Class D Illustrated (lots of pics)

[Damian Coccio]

Cool stuff. Nice work.

 

[will33]

Wondering where I was June 12th and how in the hell I missed this thread.

Thanks for the explanations Bobby.

Of course, a good portion of it is over my head, but I think I can at least grasp the general concept of some of it.

With all this sample rates and on/off switching and stuff, I suppose people might not be entirely wrong when they call these "digital amps"?

 

[BbbyBld]

With all this sample rates and on/off switching and stuff, I suppose people might not be entirely wrong when they call these "digital amps"?

It's easy to see where the confusion comes from. Well, why not make things more confusing with random bits of information...

Class D amps have two main parts, the control part and the power amplification part. The control part can be analog or digital or both. The power amplification part can never be digital.

Let's say the pulse width minimum and maximum is 1% and 99% respectively. Since we are talking about an analog class D controller circuit, the resolution of possible duty cycles in between theoretically approaches infinity. In the practical sense, the resolution is determined by the smallest change in voltage the comparator can detect, which is determined by the smallest change in current detectable by the transistors inside the comparator. From the description I gave earlier, which is analog...all that stuff happens real time.

If we were talking about a digital class D controller circuit...first we would convert our analog audio to digital, or in other words a string of numbers, then we would take those numbers and run them into a microprocessor that would run the numbers through an algorithm that would calculate the correct duty cycle, and then the pulses would be spit out of the processor to be amplified. We would be bound by a specific resolution such as 16, 24, 32 bits, which would mean that we would have to divide up our minimum and maximum duty cycle range to fit that resolution within the digital algorithm. Let's take 16 bits for example. 16 bits means we would have a maximum resolution of 65,536 "points", so in the case of 1% - 99% min/max, we take our total range of 98% and divide that by 65,536 and find that 0.0015% is the smallest change that our system will be able to handle.

Hopefully that helps to clear up the analog / digital difference. The fact that there are sampling rates and on / off pulses don't necessarily mean digital. When we start using those on/off pulses as numbers, then it's digital.

 

[D.A.R.K.]

This is great, thank you. 20 reviews later and i'm a little closer to understanding how you guys make stick go boom!

 

[agedhorse]

The amplification part is nothing more than a very carefully timed set of very large level shifters.

 

[AlexanderB]

Calling a class D amp "digital" is about as accurate as calling a car "electrical" because it has electronical fuel injection. Still that "digital amp" expression seems very well rooted...

 

[agedhorse]

Still that "digital amp" expression seems very well rooted... in marketing.

There, I fixed it for you!

 

[Stealth Fighter]

Digilog?

 

[will33]

Digilog?

Anatal.

 

[BbbyBld]

Still that "digital amp" expression seems very well rooted...

I think Peavey has to take some of the blame for that...

At one time, digital engineers had to do a lot more with discrete circuitry. Anyone who has ever seen a PC motherboard from the 80's will know what I'm talking about. Anyway, a lot of the principles of class D amps are based on discrete digital hardware fundamentals. A lot of the types of parts used in class D control circuits are digital devices. Peavey started working on class D amps around 1983, within the digital engineering group for some of the reasons I listed, and continued into the early 90's. For that reason, the projects were classified as digital. Part of that was because there was a definite line in the sand between the analog and digital engineering groups--if the product was not digital, then the digital group had no possession. The analog group didn't want anything to do it anyway, because they were working to improve class AB power amplifiers, which was their specialty. We don't have that kind of separation anymore because most products are mixed signal these days.

Peavey started selling class D amps in the 80's at a time when "digital" was a huge buzzword--it was the "HD" of the times. Peavey named their first class D amps DECA, which stood for digital energy conversion amplifier. The next development was the DPC, which stood for digital phase conversion. All those products were advertised as "digital".

So anyway, class D amps are now "digital" because of marketing and corporate politics.

 

[Sahe_Bassist]

i think i grasp the general concept, thanks for the post!

if someone was looking to understand this, and the general topics of electrical engineering that go with this, where would be a good place to start?

 

[David Jayne]

Hopefully that helps to clear up the analog / digital difference. The fact that there are sampling rates and on / off pulses don't necessarily mean digital. When we start using those on/off pulses as numbers, then it's digital.

 

[smeet]

So is the "fuzziness" of the wave responsible for some people's perception that class D amps don't sound as solid in the bass and don't "slam" as hard as other topologies? Or is there even any validity to that perception?

 

[agedhorse]

IMO, fact and perception bear little relationship to each other. The "fuzziness" that you mention is mostly the effect of lack of understanding rather than what is really happening.

The sampling and filtering occurs many times above the frequency of the highest audio frequency (typically in the hundreds of kHz) and therefore drops out of the equation as WAY to the right of the decimal point (in a well designed product).

Solid and slam (or the perception of those terms) really has nothing to do with class D, just like low end performance has nothing to do with class D. If they did, the entire (at the big boy level) pro audio world would be suffering from lack of performance as that industry is virtually 100% class D in spite of the more extreme low end and sheer SPL. The big boy pro audio LF response is -3dB down to around 25-30Hz, and budget is secondary to performance. In the pro audio world, they have gone well beyond "it must be heavy to deliver real low end". The bass guitar industry is roughly 10 years behind... so I expect these urban myths to fade away as bad memories, much as others have done.

 

[BbbyBld]

So is the "fuzziness" of the wave responsible for some people's perception that class D amps don't sound as solid in the bass and don't "slam" as hard as other topologies? Or is there even any validity to that perception?

First, I should mention that a speaker is an inductive load, so it filters the fuzz so to speak.

A wimpy amp is a wimpy amp regardless of operating class. Some amps don't have honest ratings. I hate to say it, but Peavey started to play that marketing watts game a few years ago, but I'm happy to say that seems to have stopped when we started getting called out on it.

Some amps are not necessarily wimpy, but are not optimized for bass guitar. OEM power amplifier modules are designed for playing pre-recorded music for the most part because consumer audio is where the money is at. If you give a bassist power, he/she will use all of it because it's like crack, so bassists need high performance amplifiers that sound good and loud when operating flat out. Most people who buy bass gear can't play and don't really gig so they won't know the difference. You can't fool an experienced bass player.

Many class D amps rated at 4 ohm minimum load do not put out as much power into 8 ohms as the equivalently rated class AB/G/H power amps many of us are used to.

 

[Linnin]

This thread is a gold mine of information strewn with nuggets of knowledge, and should be 'stickified'.

 

[BurningSkies]

First, I should mention that a speaker is an inductive load, so it filters the fuzz so to speak.

A wimpy amp is a wimpy amp regardless of operating class. Some amps don't have honest ratings. I hate to say it, but Peavey started to play that marketing watts game a few years ago, but I'm happy to say that seems to have stopped when we started getting called out on it.

Some amps are not necessarily wimpy, but are not optimized for bass guitar. OEM power amplifier modules are designed for playing pre-recorded music for the most part because consumer audio is where the money is at. If you give a bassist power, he/she will use all of it because it's like crack, so bassists need high performance amplifiers that sound good and loud when operating flat out. Most people who buy bass gear can't play and don't really gig so they won't know the difference. You can't fool an experienced bass player.

Many class D amps rated at 4 ohm minimum load do not put out as much power into 8 ohms as the equivalently rated class AB/G/H power amps many of us are used to.

I also feel like the light/heavy false dichotomy is reinforced by the fact that small/light amp users are more likely to be small/light speaker users. Until fairly recently most of the neo loaded speakers that were on the market were basically neo speakers that had been retrofit in boxes designed for 'traditional' speakers and often times were also light in the lows. This, put together with amps that had inflated wattage specs (based upon extremely short 'burst' measurements) further reinforced the notion that you need 'weight' to be good. That amplified the old time notion from the 'good old days' that it was an upgrade to put heavy speakers in your cabs because more magnet meant better performance at one point, since that's the only way you could really beef up magnetic flux.

 

[smeet]

IMO, fact and perception bear little relationship to each other. The "fuzziness" that you mention is mostly the effect of lack of understanding rather than what is really happening.

The sampling and filtering occurs many times above the frequency of the highest audio frequency (typically in the hundreds of kHz) and therefore drops out of the equation as WAY to the right of the decimal point (in a well designed product).

Solid and slam (or the perception of those terms) really has nothing to do with class D, just like low end performance has nothing to do with class D. If they did, the entire (at the big boy level) pro audio world would be suffering from lack of performance as that industry is virtually 100% class D in spite of the more extreme low end and sheer SPL. The big boy pro audio LF response is -3dB down to around 25-30Hz, and budget is secondary to performance. In the pro audio world, they have gone well beyond "it must be heavy to deliver real low end". The bass guitar industry is roughly 10 years behind... so I expect these urban myths to fade away as bad memories, much as others have done.

This is the answer I expected, and I believe it. So, the question becomes - is there really a difference in solid/slam between (non-crappy) power amps, or are people allowing their prejudices about brands, cost, or light vs. heavy to affect their perceptions?

First, I should mention that a speaker is an inductive load, so it filters the fuzz so to speak.

A wimpy amp is a wimpy amp regardless of operating class. Some amps don't have honest ratings. I hate to say it, but Peavey started to play that marketing watts game a few years ago, but I'm happy to say that seems to have stopped when we started getting called out on it.

Some amps are not necessarily wimpy, but are not optimized for bass guitar. OEM power amplifier modules are designed for playing pre-recorded music for the most part because consumer audio is where the money is at. If you give a bassist power, he/she will use all of it because it's like crack, so bassists need high performance amplifiers that sound good and loud when operating flat out. Most people who buy bass gear can't play and don't really gig so they won't know the difference. You can't fool an experienced bass player.

Many class D amps rated at 4 ohm minimum load do not put out as much power into 8 ohms as the equivalently rated class AB/G/H power amps many of us are used to.

My particular question is because I have a Crest Pro-Lite 3.0 arriving today. I chose the amp based on reading about specs, user recommendations, weight, and of course price. I hope to never ask this amp to put out anything close to 3000 watts, the occasional peak aside. I don't have power amp experience to draw on, so I'm wondering ... Would I have noticed a difference in the aforementioned solid.slam if I had instead bought a CA-9 or PLX or whatever? This is assuming I'm not taxing either of the amps, using it for playing bass at a loud but reasonable volume, through a couple of fEARful cabinets (15/6/1 and 12/6/1).

 

[BurningSkies]

This is the answer I expected, and I believe it. So, the question becomes - is there really a difference in solid/slam between (non-crappy) power amps, or are people allowing their prejudices about brands, cost, or light vs. heavy to affect their perceptions?



My particular question is because I have a Crest Pro-Lite 3.0 arriving today. I chose the amp based on reading about specs, user recommendations, weight, and of course price. I hope to never ask this amp to put out anything close to 3000 watts, the occasional peak aside. I don't have power amp experience to draw on, so I'm wondering ... Would I have noticed a difference in the aforementioned solid.slam if I had instead bought a CA-9 or PLX or whatever? This is assuming I'm not taxing either of the amps, using it for playing bass at a loud but reasonable volume, through a couple of fEARful cabinets (15/6/1 and 12/6/1).

Oh, you're going to love that with your speakers. Really. Get back to me once you bridge it for a bit.

 

[_Elf]

[...] Some amps don't have honest ratings. I hate to say it, but Peavey started to play that marketing watts game a few years ago, but I'm happy to say that seems to have stopped when we started getting called out on it. [...]

I always found it kind of funny when you would see a "3000 watt" amplifier, turn it around, and see a 495W draw listed on the power inlet (using the Crest Pro-Lite 3.0 as an example, at least by the pictures in the manual). Even besides that the inlet is an IEC C14 which is only rated to 10A draw (1200W @ 120VAC) and the average 15A wall receptacle it would be plugged into would only give you 1440W continuous draw until you popped the breaker after a few minutes (80% thermal trip rating), this all not yet accounting for the power factor of the device. Also, 3000W into a 2 Ohm load would be about 39A @ 77V going across your speaker cables. That would get toasty after a while on your average 16-12 ga.

Of course the point, I guess, is that we're not talking about "a while." I get the whole duty cycle and peak vs. average vs. RMS power thing (and the effect of what time window you're measuring those things), although one wonders exactly what size capacitors are sitting there on the supply rails to support all this. I'm also definitely not trying to bash; it's clear that a lot of great products out there do this. It's just that the huge disparities involved between what is probably a pretty reasonable listed average maximum draw on the back and what gets advertised on the front kind of makes me chuckle.