakers.  When you hear distortion coming though a speaker it is almost always the speaker playing the amplifiers distortion, seldom the speaker itself distorting.  Of course, tube amps clip smoothly with pleasing overtones and harmonics which is why we love them and why they do not cause damage as often as solid state amps.  Suffice it to say that over driving a transistor amp is not pleasant and I find it clearly audible long before damage occurs.  In more than 40 years of using sound re-enforcement systems of very high power and quality I have NEVER damaged a driver.  I have repaired dozens of speakers with damaged speaker elements.  Invariably, getting the customer who damaged his/her speakers into more efficient speakers (like E/Vs) or a larger amplifier (like Carver Pros) solved the problems for good.  I choose to do both.  My gig (performing) system uses active crossovers, the largest amps available and the best speakers I have ever found.  Jim

[jjasilli]

". . . watts were to speakers as horsepower is to tires.  Speakers do not have watts, tires do not have horsepower."


Generally, your points are well taken, but the quoted statement is incorrect.  Tires are not rated in terms of horsepower, but speakers are rated in watts.  Watts are not something that a device "has".  The wattage rating states how much power the device can consume, produce or transfer, without risking damage (likely to be caused by overheating in a device rated in watts.)


Ohm's Law clearly states that electrical power is: Watts = Vsquared / Resistance.  But this holds for DC current only.  It  does not directly apply to AC such as musical signal.  AC is the realm of Impedance, not Reistance.  The nominal load of a speaker, such as 4 or 8 ohms, is often used.  But the wattage so stated wil only be true for a linear waveform at a frequency which actually "sees" that specific load.

[2deaf]

Power is measured at one specific frequency, from a signal generator.  --  ea. frequency will be at different voltage

You can measure power at any frequency the power tube will handle as long as you know the load impedance.  The amp doesn't care about the original source of the signal.  The maximum voltage will only vary with frequency if the load varies with frequency.  So if you have a constant load (like a resistive dummy load), the voltage is going to stop at the same maximum point whether the signal is a 1KHz sine wave or white noise.  That point is what amp manufacturer's are going to use when they rate their amps since it is the maximum before a wave clips.     

[HotBluePlates]

". . . watts were to speakers as horsepower is to tires.  Speakers do not have watts, tires do not have horsepower."

Generally, your points are well taken, but the quoted statement is incorrect.  Tires are not rated in terms of horsepower, but speakers are rated in watts.

I think Jim meant that speakers don't have watts (the amplifier does), any more than tires have horsepower (the engine does).

Ohm's Law clearly states that electrical power is: Watts = Vsquared / Resistance.  But this holds for DC current only.  It  does not directly apply to AC such as musical signal.

That is a false assertion. The Equation for Power is a fundamental electrical relationship which always holds true.

Ohm's Law is Current = Voltage / Resistance (or any algebraic permutation of that statement).
The Equation for Power is Power = Voltage * Current (no mention of alternating or direct current).

You can replace the term "Current" in the equation for power with the Ohm's Law statement above it, and yield Power = V²/R.

"Impedance" at the end of the day is "opposition to current" same as "Resistance." The only difference is impedance includes the frequency-variable quantity of opposition, Reactance. If you need a universal equation, replace "Resistance" in any of the above formulae with "Impedance;" they'll all still work exactly the same, and if you calculate the reactance at any frequency of interest and use vector algebra to combine that with resistance to get the correct resulting impedance and phase angle, then all the math still works the same.

... The nominal load of a speaker, such as 4 or 8 ohms, is often used.  But the wattage so stated wil only be true ... at ... that specific load.

The edited quote above is an accurate statement (though the complete original statement was not).

It's like this:

If you change the load of an amplifier while changing nothing about the amplifier, the power transferred to the load will change. This is described by the Maximum Power Transfer Theorem. The issue gets complicated when you move from "Maximum power is delivered when an 8Ω load is attached to the 8Ω tap" to what's happening inside the amplifier with loading of tubes. So stick with what's in quotes in the previous sentence.

Also, the amplifier itself has high and low frequency roll-offs, so above some high frequency power output will fall, and below some low frequency power output will fall. So you need to pick a point mid-band to measure output power.

Now, acknowledging a speaker does not have a single, constant impedance, and that changing the loading away from some ideal value reduces power output, you land at the recommendation to measure power by using a resistor as the amplifier's load. There is still a.c. across that load resistor, but now you don't need to worry about the resistor's effective value changing. All the math above still applies (though you may have to be smarter than the math to recognize complicating factors, such as a big mid-range dip due to the tone circuit at the 400Hz you planned to use as a test tone).

[HotBluePlates]

But the wattage so stated wil only be true for a linear waveform at a frequency which actually "sees" that specific load.

What is a "linear waveform"?

What I think you meant was a non-distorted sine wave. Really, you can use any stable, repeating waveform, though we typically use sine waves because 1) it makes the math easier, and 2) it's easier to pick a frequency where the amplifier's overall frequency response doesn't skew your results. You could also restate the 2nd point as a clean sine is a single frequency instead of a composite of multiple frequencies, and so a swept sine wave makes frequency response measurements easier to interpret.

Now maybe you were referring to Wikipedia's article on Audio Power. The wild-looking integral in the 1st equation under "Power Calculations" in the end says "Power = Volts * Current" but uses the integral to convert any waveform into an effective-average over time. Which is the same thing we do when we convert alternating sine-wave voltages to RMS; at the end of the day, we're still converting the a.c. to a "... value ... equal to the direct current (DC) that delivers the same average power to a resistor ..." so that the simple to use equation at the top of this post can be used.

So I'll use a sine-wave test tone and a load resistor to measure output power. Or you can use any waveform at all, if you're willing to do integral calculus on that waveform to get an average over time. If you have to be able to use any signal shape, you can again use a load resistor and an IR temperature sensor to measure heat thrown off by the resistor, because that will still be proportional to power output; the resistor does the integral calculus (average power in the form of heating of the load) for you.

[HotBluePlates]

You can measure power at any frequency the power tube will handle as long as you know the load impedance. ... the voltage is going to stop at the same maximum point whether the signal is a 1KHz sine wave or white noise.

I agree with you, but using a sine (single frequency) will be easier for interpretation (as I mention above).

The problem becomes an issue of test-tone bandwidth vs. amplifier bandwidth. And if you throw enough power through a transformer, it will constrict bandwidth more, and now you have yet another factor to interpret.

There are good reasons, based in fundamentals of electronics and a knowledge of how amps work, which caused the "usual test methods" to become the usual test methods.

[jjasilli]

Yes, I meant an output waveform that matches the shape of the input waveform without deviation (distortion).  I.e. the "lines" of the waveforms are shall we say "congruent", differing only in amplitude but not in shape.  So, sometimes described as linear.  The "lines" will be curvy except for square or sawtooth waves or the like.  The difference in amplitude is a given -- it belies the need for amplification of the input signal to begin with. 


But even the mere change in amplitude is an issue.  As amplitude increases significantly even a sine wave begins to approach the characteristics of a rectangular wave.  Put another way, a sine wave in ouput, stretched very tall in amplitude, will no longer be linear with respect to the initial, nicely rounded sinewave at the input.

[jjasilli]

Power is measured at one specific frequency, from a signal generator.  --  ea. frequency will be at different voltage

You can measure power at any frequency the power tube will handle as long as you know the load impedance.  The amp doesn't care about the original source of the signal.  The maximum voltage will only vary with frequency if the load varies with frequency.  So if you have a constant load (like a resistive dummy load), the voltage is going to stop at the same maximum point whether the signal is a 1KHz sine wave or white noise.  That point is what amp manufacturer's are going to use when they rate their amps since it is the maximum before a wave clips.

Yes, but as hotblue intimates, if you strum a chord, all the notes (frequencies) may not be at the same voltage.
And the use of a dummy load to measure output power is fine, but is another form of compromise.  It does not measure the performance of the amp as actually used into a dynamic speaker load. 


Moreover, it takes more power (wattage) to raise bass notes to the same voltage as treble notes.  This is because power is equal to the area under the curve of the waveform.  And there's always more area under the bass note curve (given the same voltage), because bass notes have a longer baseline due to their longer frequency.  So it takes more watts to output a 12vac bass note than a 12vac treble note. 

[PRR]

> Tires are not rated in terms of horsepower, but speakers are rated in watts.  Watts are not something that a device "has".

Horsepower and Watts are the same thing.

1 Horsepower is 746 Watts.

You often see European engines rated in W (or kW) instead of HP.

We rarely rate audio amps in HP because it would look bad. A Fender Twin is 0.06HP. Less than most model airplane engines.

Tires "could" be rated in HP, but it isn't simple. One big difference is that 98% of the power you put in a tire goes to the road. 98% of the power you put into a speaker stays IN the speaker.

If we used tires by jamming the bumper against a wall and spinning the tires, then there would be some use in a Horsepower rating. But there isn't a lot of use for that. Yes, exhibition "drifting", but those guys need more than just smoke, and an HP rating is not much direct use.

[PRR]

> power is equal to the area under the curve of the waveform.

The treble wave does the same power *more* times per second. More power?

Assuming resistance, the power is the same either way.

[HotBluePlates]

But even the mere change in amplitude is an issue. 

Well, everything is, or can be, an issue. The real question is will you settle for it all being unknowable, or is there value in knowing at least something about what the amp does?

For instance, in any car, there are hundreds or thousands of tradeoffs and interactions happening when you push the gas pedal. But most will not look at it as unknowable magic, but see a simple fact of "I push the gas pedal further, the car goes faster" (up to some limit).

And the use of a dummy load to measure output power is fine, but is another form of compromise.  It does not measure the performance of the amp as actually used into a dynamic speaker load. 

We can stipulate this fact; a speaker load will behave differently than a resistor load. The answer to understanding an amplifier's capability is to have a simple starting point, then layer on the additional factors which will alter the answer.

The simple starting point is a resistor load and a sine test frequency, at some maximum level of distortion measured at the resistor. To get the ball rolling, assume distortion is zero when power output is measured, and the load resistor matches the OT tap's specified load impedance.

One can increase the test signal strength until distortion appears, and then back it down just until the point of maximum voltage output and zero distortion. Now you measure the RMS voltage output and apply the math to get the maximum clean output power. You could also look at the size of your test signal and say, "For the point where this signal is injected, it represents the largest signal which can pass cleanly to the final load." That might be important if something in the entire amp circuit could distort before the output tubes. It might also be useful to apply the test signal directly at the output tube grids, if you have a signal source which is clean and can generate those large voltage swings.

Now if you look at a plot of speaker impedance (the faint line on the graph linked), you'll notice that the rated 8Ω is the lowest impedance in the usable frequency range; when you move away from ~300Hz where the speaker is truly 8Ω, the impedance rises. This may have an effect on the amplifier's output power. But to know if it will, you have to know something about the amplifier. Knowing nothing else, we should assume power transfer will b less than perfect because the load isn't matched, so clean output power is lower than our first measurement with a resistor. If nothing else, you will know clean output power cannot be any higher than the value measured with the resistor.

If you used a sine wave for the resistor measurement, you also know something else: Peak power output is double the measured RMS output power (take this on faith for a sine wave, or I can bore you with the math derivation). If the sine wave was so distorted it became a square wave, then the peak power output would be equal to the RMS power output, so you just found something else useful from using a sine and a resistor load: worst-case speaker heating is double your maximum measured clean RMS power output, and you can rate the speaker accordingly if you want to prevent it from blowing.

[HotBluePlates]

I said the changing impedance of the speaker "may" effect the amp's output power for 2 reasons. Your amp may have negative feedback around the output stage, which will minimize or eliminate the effect of varying load impedance on output power (until you drive the output stage so hard the feedback is overwhelmed). There may also be a less-than-ideal designed load impedance due to available output transformers. The GR 1840-A I mentioned earlier in this thread has a transformer to reflect 42 different impedances from 0.6Ω up to 32kΩ. It does this so you can switch loads quickly to determine the output impedance of the device under test, where the greatest output power is delivered.

So the raising of load due to changing speaker impedance could result in more or less output power, and more or less distortion. If the amp was designed well with the perfect OT and no negative feedback, the output power should drop and/or distortion increase when the load impedance increases.

Now back to the notion of injecting the test signal directly at the output tubes. Say you did that, and arrived at a certain maximum clean power output. Then you move the signal-injection point to the input of the amp, and reduce the signal by the amount of amplification you expect between the input jack and the output tube grids. But something weird happens: the measured maximum clean output power is lower than the first test. Now you know the amp circuit distorts somewhere before the output tubes, either by design or by error.

You could also start at the output tube grids with your testing, and sweep the signal frequency from 80Hz up to some maximum (maybe 10-15kHz, if you like). If you carefully recorded each resulting maximum clean output power and plotted these on a graph, you'd have the frequency response for the output stage. You could repeat at all earlier stages, and see how each stage contributes frequency-shaping to the final amp output (like the tone circuit impacts I mentioned earlier). Or you could just start at the input jack and skip the intermediate stuff, or measure voltages to graph the frequency response at any stage other than a power output stage.

Now, knowing the biggest output tube grid-signal input that will allow clean output power at the resistor load, you might evaluate how/whether the pickup's peak output tends to momentarily distort, and where in the amp, when you're playing your guitar at your fave settings. It won't change the limiting values you found earlier, and it won't change the maximum clean output power the amp is capable of producing. It might inform your understanding of how you get "fuzzy clean" when the initial transients of your playing distort momentarily while the bulk of the signal is cleanly amplified. Maybe that will influence how you design an amp. Or not.

There's probably a hundred other things we could get into about what will limit or impact clean power output, but I don't believe that's important beyond informing how you should set up the test. At least not when the question was "How to Measure Output Power?"

[jjasilli]

"Well, everything is, or can be, an issue. The real question is will you settle for it all being unknowable, or is there value in knowing at least something about what the amp does?"


Personally, I'm happy with the Ohm's Law DC power calculation, without measuring signal.  But this thread is originally about measuring, not bare calculation.  In the end I think we're stuck with apples vs. oranges comparisons, because often, test criteria are not fully divulged.

[sluckey]

Maybe the good doctor will come along soon and provide some obscure university library reference to enlighten us.   

[HotBluePlates]

Personally, I'm happy with the Ohm's Law DC power calculation, without measuring signal.

Well, if I can't convince you Ohm's Law and the equation for power apply to a.c. and not just d.c., then I don't think we'll get very far.

[jjasilli]

To clarify, I'm happy calculating an amp's output using the plate voltage squared / the OT's prmary impedance, without measuring signal. 

[HotBluePlates]

Oh, gotcha. Now your replies make sense in context. Sorry...

[shooter]

I think tires would be better rated by applied torque, I could get my engine to last 5 years, but my tires only lasted about 4mths!

Good discussion btw!

[jjasilli]

> power is equal to the area under the curve of the waveform.

The treble wave does the same power *more* times per second. More power?

Assuming resistance, the power is the same either way.

This is true but does not tell us the Power that the amp puts out to produce bass vs. treble frequencies at the same voltage.


Empirical example: a 2-way hi-fi speaker has a woofer & a tweeter.  The crossover sends only about 1/3 of the power to the tweeter, and still the tweeter may be to loud with respect to the woofer and need more attenuation.  Next, the speaker may be bi-amped with a separate power amp driving each speaker-driver.  The tweeter's power amp need be no more than 1/3 the power of the woofer's power amp.


I think this is roughly analogous to leverage in a mechanical system.  It takes 100 ft-pounds of energy to raise 100 pounds one foot.  But if we use a pulley system we can pull one foot at a time, do it 10 times, and expend only 10 ft-pounds of energy per pull.  Likewise an amp may put out only 10W of power to drive a treble frequency to a certain voltage level (the peak of the curve), but need to put out, say, 30W to drive a bass frequency at that voltage.

[HotBluePlates]

> power is equal to the area under the curve of the waveform.

The treble wave does the same power *more* times per second. More power?

Assuming resistance, the power is the same either way.

This is true but does not tell us the Power that the amp puts out to produce bass vs. treble frequencies at the same voltage.

You're lurching towards the unknowable again with complicating factors. 

Equal-voltage across a resistance load is the same as equal-power. So if the amp puts out power to the load resulting in equal voltage for some high- and some low-frequency, then there is equal power at those 2 frequencies.

Your example about reduced power to the tweeter really has more to do with Fletcher-Munson Curves, and the fact that your ear doesn't respond to all frequencies the same. Moreover, your ear's relative response to different frequencies changes for different loudness levels (with more variation at softer ear-loudness).

Really, this has little bearing on an amplifier's clean power output, except with regard to measuring at high- and low-enough frequencies to determine where the amp's overall frequency response rolls off (for instance, due to OT characteristics and/or roll offs resulting from the use of caps at different places in the amp).

...  Likewise an amp may put out only 10W of power to drive a treble frequency to a certain voltage level (the peak of the curve), but need to put out, say, 30W to drive a bass frequency at that voltage.

You haven't stated an impedance for the woofer & tweeter. If they are the same, then equal voltage means equal power (proved by the equation for power). If they're different, the driving voltage is different for the same-power; if you need different-power because your ear hears upper-mids much better than lows (proved by Fletcher-Munson), then you can't make an easy comparison.

Now before you say "A-Ha! The speaker's impedance is all over the place, so now we really don't know," keep in mind the speaker's impedance is rising at upper mids and highs, which we assumed before would cause a well-designed amp to have reduced output power.

Anyway, we are still going far, far afield of the point: What's the maximum clean output power the amp can deliver into a specific load? Which is what we'd be measuring. If you need to determine without doubt the performance of an amp/speaker combination, you go get an anechoic chamber, a measurement microphone, and set it up to measure the resulting performance. But unless you're manufacturing hi-fi or speakers, you don't need anything approaching that level of complexity.

[Jim Coash]

Greetings:  I love this discussion!  Your comments are exactly the kind of responses I have expected and have been dealing with all my life.  I am not an engineer with a degree, just an old man with a lifetime of experience.  So many times others have disputed my statements about power.  All I am really saying is that I have sold hundreds (thousands?) of pairs of speakers and amplifiers and found that  the people who ultimately damaged their speakers did so because of a lack of power, not by having too much.  So many times I have sold a very large amp to a person who destroyed his speakers with a 60/WRMS/channel receiver and then never had a failure again.  This includes setting them up with a power amp rated to deliver far more power than their speakers were "rated" at.  Music is a difficult thing to describe, quantify or replicate on a test bench.  A speaker load is far more complex than a resistor.  Specifications on amps and speakers are all over the board and seldom reliable.  My main premise is simple;  virtually any well designed speaker or speaker system can handle any amount of clean, un-distorted power but none can handle clipping distortion for long.  Solid state amps using the original transistors available in the late 60s and early 70s (prior to MOSFET) were notorious for damaging speakers.  Few people were equipped to actually tell when the clipping was occurring and they eventually overheated the voice coils and damaged speakers no matter what the power rating might have been.  On the other hand, I have been routinely using very large power amps to drive all manner of speakers for more than 40 years without EVER damaging a driver.  This is thoroughly tested in sound re-enforcement situations because I have been a working DJ since the late 60s.  Frequently someone comes up to me during a gig and wants to know "where my subwoofers are".  I point to the E/V 12" unit in front of my table and they shake their head in disbelief.  The secret is a Carver PM-600 run mono driven by the E/V System 200 controller (an active crossover at 100hz) and a pair of E/V SX-200 speakers driven by another Carver Pro.  Incredible SPLs.  No distortion.  Jim

[jjasilli]

Jim, you're right for sound systems.  But electric guitarists in certain music genres love overdrive and distortion and are willing to risk breaking their amps or speakers to get it!


Back to practicality re amp output power.  To reiterate, to me the simplest way to get an "indication" of the amp's performance is to calculate it using the Ohm's Law Power Formula: Watts = Voltage²  / Resistance.  Voltage is the plate voltage; resistance is the nominal primary impedance of the OT primary. 


Next per tompagan123's 1st post above: is the Gerard Weber method measuring the AC volts of a clean signal, just before it "goes dirty" across the speaker and applies ohms law to calculate the power.  One obvious problem is that the signal may be going dirty in the preamp or PI, while the power tubes still have more clean headroom of watts to put-out.  This could be further investigated with a 'scope or a listening amp.  The industry standard is 1KHz; lower frequencies may be more useful for guitar, but there are no universal standards.  A lack of standards raises the issue of apples & oranges comparisons.  Also the results will vary from speaker to speaker.

Next, also mentioned above is to use a dummy load resistor, usually = to the nominal impedance of the speaker. But, here's a typical Eminence guitar speaker @ 8 ohms nominal:  http://www.eminence.com/speakers/speaker-detail/?model=GA_SC64#   Note that the pdf spec sheet puts most of the guitar frequency spectrum at about 90 Ohms!  So the standard use of an 8 Ohm dummy load resistor will give an "indication" of the amps performance which could be used to compare one amp to another; but may not specify the output performance of any particular amp > a speaker.  Note also that Eminence is kind enough to give reasonably detailed info about how it's measurements are done, without which its numbers & charts would be largely meaningless.

There is no way around these things.  Any particular method results in a "knowable" number.  But there is no universally accepted method or standard, so the comparison of these numbers is frustrating.

[2deaf]

How do you all measure output power?

If you take a typical transformer output tube guitar amp rated at 100Wrms, set the selector for 4 ohms, put a big-ass 4 ohm resistor in place of the speaker and turn the amp up, then the thing puts out 100Wrms from about 50Hz to 15KHz before clipping (usually right before clipping).  If you do the same with a 50Wrms amp, it puts out 50Wrms.  The maximum non-clipped RMS output wattage of your amp is a really handy thing to know and all of that other stuff doesn't matter. 

[Jim Coash]

Excellent responses all.  Over 40 years my interest in discovering ways to accurately portray the relationships between, amps, speakers, distortion and listening preferences left me wondering how I could possibly develop some kind of "unified theory".  Just like Hawking, Einstein before him and Newton before him, I finally concluded that the best I could do was to quantify the variables, nail down the parameters and present my thoughts with a grain of salt.  A large grain of salt.  I also am a player and I just love the sound of an over driven tube amp.  Of course it proved much easier to do that with my Fender PR (Princeton Reverb) than with the TR (Twin Reverb), depending on the venue.  My wife could not stand my driving the big amp that far, especially with the E/V-12G speakers in place of the original Oxfords.  Fortunately, we all have access to many different stomp boxes that provide the distortion we crave at the pre-amp level.  I also added a switch on the TR that allows one speaker at 8 ohms or two at 4 ohms to be selected.  I also replaced the ground switch with a NFL selector so I could choose either normal, none or 1/2 of the usual amount.  My goal, probably due to my audio influences has always been to make my guitar amps as clean, quiet and dynamic as I possibly can.  That is my baseline.  Even in a large room I can generate massive volume levels with clarity.  I also added a pre-out jack to both Fender amps.  That way I can patch either into my PA/gig rig when needed while still getting the crunch I like for most of the music I play.  I sold home A/V products for many years including speakers from at least two dozen highly respected companies.  I could find no correlation by comparing "spec" sheets.   When I read the E/V and then the JBL white papers on power handling I found the answer I was in search of.  No matter what a company says about their speakers ability to handle power, they can be damaged by over diving a small amp into clipping and still remain perfectly safe when used with an amp rated for several times more power than they are rated for.  I run that experiment every time I perform.  Even my Electro Voice system is not rated to handle the kind of power I have available with the amps I use.  The reason I have NEVER damaged a driver is simply because I remember the old adage from a Clint Eastwood movie;  "A man's gotta know his limitations"!  No distortion.  Ever.  Jim

[2deaf]

calculate it using the Ohm's Law Power Formula

There is an Ohm's law and there are power formulae, but no "Ohm's Law Power Formula."  Wikipedia has no match when searched.  Didn't HBP already set you straight on this?

Watts = Voltage2  / Resistance.  Voltage is the plate voltage; resistance is the nominal primary impedance of the OT primary. 

Then how come I can flick my 60W/100W switch and get 60W instead of 100W when the nominal impedance is still the same and the plate voltage actually increases a little?

One obvious problem is that the signal may be going dirty in the preamp or PI, while the power tubes still have more clean headroom of watts to put-out.

Then don't do that.

The industry standard is 1KHz; lower frequencies may be more useful for guitar, but there are no universal standards.

Which industry standard is that?  The maximum wattage is the same for all guitar frequencies.

[jjasilli]

By the Ohm's Law Power formula I mean the one in which you're solving for Watts and know Voltage and Resistance.  The Pie Chart is attached.



Then how come I can flick my 60W/100W switch and get 60W instead of 100W when the nominal impedance is still the same and the plate voltage actually increases a little? 
Because the formula is voltage squared.  So a small change in voltage makes a large difference in watts.  (Depending on the amp, other circuitry may also come into play.)


Then don't do that.
    Not an easy choice.  You need to give the power tubes enough signal to start getting dirty, then dial back a tad.  By then, depending on the amp, the preamp or PI may already be distorting.  So if yo begin hearing "dirt", it may not be from the power tubes.  The power tubes may be providing a clean boost to a signal that's already gotten dirty.  So, the power tubes may have more clean signal voltage to put across your speaker load that isn't obvious just by listening for dirt.


Which industry standard is that? 
The audio industry.


The maximum wattage is the same for all guitar frequencies.
Yes, but an amp will it the max wattage for bass frequencies long before it hits the max wattage for treble frequencies.  That's one reason for bi-amping.

[Jim Coash]

I recommend bi-amping or even tri-amping for every application using more than one driver element.  The current trend in powered speakers bears out my thinking.  Using multiple amps and active crossovers eliminates much of the distortion and power losses passive crossovers introduce into any system.  Naturally, most combo amps use either one speaker or more than one of identical types so no crossover is needed.  Consider that in a two way system, as much as 90% of the power is actually devoted to the low frequencies.  That is why woofers are always much larger than tweeters.  Low frequencies appear to be a much smaller percentage of the spectrum but they require much more energy to reproduce.  The best place to use bi-amping is always the lowest crossover point; the subwoofer.  That frequency is usually 100hz or lower.  Removing all of the frequencies above 100hz takes a tremendous load off of the amp that can be devoted to driving the frequencies above 100hz.  If practical, another active crossover point between mid and highs and two more amps will yield a major improvement.  The new compact sound re-enforcement speaker systems with on board amps are made that way with a separate amp for midwoofer and tweeter making use of a low level active crossover at signal level.  My main home A/V system uses a Carver Signature Sub with high pass filter and a pair of Carver Pro power amps, one driving highs, the other mids.  The main speakers are KEF Reference Ones.  It is easy to think about a speakers voice coil as just a coil of wire of a certain gauge with a fixed resistance.  When you place that coil in a strong magnetic field, wrapped around a hi-temp Teflon former with Pro-Tef technology and drive it with music it becomes much more difficult to quantify.  The motion of the voice coil makes it an air pump so cooling occurs normally.  That holds true only if the voice coil is constantly moving with a true sine wave input.  It breaks down when those waves become square as amp clipping is reached.  A speaker can reproduce clipping distortion if it occurs at pre-amp level but not for long if it is the result of amp clipping.  Jim

[sluckey]

By the Ohm's Law Power formula I mean the one in which you're solving for Watts and know Voltage and Resistance.  The Pie Chart is attached.

Technical foul... Ohm's law has nothing to do with power. Never did. It deals with voltage, current, and resistance ONLY.  5 yard penality! repeat 1^st down. 

"OHM’S LAW—The current in an electric circuit is directly proportional to the electromotive force in the
circuit. The most common form of the law is E = IR, where E is the electromotive force or voltage
across the circuit, I is the current flowing in the circuit, and R is the resistance of the circuit."

     NEETS Appendix 1, page A1-5.

[jjasilli]

I stand corrected and plead guilty with an explanation.  I picked up the Pie Chart long ago, with a reference from this Forum to these nice people:  http://www.the12volt.com/ohm/ohmslaw.asp#pie  who call the entire set of equations on the chart the Ohm's Law Pie Chart, and that's what I've been doing ever since.  So it seems the Power Formulae belong to Watt's Law. 


Sorry for the confusion in the terminology.  However, a power formula by any other name is still a power formula, and the points made remain the same.

[2deaf]

By the Ohm's Law Power formula I mean the one in which you're solving for Watts and know Voltage and Resistance.

I know what you mean, but it is not titled "Ohm's Law Power Formula".

Then how come I can flick my 60W/100W switch and get 60W instead of 100W when the nominal impedance is still the same and the plate voltage actually increases a little? 
Because the formula is voltage squared.  So a small change in voltage makes a large difference in watts.

Yeah, but the voltage increased by the square making the wattage on the 60W setting even higher.  Let's just say the voltage stays the same.  60W is still inconsistent with your power hypothesis.

Then don't do that.
    Not an easy choice.  You need to give the power tubes enough signal to start getting dirty

I don't give a FFF about dirt and apparently guitar amp manufacturer's don't care much more about it than me.  It's just the maximum output signal that determines what these things are being rated at, regardless of the nature of the signal.  Hi-Fi cares about THD and other distortions that can cause their ratings to be less than the clipping point, but guitar amps don't.

Which industry standard is that? 
The audio industry.

Any particular standard within the overly-broad category of "audio industry"?

The maximum wattage is the same for all guitar frequencies.
Yes, but an amp will it the max wattage for bass frequencies long before it hits the max wattage for treble frequencies. 

So?  Even if it were true (which it isn't), there is still the maximum wattage and that is all that we are concerned with.   

[jjasilli]

2deaf:I think at this point we're going in circles.


Watt's (not Ohms!!!) has a law; it's not my hypothesis.


I can't speak to your particular amp without a schematic.


I don't give a FFF about dirt and apparently guitar amp
This is the Gerald Weber method, posted above by tompagan123.  This method is fine with me. To me, it's another form of useful compromise to measure output power.  You don't have to like this method, but you should take that up with Mr. Weber.


there is still the maximum wattage and that is all that we are concerned with
Not necessarily true, especially when playing full chords.  If the bass notes breakup 1st then the chord will sound dirty even if the hi notes are still clean.


At this point we may wish to start a thread on how to measure bias.  ANY TAKERS? 


 

[2deaf]

2deaf:I think at this point we're going in circles.

Fine.  I notice you kept on going, so I guess we can ignore the geometry.

Watt's (not Ohms!!!) has a law; it's not my hypothesis.

You're hypothesis was that you can determine the wattage of an amp using the plate voltage and the nominal impedance of the primary winding.  I presented a case that was inconsistent with this hypothesis.

there is still the maximum wattage and that is all that we are concerned with
Not necessarily true, especially when playing full chords.  If the bass notes breakup 1st then the chord will sound dirty even if the hi notes are still clean.

I don't think this is circular.  This is responding to something other than what was stated.

At this point we may wish to start a thread on how to measure bias.

I always love this one and the confusion about which negative number is larger.

[Jim Coash]

Well, I am interested in adjusting bias.  My PR now has a solid state rectifier and Iv'e replaced all the caps and resistors in the amp with better new ones.  I also have incorporated the bias circuit found in EL34's notes on Fender amps.  I increased the size of the bias cap, drilled a hole in the chassis and mounted a pot, grounded the bias resistor that was fixed to the pot and now I am wondering if I dare plug the thing in.  When I did many of the same things to my TR I took it to a good tech I know and he started it up on his Variac.  He found a couple small problems but after he addressed those it came up just fine, new tubes and all.  A little tweak with the bias and all done.  Wow is that amp great!  I can select either one or two EV-12G speakers, either factory NFB, none or half of factory with a switch and add whatever distortion I want with a stomp box.  That Twin Reverb can go from sounding like a small wattage amp to a screaming clean monster in just a moment.  With the pre-out I can put whatever I want into the PA mix too.  I just wish it wasn't so heavy.  That's why I'm doing the Princeton Reverb.  Jim

[drgonzonm]

Just finished calculating gains for each stage n overall Amp pwr.  Pre = 59, Pa=16.5  pwr =21.4 which shouldn't be since it's PSE el84's

I think power ratings are nominal by necessity.  Without measuring, output power can be calculated using the Ohm's Law power formula, and plugging in the plate voltage, and primary impedance of the OT (assuming a correct speaker load).


If you measure per a real speaker:  Does it's nominal impedance match the OT secondary?  If NO, then you get a different power reading.  If YES, still the speaker shows Impedance -- i.e., a different fixed resistance for every frequency.  The industry standard is 1000 Hz with a clean (undistorted) output wave; but different speakers will have a different resistance at any given frequency.  So the amp's output power reading may change with different signal frequencies, and with different speakers at the same frequency.


If you measure per a fixed resistor, similar complications persist.

Could you quote your source for industry standard.  It is my understanding that speakers are rated at its nominal ohms at 400 Hz, while the amplifier power readings are conducted at 1KHz, as you have stated.  (It is my understanding that the 400 Hz, is geometric average for the human voice, while the 1KHz is the geometric average for human hearing.  )  (The reference for 400 Hz is an old Western Electric publication). 

I agree with you that power readings should be conducted at 1KHz.  A resistive load is acceptable. Like most people, I don't really want to be around a blaring 1KHz sound.   

Will I see a "standard response"?

Remember people, Ohm's law is empirical.  For the most part, it works.

[jjasilli]

I was trying to gracefully end this discussion -- at least my role in it, due to a lion's share of the attention.  But it seems there is still interest & fair questions are being asked.


It seems to me that the stated output of an amp is akin to the stated horsepower of, say, a gasoline engine -- a topic I'm sure is rife with discussion on such other forums.


Anyway:  here's some corroboration of the 1KHz industry standard:  http://www.diyaudio.com/forums/tubes-valves/15489-measuring-amp-output-power.html  For further corroboration I call to the witness stand Hotblue.  This topic has come up before on this Forum.  My recollection is that he has recommended it more than once, along with 400Hz (440 maybe??) as more appropriate to a guitar amp. 


But audio industry standards may be too rigorous or barely applicable to the rough & tumble niche of guitar amps.  Hence the rough & tumble method of the venerable guitar amp guru Gerald Weber, about which 2deaf states he couldn't give "FFF" about, because 2deaf is a fixed-resistor dummyload measuring guy, end of story -- thereby making my point. 

[Jim Coash]

Ohm's law is empirical and it works like a simple, reliable formula for most circuit work.  Mathematics is difficult for me.  I always reached a point in college where the math became more than I could deal with.  Dyslexia, I think.  All those years in the audio business forced me to deal with printed specification sheets that varied wildly, did not seem to apply in real world comparisons and were often more harm than help when trying to help someone select the right equipment for their needs.  Common questions sometimes could not be answered with simple answers.  "How many watts are those speakers"? was one of the most common.  If I quoted the printed wattage from a factory spec sheet, say 100 watts, the inevitable conundrum was did that mean they had to buy an amp or receiver with less wattage?  Those of us who love tubes know that trying to compare two instrument amps, one with tubes and the other with transistors just doesn't really work.  I own several fine tube amps ranging from a Fender Princeton Reverb rated at 12 watts (a pair of 6V6 tubes) to the Fender Twin Reverb rated at about 100 watts (four 6L6 tubes).   My Ampeg with a pair of 6L6s is rated at 60 watts.  All of those amps produce impressive volume levels.  For playing music at weddings and parties I use a stereo 3-way speaker system with an active E/V S-200 System Controller and a single Carver solid state amp, run mono, that is rated at over 1000 watts for the 12" sub and either a single Carver at 600 watts/channel for the mains or two of the Carver amps run mono for twice that much.  As I have previously stated, I have never damaged a single driver element in more than 40 years of DJ work.  That includes playing in very large venues, including out doors, often for more than 4 hours continuously at very high volume levels.  Technically, even the E/V speakers are not rated to be safe with that much power.  I contend that any good speaker is safer with a larger amp rather than a smaller one.  How can you accurately measure power other than by using a fixed frequency or frequencies and a load resistor?  Reactive loads are just that; reactive.  Impedance may vary wildly from well below rated "nominal" to way above and content material in music is impossible to quantify.  Trying to specify power numbers and ratings for amps or speakers is like trying to heard cats.  Next to impossible and not really relevant in the real world.  Agreed?  Jim

[sluckey]

Cross examine?

"L.A. Law" TV Intro

[HotBluePlates]

... to me the simplest way to get an "indication" of the amp's performance is to calculate it using the Ohm's Law Power Formula: Watts = Voltage²  / Resistance.  Voltage is the plate voltage; resistance is the nominal primary impedance of the OT primary. ...

Yes, you can get an indication that way but you have to be aware of pitfalls.

The main one is you have to know something about the operation of the amplifier. Absolute theoretical maximum output power you could come up with using a knowledge of supply voltage and OT primary impedance is (B+²/√2)/(Primary Z/4). This equation assumes the output stage moves into Class B territory (one side of the push-pull stage is cut off) and the output tubes can pull their plates all the way down to 0v.

But no tube can pull its plate to 0v (though some can get "close enough"), and you may not be running the output stage in Class AB or Class B. If the output tubes are running class A (not class A for small signal and crossing into class B for large signals), then the bottom of the equation becomes Primary Z/2. You can fool yourself thinking you'll run a tube with high B+ in Class A and estimate a very high power output, when in reality the tube plate will melt. You can also fool yourself about the output capability of a class AB stage and overestimate the capability of a stage that's biased just a bit beyond class A (the high peak currents working far into class AB, which yield high output powers, also require large bias voltages to keep the tube within dissipation limits).

I bring that up because the reduction of clean output power as you take a simple class AB amp and bias the output tubes hotter, to get closer to Class A operation, is both a mathematical certainty as well as readily observable with any amp using no math and only a try-n-see approach.

The 1/√2 factor converts from peak voltage swing to RMS volts (assuming sine wave a.c.). Otherwise you'll calculate peak watts and (with a sine wave output to the load) estimate double the RMS power output.

I think you'll see without care, it's easy to mess up.

... Next, also mentioned above is to use a dummy load resistor, usually = to the nominal impedance of the speaker. But, here's a typical Eminence guitar speaker @ 8 ohms nominal:  [/font]http://www.eminence.com/speakers/speaker-detail/?model=GA_SC64#   Note that the pdf spec sheet puts most of the guitar frequency spectrum at about 90 Ohms!  ...

Frequency of Musical Pitches

A440 is the 1st string, 5th fret on your guitar if you're tuned to A440. For my 21-fret guitar, that puts the top note, D, at just shy of 1.2kHz. The usual rule of thumb is to say there are few, if any, fundamental pitches above ~1kHz, leaving everything above as overtones of the notes you play (or harmonics due to distortion).

Comparing to Eminence's graph, that puts almost all guitar notes at/below 15Ω, except for that huge spike at 88Hz (pretty much right at the F on your low-E string). Catch is, this is the resonant frequency of the speaker, the frequency at which it moves most freely. Power delivered from the amp to the speaker may be well-reduced due to the impedance rise, but the speaker will play that note very much louder than all others. The 2 situations offset each other, to the point it's not worth worrying about.

[HotBluePlates]

I was trying to gracefully end this discussion -- at least my role in it, due to a lion's share of the attention. ...

I hope you don't feel like we're beating up on you. That's not my intent, anyway. 

... here's some corroboration of the 1KHz industry standard:  http://www.diyaudio.com/forums/tubes-valves/15489-measuring-amp-output-power.html  For further corroboration I call to the witness stand Hotblue.  This topic has come up before on this Forum.  My recollection is that he has recommended it more than once, along with 400Hz (440 maybe??) as more appropriate to a guitar amp.

I didn't read the linked thread, so I'll apologize in advance if I look silly.

The chart I linked shows 1kHz is above most of the fundamental tones of the guitar amp, so it may be less relevant. It sits pretty-near perfectly in the middle of a log-plotted frequency spectrum of human hearing. I'm thinking of a 100Hz-1kHz-10kHz range. Top of human hearing is around double the top end of that (at 20kHz), while the bottom limit of hearing is a bit beyond half of the bottom end (at 50Hz). Yes, humans can hear lower (my limit, tested when I was young was around 16-18Hz), but you quickly get into rumble which is very difficult to distinguish as distinct musical tones. And humans have been shown to be able to sense the presence of frequencies up to around 48kHz, even you're not aware of "hearing" them (scientific tests reported by AES on this were part of the argument for higher sampling rates in digital recording than the CD standard of 44.1kHz).

So to  pure scientist, 1kHz is "mid-band" of hearing.

EDIT: Oops! Forgot to mention Fletcher-Munson again; your ear is most sensitive near 1kHz, so there's another reason.

The guitar range of ~80Hz-1kHz (not counting overtones or distortion) implies a lower pitch would be better for testing in the "mid-band" of a guitar amp. I think I even argued for 400Hz in the past after noticing that being a recommended test pitch somewhere in RDH4.

I now think you might want something more like 200Hz or 500Hz might be better, though you still need to know something about your amp's circuitry. Why? After looking at Duncan's Tone Stack Calculator, your Fender blackface tone stack has a nice deep notch at about 500Hz, so your power output measurement might be low. The exact frequency of such a mid-notch varies a lot depending on your particular tone circuit.

All of the above may be an argument for testing at a number of different frequencies, or for injecting the test tone at the phase inverter or output stage itself.

[Willabe]

At this point we may wish to start a thread on how to measure bias.  ANY TAKERS? 

Ooh, Ooh, pick me! With an O scope!   

I always love this one and the confusion about which negative number is larger.

Ooh, Ooh, I know this one too,

-45 is larger than -58 because it's less negative.


                  Brad         

[sluckey]

One outta two ain't bad. 

[Willabe]

There BOTH correct!!!!!

Which do you say is wrong?


                    Brad     


(Sluckey, Shhhhh, I'm trying to start up some fun!)