How to test an amps true power output

[bobdabilder]

Enough with all this technical mumbo jumbo just show us the video once you've tried whatever it is you're going to try. Please video it.

 

[Spectre Gunner]

There are many sound testing apps such as an SPL meter available to download to your smartphone. They are probably not going to be as accurate as a dedicated device but are much cheaper (some are free) and may suffice for basic tests. To get better results with your phone app you can plug in an external microphone which is likely the weakest link in your phone.

 

[TheReceder]

Well.....if I did that, it would blow everyone out of the room. I use an 800-watt Eden amp (WT500/800) and I run it bridged at 800-watts into either my DNS-210 cab for indoors, or my DNS-410 cab for outdoor festivals. The highest I ever turn the master up with either of those cabs is 10 o'clock. I turned it up to 10:30 on my DNS-410 at a couple of outdoor festivals and immediately got told to turn it down. Never turned it above 10 o'clock on my DNS-210. Both cabs would handle the power of turning it up half-way (12 o'clock), but you couldn't stand to be anywhere near those cabs at those levels.

You have to remember, that SPL measurement (how loud it is) is a mix of both the amp's power and the cab's sensitivity. I'd be amazed if you could turn it half way up with Mesa amps and cabs either.

Hmmm... a definite flaw in my theory... don't tell my wife ... I have her convinced I'm never wrong... wait... sorry... it's the other way around.

 

[RickyTee]

I had my Trace Elliot AH350x head tested after a repair. The amp is a bi-amp that is actually a hi pass system ( full range bottom/250hz and above for the top amp) used with 18/10 cabinet and or a 15/10 cabinet. The lower amp had a few fetts replace so we tested it with i believe a pink or white noise generator for full freq, The bottom amp was rated at 250 watts at 4 ohms and the top was 175 watts at 8 ohms. We desided to test the bottom full range section we repaired. The amp was put on a watts meter and loaded with 8ohms and it went to 415 watts full range before it square waved. The tech that did it was stunned! It was only suppose to make about 175 watts at 8 ohms and made more than 2.25 times what it was suppose too. He said thats British watts! he was tested Marshall JCM 800 that were suppose to be 100 watts and they made over 400 watts and he test a Marshall Major he did and he said it made close to 600 watts. This tech is well know in southern Ontario as one of the best. He estimated that my amp would easily make 800+ when it is rated for 425. Thats why even in concert I never have it turned up higher that 2-2.5/

 

[agedhorse]

I had my Trace Elliot AH350x head tested after a repair. The amp is a bi-amp that is actually a hi pass system ( full range bottom/250hz and above for the top amp) used with 18/10 cabinet and or a 15/10 cabinet. The lower amp had a few fetts replace so we tested it with i believe a pink or white noise generator for full freq, The bottom amp was rated at 250 watts at 4 ohms and the top was 175 watts at 8 ohms. We desided to test the bottom full range section we repaired. The amp was put on a watts meter and loaded with 8ohms and it went to 415 watts full range before it square waved. The tech that did it was stunned! It was only suppose to make about 175 watts at 8 ohms and made more than 2.25 times what it was suppose too. He said thats British watts! he was tested Marshall JCM 800 that were suppose to be 100 watts and they made over 400 watts and he test a Marshall Major he did and he said it made close to 600 watts. This tech is well know in southern Ontario as one of the best. He estimated that my amp would easily make 800+ when it is rated for 425. Thats why even in concert I never have it turned up higher that 2-2.5/

I can say with 100% confidence that these claims are false. Either your star tech didn't know what he was doing, or the claims were made up by either your tech or maybe you. It's simply impossible, I am intimately familiar with the technical details of Trace Elliot line (having worked as an engineer for the company that owned TE during this time period) and there is categorically no way that the AH-350X head can possibly deliver anywhere near 415 watts RMS into 8 ohms. The PM6S module that is in the AH-350X does not have the voltage swing to get anywhere near that power into 8 ohms, TR7 is 92V CT, which has a theoretical maximum power using the lateral FETs of 42Vrms unloaded, which is 220 Wrms into 8 ohms. There is no transformer in that entire series of amps that has a higher secondary voltage either, therefore the claim is impossible.

I am also familiar with the JCM-800, it too can not possibly, under the most favorable conditions in the world come even close to 400Wrms, in fact it's not physically possible for it to exceed 150 watts.

The Marshall Major, with 4 x KT-88 tubes (another design that I am quite familiar with) is capable of about 275 Wrms at the very best, again due to limitations of the output tubes and the voltage needed to achieve this power.

There's no doubt that each one of these amps is capable of meeting their claimed and published specifications. It's your claims that do not meet reality.

Because of this, every one of your claims is suspect.

 

[fdeck]

I've measured the output of my amps. But I have to mention a couple of things. First, I'm qualified to do this.

Second, I don't do it for the purpose of verifying the manufacturer's specs. That's too hard, because I don't know what the testing conditions were, and in any event I can't replicate those conditions. There are things like the precise line voltage and how it sags under load that I'm not controlling. Also, sharing the results of home based tests could be controversial to say the least. What I can say is that I haven't encountered an amp from a reputable vendor that made me want to dispute their specs.

Instead, I test amps just to get a better understanding of how they work, for instance how they behave when they're operated near their limits. I also use those results to help me understand what's getting fed into my DIY speakers. So it's mainly for personal education.

Fortunately, I don't own any super powerful amps, so my gear can be pretty modest. I have one of those huge ancient brown devil resistors, a voltmeter that I trust, a reasonable scope, and some other odds and ends.

I'm not an engineer. But by day, I help design measurement equipment, so I spend a lot of time thinking about what a measurement is actually measuring, and how to know that it's believable.

 

[agedhorse]

I've measured the output of my amps. But I have to mention a couple of things. First, I'm qualified to do this.

I'm not an engineer. But by day, I help design measurement equipment, so I spend a lot of time thinking about what a measurement is actually measuring, and how to know that it's believable.

Francis - you bring up a really good point that I get questions about all the time. "How do you know that the measurements are believable?" Often I get these questions because folks are really trying to understand the "why" behind something, the science behind a commonly misunderstood principle, claim or statement. Occasionally, the questions is not really a question but a statement that I/we can't possibly know that and I am just arguing a point that I am wrong about. Generally, when looking at a claim to determine validity, we will look to see if the components themselves can support the claim. If the claim is "bogus", usually we will find that not just one aspect of one part doesn't support it, but that multiple parameters of multiple parts can not support it... which is more or less the gold standard of "impossibility".

Since I do this every day, and I happen to be an EE (for better or worse ), Perhaps it's time again to answer this question about how much power an amplifier can deliver based on the science and engineering behind the principles of power. Since I already started with the Trace Elliot AH-350X (Mark V series), I will take this example step by step through the process so that you all can see that it's not magic, there's no BS, there's no snake oil, and it's the same things that every test lab in the country does when evaluating a product for safety. Note (to be crystal clear) that I am in no way disrespecting Trace Elliot, as I worked for one of the companies that owned the brand and have done fairly extensive engineering analysis on this model/series so it's something that I am quite familiar with and have the data easily available.

1. Power is related to voltage and current, voltage and current are related to resistance/impedance. All of these elements are related through Ohm's Law. For a sine wave, peak voltage is 1.414x the RMS voltage. Since Power = V^2/R, and 1.414^2 = 2, peak power is by definition 2X the RMS power due to the squared factor in the power equation.

2. The operating voltage inside the amp that powers the amplifier's main supply rails defines the power that the amp can deliver. This rail voltage is in turn defined by the secondary voltage of the power transformer.

3. Every real amplifier has internal losses at rated power. The predominant losses are sag of the power supply (mostly the transformer) and the losses of the amplifier itself (primarily the losses across the output devices, which in the case of Power FETs are higher than for bipolar power transistors). The best transformers used in bass amplifiers typically show a sag of ~7.5% at rated load at the lowest impedance (say 4 ohms in this case), and about 1/2 of that (3.75%) with the higher load impedance of 8 ohms. For the output devices, this amp series uses the 2SK135/2SJ50 devices which are lateral geometry and has a Vds saturation voltage of about 7 volts approaching the drain current of 7 amps (maximum rated drain current for the device, more on this later because there's a unique aspect to this type of device that comes into play)

Here are the calculations that determine the maximum rated power of an amplifier using 2 pair of these devices in push-pull arrangement into an 8 ohm load:

Power Transformer: 92 volts center tap, conventional power supply and rail configuration. For a 92V CT, that would be 46V-0V-46V and the unloaded rails will be 46 x 1.414 = +/-65V (DC). Now subtract 3.75% (+.6V bridge drop) from this number and you will have the best case loaded rail voltage of (.9625 x 65V)-0.6V = +/-62V

Power Amp: Taking the available rail voltage of +/-62V, now deduct the amplifier losses across the output devices. We know that at the rated current of the devices (7 amp), there will be ~7V drop but this drop varies with current (roughly 1V of drop per amp, but it's really a non-linear drop). So, we need to determine the calculated current to know what to use for this drop... going back to statement #1, power, voltage, current and resistance are all inter-related. We are looking for total current, knowing approximate voltage and resistance so will use the formula V = I x R which becomes I = V/R = 62V/8 = 7.75Apeak but since we have 2 devices, this is ~4 amps/device and we can now use ~4V as the output stage drop. (note that we have to use the peak voltage due to the saturation occurring at the signal's voltage maximum)

Ok, we now have all the information needed to determine the theoretical maximum rated power of this amplifier. The maximum peak voltage of the output signal is defined as the loaded rail voltage minus the drop across the output devices, which would be +/-62V minus +/-4V = +/-58Vpeak.

Now, converting this to RMS voltage, 58V/1.414 = 41V RMS. Power = V^2/R = 41V^2/8 ohms = 210 watts RMS at 8 ohms. In practice, it will be a little less than this because there are other losses that have not been accounted for. The amp was rated at 200Wrms into 8 ohms and 350Wrms into 4 ohms, so this agrees closely with the science based calculations.

As a check and balance, why can't the amp deliver more than this? Several reasons. In order to deliver 400Wrms into 8 ohms, the amp would need to swing 57Vrms into the load which is 80Vpeak, now adding the losses that would be (80V/.9625) + 7Vsat which would be a rail voltage of +/-90V. The output devices are rated for 160V maximum, and +/-90V is 180V, therefore we are well above the production voltage survivability capabilities of the devices. As a second check & balance, if the voltage was that high, the amp would try to deliver ~700 watts RMS into 4 ohms (again, that pesky Ohm's law) the RMS current would be I = P/R = sqrt(700/4) = 13 amps RMS, which is ~19 amps peak. Since the devices are rated for 7 amps peak, two devices in parallel would be ~14 amps peak and there is simply not enough current capability available... the amp would either fold back and collapse due to the gate protection current limiting mechanism or fail. So there are 2 separate mechanisms in addition to the basic calculations that support the manufacturer's ratings as being true, accurate and honest.

This is why science is important, just stating something as true, as fact, that flies in the face of the underlying science, does not make it true. Hope this helps answer the question of how something can be determined to be accurate and believable, rather than just a random statement based on nothing but myth, rumor or opinion.

 

[BassmanPaul]

Thank you for a concise, educational post.

I have to argue a bit. I have a home designed amplifier that has 400W output at 8Ω. It's powered from two 9V batteries! Honest!

 

[agedhorse]

Thank you for a concise, educational post.

I have to argue a bit. I have a home designed amplifier that has 400W output at 8Ω. It's powered from two 9V batteries! Honest!

Those had to be BFB's really big

 

[BassmanPaul]

Those had to be BFB's really big

BTW Where do I hand in my Thesis on the subject?

 

[AudioTaper]

I have a fairly basic setup at home.
Frequency generator
8 Ohm 600W dummy load
Analog oscilloscope.

I did some work on my Eden WT400 a few months back to improve reliability (eliminate known problem areas) and bench tested it before and after. There was no change to the power output, but the front end ran cooler.
Signal gen put out a 500Hz sine wave and I brought the gain up until right before the signal light comes on (about where I have it set for my active bass) then I brought the master up until right before any perceivable distortion or clipping (about halfway)
I used a T on the coax input to have channel 2 show the input. (Why you see two waveforms on the screen)

Sine wave was a clean 36V P-P which translates to about 113W RMS. (Please, check the math)

I have another 8 Ohm resistor so I can run it at 4 Ohms if I want, but I would expect to see about 229W RMS if everything else is the same. The heatsink needs fans, it got hot real quick.

 

[agedhorse]

36V peak is 25.5Vrms which is 81Wrms into 8 ohms. I suspect that this is below the clip point, and perhaps will be lower if you have an AC line voltage below 120V too. If that's peak to peak, you are WAY low.

Generally amps do not quite double their power with halving of impedance due to limitations within both the power supply (sag) and power amp (Vsat of the output devices). I would expect the 4 ohm output to be about 1.85 - 1.9 x the 8 ohm rating because of this.

 

[beans-on-toast]

I have a fairly basic setup at home.
Frequency generator
8 Ohm 600W dummy load
Analog oscilloscope.

I did some work on my Eden WT400 a few months back to improve reliability (eliminate known problem areas) and bench tested it before and after. There was no change to the power output, but the front end ran cooler.
Signal gen put out a 500Hz sine wave and I brought the gain up until right before the signal light comes on (about where I have it set for my active bass) then I brought the master up until right before any perceivable distortion or clipping (about halfway)
I used a T on the coax input to have channel 2 show the input. (Why you see two waveforms on the screen)

Sine wave was a clean 36V P-P which translates to about 113W RMS. (Please, check the math)

I have another 8 Ohm resistor so I can run it at 4 Ohms if I want, but I would expect to see about 229W RMS if everything else is the same. The heatsink needs fans, it got hot real quick.
View attachment 3934057

Nice setup. Just a couple of comments. Your measurements should be good enough for your needs. They should be repeatable, which is good, but they may not be accurate because of limitations in the equipment. Just be aware of that.

If you haven't used heat sink compound between the resistor and the heat sink, it can help with thermal transfer. Since the assembly is getting very hot, the heat transfer compound may not help enough. Still, it's good practice to have optimal transfer. A resistor will change in resistance with temperature. Use an accurate ohm meter to verify the resistance at the operating temperature. This can slightly affect the results of your calculations if the true resistance is off from the nominal 8 ohms. If you are using different frequencies to evaluate your amp, ideally the load resistor should be non-inductive. These resistors read consistently over a range of frequencies.

A low noise at all frequencies signal generator is important. You don't want to be injecting noise at the front end with the sine wave.

You know how a scope probe can be adjusted for a square wave. Similarly, BNC terminators and cable compensation adjustments (like with a scope probe) can be used when using T-junctions to connect multiple pieces of equipment. This will ensure that there is minimal signal loss and the levels are correct.

 

[agedhorse]

At <1000Hz audio frequencies, probe compensation will not be an issue.

 

[seamonkey]

This is a great video on how at one point in time they could make a standard watt meter


Moving coil meter for the meter arm, and and electromagnet fed by the current.
Really smart for how it works. It has a certificate for up to 400hz. That's a good frequency.

I'm sure they could be made very affordable these days. In fact the shields look like 3d printed metal made today.

 

[fdeck]

@AudioTaper the only thing I'd add, which might just not be in the picture, is a multimeter that I really trust. My rationale is that I don't really trust the voltage scale of oscilloscopes any further than I can throw them. I owned a scope of the same model as the one in your picture, and then a used digital scope. Companies dispose of the old scopes when the calibration and service contract cost more than a new scope. Eventually, switch and pot contacts get flaky.

Also, as a word of caution to everybody: The output terminals of an amplifier can produce voltage and current levels that can do serious damage. You can't trust either speaker terminal to be "ground" or to be referenced to zero volts. Many of the modern Class-D amps are permanently wired in bridge mode, and some are a floating bridge, meaning that the two output signals are referenced to a voltage above ground.

 

[BogeyBass]

Depending on topology and output devices you very rarely see a double of power at 4 ohms.
Likewise depending on power supply it's usually no where close to double the power.

Sine wave was a clean 36V P-P which translates to about 113W RMS. (Please, check the math)

I understood your readings but be careful with terminology. Far as voltage peak or peak to peak. Or Vp and Vpp

Anways 36 Vp or 72 Vpp is realatively low. Is this a mosfet or BJT amplifier. That is 25.45 Rms
I'm guessing your just eye balling a clean sinewave. What voltage does it get too if you just start hitting the rail?

36 volt peak would assume maybe 40 to 42 volt rails. Depending on how good the output devices swing to the rails. Which would still be not very good at 8 ohms. Unless it's mosfet. Basically a 150/160 watt amp at 4 ohms with 42 volt rails.


Maybe it's the way it was measured. Or it's biased very high. If the amp is supposed to be in 200 watt territory.
I'd assume at least 49 to 52 volt rails. And 36Vp would be extremely awful especially at 8 ohms

 

[beans-on-toast]

This is a great video on how at one point in time they could make a standard watt meter


Moving coil meter for the meter arm, and and electromagnet fed by the current.
Really smart for how it works. It has a certificate for up to 400hz. That's a good frequency.

I'm sure they could be made very affordable these days. In fact the shields look like 3d printed metal made today.

I have Weston meters like that, voltage and current. A work of art.

 

[AudioTaper]

36V peak is 25.5Vrms which is 81Wrms into 8 ohms. I suspect that this is below the clip point, and perhaps will be lower if you have an AC line voltage below 120V too. If that's peak to peak, you are WAY low.

Generally amps do not quite double their power with halving of impedance due to limitations within both the power supply (sag) and power amp (Vsat of the output devices). I would expect the 4 ohm output to be about 1.85 - 1.9 x the 8 ohm rating because of this.

Yeah, this was a fairly non-scientific. I was verifying operation and didn't monitor the input voltage. I measured the rails, I think they were +/- 72V.

The power output seemed low, but I wasnt pushing it as hard as I could have. The waveform wasnt as clean once I pushed it up into the 40s.

When I get more time I should set it all back up and use my Tektronix digital scope for a more accurate measurement. (Out of cal, but not out a lot)

Rated power is 250W into 8 and 400W into 4.

 

[BogeyBass]

As the title says, how would I go about this? I'm sure I need some kind of multimeter, but where on the amp would I take the reading?

I'm sure there is various videos showing guys doing bench tests on audio amplifiers.

You basically need a oscilloscope and realated probes. A dummy load which can handle the estimated wattage from the amplifier. And a sinewave signal generator.

You just apply sinewave signal till it starts to clip the waveform and that is the end of the road. You'll get a voltage peak or voltage peak to peak measurement in voltage. And calculate the RMS voltage. And apply the RMS voltage using ohms law to whatever impedance dummy load your using.
Usually a 8 ohm or 4 ohm load.

All the can of worms from a extremely basic measurement is. What percentage of distortion your measurements are at. And how accurate your dummy load is at rated wattage.
When your signal starts to flat top or starts distorting. Percentage of distortion increases. So if you use higher level of distortion readings the wattage can be slightly higher. Likewise they love " burst" test which can have extremely high unrealistic wattage readings. But as a designer a good burst reading gives you a good idea of how good the power supply is.

Most use " clean" power for a basic bench test and without a distortion analyzing equipment to know how much distortion your at. You basically raise the signal till the sinewave starts to clip and then slightly back off for a clean waveform. And then get you Vp or Vpp voltage.

To be more precise or to know real world distortion levels even with a pretty sinewave. You'd need a distortion analyzing equipment. For basic test you'll see the sinewave on the scope hit the rail or start clipping. And that's all the voltage the amps got.

Back in the old days. It's not a dead accurate measurement. But if a amp claims outrageous power or certain level of power. It's going to have a fuse rating that at least makes sense. If a amp claimed high power but had a realatively small rated fuse you could call BS.

Having built boost/buck converters for 12 volt car amplifiers. I can call BS pretty quick just looking at fuse ratings. It's not accurate. But that market likes to claim amazing 600 watt or 1000 watt amps. And I can tell right away many are way way way below that by the fuse. Then again that was years ago. Things have changed