If DC is amplified the same amount as the musical signals, which are AC, then any little bit of DC entering the unit will be amplifier by 10 to 30 times and that can spell big trouble for connected equipment – in particular loudspeaker woofers.  What this all means is that 99.9% of all amps and preamps aren’t really flat at the bottom end – because if they were they’d amplify DC – so by default these amps and preamps are rolled off on the bottom end.

This is an area where great differences in sonics can occur.  Ever heard a product that reviewers and yourself exclaim has “thunderous bass”, while others have “ok bass”?  When you look at the specifications of these two products no doubt they are ruler flat to below 20Hz and therefore it’s a mystery why they have such good bass.  Well, let me unravel some of that mystery for you.

When we design a product like a preamp, power amp, DAC etc. we roll off the bass starting at about 2Hz and below.  The point the designer chooses to begin the rolloff has EVERYTHING to do with how “thunderous” and “solid” the bass you hear on your music.  I have helped a number of younger audio designers just getting into the field with this knowledge because conventional engineering would tell you that as long as you’re ruler flat to 20Hz, you should be fine.

There are two ways to roll off the bottom end: capacitors and servos.  Let’s focus on capacitors and then we’ll move on to servos.

I explained earlier that placing a capacitor directly in the path of the audio is something to be avoided if at all possible; on a solid state design it’s pretty easy to do that.  But if we don’t use capacitors in the signal path to roll off the bass at low frequencies, where can we place them?  In the gain setting elements – where they are still technically in the signal path – but not directly.

Using a capacitor in the right place and out of the direct signal path, allows designers to have the product’s gain variable with frequency – DC has a gain of 1 and all the music has a gain of 30 – thus we can pass DC unimpeded yet unamplified and still have great bass response.

But capacitors are the easy way out and if you REALLY want the best bass impact and slam without any hesitation, you need a servo.

If a preamplifier or phono stage uses only one power supply voltage to feed the circuit, as is typical in many classic tube designs, a blocking capacitor smack dab in the middle of the musical signal is required to interface to other kit.  The reason this is true is that 1/2 the entire power supply voltage is on the output of the preamp or phono stage.  In the case of a tube, with power supply voltages in the 200 to 300 volt range, that’s a lot of DC sitting on the output.

Transistor preamps and phono stages have a much easier go of it because we have a bit more freedom to run everything with two power supplies – which enables us to have zero volts on our outputs and inputs – thus we don’t really need blocking capacitors in the music’s path.  The lack of these components means cleaner sound.

But nothing is perfect and when I write “zero volts” it isn’t entirely accurate.  To be perfectly accurate there’s just really “low DC” on the output.  Let’s call it on the order of 1/100th of a volt – chump change to be sure – but this small voltage can cause problems and here lies the rub.

Imagine a chain of three products all tied together and DC coupled without coupling capacitors.  The first in our chain is a phono preamplifier, followed by a preamplifier and finally a power amplifier.  And for argument sake let’s say the phono preamp has a DC output of 1/10th of a volt (0.1 volts).  The preamp has a gain of 20dB (10X) and the power amp has a gain of 30dB (30X).  Now, multiply the 0.1 volts out of the phono preamp through the preamp and it is now putting out ten times more DC than you put into it – or 1 volt of DC.  All of a sudden, that’s a lot of DC.  Now put that 1 volt of DC into your power amplifier – with its 30 times gain – and guess what?  Yup, you now have 30 volts of DC on the output of your power amp, enough to fry any decent woofer.  Not good.

So the problem isn’t so much the DC the units produce but what the attached products do with that DC that can cause the problems.  So what do we designers do about it?  Well we have a few solutions, the most common of which is to stop our designs from amplifying DC and just passing it along.  How do we do that?  The simplest method is to use our old friend the capacitor – but not directly in the signal path – indirectly in the path instead.

Tomorrow, rolling it off.

We know that many are simply a holdover from when tube circuits dominated the landscape and as soon as solid state came into the forefront designers of those circuits just kept doing what they knew how to do.  After all, nearly all solid state designers in those days converted from tubes to the new fangled transistor without having much in the way of focused training on the new devices.

But in the early 70′s a new crop of designers came onto the scene and they, like Stan and I, had never designed a tube circuit and knew only about transistors and designing with them.  When we came onto the scene we started questioning “the old guard” about their practices and, like many young turks, did things our way because it worked better specifically for what we were doing.

Getting rid of extra components in the signal path was high on our list of ways we could build better sounding products.

The problems that we faced wasn’t so much from the circuits themselves – in solid state design it’s pretty easy to get rid of the coupling caps at the input and output of preamps and amps – but with other pieces of gear that might lead to trouble.  So these coupling caps were needed to guard against products that output DC.

I’ll go into more detail about this subject tomorrow.

Our quick response to these questions would be to treat the cause rather than the symptoms – we’ve seen many examples of this in our lives and intuitively the answer seems obvious – but does it apply to our rooms?  Should we be running digital correction, processors, equalizers to force our systems to work within our rooms or the opposite?  And why?

If you’re a subscriber to my YouTube series on building a music room you’ll know that my choice is clear: fix the problem, don’t try and cleanup afterwards.

The problems we find in our rooms are specific to the room as well as where in the room they are occurring.  Let’s take, for example, my room with dimensions of 22.8L x 15W x 9.5H which are pretty danged good overall.  The fact that my walls, ceiling and floor are parallel surfaces to each other means that I have a myriad of problems: bass peaks, slap echo, unnatural reverb and decay.

But where do these problems occur?  At the listening position or somewhere else?  The answer is both – and that’s really part of the problem.  I don’t want to build a single seat head-in-a-vice music room.  Rather, I want a room that can handle perhaps 6 to 8 people and sound just as sweet as if there were only one in the sweet spot.  Were I to try and use an EQ or digital correction system to fix these bass peaks, then wherever the peak was flattened out, there would be a hole – the same size as the correction – and we’d have a suckout. So all we would have accomplished is to fix it in one place and another, perhaps just as bad, problem pops up somewhere else.  Remember the game Whack-a-Mole?  That’s what you’ve created.

Another example of this is AC power.  I am a staunch advocate of fixing the power problems in the first place rather than filtering out a few issues and then adjusting the rest of the system to try and hide the sonic issues caused by poor power.  Fixing the room rather than dealing with its deficiencies is no different.

As you’ve seen in the video, we can “easily” adjust the bass peaks in the room by building a Helmholtz Resonator in the corners.  This correction of the room itself will always yield better results than taking a microphone and measuring the problems at the listening position and “fixing” them through correction.  And I am not even venturing into the hornet’s nest of what other havoc EQ and correction can wreak on the purity of the signal chain itself.

For today let me suggest it’s better to cure your ills rather than treat their symptoms.

Many of you requested links to the calculators we used to determine the room modes peaks and where those peaks are located.  The calculators we used for determining the Helmholtz equations are bit more complex and I had to engage our chief engineer for help as the math is somewhat complex – but I’ll bet there’s some good online programs one could use if you wanted.

Ask yourself the following question: is it better to fix the room or the source?

There are two schools of thought embedded in this question: do you address the cause or the effect?

An untreated room has many problems that are caused by unfavorable dimensions, parallel walls, surface treatment and so on.  The problems of a room don’t really matter much until you place a pair of loudspeakers in the room that stimulate and exacerbate the problem areas of the room.  So one school of thought suggests a digital correction filter should be employed to smooth out the room and speaker problems.  In fact, there are a growing number of digital correction filters for the system as well as subwoofers with analyzers and DSP circuits to fix the output of the system to compensate for the room.

Watching the video you know that in our room we have a 25Hz and 37Hz bass peak that’s pretty severe.  Left alone, any loudspeaker in the room would stimulate these frequencies and cause a lot of boominess.  If we add a digital correction system we could simply turn down those two frequencies so whenever they play the result would be flat.  That would constitute treating the symptoms and it would be very effective.

Or, we could treat the cause by fixing the room.  In medicine it’s always best to treat the cause rather than the symptoms but I wonder if that same reasoning can be applied to audio?

Let’s spend some time on this subject and I welcome your thoughts.

If we have a choice in our designs we’d prefer to have as little in the signal path as we can – the proverbial straight wire with gain would be ideal – but of course that’s just a myth.

It’s been pointed out that perhaps this idea of minimal obstructions in the music’s path doesn’t matter since the recording chain we listen to is littered with many components and devices in that same path.  However, I would argue that regardless of how many barriers are in the music’s path, adding more only serves to make things worse.  Truly this is a case of less is more.

Direct coupling of an amplification circuit means we have less in the path, specifically the lack of a coupling capacitor at either the input or the output of the device.  But many designs today still rely on these coupling capacitors and the music we’re so interested in preserving is forced to pass through them.  While we don’t subscribe to their use (and I’ll explain the alternatives later) I think it’s instructive to understand the different types of coupling capacitors available to audio designers – because some designs (in particular tubes) require a coupling device both on their input as well as their outputs – but their use is certainly not restricted to tubes.

There are many types of capacitors on the market but I would suggest the two most common coupling capacitors are electrolytic and film.

If you’ll remember our post on what capacitors are  we have two conductors and an insulator between them.  When you put a signal into one conductor it develops an airborne field that jumps across the insulator and transfers over to the second conductor.  The good news here is there’s no physical connection between the two conductors – imagine a wire that you cut in half and the musical signal simply jumps across the air gap you’ve created in the wire - now that’s freaking magic!

The cheapest and most common capacitor of the two is the electrolytic.  Pretty much all power supplies use these capacitors to smooth out the AC in DC but they are also used a lot as coupling capacitors.  These work by using some chemical goo (called an electrolyte) as the conductor and a thin strip of anodized aluminum for the insulator.  The goo and the aluminum are rolled up like a jelly roll and leads attached for the signal.  Electrolytics work best at lower frequencies and aren’t particularly linear when it comes to passing music – yet their use in audio circuits as coupling capacitors is ubiquitous because of their very low cost.  Their performance can be helped if you place another type of capacitor in parallel with them – or replace them with – a film capacitor.

Film capacitors use a thin film (plastic sheet) as the insulator with an equally thin sheet of metal between the film, forming the conductor.  The plastic film and the metal sheet are rolled up together to make the capacitor.  Depending on the type of film used – mylar, polypropylene, polystyrene, Teflon – very linear audio performance that extends out to high frequencies can be achieved.

You might notice that the names of the insulating plastic material are also the names of the film caps.  Their downside is they can’t be made very large – thus limiting their low frequency (bass) response – and they are relatively expensive.

So if less is more – and it is in audio – then how do we move away from these intrusive rolled sandwiches of goo, metal and plastic to couple our audio kit to the outside world?

Ahhh, let’s wait until tomorrow.

So, just how does a capacitor decide what is AC and what is DC?  Frequency.  If we remember that it is the alternating movement of our loudspeakers – in synch with the music – that creates sound in the first place, then we have to figure out where we draw the line between moving and not moving signal.  The output of a battery does not move – but the output of a microphone or phono cartridge does – so clearly we can say that a battery is DC and a phono cartridge is AC.  How do we differentiate between these two extremes and where does one draw the line?

For a capacitor to transfer music through it that musical signal must be moving back and forth at least a little.  If it is moving back and forth between + and – once per second that is enough for a capacitor to transfer this signal from its input to the output IF the capacitor is the right size.

Capacitors come in different sizes and they are rated in what we call farads – a measurement of capacitance – and for our purposes we use capacitors that are significantly less than a farad, typically a microfarad (one millionth of a farad).  The bigger the capacitor the lower the frequency of sound it will pass.

Are we happy with one cycle per second?  The reason I ask is that as capacitors get bigger they get more expensive, pass audio with less accuracy and take up a lot of real estate on a circuit board.  Here’s a case where bigger isn’t necessarily better – and if our goal is to keep the number of parts in the signal path to a minimum – and make sure the quality of the parts in that path is the best – then we need to make sure we actually want to pass 1 Hz.

You cannot hear 1Hz – but you can hear 20Hz and you can feel 16Hz and even below.  So if we’re designing a product with a blocking capacitor, we need to figure out how low we want to pass signal – and not go too low so as to cause problems.  As a designer I typically use a 10X rule of thumb.  If the lowest I want to hear is 20Hz, then I use a rolloff of 2Hz if there’s only one capacitor in the signal path.  That is, unless it’s a turntable preamp I am designing and then we have to be careful about letting unwanted table rumble through the system.

The take away here is that capacitors in the signal path need to be carefully noodled out by the designer.  The quality of the capacitor and its size are critical elements when designing a high end product.

We’ll look at types of capacitors tomorrow and we’ll also be working up to our original subject; direct coupling.

Direct coupling describes a direct connection between two pieces of audio electronics without an intermediary component such as a capacitor or transformer.  Why should that matter if there’s a capacitor in the signal path?  Because less is more in audio, certainly if you’re not giving up anything in the process.  We all like listening to shorter paths with fewer components between the music and our ears.

Most solid state audio products manufacturers direct couple many of their products today without much fanfare; pretty much a given for high end audio products.  Tube audio manufacturers are the opposite; they routinely require either a capacitor or a transformer to couple the tube’s input and output to whatever the user wishes to connect to.

The main reason a designer would want to include a capacitor is to block DC and allow only AC to enter or leave the audio device.  As you will remember music is AC, meaning it must continuously move between + and – to force the loudspeaker cone back and forth so we can hear sound in our room.  Put DC into a power amplifier and the woofer cone of your loudspeaker will jump forward and stay pushed outwards as the voice coil of that speaker heats up and then burns out; not something we’re interested in.

Capacitors will not pass DC and will only pass AC, which is a nice feature for a musical amplification device – since we certainly want nothing to do with DC going anywhere in our musical systems.

So, what is a capacitor?  Two conductors and an insulator.  One conductor is on the input of the capacitor and the other forms the output of the part.  The two conductors are separated from each other with an insulator, sometimes called a dielectric.  For those of you following these posts this arrangement might sound familiar – and that’s because it is – we call the other device a transformer.  Neither the capacitor nor the transformer will pass DC but both will pass what we want, musical AC.  So far so good.

If you remember, a transformer is two coils of wire separated by an insulator (insulators won’t conduct electricity – think wood, plastic, paper).  Just like our friend the capacitor, there’s an input side and an output side.  Place a signal on the input side of a transformer and it becomes a magnet – generating a magnetic field – that is then sensed by the output coil and converted back to its original electrical form.  The key to this device is that we removed the physical connection in our signal path – and replaced it with a magnetic connection that travels through air.  Were you just to have a wire between a source with both DC and AC present at the same time, you’d get both passed right on through.  But with a transformer or capacitor between two pieces of equipment – only AC makes it through while DC stays behind.

Capacitors are similar but instead of generating a magnetic field, they generate an electrostatic field.  Place a signal on the input conductor side of a capacitor and a small electric field is developed (think of the charge you generate and store on a hot dry day when you get static zapped touching something).  The field hangs out in space (just like it does in you when you’re all charged up) until the second conductor is attached to a different potential on a battery (think of the difference between + and – of the battery) and then the field jumps across the insulator and goes to the output conductor completing the transfer of energy,  Here’s a picture:

Again, just like the transformer, we’ve eliminated the physical connection (the wire) between input and output and our music travels merrily through the air leaving the unwanted DC behind in the dust.

The amount of energy storage and the frequency that can excite this electrostatic field is dependent on the size of the conductor and insulator’s surface area; more surface area, lower frequency and great energy transfer/storage.

Tomorrow I’ll explain why we need this at all, then we can discuss the various types of capacitors that you’ve no doubt hear of like film caps, electrolytics, oil and paper, etc.

Most tubes circuits of the day required them but solid state designs didn’t – they were mostly added as “good practice” by engineers who were making the transition from tube design to solid state.  And it was at the beginning of the second golden era of audio.

It was at about this time many designers were starting to figure out the idea that “less is more” that everything in the signal path mattered and that certainly included the capacitor.  Further, we started to discover that if you had a capacitor in the music path the quality of that capacitor mattered greatly – although there was no capacitor like no capacitor at all.

“No capacitor at all” earned itself a marketing phrase which we still use today: Direct Coupled.  I say marketing phrase because it’s not like you’re adding something extra or special to the circuit, in fact, you’re removing something that really didn’t need to be there in the first place.  Direct Coupling simply means the lack of an unnecessary component in the signal path – as marketers when we remove something we call it out as a feature – simply because ours sounds better without it.

Just what is a capacitor and why would we have used one in the first place?

Tomorrow we’ll dig in.

We don’t see this added energy in typical audio measurements such as THD and IM because these distortion measurements are made with steady state audio – meaning they do not stop and start – rather they are recorded with steady tones which do not change states.  So when we see an amplifier with a THD reading that’s 0.001%, way below anyone’s ability to hear if a few of those zeros are missing, we haven’t any clue how it responds to transients – and transient response and distortion play a HUGE role in how our systems sound.

Transients provide essential clues to our ear/brains on the nature of what’s being presented to us in our music systems; information that is absolutely critical for us to identify what’s going on and whether a system is believable or not.

The sound of live unamplified music is full of transients, each with their own set of “ring tones” attached to the original impulse.  The audio equipment’s ability to faithfully reproduce these transients, without adding more ringing, is critical to our system’s success if we’re trying to build a credible high end system in our music rooms.

Designers and reviewers alike sometimes use what’s called impulse responses to see how their systems respond.  Here’s a picture of an impulse response:

Note the big spike in the middle, flanked by multiple smaller spikes on each side of the big one?  The big one is the impulse and the smaller versions of it are the ringing that it produced.  Just to be clear, the big one is all we want to see – and all the little ones are added and inaccurate distortions.

This particular impulse shows ringing before and after the impulse – a strange feat indeed – and this is typical of many digital circuits called “pre-ringing”.  Pre-ringing pretty much never happens in analog circuits.

We can also see this ringing when we place another type of impulse through our system called a square wave.

This should just look like a perfectly flat line but instead note the ripples.

What’s all this mean?  It means that whenever a musical transient is playing through your system you are getting far more inaccurate information than you ever wanted.

These are the areas that we look at very closely when we design a product because too much ringing sounds hard and bright – a characteristic none of us are happy with.

So, the next time someone says their amp or loudspeakers are perfect because “all the measurements show no distortion” you’ll know better than to believe them and trust your ears a bit more instead.