continuous power handling

[nickb]

Are there ventilation holes in the bottom? If so, what happens if you raise it off the bench an inch or so? The construction looks at though the fan doesn't do much to cool the main heat sinks - they appear to rely on convection. Another suggestion, could the quiescent current be too high?

Maximum dissipation occurs when the output power is about (2/3)^2 of max. Is this one channel or bridged during this test?

[Rhodesplyr]

But here's a concrete example -- an Eden WT800...

Considering what I've seen inside the couple of Eden amps I've worked on, I wouldn't push my luck very hard in terms of seeing how much continuous power I could get out of it . . .

[km6xz]

What specifically are you seeing wrong with the design? They work fine and hold together.

[J M Fahey]

I tried running one side with 440Hz continuous at quarter power (137W.) It seemed reasonable because at that power, each transistor only has to handle 23W,

No offense implied but, sorry, it's not that way.

You are confusing *output* power with *dissipated*power, a completely different thing.

And heatsink temperature , which you use in your derating calculations, with chip temperature.
Etc.

In a very simplified way :

1) transistor dissipation is a fraction of output power, but the fraction is not constant.

Dissipated power will reach a maximum at a certain point , around 40% of rated power, if we use a sinewave and rail voltages are constant.

2) chip/die temperature will be *much* higher than heatsink temperature.
We have a thermal resistance between die and case, another (in series) between case and heatsink, plus more between heatsink and air.
Not forgetting that air usually will *not* be at 25ºC but higher, at least in the rack/enclosure or where the amps are.

Best would be to have a thermocouple or some diode sensor inside the chip (a few advanced transistors do) , but usual practice is, say, to have a 50ºC sensor to turn fans on, and a 65ºC to 80ºC safety sensor to turn things off if things go wrong.

As of the "old" power rating, amps had to stand 1 hour (might be non continuous) at 40% power (if I remember right) into the rated load.

[woodyc]

...As of the "old" power rating, amps had to stand 1 hour (might be non continuous) at 40% power (if I remember right) into the rated load.

Juan, I'm just going to ignore the nit picking and agree with you about the hour idea.

I think maybe continuous power handling at 30% of rated power for an hour might be a useful design standard. At the end of the hour, the temp should be stable and nothing should be over its ratings. The hour could alternately be full power bursts with a 30% duty cycle. That would eliminate the need to separately make sure the amp can put out a burst at full power.

The reason for the 30% is because...

I know that Eden is a total fail. I also have another bass amp here -- a GK 1001RB-II. Its vastly better than the Eden, but I wouldn't call it a great amp since they break a lot too. But it is a 700W amp designed for a 16% duty cycle, or 113W continuous and it runs pretty cool at that level. So I figured a solid number has to be higher than 16%.

Another factor is the Crown datasheet mccollum put up which says compressed rock is 40%. I assume they know and care about this because people play CDs loud with their amps on a regular basis. So maybe 40% would be the high end of the range for instrument amps.

[Steve Conner]

It's important to remember that the maximum dissipation in a Class-AB amp isn't when it's putting out its maximum output voltage. As you turn it down below clipping, it dissipates more, down to about 40% power at which point the dissipation starts to fall again.

So the worst case scenario for a bench test is when the amp is loaded to its minimum rated impedance and putting out 40% power. I don't know of any amp that will handle this without tripping the thermal protection. (Or just blowing up if the design is bad.)

The easiest scenario is running just short of clipping into an 8 ohm load. I would expect just about any amp to handle this indefinitely. If you feel even this is too hard, remember, turning the output voltage down does not make the test easier!

The rise in dissipation at lower voltages is mainly a concern with sine wave bench tests. It doesn't manifest with music or white noise, because the random nature of the signal averages together all the possible output voltages. However as Rod Elliot points out, it certainly does manifest with a square wave from a high-gain master volume guitar preamp, even more dramatically than with a sine wave.

[woodyc]

It's important to remember that the maximum dissipation in a Class-AB amp isn't when it's putting out its maximum output voltage...

Steve, its an interesting point in theory, but in practice the temperature rises with the output power. Maybe the details of the output devices are swamped by the fact that there are lots of other resistive elements in amps?

In any event, its not all that relevant to what I'm getting at here. I just made a reasonable argument for rating an amp at 42W when the manufacturer claims its good for 1100W. I think we are in desperate need of some kind of standard.

[Steve Conner]

The transformer, rectifier etc. will certainly always get hotter with increasing output power. But above 40% output power the output transistors will start to cool down. This is a scientific fact that you can verify for yourself. Since the output transistors are the weakest link in a cost-optimised design, the temperature of them is what matters most.

I think it is absolutely relevant. The assumption behind your original post is that increasing the output power always makes the output stage hotter. This assumption is false, so there's no point even trying to answer your question.

I prefer to put it in the following terms: With the amp driven just short of clipping by a sine wave, what is the minimum load impedance that it can drive continuously at a reasonable temperature? My answer was 8 ohms. Decreasing the load impedance always makes the output stage hotter, there are no surprises here.

My own hi-fi amps (I've designed two) will run all day on a sine wave test at 40% power into 8 ohms or 100% into 4, but they will thermal out at 40% into 4 ohms. They are pretty conservatively built compared to musical instrument amps.

[woodyc]

...I think it is absolutely relevant. The assumption behind your original post is that increasing the output power always makes the output stage hotter. This assumption is false, so there's no point even trying to answer your question...

Sorry my question is so flawed and pointless. I'll try to do better next time.

[Steve Conner]

Good! See you next time.

[bob p]

It's difficult to come up with a universal answer that says how long any particular amp can be expected to reliably deliver it's full rated power. So much depends on the design of the amp, and how well it's designed in terms of heat dissipation.

Bear in mind that there are several contributing factors that are the major determinants to an amplifier's cost. The primary offenders are the transformer iron, the heatsinks, the capacitors and the semiconductors. And the box, depending on how HiFi / fancy it has to be. All of the other parts are pretty insignificant as far as total amplifier production cost is concerned. It shouldn't surprise anyone that the companies that build amplifiers have spent a lot of time trying to decide where they can cut corners to make amplifiers more inexpensive to build and more profitable to sell.

A good way to illustrate these differences is to compare amplifier classes. One reason that high powered Class A amps (like an old ML series Levinson or a Krell) are so expensive is because they require a lot of transformer iron, a whole lot of heatsink aluminum, lots of semis, and a fair amount of capacitance.

Compare that to a more efficient Class AB design, which requires a lot less heatsinking, and the cost is reduced. Other expensive parts also become more economical, so it's no coincidence that a Class AB amp is cheaper to produce than a Class A amp because it's not designed to dissipate so much heat.

Jump ahead a few years and you'll see Hitachi-type rail-switching designs that were made famous by Bob Carver in his M-series amps. By using several voltage rails and commutating between them, the power consumption of these amplifiers and the dissipated heat became drastically reduced -- to the point that these amplifiers have almost no heatsinking requirement. Take a look inside of a Carver M-1.5t and you'll see no finned aluminum heatsink in the amp. The heatsinking in this amp is limited to bolting a dozen or so TO-3 transistors to the metal box. These amps aren't designed with much heat dissipation in mind because "they don't need it." The prospect of operating a gain stage at or near the limit of the voltage rail increases efficiency dramatically, and switching between 3 voltage rails allows the amp to always operate at near-max efficiency. This results in far lower requirements for heat dissipation in the amplifier design.

Of course, that's because the rail-switching amp is only designed to be reproducing musical signals, which are subject to short duration bursts; they're definitely not sine waves. As an example of this, you can drive a rail-switching amp with a musical signal all day long at it's rated power output and it's not going to fail. Drive it at it's rated output with a sine wave, and it's going to get so hot that it's either going to shut down or fail. Why? Because these amps are not designed to reproduce full power with sine wave drive. Bob Carver was a genius at figuring out how much money needed to be spent to build a musical-reproduction amplifier, and removing every part that cost money that wasn't absolutely necessary.

Class D amps take the efficiency rating to the next level.

At the opposite end of the spectrum, we have military radio transmitter amplifiers that are heavily heatsinked and are rated to work continuously under full sine wave drive conditions.

My answer to the OP's question is that there isn't really a universal answer. There are two variables that have to be defined when considering a power rating specification: 1) type of signal being reproduced (sine wave drive, music signal, or something in between), and 2) how long the test period will be conducted. There's a huge difference in the design requirement for heat dissipation for an amp that is rated to run under sine wave drive at infinite duration, and a home stereo amp. When it comes to power ratings, at one end of the spectrum we have sine wave drive for infinite duration, and at the other end of the spectrum we have peak music power. Somewhere in the middle is RMS power of indeterminate duration, and all sorts of other non-standardized tests that have indeterminate value.

Clear as mud?

As an example of this, I restored a pile of old Carver Professional PM-1.5 amps for live sound use. After rebuilding them, I attempted to test them under sine wave drive. They met their rated power specs, but they got really hot and their cooling fans went into panic mode. I had an opportunity to speak to Bob Carver about this, and his response was quite candid: Those amps are designed to reproduce music -- you can't test them under sine wave drive, they'll overheat.

The moral of the story is that it all depends on an amplifier's individual design, and there are no universal design standards.

[Mick Bailey]

I've just had an Eden WT800 for repair, but the older model - series A. When I get these I cut away the fan vent perforations and install a wire guard. The amps will run comfortably hand warm at 360w into 4 ohms with flat EQ.

The Eden referred too looks like it could have a fault - it seems to me to be running too hot for the measured output.

[J M Fahey]

FWIW my own B300 amps put out 305 to 320W RMS into 4 ohms loads, all day long; a continuous Month or forever if you don't care about the electricity bill, and heatsinks stay warm to the touch, quite bearable.
One important detail is that my side mounted fan blows *first* into the conventional EI power transformer, which to boot is mounted sideways so air can go around it *and* around the bobbin, between copper and iron; then cools the heatsinked bridge rectifier , then the electrolytics and finally the heatsinks.
The transformer itself never rises above, say, 35 to 42ºC, depending on outside temperature.
Just in case a carelessly thrown sweater or jacket blocks the fan opening, I add a 65ºC thermal cutoff to the heatsink, but it never trips.

So yes, it's no big deal to make an amp which meets rated power specs continuously.
It isn't even expensive.

Most which don't, simply didn't even try, being for Hi Fi or light use.

[km6xz]

Each standards organization and many national governments specify how power is to be measured and reported in promotional material. In the US we got the FTC involved when they got so many complaints in the 60s of highly inflated power claims. There were dozens of power ratings, each more exaggerated than the next, particularly for console stereos and TVs of the error so the US Federal Trade Commission established a test and advertizing requirement. An amplifier was to be run in a preconditioning phase of 1 hour, at 1khz running 1/3 of rated power, before a measurement run that involved specifying the highest RMS sinewave power that would result in no more than a claimed THD level across a claimed bandwidth.
It became illegal to advertise an amplifier power rating without that form: xx watts rms into x ohms(standardized on 8 ohms but other stated Z could be used) from xx hz to xx khz at no more than 0.xx% Thd. The brown-goods makers protested loudly that their 400 watt TV output using their IPADP rating(instantaneous peak absolute destruct power) could only be advertized as "Glorious power of 6.45 watts RMS at 8 ohms load at not more than 3.2% THd from 60hz to 9khz" after the new FTC rules went into effect. Heat sinks got larger, power claims got smaller and everyone got better, more honest ratings that could be compared between various products for the first time.
In the late 90s consumer gear makers proposed that the rating was too harsh and Thd was dropped and power for preconditioning was lowered to 1/8 power. UL testing labs tested for heat and safety at 1/8th power for a long time before that.
So there has been a standard for years, I am surprised no one mentioned that it is not so arbitrary as suggested in some of the posts.

[km6xz]

Power ratings are misleading in that it replaces the final goal of perceived loudness with technical requirements of one component of the chain based on a fast fading technical limitation of AB amplifiers. Few consumer and even fewer portable consumer gear are AB amps now.
With processing to enclose the whole system in a feedback loop where perceived loudness and sonic character is the goal, and involves transducer efficiency, program material, environmental noise, we are fast approaching a period in what you hear or think you hear is all that matters. A battery powered 3 watt amp with 93-94% efficient Class D amp running on a few AA cells, driving a 1950s Klipschorn can force a lot of people to rethink their concepts of power relating to loudness.
We are going through an era where music has been augmented by subsonic components that were never used in music before that complicates things a lot, and in no small measure related to the downturn in importance of music to the current western cultures. It has certainly lowered access to live music and made new artist less accessible to live audiences, for a number of artistic, financial and logistics reasons. It is a fad that will pass. New styles and tastes will replace the current habits and probably see an increase in acceptance.
Powered speakers and sealed instrument amps make a lot of sense since they allow a designer to focus on results of the entire system instead of components, each having to interface with unknown other equipment. For example, for mass and efficiency, what is so optimum about a nominal 8 ohms load z when we know that z varies all over based on frequency. We have non-linear mechanical drivers/cabinets but expect the driving amp to be optimum driving purely resistive linear loads. A self powered speaker, from a designer's point of view opens lots of options to get the design goal that can improve performance, efficiency, lower unwanted artifacts and lower cost, and have it all adaptable to the acoustic space it is in. Just the option of using real servos for full system feedback loops, where the actual output of the speaker is the signal compared in the digital or analog feedback, instead of having the limited window of performance monitors that back emf from the speaker terminals now are used in high damping factor amps as feedback signal.
When a guitar or bass amp is rated at 1200 watts with its matching speaker, it says nothing about the loudness and the only factor that we absolutely know from that rating is that it is terribly inefficient, intentionally, to generate and then waste that much power for the marketing claim that it is that powerful. Big amps are fragile, expensive, hard to handle and less reliable than lower powered rigs since both are constrained to the same contemporary technology.

Regarding the Eden WT amps, what failure rates are being seen? My shop was warranty station for them and many of our clients and stores around the country used them and never had a reliability problem. Maybe a bit of diagnostics of the failure mode is needed to find out if something has changed, if subsequent repairs on particular units is skewing the opinion that they are generally not reliable. If you are doing 200 a year, a pattern would develop that would be much more valid than a pattern based on 3 a year.

I read many comments from tech issuing blanket statements about a brand being junk but almost always those opinions are coming from a very small sample of repairs, in low volume shops, and usually non-warranty so a higher percentage of repairs are caused by improper prior non-professional repair. Anyone can call themselves a tech. I kept a database of the last 60,000 repairs done before I moved here and there were patterns but usually not with models and brands complained about here. Some types of units had a predictable service need, usually those with mechanical systems such as DAT or ADAT decks, or SVT amps used on the road, for Fenders with complex switching input jacks that were using very thin metal contacts. Each of these and others had predictable wear and tear related service cycles.