Conversions

Yesterday we traced the digital audio path from the input of the DAC all the way to the output of the DAC itself, which outputs a series of current steps, and are in the process of converting those current steps to voltage steps.

You can use a simple resistor to make the current to voltage conversion and several manufacturers do, but it’s not an ideal load for the DAC chip. What the DAC chip wants to see on its output is zero impedance: no resistor or resistances at all. But in order to convert current to voltage we’re going to need something that either is a resistor or simulates one and that simulation is typically handled with an op amp.

As we learned, an op amp has two inputs: a plus and a minus. The plus input is high impedance and the minus input is essentially zero impedance. So we use the minus input to connect the output of our DAC up and voila, we get voltage on the op amp’s output. But a few of us nutso high end designers would rather not place an op amp right at this junction because they simple don’t sound great.

The first clue that an op amp right at this position is a sonic no,no happens when you start to swap op amps out to see if the sound changes. Oh my, it changes mightily depending on the op amp’s characteristics. The faster op amps sound better than the slower ones, generally speaking, but all sound a bit bright and transistory in my opinion. Speculation as to why this should have such an impact on sonics usually revolves around the rapid speed at which the DAC is changing it’s current output and how those rapid step responses affect op amp performance.

The whole issue is, however, easy to skirt if the designer wishes to – but unfortunately most don’t. Of the few that do, Nelson Pass and I prefer to use a pair of transistors instead of an op amp – others may have some different scheme I am unaware of. Nelson prefers MOSFETS and I prefer bipolars here, but that’s a matter of choice that is really dependent on everything else in the chain.

The cool thing about using a simple pair of transistors is that if you connect them properly they are still extremely low impedance but don’t have any feedback related hardness issues to contend with as do op amps. Further, we can ensure that they produce low noise, high speed, low distortion outputs that are perfect for feeding into the next stage of our chain: the analog filter.

Tomorrow we get rid of what we don’t want.

Paul McGowan – PS Audio, Intl.

Amperage alchemy

We’ve briefly discussed getting the digital audio stream into the DAC and then decoding those bits back into a form of analog that gets us a lot closer to being able to play it on our systems.

At the output of the DAC chip we have a series of current steps that need to be converted to voltage steps – a sort of alchemy process like turning lead into gold. These current steps can be thought of like different wattage light bulbs if we want to understand them more easily.

Imagine for a moment that we have 24 light bulbs (like 24 bits), each bulb capable of consuming twice the power of the preceding one and in the process getting twice as bright. We start with a 1 watt light bulb, next we have a 2 watt, then a 4 watt and so on. Each of these bulbs consumes more power – or current – than the other and the combination of brightness gets us a large range of light: from the very dimmest to the very brightest and everywhere in between.

This is similar to what is happening at the output of your DAC chip. Now take your imagination one more step forward: imagine instead of 24 light bulbs you have only a single light bulb and you feed it different wattages. The total wattage you feed this single bulb creates a unique brightness level that spans a 144dB range which, in audio terms, covers the range from a single air molecule hitting your ear to the noise of a jet blast and everything in between.

Preamplifiers and power amplifiers need voltage on their inputs, not current (power). So we must convert these power steps to voltage steps at the output of our DAC chip. The easiest way to convert current to voltage is to simply pass it through a resistor. Place one end of a resistor at the output of our DAC and the other end goes to ground. Voila! You have a simple current to voltage converter.

One of our very first DACS used just this approach and, if I remember from reading John Atkinson’s review of the Devialet integrated, that is also what they do. While this works and is simple, it isn’t always the best way to do this because not every DAC appreciates working into a resistor and you start losing linearity and gaining noise.

Most DACS simply place the output of the DAC into a chip op amp’s inverting input. Now, I understand this is technical sounding stuff, but it’s actually fairly simple. An op amp’s inverting input looks like ground to the DAC and it is very happy – somewhat like a zero Ohm resistor. This is the point where a lot of audible trouble can begin to take place.

So important is this conversion process that tomorrow we’ll look a bit deeper into it’s workings and problems.

Paul McGowan – PS Audio Intl.

DAC attack

Yesterday we detailed the layout of a PCM DAC and its 4 components that make it work: the input receiver, DAC itself, Current to Voltage converter and finally the analog stage: the subject of this series.

We want to try and understand why the output analog stage plays such an important role in the way a DAC sounds – which may run a bit counterintuitive because 1/2 the DAC components are digital – the other half analog, yet they are definitely not equally weighted when it comes to how good a DAC sounds.

We also reviewed the 1st of the 4 components in the DAC, the input receiver, and learned that one way or the other, we need to make sure our digital audio data has been converted to I2S – the native format for digital audio within the DAC. I2S has 3 separate clocks and a data line. Now we have the I2S data in the form we want and we’re ready to let the 2nd component in the DAC chain work its magic.

We have covered the DAC itself in several series, starting as far back as November 9th. You can go back in the series here if you wish and catch up to us or just hang in there till we get to the meat of this series.

We know that most modern DACS are 24 bit and to achieve this level of resolution they are actually not traditional ladder DACS, but rather single bit DACS (sometimes 2 or 3 bits) similar to how DSD is handled.

Here’s what’s important to grasp in this phase of our understanding: once the data has been organized the way we want it is sent through the DAC and processed through upsamplers, digital filters, sigma delta converters and all sorts of crazy modules few of us will ever understand or need to. What comes out of this mess is a conversion of what we started with – numbers representing voltage levels – into 24 current steps. Each step is twice the current as the next and between them we have managed to almost get back to where we started when we encoded the music in the first place.

Tomorrow we will magically change these current steps into voltage steps – exactly where we started – and then get to work on turning it into music once again.

Paul McGowan – PS Audio, Intl.

Inner Workings

We’re starting down the path of trying to unravel a bit of the topic why the analog output stage of a DAC is so important: perhaps more important than the DAC itself. To begin let’s first take a DAC apart into its 4 main categories so we can start with a clear understanding of what’s inside these devices.

The purpose of a Digital to Analog Converter is just what the name implies: converting digital audio to analog audio. We’ve learned in past posts that one type of digital audio, known as DSD, can be used directly to feed an analog power amplifier without decoding (although no one does this because it’s quite noisy without further processing). When we refer to digital audio, in this series, we will be speaking about PCM which cannot be listened to directly as the code is based on numeric representations of the audio.

DACS have 4 main components:

Input receiver
The DAC itself (including an upsampler, digital filter, ladder or DSD engine)
Current to voltage converter
Output analog stage

Today we’ll cover the input receiver in a bit of detail.

The goal of the input receiver is to convert any incoming data to I2S, the native format for all PCM DACS. I2S has 4 components, 3 of which are just clocks, the 4th is the actual musical data:

Bit clock
Word clock
Master clock
Data

PS Audio DACS, as well as a growing number of other manufacturer’s DACS, have adopted the PS open standard for I2S over HDMI. This method eliminates the need for the input receiver, so we’ll ignore this for right now. The vast majority of all modern DACS get their data through a single stream called S/PDIF (Sony Philips Digital Interface), USB or Ethernet/WIFI . In any of these cases we need to break apart the data and wind up with our 3 clocks and data.

Since the scope of this series is limited, let’s just focus on what most people have coming into their DACS: S/PDIF from a CD player or a computer sound card.

When a CD player or computer sound card gets the information off of a disc (optical or hard) that information is sent using the same I2S components that we need to get into the DAC. However, Sony and Philips decided (in their non-Audiophile friendly) collective wisdom that it would be inconvenient to have a 4-conductor cable to send I2S to an external DAC. Instead, thought they, it would be simpler if they just smooshed all the clocks and data together so it could be sent over a single cable and then decoded at the other end.

Certainly there are positives and negatives with this approach, but it is what we have to live with. So the job of the input receiver on a DAC is to disassemble the smooshed clocks and data as best it can and send that data to the DAC. In the process of recovering the clocks and the data we have our first signs of trouble – which years ago we solved with an accessory product called the Digital Lens.

Today, however, many DACS have well designed receivers that, while not perfect, get the job done with good results.

USB, popular today, is a little friendlier if the DAC has an asynchronous input receiver because the master clock is not included in the data and is generated by the receiver itself.

So now we have our data in the form we wish, 3 clocks and 1 data stream. Tomorrow we’ll convert the numbers into audio with the 2nd part of the DAC, the DAC itself.

Paul McGowan – PS Audio, Intl.

Analog matters

I thought we might delve into a new subject this week: the all important analog stage at the output of a DAC. I think this may make for some interesting posts as a disproportionate amount of attention is paid to the digital side of the DAC relative to the analog side.

I smile every time I read about the core of a DAC being a Sabre or a Woflson or a whatever and almost nothing mentioned about the analog output of the DAC itself. To me that’s out of balance and here’s why: a modest DAC feeding an awesome analog stage will always outperform an awesome DAC feeding a modest analog stage.

Tomorrow we’ll get cracking on this notion and cover a number of topics that relate to it. In the meantime, here’s some food for thought.

Why is it none of us would be surprised to learn that a harsh sounding preamplifier would be the weak link in any system, even if fed from the most expensive and best performing source? Is it not obvious that a great source would sound harsh running though such a preamp?

Paul McGowan – PS Audio, Intl.

Analog Coding

One of the reasons Class D amplifiers are considered by most to be digital (technically they are not) is the need to encode the continuous analog input signal into another language to work. So it may be correct, in many minds, to classify this amplifier type as something other than analog because of the encode/decode process. However, I would disagree.

Class D’s language, Pulse Width Modulation (PWM), is very far removed from true digital audio (called PCM) and is more closely related to DSD or 1 bit audio. In a PCM language the analog voltage is converted to a number which can then be stored in a computer. Without a magic decoder ring that understands this numeric language, true digital audio cannot be understood or reproduced. When we think of DSD, we get closer to analog because the stream of bits can be converted back to analog without a magic decoder ring.

In other words, if you take PCM and put it into your preamp you get nothing out but nasty noise. DSD and PWM (very closely related) fed directly into a preamp produce music! In fact, when DSD is recorded onto a disc a type of PWM is used to record that DSD data and, if you take either out of the disc or hard drive they are stored on, you can play them directly onto a stereo.

So we don’t typically refer to something as digital unless it’s based on a binary (2) system of on and off. With only on and off you must use a numeric system that represents values that must be decoded. I have always thought it a mistake that DSD has the word digital in its naming – because you can make an argument either way – and the fact both DSD and PWM can be streamed as analog without conversion is the key for me not referring to it as digital.

If you remember from yesterday’s post we learned that PWM (and DSD) actually have 3 states, as opposed to 2. It is the addition of this third state (time) that clearly differentiates both PWM and DSD from digital audio.

Why is today’s post focused so heavily on our understanding of both DSD and PWM as coded analog instead of digital? Because getting a clear picture of the difference between the two will help us understand just how this works.

Lastly, I wanted to point out the similarity between DSD and PWM because these are the two most analog sounding technologies I know of.

Just in case the world doesn’t end

Sorry, I am writing this yesterday and when I went to input the publish date I saw it would be the 21st, winter solstice, my son Sean’s birthday and “the end of the world”. So I am guessing that’s a bunch of bunk and I better have a post ready.

We’re now ready to start understanding class D amplifier technology which can be a bit bewildering to many but, actually, it’s rather simple. I’ll make sure by the end of a few day’s of posts you get a very clear idea of how it works, what issues it has and the plus and minus attributes it brings to the table.

Class D exists as a means of optimizing efficiency. We know that a traditional power amplifier makes heat, class AB a decent amount and class A an obscene amount. We know this because of the large heat sinks on the sides of most power amplifiers: there to dissipate whatever heat is generated. We can quantify heat in terms of percentages: a typical AB class amp is 50% efficient and a typical class D 90% – meaning 50% of the power consumed by the amplifier is wasted, producing heat, in an AB amplifier. In a class D amplifier only 5 to 10% is wasted on heat production.

The reason A and AB amplifiers waste so much energy can be found in their method of producing music. Take a look at this picture of a sine wave:

Sine wave1 300×225 Just in case the world doesnt end…

See the line going right through the middle? This line represents what we call the zero crossing point and it is where no power is being delivered to the speaker nor converted to heat. Now follow the sine wave up to the highest point it goes (as well as the lowest): called the peak (like mountain peak). This is the point where the maximum amount of power is sent to the loudspeaker – but not the point of the most heat. This is important to understand so hang in there with me.

The zero crossing point and the peak are the two points on the sine wave where the least amount of heat is generated, relative to our efficiency figure I quoted you earlier. In the case of the zero point it’s rather obvious because essentially the amp is turned off. However, the peak is where we get the greatest efficiency – meaning we deliver 95% of the power to the speaker and only 5% goes to producing wasted heat.

Everywhere in between the zero point and the peak generates more heat than it does power to the speaker. Therefore, everything in between the peak and the zero is inefficient. What would be perfect, if we wanted to create a power amplifier, would be to have only the two most efficient states in use at any one time.

Class D amplifiers have three states:

Off (zero)
On (peak)
Time (width)

Let all this seep into your being today and if we’re all here Saturday we’ll get to the bottom of it. If not it doesn’t matter anyway. 🙂

Paul McGowan – PS Audio, Intl.

Amp classes

In the past few days we’ve covered a lot of ground learning about the various amplifier classes like A, AB, B which are the main amplification classes, referring mostly to how the output stages of the amplifiers use power: efficiently or inefficiently. We also covered Complimentary, Push Pull and Single Ended output stages.

There are others, such as Class H and T: H is a Class AB output stage with a variable power supply that goes up or down in stages depending on the size of the input signal. Class T is a marketing term used by TriPath to differentiate their Class D amps which have some unique feedback characteristics. I am sure there are more, but probably none very interesting.

Perhaps the most interesting of them all is Class D, sometimes known as a Digital Amplifier. As I have mentioned before, Class D is not digital it is, in fact, analog but in a very different form.

Traditional power amplifiers take a small input signal, make it bigger and then provide enough horsepower to drive your loudspeaker. To do this they are designed in two successive stages: the input voltage stage and the output current stage. The first stage of a power amplifier takes a small input signal and amplifies it typically 30 times: put one volt of music in and you get 30 volts of music out. Simple.

The second stage of a power amplifier takes the large voltage output of the first stage and adds current capabilities so that large voltage has enough power to drive a loudspeaker. Put 30 volts into this stage and get 30 volts out – the difference is this second stage can also deliver power – measured in watts, to the loudspeaker.

In a Class D amplifier we also have two stages, but the first stage is very different than a traditional power amplifier as is the second stage. In a very weird scenario the input stage compares the music to a reference signal, unrelated to the input, then feeds that information to the output stage which then connects all the watts in the amp’s power supply up to the speaker – for varying amounts of time. This process is called PWM or Pulse Width Modulation.

Tomorrow we will make this strange and wonderful process make sense for you.

Paul Mcgowan – PS Audio, Intl.

Different strokes

Yesterday we explained how a single ended power amplifier worked and what the differences are between our traditional power amplifier output stage and this unique topology.

Today I want to wrap up the background series on traditional amp topologies so we can move on to learn about Class D – a topology that is still analog – but way different than anything we’ve discussed so far.

Several readers have asked me to explain the term “push, pull” and since we covered the opposite of push pull yesterday, I’ll explain what that means and then tie them all together with the different classes of amps.

If you’ve been following along in the series you’ll remember I have referred to the output stage of both Class A, B and A/B as having two transistors called a complimentary pair. One transistor handles the upper part of the output and the other other the lower. This arrangement is also known as push pull and because I was trying to keep things simple and easy to understand, I didn’t go into details.

It’s probably somewhat misleading suggesting that each of the two transistors in our complimentary output handles half the waveform because, what’s really happening, is one is pushing current while the other is pulling current. Imagine for a moment you’re operating a hand pump for water. The handle on the pump moves up and down and when you push the lever up, it pulls water out of the well. When you apply force in the downward direction, you force the water you sucked up out of the well into a pipe for delivery to a storage tank. This is a good metaphor of push pull: applying equal energy to pull something up and push some out.

Now imagine that same well, you pull the lever up and load the water, but then you simply release the handle and let gravity slowly push it down to get the water out. This is what a single ended amplifier does and, as you can see, it is asymmetrical: meaning it pulls up but it doesn’t push down.

Here’s a picture I grabbed of a push pull output stage:

Push pull Different strokes

In this drawing the circle on the far left labeled VS3 is the incoming audio signal, the triangle labeled X0P1 is the voltage amplifier and the two transistors, Q21 and Q22 are the push pull output stage. Q21 pushes up and Q22 pulls down.

If this were used in a class B mode, both transistors would be off when there’s no signal. In class AB mode, both would be always on, even when there’s no signal, perhaps 10% of their total capability. In class A mode both would be on 100% of the time, either creating heat or sending power to the loudspeaker.

Here is a picture of a single ended output stage:

emitter follower Different strokes

Sorry it isn’t more complete, I had to grab and go this morning. Just mentally add the signal source and voltage amplifier of the first picture to the left of this drawing to complete the picture. Notice there is but one device and, where it says Vout is where you would connect your speaker.

Notice it is the same as Q21 in our first picture? It’s missing Q22. What that means is it can push but it cannot pull – so any music falling below the center line is asymmetrically feeding the lower half of the signal to the speaker.

To compensate for this designers run a lot of current through this device constantly and, like Nelson Pass’s Aleph amplifier of yesterday, that can mean either the ability to fry an egg on the amp or very few watts available to send to the speaker to keep the heat in check.

One last thing: many of you have asked what a SET amplifier is. SET stands for Single Ended Triode and it is exactly what we’ve been talking about. Replace the single transistor with a type of vacuum tube called a triode and you have it.

Paul McGowan – PS Audio, Intl.

Single ended frying pan

We’re in the middle of our series on amplifier classes (not learning classes, but output classes) and getting on to understanding one of the very different classes of analog designs, Class D.

Along the way we’ve covered Class A, B and A/B. Today I thought we’d take a side street over to one of my favorite designs, Nelson Pass’ Aleph series. The first I knew of this was from the cover of a Hi Fi magazine showing the amp frying an egg.

Hot Aleph3 2 Single ended frying pan

I am sure Nelson wasn’t thrilled with this picture, but I also imagine he smiled as he’s got a really good sense of humor.

The point of this is heat. Here’s a Class A design that does not use two transistors in its output as I explained yesterday, solving one of the big problems we discussed.

Instead, this design uses a single transistor arrangement (there are actually multiple transistors but they all work together doing the same thing) to cover the entire signal. This type of design is even less efficient than what we described as Class A, yet has the advantage of not dividing the amplification duties between two devices (upper and lower halves of the signal).

If you’ll recall, a typical output stage of a power amplifier handles the signal in halves, the top half has one type of device and the bottom half another. These devices, in a solid state design, are very different transistors known as NPN for the plus half and PNP for the bottom half. If we were using a different type of solid state device called a MOSFET, the terminology would be slightly different: N Channel for the top half and P Channel for the bottom half.

The point is the signal is handled by different styles of devices and each style has different characteristics. When a designer puts together a pair of these device (known as a complimentary pair) he uses pre-matched types that are close to each other but not identical.

What Nelson did was use a single N Channel device to cover the entire range from plus to minus, thus never suffering from this handoff between the two device types. Many people love the sound of this design (as do I) but are a little hesitant as it produces a lot of heat for not many watts (as you can see).

You may have heard of this type of amplifier, it’s called Single Ended.
email Single ended frying pan.

Paul McGowan – PS Audio, Intl