Explain Class D amps to "non audiophile friends"


Hey folks, I was really enjoying the thread explaining class A amps to class B and AB, wheni was wondering where Class D comes in. I would love to be a snob and say its for my friends, but I have to admit I am doing the learning here! (I am not implying that the other guy who posted was asking the question for his benefit)

Thanks for your help!
Jeff
jeffatus
In general efficiency is inversly related to sound quality. Class D gives you a lot of power in a lighter, cheaper, more efficient package. Good for sub use. High quality Class D amps are not cheap so there is no free lunch.
Imagine switching two wires between +V and -V voltage at very high non-audible frequency. You will get high frequency square wave with average value of zero as long as this switch stays in each position 50% of the time. Change this percentage (ratio) and average value will change from +V to -V. Filtering this square wave (inductor and capacitor) will convert its average value to DC voltage that feeds speaker. Changing symmetry of this switch according to input voltage is what class D amp does. It is called PWM (pulse width modulation).

Efficiency of class D amp is less than 100% because ON resistance of this switch and transition time are not completely zero while output filtering inductor has also some resistance, plus some small power is used by modulator. Efficiency of my Rowland 102 class D amp is 90%.

You will find PWM modulation in other places where it has nothing to do with efficiency. For instance SACD is the same as class D before filtering (operating at 2.8MHz). Also Sigma Delta DAC operates on the same principle. In fact modulator used in modern class D amps is similar to Sigma Delta modulator. Main problem is to make switch (Mosfet transistor) that can switch extremely fast big current and has close to zero resistance. Faster switching allows to use higher switching frequency that can be filtered out better while bandwidth of amplifier can be extended further.
>>05-22-12:Stanwal
In general efficiency is inversly related to sound quality.<<

That is simply not true.

Ignore the inexperienced post(er).
You will find PWM modulation in other places where it has nothing to do with efficiency. For instance SACD is the same as class D before filtering (operating at 2.8MHz). Also Sigma Delta DAC operates on the same principle.
Kijanki, what you wrote is absolutely correct but what you'll find is that a class-D amplifier is a very minor tweak of a switched-mode power supply (SMPS) where the input (DC) voltage reference is replaced by a (modulating) music signal. Do this simple swap & you have a class-D power amplifier. Compare a class-D audio amp & a SMPS & see for yourself.....
STOP THE PRESSES: News flash...To the astonishment of the audio geek community it would appear that some Class D amps sound great, and others less so. Details to follow.
Bombaywalla - it is the other way around. Class D was invented tweaking SMPS. Engineers demoing SMPS showed that load regulation is so fast that SMPS can even play music.
Any linear power supply is in fact primitive switcher operating at 120Hz. Primitive because it requires huge transformers, big capacitors (to filter out 120Hz) has no line or load regulation, switches at max voltage, is susceptible to DC on line etc. That's why Rowland uses SMPS in preamps where efficiency is not important or power amps (model 625) that are class AB. Bad rap that SMPS has come mostly from primitive computer supplies.

Wolf_garcia, They sounded great from the beginning (at least to my ear). Second "version" was designed to make it easier for some critics to remove foot from the mouth.
I've read all the posts and I still don't understand how a class D amp works. What are the devices that do this "switching?" I get the difference between class A and class A/B, that thread was great, but for this one I think I need some kind of dumbed down explanation...more layman terms please?
Realremo, Imagine just one mechanical switch that switches between +10V and -10V. Output of this switch after averaging (filtering) will be zero volts if switch was the same amount of time (for instance 1s) at +10V and -10V positions (total cycle time of 2s). Manipulating ratio of times changes output voltage.

For instance:
1.5s at +10V and 0.5s at -10V will result in output of +5V because (1.5*10+0.5*-10)/2=+5
or
0.5s at +10V and 1.5s at -10V will result in output of -5V because (0.5*10+1.5*-10)/2=-5.

Real frequency of switching is very high (in order of 500kHz) where mechanical switches would not work - hence Mosfet transistors (fast switches). Mosfet transistors are even manufactured in two different technologies for use as switches or linear transistors.
Fascinating technology!

The main thing I wonder about is how much better even will it get over time for hifi audio purposes.

There would still seem to be some practical upside in maturation of this particular technology though I would say it has clearly evolved to the point of being highly competitive with others already in hifi audio. I expect it could widen a lead practically over time moving fore ward if it has not to some extent perhaps already. The bigger question will be how much will it matter? Similar to the discussions about value of high res digital audio sources I would think.
Mapman, I suspect that it will get much better because of semiconductor speed. Things based on timing scheme like Sigma Delta DACs are limited in resolution by timing accuracy while traditional DACs are limited by tolerance of components. Semiconductors are getting much faster every year including high current Mosfets.
"Semiconductors are getting much faster every year including high current Mosfets."

Exactly!

Much technological progress over the last half century has evolved around the speed of semiconductors as evident in the myriad of increasingly functional electronic devices that never stop coming and likely will not in the foreseeable future either I would bet.

A quick look around today versus 5, 1, 20, 50 years ago quickly confirms that!

I look forward to the day when I might consider replacing my current Class D amps with something that is even better perhaps even for a fraction of the current cost.

Might high end audio be the one frontier so advanced that advances in semiconductor technology might fail to conquer it along with the rest?

I think not.

Time will tell even if it has not already started to do so for some.
is that a class-D amplifier is a very minor tweak of a switched-mode power supply (SMPS)

05-22-12: Kijanki
Bombaywalla - it is the other way around. Class D was invented tweaking SMPS.

Kijanki, I think that we are saying the same thing.... :-)
Jeff, this Wikipedia writeup should help to clarify what has been said. The entire article is well worth reading, but note particularly the block diagram, and the paragraph headed "Signal Modulation" that describes it.

As noted in the article there are ways of implementing a Class D design other than the one depicted in the block diagram, but it illustrates the basic concept.

Regards,
-- Al
Thanks for the info, guys. Us non-engineer types need to read and re-read the technical info a few times before we start to pick it up. Question: it appears to me that class A, B or A/B share some similarities, yet class D seems worlds apart from the rest. Is this correct? Could you compare the differences between class D and class A or B amplification?

Thanks!
Jeff
05-22-12: Jeffatus
Question: it appears to me that class A, B or A/B share some similarities, yet class D seems worlds apart from the rest. Is this correct? Could you compare the differences between class D and class A or B amplification?
Yes, it is a fundamentally different approach.

Put simply, at any instant of time each point in the signal path between the input and the output of the traditional amplifier classes has a voltage that is (to a close approximation) proportional to the voltage that is present at the input of the amplifier at (approximately) that instant of time.

A Class D amplifier does not work that way. At some of the points in its signal path, including the high power stage, what is proportional to the voltage of the input signal is the width or timing of a series of pulses. A "pulse" in this context refers to a voltage that (to a close approximation) switches between two (and only two) possible values.

An analog output signal whose instantaneous voltage is proportional to the instantaneous voltage of the input signal can be (and is) recovered from that width or timing by simply filtering out frequency components of the power stage's output signal that are far above the audible frequency range, and that are unrelated to the audio information itself. As indicated in the Wikipedia writeup I linked to, those high frequency components "serve no purpose other than to make the wave-form binary so it can be amplified by switching the power devices."

Regards,
-- Al
OK, light analogy. In order to achieve certain brightness of 100W light we can either adjust brightness in linear fashion by increasing bulb supply voltage or we can apply full voltage and blink it very fast. Continuous brightness that our eyes see depends on amount of time the light is on to amount of time it is off (average brightness). If we pulse it symmetrically (50/50) ON and OFF our eyes will perceive half of the brightness. That's how light dimmers work.

First method of achieving desired brightness corresponds to traditional class A or AB amps while the second method is class D. Dimmer is class D regulator. Thermal inertia of the filament is equivalent of output filtering.
Ok my brain just exploded all over my iPad.
"Ok my brain just exploded all over my iPad."

Lol! I can't stop laughing....
Ignore the inexperienced post(er). True; pay no attention to Audiopile who has had far less experience in audio than I have. His knowledge is inversely proportional to his ego.
"Ok my brain just exploded all over my iPad."

Realremo, we are seriously trying to explain but I can see that making jokes is more important to you and Jeffatus (who finds it amusing). I won't make this mistake again.
Over here http://www.hypex.nl/ and at their shop site https://www.hypexshop.com/, this Class D designer has been creating quite a stir.

I considered buying the modules but funds are a bit short now and I have "0" confidence in myself to build one but what reviews are out there have been overwhelmingly positive.

Their stock sells out as soon as it's posted so it's best to pre-order. They only sell the components to the DIY crowd and not OEMs. What the OEM crowd gets are made to order along with the usual outrageous mark up afterwards.

I heard a Genesis amp that used an earlier version of the Hypex unit and it floored me. The newer Ncore is said to raise the bar considerably. If you can build it yourself, it's conceivable to build two mono amps for just under $2K.
(the Genesis went for over $6K)

All the best,
Nonoise
Kijanki, in my haste to blow off some steam after a long day filled with last minute deadlines and draconian utility company notices, I neglected to remark about how your and Almarg's last posts really helped explain all of this for me. I won't make this mistake again.
Realremo, No harm done. I'm just overly sensitive today. Don't hesitate to ask if you want to go more into details. I'm not the best to explain but Al does it very well and he is forum Guru (in case you don't know it yet).
"Realremo, we are seriously trying to explain but I can see that making jokes is more important to you and Jeffatus (who finds it amusing). I won't make this mistake again."

Hold on a second, I find the posts extremely helpful, which is why I always make sure to thank everyone for their posts. You have to understand though, what may be easy for you guys to understand would take a layman a while to pick up. It seemed obvious to me that Realmro was poking fun at himself, which is exactly how I felt (about myself, that is). I mean, here I was trying hard to take all this in and the exploding brain post came up...I seriously couldnt stop laughing.

I wouldn't say I feel stupind asking these questions, but I certainly don't feel like the smartest guy on the web. Either way, you guys are way ahead of me and your help is greatly appreciated.
If I may venture a comment or two; This technology is prevalent in the automotive world. It is used to control the output of the fuel injectors and fuel pumps. I'm guessing that it also is used in the audio amps in many automobiles for the weight and energy savings it provides. I'm sure better quality computer speaker amps use Class D as well and they don't sound bad for what they are. All that being said, and I'm sure that I am not alone in saying this, I'm just not ready to accept Class D amplification as my final output stage in my HiFi. Maybe some of it is stubborness on my part, although I enjoy my CDs but I still like vinyl better. What I have seen in Class D designs is circuit board amps with lots of electrolytic capacitors. This goes against my conventional hifi wisdom of discrete components, isolation from power supplies and roll film capacitors/wire wound resistors in the music signal path. Am I wrong about that?
Tony,

If you are not accepting due to a pre-conception about what a good amplifier looks like, or even how it works, then I might venture to say you are on the wrong path.

IF you have heard a Class D amp and are not convinced, then I would say your conclusions at this time are justifiable based on your personal needs and preferences. If you hear several over time and still are not a fan, then you have done due diligence in making a valid determination.
My Wyred 4 Sound ST-1000 is excellent
It drives the B&W Nautilus 802S speakers to excellent sound and control of the speakers

W4S makes modifications to the stock B&O ICE modules, they also have their own input circuit