Measuring output power?

[2deaf]

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.

Heh, heh, heh.  I don't even know what your point is let alone how I made it for you.  I don't know who Jerry is or what his method is, but when you can consistently derive the manufacturer's specifications with a particular experiment, it really lends a lot of credibility to the method.

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

I have no doubt that if I gave you all the values of all the parameters of a PA you could calculate the actual output wattage and it probably wouldn't take you very long, either.  However, just knowing the theoretical maximum isn't going to do much more than give you an "indication" and I am talking about actually determining the wattage.  Weighing a push-pull OT is another way to get an "indication". 

[jjasilli]

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         

You're reading my mind, brother.   

[Willabe]

      

[PRR]

> 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.

There is a total dichotomy between "music reproduction" systems and "music production" systems.

Yes, often over-powering is safer if you NEVER clip. And in a fixed installation the weight of the hyper amp is moot.

Stage musicians in many genres USUALLY push FAR past clipping.

Partly because the dynamics of a bare string are boring. Lack the emphasis of hard-pushed voice or reed. The rise of overtones in clipping gives the guitar some of this dynamic expression.

Also because stage amps often must be _carried_ twice a night. Watts are heavy. Nice-clipping Watts (tubes) are extra heavy. I'm not gonna carry 1,300 Watts and then not-clip it, if 50 Watts of heavy clipping does the same job and probably lighter.

Now, how do speakers survive?

Guitar speakers ARE different from Hi-Fi speakers.

A Hi-Fi speaker may say "100 Watts", but a steady 20 Watts will cook it.... the maker relies on the >~10dB peak/average ratio of not-badly-clipped "music". Also a woofer with extended bass can be over-excurded with 10 Watts at 30Hz. While many listeners will push to clipping, gross clipping of full-band material throws so much intermodulation that it becomes unpleasant.

OTOH, a guitarist does not cover the whole spectrum. The guitarist can only play 6 notes at once, and selects them so their IM products suit musical tastes. So he can tastefully(?) play far into clipping.

How do our speakers stand it? There are *limits* to our abuse. A "25 Watts" (sine at clipping) amplifier will deliver almost 50 Watts of hard-clipped Square Wave. Because speaker efficiency is 6%-1%, essentially all of this heat goes to the coil. Modern speaker glues will hold-together at 50 Watts of raw heat in the coil. Excursion is limited by very *stiff* suspensions-- typical Hi-Fi Fs is 40hz, typical g-speaker Fs is 80Hz-- *four* times the stiffness. (Maybe 3X because cones are lighter.) Also coils are short. Typically the coil can be driven out of the magnet without damage. (Once the coil gets out of the magnet it stops trying to move more.)

This is a careful compromise and there IS a way to blow-up many guitar speakers. Play Bass through them HARD. The octave-down stimulation is twice as likely to stretch things beyond limits. This is not an invariable rule, some guitar speakers handle big bass, but many have been torn this way.

This type of use/abuse is reflected in the speaker ratings. A "100W" Hi-Fi speaker may melt at 20 Watts steady power. A "25W" guitar speaker will absorb *50 Watts* of thermal abuse all night. I have in fact used an older "25W" EV speaker as a 36 Watt DC dummy-load.

As for death of Hi-Fi systems-- many users (dancers are the worst) will turn-up to some high level of intermodulation. Often the "best" speaker for them is a small 2-way with a very high crossover. The gross bass excursion IMs the midrange clarity to get the "haze" they associate with "loud", at somewhat modest thermal power. Yes, excess electric power is needed or the tweeters will burn. (And these users often do not miss them.)

[PRR]

> a total dichotomy between

Just hit an analogy, though it may not be common-experience.

How heavy should an engine be?

In a "car", engine power is good, engine weight is very bad. Engines should be as light as possible. (Moreso in airplanes. Less so in long-range trucks where weight buys extended life.)

In a Dirt Tractor, a light powerful engine will just spin its wheels/treads. Working on dirt you need WEIGHT to get Traction. Of course you could throw dead weight on the tractor, but if there is ANY excuse to build the engine heavier, you build it as HEAVY as you like.

Real dichotomy between the two applications.

[Jim Coash]

My first really good audio amp was a McIntosh 2100.  It was heavy for its rated 105wrms/channel.  I carried it around for years.  When I worked at "The Sound Room", first in the service department, we had a clinic put on by McIntosh each year.  They would measure any amp brought in by our customers using a scope, load resistor, signal generator and just run a 1kz tone through the unit, observe the waveform and measure its peak to peak and then generate a few inches of graph paper that showed the THD from 20 to 20k.  It was consistent and repeatable.  Some companies. Harmon/Kardon comes to mind, disputed the usefulness of this test.  The claim was that  their amps were wideband designs that were essentially flat to a much higher frequency.  The McIntosh people said that was irrelevant given the limits of human hearing.  I eventually owned two H/K Citation 19 amps that could be run mono at nearly 4 times the power of the 2100.  Going from 100 to 400wrms did give me more volume; 6db more, which was useful.  What made much more difference was changing from speakers that sounded great but were approximately 92db/w/meter to E/V speakers that were over 102db/w/meter.  That 10db difference was  a bigger usable increase in output.  When I eventually tried Phase Linear 700 amps I realized not only another increase in power but a big reduction in the weight I had to carry to gigs.  After using many different amps and speakers I finally arrived at the Carver/Electro Voice system I still use today.  Another big change was changing from carrying three TEAC reel machines for sources to DAT and finally CD.
A similar evolution occurred with my guitar amps as it did all over the world.  Musicians went from needing huge stage amps like Marshall double stacks and Ampeg SVT amps to using smaller amps that sounded great and then patching them into huge PA rigs which finally became available as technology caught up with the demand.  Imagine what The Beatles would have sounded like using today's sound systems!  If you have heard one of the clone versions like "1964: The Tribute", you know.  Jim 

[shooter]

I'm no math wiz so laugh, I'm liken -58 as the larger  number especially if you are trying to get to -100, your closer than the guy in the 40's........anyway       I use 2 meters, 1 scope, n 1 10ohm 250watt 1% resistor, and 1 ToneGen software, not to try n conform to the worlds standard, just so when I get the next amp done, my  previous numbers will already be apples.

[SoundmasterG]

"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.

But this is only a theoretical number and not representative of what is coming out of your amp. I built an amp that has four 7868's, 460V, and a 2500 ohm plate to plate impedance. It should be almost 85 watts....but it is around 50 watts instead and is functioning fine and correctly. Just one example of my point.


Greg

[jjasilli]

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

Quoted above is how this thread started, and why the Weber method is relevant.  Everyone is entitled to make their own point.  In essence my point has been that there are 3 ways to determine output power: 1.  calculate it using the power formula; 2.  measure it into a resistive load; and/or 3. measure it into a reactive load (Weber being a simplified approach).  Ea method is "valid", but has its pro's & cons.  Also, they are likely to yield different outcomes from one another.  And, the latter two methods may yield inconsistent results with themselves, depending on exactly how the test is done.


This I submit is a statement of objective reality and should be non-controversial.  Any controversy seems to arise from the person's subjective preference, possibly to the degree of undue bias (pun intended per my prior post, but try measuring that  ), as to which method they prefer. 

[2deaf]

there are 3 ways to determine output power: 1.  calculate it using the power formula; 2.  measure it into a resistive load; and/or 3. measure it into a reactive load

What about weighing the transformer?

[Jim Coash]

Actually, when people ask me how they can know that an old chassis is worth buying, I tell them, the heavier it is, the more likely it is worth something.  That way if they can barely pick it up, they know I want it.  If it ways little or nothing, leave it alone.  Jim

[2deaf]

Actually, when people ask me how they can know that an old chassis is worth buying, I tell them, the heavier it is, the more likely it is worth something.

I was just joking, but if you know how many primary wires there are and the weight, you really can make an educated guess about the wattage.  If you also happen to know the nominal primary impedance, you can make an even better guess.  The size of the main fuse and how much the lights dim in your house when you first turn it on can give you a clue.  How many output tubes and the size of the bottles can give a clue.  The number incorporated into the name of the model frequently gives a clue.  The number near the output jack(s) usually accompanied by the letter "W" or the word "watts" is usually a really good clue.  If you want to know what your amp can really put out, you're gonna have to measure it.

Merry Xmas 

[darryl]

If the amplifier is factory built, always keep in mind that the watts quoted by the Sales Department can be distinctly different to the watts measured by the R&D Department.

[Jim Coash]

During my years in the A/V business I represented many different brands, most of them, the more highly respected ones.  Over the years most of those brands went through changes of ownership and/or philosophy.  Today, when someone tells me they just bought something based on brand reputation I cringe.  There are many examples, but recently I was contacted by a client who could not figure out how to hook up his new TV.  The name on the unit was one he knew from days long ago.  A name that once meant American made, quality and local service availability.  His new TV bore no resemblance to those ideals.  There was only a single RF stud on the rear.  No HDMI, composite or other input, no audio out, not even a headphone jack to connect it to his A/V system.  I was able to use the RCA outputs on his cable box to get audio to his system but there was no way to connect his DVD player to the set.  It had RGB, HDMI, composite outs but no RF in/out.

My point about measuring power is that in my career, spanning more than 50 years, every brand of amp, receiver or integrated used their own method of measuring power making it impossible to compare.  The most conservative methods specified were using a system based on WRMS/chan/20to20k/.1%THD (or less)/w3db headroom (or more) and those numbers were from companies like McIntosh, Carver, NAD, Harmon-Kardon and the like, at least when those brands were respectable.  Onkyo, Yamaha, Electro Voice, QSC, Crown and a few others were also reasonable.  Most others, not so much.  Some were downright ridiculous.  Giving a number for one channel, at 1000hz, 1%THD with no headroom specified yielded specifications that were of little practical use.  These were all solid state ampsby the end of the 70s, generally used for home audio but some also usable in sound re-enforcement apps.  Classic tube amps are really not comparable in a meaningful way because valves and transformer coupled outputs are so different.  MOSFET and some other later solid state devices did permit some comparisons but direct coupled bi-polar output stages simply are a different animal.

I own some very good solid state amps that I do use for instrument work.  My Ampeg BA150 is a respectable unit.  Using my tube pre-amp Hafler Hellraiser to drive a large solid state amp like my E/V 1.5 or any of my Carver Pro amps sounds good through my many E/V cabinets either factory or custom made.  I have an E/V loaded Mesa-Boogie box with an EV-15B in the bottom, an EV-12G and an EV-10 in the top that is really great and so flexible now that I re-built it with a jack plate that allows the top and bottom to be separated or used together or bi-amped.  But the underlying truth is this; the more "real" power I have, the better my speakers sound and the safer they are.  Jim

[2deaf]

I suggest we take a clue from component suppliers, take Hammond for example, a lot of their  OT's reference 1K Hz as a reference.   (While the engineers at Jensen will say Hammond does not provide enough information, it is starting point.)  Hammond then provides an operating range for the OT.

Hammond typically rates their wattage within a range of frequencies and that wattage will only deviate a specified amount plus or minus as compared to the wattage at 1 KHz.  I don't know if that's what you said or not.

[jjasilli]

Yes, different test conditions will produce different results.  E.g., the Stromberg Carlson 1100 APH manual (100W mono bloc PA power amp) states that the amp produces 100W @ 5% Harmonic Distortion; 80W @ 2% Harmonic Distortion, over much of the range of human hearing.  But it doesn't state specific test conditions, such as the test frequenc(ies) used, or into what load.  I assume Stromberg Carlson means that within the stated frequency range, no frequency will have more than the stated distortion at 80W or at 100W, in the actual & proper use of the amp and its speaker load.  But they don't actually say that.

Assuming that both 80W & 100W are both honest measurements of the same amp under different but valid testing conditions, that's a very large spread if you're a builder or manufacturer trying to sell amps to the general public. 

This doesn't mean that power output is "unknowable" as HotBlue has "accused" me of advocating; but that stated wattage is not an absolute in itself, and needs to be knowledgeably interpreted in terms of how the test was done.  Without that background info the stated wattage may be meaningless or unreliable.

[2deaf]

From reviewing some old books on audio transformers.  (Norman Crowhurst, seems to know what he is talking about),  I believe the 1k Hz reference and the range of frequencies, refer to two different standards. 

It is my understanding, that 1k Hz, as a standard is the close to the geometric mean of the range of human hearing, so a single point parameter can be used.  Then from some undocumented research, (I was looking info on a Western Electric handset), Western Electric gives 400 Hz, as the geometric average of the human voice.  (you would be surprised how many things are engineered using single parameter formulas). 

As a followup, 

Fender Music in its 1995 Blues Jr schematic. uses a 1kHz sine wave as an input reference.  Other conditions are stated.  Output is 13.1 watts.  The 1995 blues Jr., IMO, provides superior documentation.

Stomberg-Carlson in its au-33_ps-33 uses a 400Hz signal to define its output power.

Uuummm...Okey-dokey.

[2deaf]

But it doesn't state specific test conditions, such as the test frequenc(ies) used, or into what load.

They spell out the range of frequencies so there is that specific test condition.  You can tell they are using a constant load for their ratings because the response is damn flat and I don't think anybody would entertain the possibility that they used mis-matched loads.

[HotBluePlates]

This doesn't mean that power output is "unknowable" as HotBlue has "accused" me of advocating; but that stated wattage is not an absolute in itself, and needs to be knowledgeably interpreted in terms of how the test was done.

"Uknowable" was just a running joke. Speaking plainly, I meant that you were taking a very simple, objective issue (measure volts across a resistance load to compute actual output power) and complicating it with extra factors until it became complex and subjective. The hope was we could agree on the simple, then build outward with an awareness of added real-world complications.

Yes, different test conditions will produce different results.  E.g., the Stromberg Carlson 1100 APH manual (100W mono bloc PA power amp) states that the amp produces 100W @ 5% Harmonic Distortion; 80W @ 2% Harmonic Distortion ...  But it doesn't state specific test conditions ...[/font]

Sure they do. "100W @ 5% Harmonic Distortion; 80W @ 2% Harmonic Distortion" There is an implied, "with rated load attach to the output transformer secondary." In a testing environment, that's very likely to be a resistance load (we can debate whether this is a fact, but show me a test lab measuring power amplifiers which doesn't do it).

We've said this another way several times in this thread. If you had a distortion meter attached to monitor the output, and THD was measured as 0.000000001%, you will measure one level of power output. If you increase the driving signal to the amp until 2% THD is measured at the output, you will also measure higher Volts RMS across the load resistor, and have "more power output." Increase drive signal again until 5% THD is measured across the load resistor, and measured Volts RMS and output power are again increased.

This increasing-power effect of a distorted signal runs into a limit where RMS power is double the RMS power of a clean sine output; this is the same as the RMS power of a square-wave, where RMS and peak power are the same. A square wave is an infinite series of odd-harmonic sine waves added to a fundamental sine wave. While testing amps with a square wave has many uses, a sine wave consisting of only a single frequency is typically used for power tests because the results could be gimmicked when multiple tones are present (meaning, presence of distortion & artificially-high power measurements are harder to detect).

So Stromberg was very straightforward in their power rating: "this amp is 80w if you need 2% THD, but 100w if 5% THD is acceptable." 5% THD was once set forth as an amount that was minimally-audible and a benchmark for "Hi-Fi" performance.

In the flyer 2deaf posted, Stromberg implies the AU-33 will output 25w, at 5% THD over a bandwidth of 50Hz - 10kHz.

There was a game some transformer and hi-fi makers played with bandwidth numbers at one time. A transformer's power through-put tends to be pyramid-shaped: if you reduce the power applied below the transformer's rated power output, parasitic losses are reduced and the transformer appears to reach both lower and higher frequencies. So a transformer maker could spec a part as "50w output power" and "20Hz to 50kHz" but not tell you the frequency range quoted was measured for 1w of through-put. Then when you apply a full 50w, you get something more like 100Hz-10kHz.

Hammond is very proud of their full-power bandwidth specs, so when they say "30Hz to 30kHz" they mean at the full rated power of the part. You could really pass more power through one of their transformers if you're willing to forgo full power, as low.

But the point I tried to make repeatedly is that these are all side-issues if the question is "how do I measure the amp in front of me." The answer still remains, "attach a resistor as a load, inject a test signal, measure volts across the resistor, mathematically convert to power output." If you have the ability to measure or delineate distortion, then add to the procedure, "... at ___ measured % THD." If bandwidth/frequency response is desired information, then add to the procedure, "... tested at W, X, Y, Z frequency, or over X-Y frequency range."

In reality, the low end of response is going to be set largely by the value of the coupling caps used, followed after by the capabilities of the output transformer. The upper end of frequency response is going to be set by any shunt capacitances used (e.g., plate load bypass caps, "enhance caps," etc), followed after by the parasitic characteristics of the output transformer. All of the above might be modified by the use of negative feedback around the output transformer & output stage, to widen frequency response to a limit.

[jjasilli]

SC's range of frequencies is 50 - 15,000Hz @ 5% distortion; 40 - 20,000Hz @ 2%.  Did they test thousands of frequencies?  Or just 1KHz, and assume the rest?  Into a resistive load, or a speaker load?

[jjasilli]

But the point I tried to make repeatedly is that these are all side-issues if the question is "how do I measure the amp in front of me." The answer still remains, "attach a resistor as a load, inject a test signal, measure volts across the resistor, mathematically convert to power output." If you have the ability to measure or delineate distortion, then add to the procedure, "... at ___ measured % THD." If bandwidth/frequency response is desired information, then add to the procedure, "... tested at W, X, Y, Z frequency, or over X-Y frequency range."

Fine with me, but you too will need to take that up with Gerald Weber, whom advocates an actual speaker (reactive) load!   

[darryl]

Fine with me, but you too will need to take that up with Gerald Weber, whom advocates an actual speaker (reactive) load!   

That method would allow an amp builder to shop around for a speaker which had an impedance curve which optimised the output of the amplifier under test.

I agree with HotBluePlates. For comparing like with like, measure the output voltage across a load resistor at the onset of clipping, when driven with a continuous 400Hz or 1kHz signal.

If you have the necessary test equipment, quote additional information, such as the distortion. If you consider your amplifier has particular attributes which enhance its perceived output, quote those as additions to the raw output calculated with a resistive load.

[jjasilli]

Yes, I'm sure Weber advocates his method 'cause it's "Easy peasy" as tompagan says in the initial post to this thread.  All you need for equipment  is your amp hooked up to its usual speaker, your ears, and a volt meter.  Back to pro's & cons again.

[HotBluePlates]

SC's range of frequencies is 50 - 15,000Hz @ 5% distortion; 40 - 20,000Hz @ 2%.  Did they test thousands of frequencies?  Or just 1KHz, and assume the rest?  Into a resistive load, or a speaker load?

Stromberg almost certainly used a resistor load (again... see the General Radio Power Output Meter I linked at the start of this thread; GR was one of, if not the, biggest test equipment manufacturers back in the day).

Stromberg said also "50Hz to 10kHz, within +1db, -2dB." That implies they swept the test frequency over that range while measuring voltage across the load, monitoring for voltage deviation, which then gives the number of dB above/below the nominal power output. Hewlett Packard made a Sweep Oscillator to span 20Hz-20kHz in a single dial turn, facilitating such measurements; they also made a bunch of other oscillators to cover this range and more if you were willing to flip a range knob.

Once you realize the test equipment available and the intended use, these arguments over "1kHz or 400Hz?" and "1 frequency or many?" start looking very silly.

But the point I tried to make repeatedly is that these are all side-issues if the question is "how do I measure the amp in front of me." The answer still remains, "attach a resistor as a load, inject a test signal, measure volts across the resistor, mathematically convert to power output." If you have the ability to measure or delineate distortion, then add to the procedure, "... at ___ measured % THD." If bandwidth/frequency response is desired information, then add to the procedure, "... tested at W, X, Y, Z frequency, or over X-Y frequency range."

Fine with me, but you too will need to take that up with Gerald Weber, whom advocates an actual speaker (reactive) load!   

No, at this point I step back and ask "Who is Gerald's audience?"

Guys who have test gear and know how to use it already know more than Gerald is teaching. A guy just getting into amp-tinkering seems like the target audience. This guy has an amp but little test equipment.  He's probably lucky to have an oscillator at all, and probably no dummy load. So just measure across the speaker terminals. It works.

Friendly Dare: Try listening to a full-power 1kHz sine wave when checking the output power of a 100w amp. Then tell me how many milli-seconds it takes you to decide that sucks, and that you should invest in a dummy load for power measurements.

I did a fair amount of multi-track tape machine alignment when I was in school for audio engineering (like at the start of every recording session). Listening to a 1kHz tone will make you want to put an icepick through your eye. Listening to it at 50w or 100w will make you swear off power measurements without a dummy load.

Just try it. 

[jjasilli]

Listening to it at 50w or 100w will make you swear off power measurements without a dummy load.


That's the Con's alright!   

[PRR]

> Did they test thousands of frequencies?

No. Impossible and "obviously"(?) not necessary. Why would a simple broad-band amplifier pass 456Hz but not 457Hz? It isn't smart enough to do anything so refined.

> Into a resistive load, or a speaker load?

Assuming we test at the 8 Ohm tap, then into *8 Ohms*. Something which is a KNOWN resistance over any broad range of conditions (level, frequency) IS a resistor. Loudspeakers are not resistors (let's avoid that can-o-worms for a moment).

These amps have several taps. By ideal transformer therory, we get the same power at any tap if loaded in the design resistance. In real life we may see small differences, but SMALL. It will surely make 15 Watts in 4 or 8 or 15 or 500 Ohms. It is probably a 20 Watt amp; good PA amps use a large CYA factor to allow for tolerances and rounding errors.

An audio power amplifier has a Flat Band in the middle of the audio spectrum and Roll-Offs at high and low frequency. The choice of roll-off often depends on Price. Note the low-price AU-32 claims 75Hz, the high-price AU-33 claims 50Hz. We could buy tube amps rated for full power to 20hz or lower (shaker-table amps), but they cost and weigh more.

Guitar amps have to go to 80Hz pretty-much. Some (say a $400 Fender) maintain full power much lower and cleanly. Some (say a $40 Kent) get pretty crappy at full power anywhere near 80Hz. There's a Right Way and then there are compromises with customers' budgets.

> Into a resistive load, or a speaker load?

Speakers are complicated contraptions. And all practical speakers face fundamental problems. Any air-pusher is too small or too heavy (invariably both) to cover the audio band. They have to balance 2 or 3 reactances against each other to approach wide-range operation. Impedance seen at the terminals is usually ALL over the place as these reactances wrestle each other.

The saving grace for dynamic speakers is that the copper resistance is in series with everything else. So there is a *minimum* impedance which is reliable and can be designed against. Most power sources (including audio amplifiers) can be designed for a load which is not-less-than some specified Ohms.

The speaker designer (not the amp designer) is responsible to ensure the speaker gives good results even while its impedance varies higher than nominal. This actually works-out nice. At bass resonance the impedance goes up to 50 Ohms. The power sucked from an 8-Ohm tap may be 8/58 or about 1/8th of the rated power in 8 Ohms. The amp is happy delivering 1/8 full power. The speaker input is less but at resonance the speaker self-efficiency is very high, easily 16% on a nominal 2% efficient speaker. So the acoustic output is similar on or above bass resonance.

The AMP designer works to resistor loads. Oh, to cover his butt he should consider other likely loads. But comparisons are done on resistor loads, because they are valid-enough for most loudspeaker loads. If a loudspeaker causes a problem, blame the speaker designer.

[Jim Coash]

I agree that finding and sticking to a "method" of measuring power is a good thing provided you remember to take any numbers with a grain of salt.  My method is to look for a number that specifies all of the terms.  A few years ago I needed a small amp to drive a set of KEF speakers I installed in my Jacuzzi room for my wife, primarily.  There was no room for a conventional receiver or integrated amp.  I bought a Chinese unit not much larger than a paperback novel with a pair of mounting brackets and a wall wort A/C supply.  The box clearly says in large letters "180 Watts X 180 Watts".  Patently ridiculous.  Maybe 12 watts/channel at 1000hz.  The problem I dealt with every day at my audio store was people making comparisons based on printed "specs" or what an appliance store salesperson told them.  Often someone would tell me that they could get an XXX watt amp for half the price of the one I was demonstrating.  When I asked them if the amp actually sounded any better (or even louder) than mine they often would say that they hadn't even listened to it.  They were simply buying on misleading information, not sound.  The Carver or NAD amp I recommended appeared to cost more, until you brought the actual power rating into perspective.  An NAD basic receiver at $300 was rated at 25 watts/channel/@>.1%THD/both channels driven/at any frequency 20 to 20K.  They even specified at least 6 db of headroom.  That means that the amp can cover 100 watt peaks for musically significant periods of time and those amps were stable into less than 2 ohms!  Speakers are a very complex load and music is nothing like sine waves.
Yesterday my son and I connected two of my recent speaker projects for an evaluation.  For an amp we used a rack with a Hafler Hellraiser
tube guitar preamp run through a DOD compressor into an E/V 1.5 power amp.  My Fender Dual Showman "clone" box with a pair of EVM-15 speakers is awesome.  The old Ampeg cab I bought for $50 on eBay and re-built to eliminate the place for the missing amp chassis and fitted with an E/V-15G recone I bought for .99 cents and fixed for $75 sounds amazing.  The E/V SRO-15 that came in that box will go into a new cabinet I am nearly done with that will be my Stevie Ray Vaughn box.  Just like he specified for his Fender Vibrolux.  What point is there in discussing power?  It is nearly impossible to make any valid comparisons given all the variables.  Jim

[HotBluePlates]

... The problem I dealt with every day at my audio store was people making comparisons based on printed "specs" or what an appliance store salesperson told them.  ...  What point is there in discussing power?  It is nearly impossible to make any valid comparisons given all the variables.

The point within this forum is that by & large, the folks here will be building an amplifier from scratch. Possibly a completely original design. There are no printed specs, nor salesman.

[The exception is building a copy of a known-plan. But in that case, the power output is largely known. And there's very little performance difference if the claimed "40w amp" turns out to be 30w or 45w in actual use. Known-plans have also taught most builders that 100w from a pair of 6V6's is impossible, and what the likely power output will be.]

In that context, they may wish to evaluate their new design against previous efforts, as a way to evaluate any new changes. If the builder cares about the validity of the numbers, they will pick a procedure to be replicated with every measurement; control of the process is absolute, because they will be the one performing the process.

And a simple measurement might be preferable to looking at a pair of push-pull 6V6's and saying, "Ehh.. something between 10-24 watts." The builder should know enough if they're creating their own design to rough-in the numbers at least that much. It should also be obvious to them that if they measure 40w output there is a problem in their measurement; if they measure 4w, they should know there is a problem in the measurement, or in some failing in the amp's design or construction (which would be handy information for future designs/builds).

[jjasilli]

What point is there in discussing power?  It is nearly impossible to make any valid comparisons given all the variables.  Jim


We have lost track of the original question posted in this thread, about the Gerald Weber method.  There is nothing wrong with measuring the actual output voltage of particular amp into the speaker with which it always, or often used.  And using that speaker's nominal impedance to calculate watts.  The fact that there are other methods, or that other methods are used by testing laboratories or R&D dept.'s, does not invalidate this method.  Inquiring minds want to know! 


What point is there in censoring power out of consideration?  Do you want a bedroom amp? living room amp? stage amp?  Do you want the sheer power to affect the tone of the guitar on stage per Jimi Hendrix? 

[jjasilli]

It is impossible to get a wrong answer by doing a test, any test, correctly.  The fact that different test methods produce different results does not invalidate a test method.  The differences in test results can be accounted for scientifically. 


The Fender BJr revD schematic states 10.25V rms output into an 8 Ohm resistive load; 1KHz input signal @ a specific mV rms, with controls @ certain positions.  5% THD.


Presuambly:


* if we change control positions we'll get a different output voltage
* if we use a reactive load we'll get a different result (a reactive load doesn't have to be a speaker.  It could be a speaker simulator circuit; or a speaker motor like a Weber Mass dummy load)


No test is invalid.  Different results can be accounted for if we know how the test was done.

[HotBluePlates]

The Fender BJr revD schematic states 10.25V rms output into an 8 Ohm resistive load; 1KHz input signal @ a specific mV rms, with controls @ certain positions.  5% THD.

Blues Junior Rev D Schematic

Correct. This measurement was given as a reference for a full-power test of this amp, for a repairman's reference. The Output Test note gives the specific conditions.

Presuambly:

* if we change control positions we'll get a different output voltage
...

No test is invalid.  Different results can be accounted for if we know how the test was done.

These statements are correct.

Look at the input jack on the schematic; it says, "For AC Voltage Measurements, Apply 1kHz Test Signal" and then indicates a 10mV signal at TP1 (the input jack). All the subsequent voltages in the ovals are the resulting voltages at those test points, with 10mV at TP1 and with the controls as specified in Note 2 at the bottom left of the schematic.

There is a clue to all this: if you look at TP16 (output tube grid) and TP18 (output tube plate) you'll see similarly-small voltages... Only ~1vac drive to the output tubes (when a peak voltage closer to the bias of -10.7v would give full output power) and only ~41vac at the output tube plate. As a result, the output at the OT secondary is appropriately-low (so TP20 does not represent full-power output, but that was known by the math and Fender's Output Test note).

I going to state an opinion, without watching Weber's  video.  IMO, Weber is not correct,  Look at the various schematics of Fender Music's Blue's Jr.  schematics.  it gives an A/C voltage on the O/T's secondaries of some what less than 3 volts.  Using the 8 ohm resistive load and the Ohm power calculation, you don't come up with 13 watts the schematic presents as power output, you need between 10 and 11 volts to come up the the stated wattage.  Bottom line, unless someone provide me with Weber's credentials, I'll take Fender Music information as being more credible.

Standard Response

[HotBluePlates]

Gerald Weber shows you how to do a Quick and Easy Wattage Check

Standard Response

[darryl]

Contributors to this thread have provided their opinions on whether 1kHz or a 400 Hz single should be used.  I don't know if the signal affect the power reading on a resistive load, but my guess is no.

If an amplifier's output power measured at 1kHz is significantly different to that at 400Hz, either the amplifier or the test method has a serious problem.

[2deaf]

You know . . . Jerry tells you how to do it at the start of the clip.  He never says that the text books are wrong and it's kind of funny how he mentions O'Scopes and function generators as the camera pulls back to reveal both right behind him.  He doesn't finish telling you his method or what the significance is of whatever number he comes up with -- you gotta buy the DVD.  You have to ass*ume (as PRR put it and as Felix Unger expounded upon) if you think that he is going to divide by the nominal impedance of the speaker to derive the output wattage.

He says that the voltage is about 5 volts right before fade out, but I saw 6.867 go by.  The wattage at 6.867 is approximately twice the wattage at 5.  Is the amp even maxed-out?  It doesn't sound like it to me.  I can't really criticize the guy from this clip since I have no idea what he is going for.  I'll tell you what, though.  He can take a helluva shock and come back with a grin.

[JB]

I've always thought 1KHz was used for measurement as it's the nearest nice decimal number near the middle of the speech band used in telephony - 300Hz to 3KHz band limited telephone circuits.  In my field of broadcast engineering 1KHz is used for all circuit line-up's, reference tone on TV test cards, mixing desk built in reference tone etc. etc.
 

[HotBluePlates]

I've always thought 1KHz was used for ...

Your explanation sounds perfectly reasonable; thanks for offering your expertise.

For the purposes of this thread, the frequency is probably a non-issue (if it is an issue of contention, a good sweep oscillator is all that's needed). Instead, the question of 1kHz vs 400Hz is only being raised repeatedly by someone notorious for trolling on this forum.

[jjasilli]

Blundered into some interesting info here, though geared to hi-fi: http://www.americanradiohistory.com/Archive-Electronics-World/50s/1959/Electronics-World-1959-11.pdf


This addresses many of the issues in this thread; how & why tests are done in certain ways; and how to interpret test results.



The Nov 1959 Issue of Electronics World:  Harmonic Distortion, p. 72; and Measuring Your Audio Output Power, p. 92. 


Thanks to PRR for posting the americanradiohistory.com website in another thread.  I've been going through the issues as I get the chance.

[shooter]

I ran into an anomaly when "spec-ing" my SE build, I used a 10ohm load as usual @ 1K, but then thought what the hell, wife is gone, got ear muffs so I did the output @ speaker and it did change.  It made sense since I changed the "wanted" load of 8ohms to 10.  The oddity though was both loads at max clean signal came in about 10.8Vac rms.  Times that by 1.414 and that should be close to max, and it was for the resistor but the speaker yielded 23W. double checked, same.  Donno what it means, or care, my standard is the 10ohm resistor @1k then all my amps are apples to apples for all MY amps.

[HotBluePlates]

Blundered into some interesting info here, though geared to hi-fi: http://www.americanradiohistory.com/Archive-Electronics-World/50s/1959/Electronics-World-1959-11.pdf

This addresses many of the issues in this thread; how & why tests are done in certain ways; and how to interpret test results.

The text of the article on distortion says, "In the early days of hi-fi, it was common to take harmonic distortion measurements at only one or two frequencies around the middle of the audio band, commonly 400 or 1000 cps." The article doesn't say why, other than that test equipment had improved.

Below is an HP 320A "Distortion Analyzer". Really, the device is an attenuator which can accurately reduce the signal at the input terminals by a specific number of dB set by the switches, and a filter which can null out either 400Hz or 5kHz.



The method for measuring distortion at the time was to apply a test signal to the circuit under test (at either filter frequency) and measure the signal level at the output with a separate meter. Then you'd kick in the filter to eliminate the single test frequency (at the output) and measure everything else left over, yielding a "THD + Noise" figure. Part of the test setup was to tune the frequency of the oscillator until you had the best null when applied only to the filter and nothing else, because you wouldn't trust the oscillator and the filter were calibrated well enough to each other to agree on what "400Hz" was. You might also need to do this setup just to measure how much distortion your oscillator had prior to being applied to the circuit under test.

Kinda clumsy, and you needed at least 2 other pieces of gear besides the "distortion analyzer" just to make a distortion measurement. But if you already had a meter and an oscillator, it was a good way to add a distortion measuring capability. The bulk and cost of the system at the time made having multiple frequencies for filtering cost-prohibitive and impractical. See the image below for an example of this test setup; the item 2nd from left is the amplifier actually being tested.



HP wound up combining all the functions in a single chassis in the 330B, below. There was an oscillator circuit which was tuned by the center knob to a frequency displayed in the center window. Also tied to that knob was a tunable filter which nulled out the single frequency produced by the oscillator. Now you could dial to any frequency in the range of the analyzer and measure THD+noise. There was an output to the circuit under test, with the output of that circuit returned to a set of binding posts which fed a calibrated attenuator and a voltmeter. Between the two, you could measure THD and noise in the form of "so many dB below the desired signal."



It took yet more time (and transistors) before the HP 302A Wave Analyzer was produced (below). This was a selective voltmeter design that let you measure the amplitude of a single frequency (at least within a fairly narrow pass-band), so that you could apply a 100Hz test signal, for instance, and find out the amplitude of the resulting 200Hz, 300Hz, 400Hz, 500Hz, etc harmonics, and be able to know exactly how much of each harmonic was being generated apart from noise.

Point being, there are all kinds of inter-related reasons for conducting tests the way they were once done, some for good technical reasons and others because of limitations of existing test equipment at the time.

[jjasilli]

Been reading up on this including articles on this site: http://sound.westhost.com/projects.htm  Now I don't remember which bit of info comes from which article.  Anyway it seems that for human beings the heart of all musical information comes from the freq range of about 300 - 3000Hz.  Even deep bass notes may be implied from their harmonics in this range. Confirming many posts in this thread, 1000Hz is conveniently right in the middle of that range.  400Hz is a good distance below yet still in that range.  Deeper notes are expected to give worse THD, so this gives more info.


THD = total harmonic distortion.  It lumps it all together and lacks specifics: like which order of harmonics; at what frequency; etc.  But, it is still useful info.  If a reputable manufacturer states, say, 2% harmonic distortion @ 40W RMS from 30 - 15,000Hz, this means that's the worst you'd expect over that entire range.  Hence you can expect THD to be dramatically lower @ 1000Hz.


Other points.  Typical volt meters are for low freq AC; and don't measure VAC well at musical frequencies.  A true RMS meter would be  needed.  The tests -- THD and Watts -- are done purposely with a non-inductive resistive load, to eliminate the vagaries of reactance.  The load value should = the amp's rated ohm output tap.  If the resistor gets hot its resistance value increases, throwing-off the measurement.  Hence the resistor should be rated for 5X the expected power in watts.  A lot of this seems like overkill for guitar amps, but it's good to know.

[HotBluePlates]

Other points.  Typical volt meters are for low freq AC; and don't measure VAC well at musical frequencies.  A true RMS meter would be  needed.

"True RMS" doesn't tell you about the frequency range of your meter. I don't know if you meant to imply that, or if it was just 2 separate statements.

The article was probably talking about old analog voltmeter movements, like a 1000Ω/volt (1mA) movement. Today, your multimeter very likely has a much wider range of a.c. measurement. My Fluke 87III doesn't seem to droop until couple-hundred kHz. HP made some a.c. voltmeters good to 400kHz and beyond. Point being, you have to know the capabilities of your particular equipment to know if this is a pitfall.

[PRR]

> a 10ohm load as usual @ 1K,
> I did the output @ speaker and it did change

The speaker is everything except 8 Ohms. It is probably 50 Ohms at 90Hz, may be 6 or 9 Ohms around 400Hz. A typical guitar speaker will be rising by 1KHz, possibly 12 Ohms.

In many guitar speakers, 20 Watts for many seconds will hot them up enough to raise the copper resistance. You can often double the resistance by abusing it until the smoke comes out (actually modern glues don't even smoke good).

Get a big 40 Ohm 10 Watt resistor to put across your 10 Ohm, for a true 8 Ohm load.

> Times that by 1.414

I do not understand this math. If you already found "max clean signal", what are you multiplying for? And where did 1.414 come from? If you alternately want to know the GROSS OVERDRIVE power output, run it up into Gross Overdrive. The problem is that the "RMS" of a severely bent wave is not easy to measure. If the overload is nearly Square, you may measure peak-to-peak and do the math for a square wave. Some meters read RMS direct, despite somewhat bent waves. An Averaging meter will read a Square fairly well, and a dumb passive meter was used at many early g-amp labs.

[shooter]

GROSS OVERDRIVE

That's sorta what I was going for, I think I got that number here for "ball-parking" max distorted amp power.  I did dime the amp and that's where my math and reality went separate, math showed 16ish, reality showed 23W.  My 10ohm 250W is what I have on hand but I think I can scrounge a || 40 for 8ohms.  The main thing for me is to measure everything the same way every time....even if it's wrong, it's consistent

[octal]

Just for fun, here are some scope shots and voltmeter readings from a little EL95 push-pull amp I built. First shot is clean, low power, second shot is just before real clipping, third shot is a bit of clipping and the last shot is dimed, squarewave city.

[octal]

HP sinewave generator into amp's input, 400Hz signal. Amp is loaded with an 8 ohm, non inductive dummy load. One channel of 'scope is connected to the input of the amp, the second channel is connected to the speaker output. (I like to do this so you can see how the shape of the output sine wave compares to the input.) It can create some ground loops testing this way though.

Weber's method is next to useless, IMO. You can get a decent secondhand sinewave generator, OR a test CD, OR download a file to an iPod and feed in into an amp... there's no excuse for using a guitar as a signal generator for basic amp checks other than laziness.