class D question

[wengr]

Hello, does the actual audio signal get converted to digital and back again with a class D bass amplifer? I'm reading up on them and can't find a definate answer.

[KJung]

I'm no EE, but there is nothing 'digital' about class D amps as far as I know.

I believe Genz Benz has a very nice write-up on class D amps in the Shuttle manuals.

Agedhorse (Genz engineer) has also done a lot of great explaining regarding how class D works... maybe do a search and read some of his posts. Edit: Check out the power amp section in the Shuttle 9 manual on the Genz site (in the 'support' section).

[rumblinbass]

a very loose attempt at an explanation, I am no expert and I'm sure there are other topologies...

AFAIK, it's not really digital. It's more of a Pulse Width Modulation.
Here's a good link.

Top of page 3 has a good basic diagram. Notice the signal goes to a comparator, and "compared" against a triangle wave. The output of the comparator goes high or low depending on the state of the inputs...signal greater than the triangle wave or vice-versa.

That output gets sent to the power output stage just like that. After the output stage it goes through a filter that smooths it out and re-creates the audio signal.

a very loose attempt at an explanation, I am no expert and I'm sure there are other topologies...

[agedhorse]

It's a little bit digital and a little bit not. It crosses into both territories and it depends on the definitions of digital and analog. No matter what definitions used, it's a decidely non-linear topology.

[DynaFlow]

Whatever it is, it works like a charm.

I just got a Genz Benz Shuttle 3.0 1-8 combo, and the amount of punch, presence, and consistent balanced tone coming out of even this tiny package (14 pounds for the head and cab together!) is nothing short of amazing. Plus, even after running it hard for lengthy periods of time, the head remains cool as a cucumber.

[mongo2]

Originally Posted by agedhorse

It's a little bit digital and a little bit not.

Digilog?

[wengr]

ok thanks for the replies. so it's not linear. but there is no a\d\a ic chip in there correct?

[okcrum]

No, no A/D or D/A per se. There's a pulse width modulator on the input and a low pass filter on the output in the simplest version of a class D amp. The stages in between these two run in the on or off state, which is where the digital moniker comes from. As agedhorse said, definitely nonlinear.

[BbbyBld]

Originally Posted by wengr

Hello, does the actual audio signal get converted to digital and back again with a class D bass amplifer? I'm reading up on them and can't find a definate answer.

Some of them do that.

There's the classic analog comparator PWM method described above by rumblinbass, there's the self oscillating method, there's a mostly digital way of doing it with a processor and the whole nine yards (which is what you're asking), and then there are ways of doing it that combine these methods.

If you are really interested in this stuff I suggest you check out the DIY audio forum. Class D discussions are all the rage.

[okcrum]

Originally Posted by BbbyBld

Some of them do that.

There's the classic analog comparator PWM method described above by rumblinbass, there's the self oscillating method, there's a mostly digital way of doing it with a processor and the whole nine yards (which is what you're asking), and then there are ways of doing it that combine these methods.

If you are really interested in this stuff I suggest you check out the DIY audio forum. Class D discussions are all the rage.

When it comes right down to it, there's no reason you couldn't do any of the class D encoding methods in a DSP, as well. That would lead directly to a modeling class D amp.

Edit: Kind of like Yamaha had.

[BbbyBld]

Originally Posted by okcrum

When it comes right down to it, there's no reason you couldn't do any of the class D encoding methods in a DSP, as well. That would lead directly to a modeling class D amp.

I think this is the future of class D because you can do filtering for crossovers/EQ, dynamics processing, and class D all in one shot, AND have it all interact. I've been interested in this for a while.

[fdeck]

Originally Posted by BbbyBld

Some of them do that.

There's the classic analog comparator PWM method described above by rumblinbass, there's the self oscillating method, there's a mostly digital way of doing it with a processor and the whole nine yards (which is what you're asking), and then there are ways of doing it that combine these methods.

If you are really interested in this stuff I suggest you check out the DIY audio forum. Class D discussions are all the rage.

+1. There are some people on that forum who are deeply knowledgeable about this stuff. Also, some of the folks have shared their designs in LTSpice format. I decided long ago that I would not try to build my own Class-D from scratch (though I built a Tripath kit for grins), but it's fun to play around with the simulation.

[agedhorse]

Even for folks deeply knowledgeable, it's tough as sh*t to go from "in theory" to "in practice", especially if reliability is important.

As a matter of fact, the more you know the tougher it appears.

[Jerrold Tiers]

The short story.....

Essentially all older class-D are "analog"..... the analog amplitude is converted to an analog "time percentage", and then that is averaged back to an analog amplitude by a filter.

The time is the width of a "pulse". Each pulse is "on " for a percentage of the time, and "off" the rest. The series of these varying width pulses have an average value equal to the original varying signal.

The big deal is NOT the modulation, which is a "means to an end". The big deal is that while the signal is an "analog time percentage", it can be handled by a simple on/off decision in the power stage, which with real physical electronics results in much lower power losses, and as a result can use smaller power stages, has less total power used, etc.

A DSP can be used to make that conversion to an analog 'time percentage"...... whether using a DSP really makes the device "digital" or not is a matter of philosophy. Ditto for the original system which results in an "on/off" signal. At some point it must become analog again in order for the signal to be useful.

The power signal is generally never represented as a digital word with class-D, although the low level signal is so represented in a DSP.

Some class-D have just two states, on/off. Others add a third intermediate state. There are various details of how the circuit works, but pretty fair uniformity on what the pulses are, etc.

It is actually possible to imagine a unit which would directly use the digital word to control binary weighted outputs and directly convert from a digital word to an analog signal on a sample-to sample basis. A "power A/D converter" if you will. Some methods of doing that might be more practical than others, but the present system of PWM seems to work well enough that added parts are not required.

if the signal was digital until it was actually in the air, the unit might be called "pure digital. At present there is pretty much only speculation on how to do that.

[KJung]

From a non-technical user of class D amps for the last 20 years or so as my primary amplification, I guess my recommendation is to not even think about amp topology when making a purchase.

All the following amps are Class D:

Walter Woods
Acoustic Image
Markbass F1, F500, 800 series and SD1200
Genz Benz Shuttle Series
EA iAmp series (except for the very early originals)
Epifani UL amp series
Most of the current Carvin line
GK Micro series
Tecamp Puma500 and 1000....
TC RH450/Staccato
.... and of course many more.

For those who have played or owned most or many of these amps, you will hear virtually NOTHING in common tone-wise among them... some are brite and tight, some are 'tube warm and big down low', some are MASSIVELY huge in the low end, some are pure and clean and hi fi, some are grindy and wooly.

Purchase with your ears if you can, and ignore what is inside the box IMO.

Kind of a different take on the question, but for the typical player, a more useful answer I hope.

[aborgman]

Originally Posted by agedhorse

It's a little bit digital and a little bit not. It crosses into both territories and it depends on the definitions of digital and analog. No matter what definitions used, it's a decidely non-linear topology.

Yeah, that.

[wengr]

Originally Posted by Jerrold Tiers

It is actually possible to imagine a unit which would directly use the digital word to control binary weighted outputs and directly convert from a digital word to an analog signal on a sample-to sample basis. A "power A/D converter" if you will. Some methods of doing that might be more practical than others, but the present system of PWM seems to work well enough that added parts are not required.

While I could never describe it as effectively, this is pretty much what I was wondering about. thanks for the reply

[wengr]

Originally Posted by KJung

From a non-technical user of class D amps for the last 20 years or so as my primary amplification, I guess my recommendation is to not even think about amp topology when making a purchase.

All the following amps are Class D:

Walter Woods
Acoustic Image
Markbass F1, F500, 800 series and SD1200
Genz Benz Shuttle Series
EA iAmp series (except for the very early originals)
Epifani UL amp series
Most of the current Carvin line
GK Micro series
Tecamp Puma500 and 1000....
TC RH450/Staccato
.... and of course many more.

Interesting, I think there are many like me for whom the only way to try em is to buy em. I have only tried one, and for the most part liked what I heard. My desire to learn more about the signal path was caused by the fact that I have never found a modeling guitar amp or digital effect that would consider acceptable.

[KJung]

Originally Posted by wengr

Interesting, I think there are many like me for whom the only way to try em is to buy em. I have only tried one, and for the most part liked what I heard. My desire to learn more about the signal path was caused by the fact that I have never found a modeling guitar amp or digital effect that would consider acceptable.

That's why you should not think of a class D amp as 'digital'. The modeling amps and digital effects have very little to do with class D power amps.

Interestingly, the only 'digital' preamp (if that is an accurate term) that I'm familiar with except for those awful modeling things is the TC (e.g., RH450)... and they sound kind of like old Ampegs!

Hearing one class D amp and making judgements is like listening to one class A/B amp and then deciding if you 'like or don't like class A/B amps'.. throwing Glock, Eden, Thunderfunk, Aguilar into the same heap

One thing I will say is that the SMPS DID seem to be an issue until somewhat recently (i.e., the early switch mode power supplies seeming to sound a bit thin down low... at least in bass guitar integrated bass heads).

Some of the interim lightweight amps like the Epifani UL502 and the iAmp500/800 used a combination of class D power amp with a toriodal-based power supply to try to achieve the 'best of both worlds' (i.e., hammering lows and reasonably low weight and efficient operation/low heat).

However, even these companies have gone to SMPS now. Zero issue at this point from what my ears tell me.

Two amps that are interesting to me are the Markbass LM series and the Puma350... combining class A/B power amps with SMPS. Both sound very good to me... very warm and full. However, the Puma350 and the class D 500 have much more in common than different regarding tone and performance.

IMO and a lot of IME. Things have gotten a LOT better in the class D/SMPS since the early Walter Woods days!

[babebambi]

Originally Posted by KJung

Two amps that are interesting to me are the Markbass LM series and the Puma350... combining class A/B power amps with SMPS. Both sound very good to me... very warm and full. However, the Puma350 and the class D 500 have much more in common than different regarding tone and performance.

I have an OT question.
Is the Carvin B2000 like these class AB power amps with SMPS?
Who else is making amp like this?