Just ordered a Trends 10.1 Class T Digital Amp - A whopping $130

[DrWho]

But they do define how my output transistors are now "switching On and Off". Eh? Is this a digital amp because the transistors are only on or off?

You are trying to use semantics to argue a point, which obviously shows you have no clue what is going on. You can pass a square wave through an analog circuit, which is why square waves don't define the difference between analog and digital.

Having "switches" is a necessary but insufficient criteria to declare a circuit digital

Really? Why? Perhaps you should start teaching and educate the rest of the world? You obviously don't understand the significance. Go take a course in digital amplification and then come back with educated responses. You obviously aren't interested in learning, despite your comments to the contrary. You would be better off arguing that the world is flat.

Just outta curiosity - do you consider aliasing an analog or digital distortion? And do you accept that an improperly sampled PWM signal will exhibit aliasing?

[DrWho]

The end conclusion of your last argument is that there is no such thing

as digital - because all digital circuits are implemented with analog

circuits.

A "digital amplifier" (not in the sense of an audio digital amp, just a

digital amp in general) simply takes the positive voltage that

represents a 1 and then increases it to a higher voltage. So if you've

got a digital signal where 3.3V = 1 and 0V = 0, then a digital

amplifier with a gain of 10:1 will take that digital signal and make

33V = 1 and 0V = 0. A digital amplifier (again, classical digital amp)

does not have a middle state. In other words, if you tried to feed in a

signal of 1.5V, the amplifier will output either 33V or 0V - NOT 15V.

The amplifier is digital because there is no middle state. Whether or

not 1.5V gets translated to a 1 or 0 (33V or 0V) is a choice the

circuit designer gets to make (which usually revolves around noise

immunity).

So though you could probably operate your class AB amp in such a way

that it achieves the same function as a digital amplifier, I would not

describe it as a digital circuit because intermittent states are

allowed. In other words, if your 3.3V input signal wasn't exactly 3.3V,

you're going to get distortion on the output (distortion defined as

anything other than 33V on the output). For example, a 1.5V input will

correlate to a 15V on the output, not 33V or 0V. This is not a digital

circuit because it is not a switch.

Granted, all digital circuits are implemented with analog circuits so

of course you could cheat the system and find an input that doesn't

output a 1 or 0 (33V or 0V), but that would be considered a distortion

of the system. In other words, perfect digital is impossible, but

that's ok because it is self correcting (slight variations from 33V are

still considered to be a 1).

[Daddy Dee]

does anyone know the wattage of the LED in the Trends amp? I've got a buddy who will do that for me. Just need to get the appropriate part.

[Z4!]

I'll probably order this next week: Tripath 2020 with PSU in separate chassis.

Dean,

Which model/manufacturer is that one?

Thanks

[mas]

The conclusion was: a device is properly "digital" when the function of the circuit depends upon converting analog information into digital information and manipulating that information to get a result, or an outcome. A digital computer is digital not because it has switches going on and off, but because those switches can add, multiply, divide, subtract, move and store DATA whose meaning is found by reading the bits properly.

So, as I showed, the first thermometer has a switch in it, but it's not fundamental to the operation of measuring the temperature, so it's not really digital. The second thermometer is converting the analog information to DIGITS as a fundamental principle of it's operation. So, it is obviously digital.

So an instrument whose input is digital, processes said information and outputs a digital result is not really digital as it must "convert analog information into the digital domain". And of course, only if it "has a switch", or "store(s) DATA whose meaning is found by reading the bits properly."

Listening to your convoluted reinterpretation of the universe is a fascinating exercise. Sort of like trying to follow the hare running about Wonderland.

Just out of curiosity, do you do this just for our entertainment, or do you really believe this nonsense?

[*-)]

[mas]

Your routine is old, boring, and childish.

Do I need you to confirm it? Heck no! Especially as you had already disqualified such a device!

Your opinion is akin to that of Lipshitz and Vanderkooy who once again stood to denounce the validity of Ken Wahrenbrock's folly that became known as the PZM microphone. To which another prominent attendee remarked, "That assures it, its a success!"

And who to know better than he who is the arbiter of things "digital".

But you are wrong Mark, it was simply intended to give credit to a truly inane comment that a few might have missed. Thanks for the entertainment.

[sfogg]

"The conclusion was: a device is properly "digital" when the function
of the circuit depends upon converting analog information into digital
information and manipulating that information to get a result, or an
outcome."

Which is what the PWM amplifier does. It converts analog information into 0 and 1, uses that information to manipulate/switch the output transistors to get the result/outcome of amplifying the signal put into the circuit.

Shawn

[Islander]

I'm no expert, but I was under the impression that Class D amps are analog, although they're described as "switching amplifiers" and at least one manufacturer prints "Digital" on the front of the housing.

So which class of amps are actually digital? The ones with DACs and/or ADCs in them?

[sfogg]

Mark,

"Heh, heh. They look like 1's and 0's until you examine them and discover they aren't that all, but rather pulses of different widths"

Again, it is pulse width modulated.... to vary the width you simply put more 1s in a row.

"If just 2 states were needed, all the ones would be a specific width, or duty cycle, and all the zeros would be also. "

That is exactly what there is. The transistors are switched at the modulation frequency. They do not switch at any other frequency then that. Each and every potential switch is exactly the same width (perfect amplifier) as the switch before and after it.

Again.. Tripath is a little different as they vary the modulation frequency.

"Now, in your view, the comparison to triangle involves only the question: is it greater or less - yes or no."

Not just in my view Mark. Read your own information you presented. Examine the circuit I linked to earlier....etc...etc. The difference in the different modulation schemes can change it from simply greater/less than to things like greater then a range/accumulator or less then a range/accumulator. But in the end it is simply a binary value.

"But, clearly the difference is being translated from Y axis amplitude into X axis time - the time the transistor will stay on. "

You are missing the forest for the trees. The further away the sine wave is from the triangle wave the more '1' answers you get in the modulator. Therefor the more ON pulses you get... therefor the wider the pulse width becomes because each on is for a fixed width. When the signal is close to the triangle wave the answer is '1' for less time before it comes a '0', therefor the less ON pulses you get and the narrow the pulse width becomes.

"This is "modulation" not digitization. If this were a PCM process, then yes, the value at each clock pulse would be quantized into a set of bits, that's your "PCM ADC" at work in a real digital process."

Ummm... PCM is modulation too... after all it is pulse code modulation.

"Sufferingly painful though, is the visual that makes the pulses look like they are bits. "

Only 'cause they are.....

Shawn

[pauln]

Just for those keeping score at home, one more thing to consider:
the Trends modulator frequency is not constant, it using spread
spectrum from 100KHz up to I MHz.

The use of the word "digital"
just means it has a chip in it that incorporates some bi-stable
multivibrators (flipflops), logic gates, and other geeky stuff. It
means no more to the amp topology discussion than the presence of metal
in the chassis of an amp - you might call it a metal amp.

[sfogg]

"You are missing the forest for the trees. The further away the sine
wave is from the triangle wave the more '1' answers you get in the
modulator. Therefor the more ON pulses you get... therefor the wider
the pulse width becomes because each on is for a fixed width. When the
signal is close to the triangle wave the answer is '1' for less time
before it comes a '0', therefor the less ON pulses you get and the
narrow the pulse width becomes."

Here is this graphed out to show it. (Didn't have colored pencils, used my 3 year olds crayons

Instead of a triangle wave I did it with a sawtooth wave but the idea is the same. Each column on the graph paper is one modulator cycle. If the sine wave is greater then the sawtooth wave the bit is 1, otherwise it is 0.

The bit values are below the sawtooth/sine waves.

Then square wave on the bottom is the bit values charted out. 1 is high, 0 is low. Note how when the sine wave is further away from the sawtooth wave the pulse width ends up being wider....

Shawn

Shawn

[Erik Mandaville]

Shawn:

You should keep that drawing and frame it! It's interesting, non-intentional artwork.

I haven't forgetton about the Lexicon modifcations in the least, and am going to experiment at length with that this summer. I worked things out a bit by lowering gain in other channels to help balance that reduced sub output from the Denon, and generally liked what I heard. I could tell something still wasn't right, though. So, who knows, I might invest in the one lower model Denon for SACD use.

Erik

PS: Sorry, I know this is way off topic here.

[sfogg]

Now the next point... Mark keeps saying they aren't bits because they don't hold any information.

Anyone... go get some graph paper. Use the same modulator frequency (one graph column per bit) and chart out

00011001111011111001111000011

and you will draw the same square wave as I have. How can that be if they don't hold information?

Shawn

PS... For the wise guy that uses more/less vertical divisions then my chart for ON/off... you still drew the same square wave... it is just on an amp with more or less total output power.

Shawn

[sfogg]

"here is no clock to determine "how many bits" - there is a continuous comparison yielding a modulated signal. No bits. "

Yet all the digital amps run at a clock/modulator frequency. The Tripath is a little unique in that it varies that frequency to try to spread its EMI radiation over a wider bandwidth for less power/frequency.

If your chart was correct the amps would have infinite bandwidth. They don't, not remotely. The noise floor of all the amps increases dramatically in the suprasonic range...until it is nothing but noise. That is quantization noise.. something your chart claims doesn't exist. If you doubt this read the Tripath papers explaining how to measure the amps by using low pass filters to try and reduce the quantization noises effects on SNR measurements....

Shawn

[sfogg]

Mark,

"I believe the so-called clock modulator frequency is simply the frequency of the triangle carrier."

I do not think so, I think it is the frequency at which the modulator makes its comparisons and the frequency at which the transistors switch.

For example some of the TI digital amp chips allow the user to adjust the modulator frequency based on need. They even have a chip that will automatically adjust its modulation/switching frequency based on the AM channel you have tuned to avoid interference from the switching noise. Check out TI tech note slea040 for that one.

Crown talks about the speed with which their amps 'make decisions' for the BCA amplifiers at:

http://crownaudio.com/amp_htm/amp_info/bca.htm

and how their BCA effectively doubles the switching rate of the transistors.

B&O talks about how their ICEPower system allows for a lower switching frequency then more conventional class D amps.

Hypex talks about their switching frequency being 400khz.

Spectron talks about their transistors being switched at a 500kHz square wave carrier.

Etc...etc...

"Now, we might be entering into a zone where suddenly we are referring to very specific designs and very specific products."

Maybe, but I'm pretty sure all the actual products out there have very specific switching frequencies... as such they have bits.

And again... the Tripath is a little different in that it alters its switching frequency as a matter of course... but it still has a switching frequency. Their application notes tell you how to apply offsets to the switching frequencies to try to reduce common noise between channels in a stereo amp.

"If there are papers with actual circuit descriptions can you point to one? All I find are the marketing "white papers" which are not very specific."

Nothing with internal circuit descriptions. If you are just looking for them in an amp there are numerous schematics available.

Shawn

[pauln]

Is it too early to bring up that when a signal comprised of sine
waves is low pass filtered that process is linear, but when a square
wave is low pass filtered the process is non-linear?

Below is a square wave before and after low pass filtering.

[sfogg]

"Switching doesn't necessarily create bits. "

No, it doesn't necessarily create bits, but it also do not automatically mean there aren't bits either.

The combination of a fixed switching frequency and the fact that there is one of two states does. Which gets back to my earlier diagram. Given a fixed switching frequency the on/off binary states can pass data.... they are bits.

Take the bits from one amp and transport them somewhere else and reproduce them at the same switching frequency and you will have the same output at possibly a different power level. Once again.. this is basically what SACD does.

"But, referring to fundamentals, there's nothing in all this "switch talk" that relates to digital bits and or quantization of the input signal."

With a fixed switching frequency the on/off states (bits) can pass data and all the digital amps out there have what amounts to quantization noise in their output. Because they are one bit systems their noise floor increases with frequency.... their SNR decreases the further you go up in frequency.

Shawn

[sfogg]

A square wave is nothing more then a sine with with a large number of odd order harmonics. If the low pass filter cuts off some of the harmonics you no longer have a square wave. That is all the diagram is showing you.

Cut off all the harmonics of the square wave and it will look just like a sine wave, because it is one.. you are seeing the fundamental. I have posted some pictures of this on the forum. If you are interested in seeing it I will dig them up.

Shawn

[Deang]

I just want to know why these amps don't have the grunge I hear in other solid state amps. No, they don't have the midrange bloom of a good tube amp, but they are devoid of the primary thing I find objectionable with solid state.

[Daddy Dee]

Mark,

I know you are teasing Dean, but I'm wondering what your take on that would be. His description, even in that lingo, sort of rings true with my take on the sound.

So what do you think is going on there?

I know years ago, I posted a question: 'Can SET "magic" be measured'

Your response to that started rolling a thread which went for several pages (with no flaming, congrats). And I thought it was most interesting.