yawn.
how about this (tell me if you recognize the pattern).
1. listener A listens to class-A amp. it doesn't sound good. listener blames it on the room, accompanying equipment or cables.
2. listener A listens to class-D amp. it doesn't sound good. listener blames it on the amp. case closed.
how about this (tell me if you recognize the pattern).
1. listener A listens to class-A amp. it doesn't sound good. listener blames it on the room, accompanying equipment or cables.
2. listener A listens to class-D amp. it doesn't sound good. listener blames it on the amp. case closed.
The batts provide low noise supply to a large capacitor - always needed to handle large transients. My point was that it's tough to find a quieter source of power than a battery.
I think if one wanted to get fancy, the chemistry in lithium polymer batts are superior to lead acid and can help with high transient loads. Still a big cap should always be used.
yawn.
how about this (tell me if you recognize the pattern).
1. listener A listens to class-A amp. it doesn't sound good. listener blames it on the room, accompanying equipment or cables.
2. listener A listens to class-D amp. it doesn't sound good. listener blames it on the amp. case closed.
+1 on that!
yawn.
how about this (tell me if you recognize the pattern).
1. listener A listens to class-A amp. it doesn't sound good. listener blames it on the room, accompanying equipment or cables.
2. listener A listens to class-D amp. it doesn't sound good. listener blames it on the amp. case closed.
+2
I think the overlooked factor here is the performance/price ratio. And in that class D wins over both class A and AB every single time. Hell, you could probably take a class A or AB amp at double, triple or even 10 times the cost (of ownership) and class D will still win.
Last edited:
I'm not sure about either of those. There have been some good posts showing that batteries are actually noisier than a good regulated supply. Yes, it surprised me, too.
As for the transients, yes for musical peaks. But a Class-D amp is always switching the full rail voltage. What about those transients? That doesn't mean it draws that RF current, of course, as the load (speaker) is low pass filtered. But working with Class-D you'll soon see that it's hard to have a clean ground rail, there is so much switching noise. Makes it hard to drain off noise. Batteries are not help there.
They can work, but need some consideration.
EDIT: I'm sure someone better versed in power supply design will chime in.
Power supply noise is almost irrelevant on any class D amp. It's completely on the output section as feedback will correct it. It's almost always irrelevant on input section as that is almost always supplied by an external or internal voltage regulator. It's only if that voltage regular on the input section is of poor quality that power supply noise matters at all.
I'm real late to this discussion, but for me these are the issues.
[1] Class D "synthesizes" the output waveform by blasting tiny pulses of power at variable width to a strip of high-current low-pass filters ... one hopes that the feedback circuitry is actually accomplishing the task of making a faithful output waveform viz a vis the input.
[2] #1 isn't in itself really all that bad, necessarily. Its like using teeny tiny jackhammers to carve wood. Thousands of them, simultaneously, so I hear. The results can LOOK like carved wood. The low-pass filters is the "sanding and buffing" that renders the carving smooth enough to let anyone think it was actually carved with awls, chisels, knives and manual labor. Sadly, it wasn't. Is this necessarily bad, especially "after sanding"? Dunno. Fear, uncertainty, doubt ... about the 'damage' caused to the waveform, should all that sanding not quite do what the design says it should do.
[3] RF and RFI - are a natural and inescapable consequence of the D method. All those little jackhammer pulses, positive and negative, have rise-times and fall-times on the order of dozens of nanoseconds (else, the D method wouldn't be anywhere near as thermally "efficient" as it generally is!). Hundreds of thousands of ultra-fast slew-rate pulses per second ... 1/P = F ... 1/100ns = 10 MHz ... there's going to be a lot of RF energy in that box.
[4] Because of #3, and because of a series of articles I read some years back where a researcher (Normal Korman) tracked down a not-so-subtle issue with RF and RFI interfering with the sonic purity of his amplifiers ... and on quashing ... replacing no other components than some wiring placement ... was bowled over listening to the same recordings, same equipment, same speakers, etc. RF/RFI was the culprit. And in the home-built and commercial Class D equipment ... is such RF/RFI suppression really all that effective? Ah... FUD perhaps, but the D stands not for digital, but doubt.
So, all I'm saying is - well, listen critically to it. if you're building your own, then make extra, extra special care in shielding all the input wires (coax from jack to board), and make sure the board layout totally isolates things. From having worked on a lot of MHz to GHz designs professionally, I wouldn't even think about implementing Class D without RFI "cages" on the output, to contain the RFI within a Faraday cage and prevent its leaking, unexpectedly, all over the place.
There you are.
GoatGuy
[1] Class D "synthesizes" the output waveform by blasting tiny pulses of power at variable width to a strip of high-current low-pass filters ... one hopes that the feedback circuitry is actually accomplishing the task of making a faithful output waveform viz a vis the input.
[2] #1 isn't in itself really all that bad, necessarily. Its like using teeny tiny jackhammers to carve wood. Thousands of them, simultaneously, so I hear. The results can LOOK like carved wood. The low-pass filters is the "sanding and buffing" that renders the carving smooth enough to let anyone think it was actually carved with awls, chisels, knives and manual labor. Sadly, it wasn't. Is this necessarily bad, especially "after sanding"? Dunno. Fear, uncertainty, doubt ... about the 'damage' caused to the waveform, should all that sanding not quite do what the design says it should do.
[3] RF and RFI - are a natural and inescapable consequence of the D method. All those little jackhammer pulses, positive and negative, have rise-times and fall-times on the order of dozens of nanoseconds (else, the D method wouldn't be anywhere near as thermally "efficient" as it generally is!). Hundreds of thousands of ultra-fast slew-rate pulses per second ... 1/P = F ... 1/100ns = 10 MHz ... there's going to be a lot of RF energy in that box.
[4] Because of #3, and because of a series of articles I read some years back where a researcher (Normal Korman) tracked down a not-so-subtle issue with RF and RFI interfering with the sonic purity of his amplifiers ... and on quashing ... replacing no other components than some wiring placement ... was bowled over listening to the same recordings, same equipment, same speakers, etc. RF/RFI was the culprit. And in the home-built and commercial Class D equipment ... is such RF/RFI suppression really all that effective? Ah... FUD perhaps, but the D stands not for digital, but doubt.
So, all I'm saying is - well, listen critically to it. if you're building your own, then make extra, extra special care in shielding all the input wires (coax from jack to board), and make sure the board layout totally isolates things. From having worked on a lot of MHz to GHz designs professionally, I wouldn't even think about implementing Class D without RFI "cages" on the output, to contain the RFI within a Faraday cage and prevent its leaking, unexpectedly, all over the place.
There you are.
GoatGuy
GoatGuy,
You forgot the interaction with a smps power supply and the chance of switching remnants passing through to the amplifier and being mixed with the signal source along with its rfi that can also be injected into the system. I am not saying that all of these issues can't be overcome, just that they do add a layer of complexity to the entire picture besides every other analog issue that still exist like any other analog circuit.
You forgot the interaction with a smps power supply and the chance of switching remnants passing through to the amplifier and being mixed with the signal source along with its rfi that can also be injected into the system. I am not saying that all of these issues can't be overcome, just that they do add a layer of complexity to the entire picture besides every other analog issue that still exist like any other analog circuit.
Hi,
Just to conclude, on your statements:
capacitance multiplier: it is very far from what we have written, becouse is incapable of transient response. (or not know)
Nice smps: I do not know a single one that I can put in a big class AB and has a fast response to transients.
I agree on the batteries then!
regards
you are still disqualifying your very fast power supplies from being in the SMPS group? or you mean bigger amp than they can supply? its just a little bit of a confusing comment from the designer of the 'SMPS' that has the fastest response of any power supply I have seen with Class AB
Lead acid batteries as quiet power amplifier supplies? really? NOT. they are quietish with low constant loading, thats all, they are neither fast or quiet with high slewing high current demanding needs
some of the new battery tech is much better, but I would and do still post regulate or at least buffer with caps.
Last edited:
Actually low pass in frequency domain means decimation in analog domain literally.
Analog->DSP->analog signal suffers interpolation artifacts in compare to original analog signal.
AFAIK Tektronix interpolation ICs for digital scopes is among some advanced NoHow for example. Wadia decoding computers is another sample of ramification that might be required "to restore" decimated Red Book PCM.
Analog->DSP->analog signal suffers interpolation artifacts in compare to original analog signal.
AFAIK Tektronix interpolation ICs for digital scopes is among some advanced NoHow for example. Wadia decoding computers is another sample of ramification that might be required "to restore" decimated Red Book PCM.
The RFI is a good point, but how is 400 kHz + freq going to affect audible spectrum? Also, if leads from chip to LC filter is smaller 1/4 wave of 400 khz antenna, reduces emission. Shielded wire on input seems like a good idea in general for high impedance high gain inputs. I like your tiny jack hammer analogy, and I would say that given jack hammers the size of 400 grit sandpaper grains, the end result is the same.
Oh, still trying to find the magic in something not measurable about audio waveforms that gets lost or dirty in PWM conversions or near a SMPS? Perhaps magic has more to do with expectations and human feelings, what about love? Personally I find more magic in setting up PA and mixing friend DJs and bands playing for friends, these are the origins of music.
Good point. The only thing digital is the fact that FET switches on or off. There is no DSP.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.