Yesterday we covered one of the advantages of the higher power supply voltages afforded by tube circuits: linearity.  Today we’ll jump into the second advantage of increased voltage, headroom in power amplifiers.

Headroom is a funny thing.  If you take a SET amplifier of perhaps 18 watts and try and get 20 watts out of it, you’ll be well into clipping of the amp – that is if you are trying to play continuous sine waves.  But if you are playing music and hit a peak that needs 20 watts, perhaps even more, headroom with high voltage allows you to exceed the amplifier’s wattage rating – if only just for a moment – and for music that’s really all we need.

A small power SET amplifier has plenty of voltage in its power supply to deliver far more watts than it’s rated at – yet it’s limited not by the voltage but by the output devices’ maximum rating and the wattage of its power supply.  Clearly, if one wanted to, you could add more and more output tubes with a bigger wattage power supply and get a bigger SET amplifier – but it may not be important if you have a very efficient pair of speakers and keep your listening levels in check.

The key here is headroom: the ability of the amp to produce short peaks of music when needed, peaks that exceed the rated power of the amp by up to 3dB (twice the power) in some cases.  Much depends on the capacitors in the power supply as these are the energy storage devices used by the amp to supply peak power.

High voltage also exists in solid state power amplifiers.  A typical 200 watt per channel amp will have a little more than half the power supply voltage of a tube – 160 volts or so – and bigger power amps have up to 200 volts approaching that of a tube.

One of the advantages of building a high power solid state amp – an amp with far more power than you might need – is the higher voltages needed to make that higher power.

As an example our new power amplifier is a brute – 350 watts into 8 Ohms, 700 watts into 4 Ohms and 1200 watts into 2 Ohms.  This is huge and far more power than almost any system really needs to play at live levels.  However, even on an efficient loudspeaker, this amplifier will sonically outperform smaller solid state amps because of the increased linear region and the utter lack of compression at any level.

Tube amps of any wattage only have less than 1/3 higher voltage than does a high powered solid state amp – but they too are saddled with something else that a solid state amp doesn’t have to deal with – an output transformer.

Let’s take a look at that tomorrow.

High Voltage

Most tube power amps have an advantage over most solid state amps: high voltage.  If we look at a typical tube vs. solid state design, that tube design might have up to ten times more voltage than the solid state design – at least in a preamplifier.  The tubes need the high voltage to operate properly while the solid state devices do not.

This has several advantages: increased linear region and headroom.

Every amplifying device has a region that it is perfectly linear – meaning what comes out is an exact copy of what went in, only larger.  This linear region varies from device to device but generally we could suggest that up to 25% of the total operating range of a typical amplifying device could be considered linear.

If the device has a power supply of 30 volts, then we could say that 7.5 volts (25%) would be the linear region and if we manage to keep the music signal inside that range, then we don’t have to use feedback and other tricks of the trade to keep it linear.  But 7.5 volts isn’t a large area for even a preamplifier, so we’re going to be relying on crutches to keep things linear.

But if the device has a power supply of 300 volts, like many tubes do, then our linear region just went from 7.5 volts to 75 volts and that’s a lot of region in anyone’s book.

This is one of the reasons why all PS Audio products are so power supply intensive.  Not only do we subscribe to bigger is better, but higher as well.  Our typical operating voltage for our DACS and preamps is not the standard 24 to 30 volts, but 60 instead.  This gives PS source products twice the linear region of most solid state devices – yet far lower than a tube would have.

From our perspective, if you can keep the signal you want well within the linear boundaries of the amplifying device, you’re fine – so 15 volts of linear region is far more than the maximum output signal is going to get to – hence we stick with that.

The additional linear region of a tube is wasted once you pass over a certain level.  Wasted, that is, on linear performance – but valuable on another front – headroom in an amp.

Tomorrow let’s cover headroom.

Paul McGowan – PS Audio, Intl.

SET down for this one

One of the quirkier and exotic types of power amplifiers are a tube based product known as a Single Ended Triode or SET for short.

A SET is a class A power amp using one output tube, typically a triode, to drive your loudspeaker.  They are generally low power devices with poor efficiency – as is any class A power amplifier.

This class of power amplifier has a small, but very passionate, group of supporters who swear nothing more musical exists on the planet.  There may be an element of truth in their thoughts.

First, however, we should clear away some of the mystique associated with these power amplifiers so we may better understand the SET.

When I suggest there may be some truth to their revered musicality claims there are three main areas where these qualities come from: single ended, 2nd order harmonics and high voltage – the first two shared by most class A power amplifiers, the third shared by most tube amplifiers.

A single ended amplifier means that all the amplification is handled by a single output device, be it solid state or tube.  While certainly the most inefficient means of delivering power, single ended designs are free from the crossover problems found by the more common and efficient output pair we generally see in an amplifier.  Score one.

When a single ended amp gets near clipping its distortion products are more heavily weighted towards second order harmonics which are very pleasing to the ear, relative to third order harmonics.  Score two.

So far nothing here is exclusive to a SET because a solid state version would have the same properties.

What can be said to be unique to a vacuum tube single ended amplifier, relative to a solid state single ended amplifier, is voltage – something all tubes have in common.

Tubes generally run at far higher power supply voltages than do solid state devices – in some cases ten times higher – and here is where some of the mystery about how you can get satisfactory loudness levels from a SET come into play.

Let’s look a little closer tomorrow.

Paul McGowan – PS Audio, Intl.

What you put in doesn’t always come out

Thank you for all the replies to yesterday’s question about loud levels and compressed sounding peaks.  I would say that the response was about 50/50 that you experienced what I was writing about and my guess about those differences would be the type of music you listen to, the room size, the type of speakers, your loudness levels and the size of your amp.

There are a lot of variables going on as you would imagine.  I’ve been to enough listening environments to have heard the gamut from great sounding SET systems of only a few watts to mega systems of many watts and they are all different.

Some of the SET setups I have heard sounded great on certain kinds of music but dreadful on others.  I’ve noticed, for example, that many SET owners choose their music carefully rarely playing anything with high average value – like my Pink Floyd example – and concentrating their selections on high peak value music like classical or jazz.

On big well powered systems with lots of drivers, like the Infinity IRS I am hoping to acquire, just about anything big sounds effortless if you have enough reserve amp power to make it sing.

And this brings me to the subject of today’s post, the speaker’s ability to remain linear with level.

Loudspeakers, just like phono cartridges, are mechanical devices with very defined limitations when it comes to extremes.  This means that as a loudspeaker driver (or phono cartridge) gets close to the maximum level the distortion rises as the amplitude becomes non-linear.

This is due to the nature of how these magnetically based motors work.  As you approach the limits of the drivers excursion a number of problems arise, but the one we’re most interested in is that the output of the driver is no longer linear – meaning it produces decreasing amounts of loudness for a given input signal – and we want it to produce a constant output vs. input.

RTEmagicC amplitude input output 300x213 What you put in doesnt always come out

You can see an example of this in the graph where the Linear System line is what you want for a given input and the sloping line is what the speaker is doing at its output.

To overcome this you can inject more power if you know where this is going to occur – or you can build a different type of loudspeaker – one that has many drivers doing very little individually but a lot as a whole.

Big line sources like the Pipe Dreams, IRS, Genesis I and so on have a better chance of never getting close to this compressed state because you have so many drivers working at lower levels you rarely ever get to the compressed state.

Tomorrow let’s start delving into SET power amps.  many of you have asked me to say a few words about them.

Paul McGowan – PS Audio, Intl.


In our quest to figure out why we might need a big power amplifier vs. a smaller power amplifier I’ve mentioned a lot of numbers: some scary big.  Many of you have written to me that the huge wattage numbers I’ve been mentioning don’t really stack up with your real world experience – where a much smaller amp seems to work just fine.

Indeed, most of what we listen to never even comes close to needing the extreme amounts of power I mentioned.  Let me give you an example.  Take a look at this snippet of music.

Screen Shot 2013 02 22 at 3.55.43 PM 300x83 Compression

This is from the loudest passages of music from Pink Floyd’s Welcome to the Machine I wrote about earlier.  The top trace is one channel, the bottom the other.  The dark blue areas are the peaks and the light blue areas are the average levels.  Remember we talked about peak vs. average? This somewhat shows what I was referring to.

But note in particular how most of the very top and bottom of each channel is basically empty – very few peaks come near the top and bottom edges, which means we’re almost never approaching anywhere near the clipping point.  That’s somewhat normal from what I’ve seen.

So, let me ask you a question – and this is something I ponder a lot.  How often do you find that you listen to a “live level” track and the sustained peaks of that track sound strained or compressed relative to the quieter passages?  I can tell you that on all but a very few systems I find that slightly compressed sound, bordering on hard and harsh, happens on many recordings during the loudest passages of music.  It’s particularly noticeable when that loud passage is sustained as opposed to a quick note or two.

As an experiment, I can turn the overall level of the sound down a bit and that compressed harshness goes away.  If I return the volume level back again to where it was sounding compressed on the peaks, and double the wattage of the power amplifier, the compression and tendency towards harshness is reduced significantly.

The first amp probably never got close to clipping and certainly the second more powerful amp hardly strained at all.

What we can learn from this is that there’s far more than simply clipping that meets the ear when it comes to wattage ratings.

Paul McGowan – PS Audio, Intl.

Before we get started on different amplifiers and why some seem to sound more dynamic than others, I wanted to clear up a couple of points about yesterday’s post showing how many watts it takes to get appropriate dynamic range.

All the measurements we described were anechoic chamber measurements: meaning there was no room involved.  As soon as you add a room to the mix, the loudness levels at your seat are very close to the loudness levels near the speaker – certainly greater than if the speakers were outside or in an anechoic chamber – the good news is the extremes I showed aren’t quite that dramatic.

The crest factor I describe is the difference between the average level and the peak level over a specific slice of time.  Many of you asked what was the difference between crest factor and dynamic range: the answer is that dynamic range describes the difference between the loudest and the softest passages of music in a complete piece – where crest factor is describing the difference between an averaged level (not the softest) and the peak level over a constant interval.

What’s really important to understand is that the needs for a power amplifier really are dependent on the type of music you listen to, the type of loudspeakers you have and the size of your room – among other factors.

If, for example, you listen to a lot of high average loudness rock music in a medium sized room with reasonably efficient speakers, then you’d want to find an amp with fairly high power, lots of heat sinks and a very stout power supply.

If, on the other hand, you listen only to chamber music or relatively low average loudness in the same size room, you could get away with far fewer watts and an amp with a fairly “spongy” power supply (like many tube amps have) if you don’t listen too loudly.

lastly, watts ain’t watts.  Loudspeakers are very complex loads and there’s far more than driving a resistor with a power amplifier and simply measuring the wattage that results.  Loudspeakers have different impedances at different frequencies, they are also inductive so the phase and current requirements can be off the charts.

Your power amplifier may be doing ok or it may be clipping or getting close to clipping without you ever knowing it.  VU meters on the front of a power amp are pretty much meaningless to measure peak current and I would never trust what I saw with them.

Tomorrow we’ll start to look at some different types of amps, I’ll try and explain about current vs. voltage etc.

Paul McGowan – PS Audio, Intl.

Thanks to reader John McRee for suggesting the title of today’s post.  Yesterday we learned we need 8 times more power to reproduce speech than we do sine waves – which may then surprise people that loudspeaker manufacturers continue to use sine waves as a basis for their efficiency measurements.  It certainly surprises me since we don’t listen to sine waves and one might question the value of such measurements.

If speech needs an 8:1 increase in amplifier power over sine waves, how about music?  Try 50:1 on for size.  The crest factor of music is generally around 20dB where speech is 12dB and sine waves are 3dB.  Remembering that each 3dB increase in crest factor requires the amplifier power to double, this means that a reasonably efficient loudspeaker, say 90dB efficient, needs 50 watts of power to reach musical peaks of 90dB, but only 1 watt to play the same loudness level sine wave.  50 to 1.

Now we should ask ourselves several questions: chief among them would be “what is loud”?  90dB is loud, that’s for sure, but if you’re listening to dynamic music, like orchestral, you can easily reach 100dB peaks when the orchestra plays loudly.

Our musical math would then show us if we need 50 watts to reach 90dB, we need 10 times that amount to reach 100dB!  That’s right, do the math.  For a 90dB pair of loudspeakers to produce 100dB peaks in your room without clipping the amplifier, you need 500 watts of power.

The average efficiency of loudspeakers is perhaps 87dB, and for such a loudspeaker you’d need 1000 watts to reach the same loudness without clipping.

So, what about more efficient loudspeakers?  Well, simply reduce the need back again.  If your speakers are 93dB efficient you need 250 watts, if they’re 96dB efficient you need 125 watts and so on.

One thing we haven’t yet covered is the distance from the speaker to the listener – because we must remember that all these figures in our musical math are based on sitting 1 meter away from your speaker (3 feet).  I don’t know about you, but I don’t personally sit that close to my speakers.  The further back you are the more power you need to reach the same level at your listening position.

Many of you are asking me if all this is true, which it is, how come my little 10 watt SET amplifier works just fine on my speakers?  Are tubes so much better at peaks than solid state that I am getting away with something?

Let’s take a look tomorrow at some surprising loudness figures and then start delving into amplifier types.

email Music mathPaul McGowan – PS Audio, Intl.

Riding the crest

Riding the crest

I promised yesterday we’d start to learn what Crest Factor is and how it applies to headroom in an amplifier.

While understanding this concept isn’t all that important to making your amplifier power decision, getting a basic understanding of how it all works and why is probably useful for most of us who have been under the impression that anything over a few hundred watts is “overkill” when I would suggest anything under 500 watts is “underkill” for any speaker system.

Crest Factor is the ratio between peak power and average power.  Sounds technical – but it’s actually rather simple.  Think of a sine wave in terms of energy distribution – where is the biggest part of the energy in a sine wave?  Right in the middle where it is fatest as viewed from left to right (the time dimension).

2008551448 sine wave 300x225 Riding the crest

And, of course, the least amount of energy viewed in the same plane is the top and bottom of the sine wave called the peak.  Now view the sine wave in the up and down plane (the loudness dimension) and the greatest amount of energy is the highest and lowest point of that wave (although the shortest width).

If we average the amount of energy in a sine wave (add the peak and the middle together and average) and then compare that average to the peak by itself, we get something called the Peak to Average ratio – the Crest Factor – expressed in dB.

For example, a sine wave has a crest factor of 3dB meaning that on average the ratio between the average energy and the peaks is double (3dB is double the power) – where the peak is twice the power of the average.

Why is this important?  For our purposes we want to know this ratio between quick short peaks and longer average power because that’s going to tell us how many watts we need to cover the peaks and the average.

A lot of these figures have meaning only in relationship to the type of music we will be listening to.  As I mentioned earlier on, we don’t listen to sine waves.

We do listen to people speaking and singing and here the ratio between the average and the peak is far higher: 12dB is typical.

So, to put this in perspective, if you listen to a sine wave on your system, aside from people thinking you’re weird, you’d notice a steady sound pressure constantly permeating the room.  But if you listen to someone speak it’s really made up of lots of short “bursts” of sound energy – as opposed to a constant steady tone like that of a sine wave – therefore we can safely say that speech has more and higher peaks than steady tones, hence the higher dB number for our ratio between the two.

So how does this relate to our amplifier power needs?  Remember when we started I explained how we measure efficiency in a loudspeaker: 1 meter, 1 watt gives us the efficiency.

If I place 1 watt of sine wave energy into a loudspeaker and it gives me 90dB of sound, that’s all the power I will need if I am only going to play sine waves at 90dB (which is fairly loud).  But if I want to then play speech from a CD, I need 9 more dB of power to cover the crest factor of speech, which is 12dB (3dB for the sine wave + 9dB more = 12dB).

Each 3dB is double the amp power so 9dB is 8 times more power!  This means I need a a power amplifier to produce 8 watts, instead of 1 watt, if I want to play speech over my speakers at the same loudness level.

OK, enough for today.  As you can probably figure out, music is an even higher crest factor than just ordinary speech.

Paul McGowan – PS Audio, Intl.

The ultimate point source

Several of you asked me why I was so interested in a 20+ year old pair of loudspeakers that I would sell everything in the listening room to get it and what makes this speaker so special?

This giant loudspeaker has a number of charms but perhaps the biggest for me is that it’s the ultimate point source.  The most amazing thing I’ve ever heard on a speaker came from this speaker’s offspring, the Genesis 1.  I shall never forget the sound of a single guitar playing in a room – no, not the sound of a giant orchestra in its full dynamic glory, but the sound of a single instrument floating in acoustic space – and that image has stuck with me for as long as I can remember.

Aside from the fact it’s eery sitting in front of a 7 foot tall behemoth listening to something as small as a single guitar, that guitar sound was as close to right in the room as I’ve ever heard – and I want that back.

The reason small things (as well as big things) sound so right is that a true line source is as close as you’ll ever get to a perfect point source.  I know, it’s counter intuitive, but that’s what makes it interesting.

A point source is a description of a mythical product that is infinitely small – a single point – that can produce unrestrained dynamics at all frequencies.  If someone was able to create such a device it would sound amazing – and oh so real – at least as real as the recording.

But that product doesn’t exist and the second best approach to achieving it is the exact opposite – a loudspeaker that is infinitely big – one that goes from floor to ceiling in the room.  While not as good as a point source, a properly executed line source is as close as one can get and, because there are so many drivers involved, you get the added benefit of huge dynamic range possibilities.

Most of us would be sleeping on a couch in the garage for the rest of our lives should we try and convince the Mrs. to add a floor to ceiling 1 ton set of loudspeakers in the living room – such is life at home.

Those few of us having the luxury of a dedicated room, like the one at PS Audio, can only hope to someday have such a “point source” to listen to.

Paul McGowan – PS Audio, Intl.

My consumer hat

My consumer hat

In Stereophile Magazine’s March issue John Atkinson opines about the goals and challenges of reviewing audio equipment: a challenging task to do it as well and as thoroughly as does John.  I have always admired his measurements plus listening approach to high-end audio as being really down to earth and informative.

At one point in the article he asks the question “how do reviewers assess the quality of products they write about?”  He goes on to point out the standard answer is “by comparing them to the sound of live unamplified music” – a concept popularized by Harry Pearson then of TAS.

That’s the same reasoning I use as well: “does a piece of gear get me closer to the real thing?”.

John argues that the whole process of two channel recording can never get us close and therefore holding equipment to this standard doesn’t make any sense.  He would prefer instead to figure out what the intent of the designer was and report on how close he got and how that product works in the reviewer’s system for a given set of music.

While I respect his opinion and certainly agree with him that our current method of recording and playback will never achieve a live presentation in the home, I disagree about his conclusion.

If I put my consumer hat on and pickup a review of a piece of kit, the designer’s intent on a piece of equipment is certainly interesting but not all that relevant to me.  And likewise, I am certainly interested in a reviewer’s opinion on how that product sounds in their system and with music I can buy and try for myself, but I still want to know the big question in my mind – how close does this get me to believing it’s right in the room?  Perhaps that’s just a personal bias but it’s a question I always ask.

I know we can’t get live music in our home with the current recording and playback chains but half the fun for me is getting closer and closer to it.

Every once in a while I am transported somewhere else when I listen to music – even if I know it isn’t real.

Paul McGowan – PS Audio, Intl.