Headroom? Need? I take headroom specifications as the surest sign of a weak supply that'll sag far too much to be considered "high end" as soon as you work it a little.
Classd is clearly the most efficient period, some more than others, and the point was depending on the exact method used, it will effect the efficiency curve slightly, but they're all fairly comparable.
Saying the curve is mostly flat though isn't such a good argument when you consider manufacturers rave about just a few percent higher efficiency than the next guy. It's a pretty big deal in that sense.. and the fact is while being mostly flat, it is typically highest towards peak output.
In it's most inefficient state though, it will still outdo a class B amp. While graphs may show it hitting 0 efficiency with 0 power output, that's too idealized. Fact is power in just becomes idle current consumed, and most of that doesnt' get consumed, but returned to the supply. Though, it switches faster, so there's some extra switching loss to be seen. It's easily handled most often with just the case as heatsink.
If efficiency were truly ever to hit 0 with it, "idle" would obviously become an illegal mode of operation.... it'd have to shut down as soon as no signal was present. As it is, it just warms up a little bit more.
Classd is clearly the most efficient period, some more than others, and the point was depending on the exact method used, it will effect the efficiency curve slightly, but they're all fairly comparable.
Saying the curve is mostly flat though isn't such a good argument when you consider manufacturers rave about just a few percent higher efficiency than the next guy. It's a pretty big deal in that sense.. and the fact is while being mostly flat, it is typically highest towards peak output.
In it's most inefficient state though, it will still outdo a class B amp. While graphs may show it hitting 0 efficiency with 0 power output, that's too idealized. Fact is power in just becomes idle current consumed, and most of that doesnt' get consumed, but returned to the supply. Though, it switches faster, so there's some extra switching loss to be seen. It's easily handled most often with just the case as heatsink.
If efficiency were truly ever to hit 0 with it, "idle" would obviously become an illegal mode of operation.... it'd have to shut down as soon as no signal was present. As it is, it just warms up a little bit more.
I have to disagree. ClassD needs additional power supply for modulator and gate driver, which can be as high as 50...90 mA. In many amps this is linearly stabilized from main supply, so there can be already ~10 W idle loss without any power device. Switching loss adds some watts further. A ClassAB can easily operate with 10 mA idle current. If I listen to music while I reading a book, with my unefficient speakers average power output doesn't exceed 40 mW. This means about 100 mA current, so 6W power consumption at 60V supply + 10mA*2*60V=1,2W idle loss. At average home use tipical ClassD is hot, ClassAB is cool. I would be very happy, if this wasn't true, but it is.
ClassD can be more efficient, if we supply it with 3 small sources plus two main voltage, but it requires many connections, and a logic circuit to disable output when any of these is not present. I do it this way in my bigger amps, but it doesn't worth the efforts in small (<800W) amps.
What do you mean? efficiency=power output/power input. 0/positive is definitely 0!
Power input is almost constant at very low output power, so efficiency is directly proportional to output power. This doesn't mean any inhibition on signal.
Not sure what you mean by inhibition on signal, but as per the equation power output is not 0, it's just too textbook.
If you have a systen capable of sustaining 1kW, where all of a suddent for whatever reason it suddenly drops to 0% efficiency, that means it's wasting the entire 1kW.... that'd be one damn hot class d amp at idle dont' you think?
So I guess I look at it more as power used (including all losses + what's delivered to load)/power available.
Graph that and I think the graph will look a little different towards idle, and probably alot more realistic, since there's no way a class d amp idle is cooking off all it's available power as heat.
If you have a systen capable of sustaining 1kW, where all of a suddent for whatever reason it suddenly drops to 0% efficiency, that means it's wasting the entire 1kW.... that'd be one damn hot class d amp at idle dont' you think?
So I guess I look at it more as power used (including all losses + what's delivered to load)/power available.
Graph that and I think the graph will look a little different towards idle, and probably alot more realistic, since there's no way a class d amp idle is cooking off all it's available power as heat.
But you're looking at efficiency the wrong way. Pafi is quite right. If a 200W (say) amplifier uses 6W as 'housekeeping' power, then if it is outputting only 100mW, then it is consuming 6.1W. This is a very low efficiency indeed. Not quite 0 but not far off! At an output of 100W it is consuming 106W which is a very high efficiency.
I know I know.... it's not the conventional way to look at it, non the less... if a 1000 watt amp hits zero efficiency it ought to be cooking off 1000W as wasted heat correct? But it doesn't does it.
Absolutely not! Nobody told that a 1 kW amp have to consumpt 1 kW all the time! (It's true only for classA.) Power input can be as low as you want, eg. 1W, but if you divide 0W by 1W, you still get 0%.
Are you joking? You know much more then this!
Pafi said:
All this talk about power consumption. Try an UCD180, stays very cool and sounds incredible. If you ask me, easily beats commercial amps that cost 10x more.
Gertjan
jmartins said:
the fact that jeff rowland, audio research or halco have hi-end products in this space does not negect the fact that class D started its life at the bottom end, mostly as subwoofer amps or computer / pda / cell phone audio amps where full-band fedility wasn't required.
recent advances in dsp and switchers and topologies made it more appealing to full-band and hi-fi applications that we are interested.
Conceptually, Class A is the most natural way of amplifying a signal: the output device is fully on.
Class D is NOT very natural in that it is trying to approximate a continueous sine wave by switching the output devices very fast.
Class B is somewhere in between: it is switching (aka Class D) from positive half to negative half. However, within the halfs, it is Class A.
classd4sure said:
I guess I don't pay attention to "headroom" specs. I tend to buy amps modules with more power than I'll probably use, and use a bigger power supply - creating my own headroom spec. And I at least think that I'm being more green over running class-ab (which I'd also buy more power and use a bigger xformer)
Efficiency question
You must bear in mind that efficiency ratings are for max power out. At low level and idle the situation reverses with typical class A/B using less power. Note I did say typical not some unit that runs class A to several 10’s of watts. This difference can easily be a 10:1 advantage for the A/B amp. Nothing new here Bruno already addressed this issue in one of the early threads for the UcD’s, unfortunately not much can be done about it yet. It will have to wait for better parts.
Roger
You must bear in mind that efficiency ratings are for max power out. At low level and idle the situation reverses with typical class A/B using less power. Note I did say typical not some unit that runs class A to several 10’s of watts. This difference can easily be a 10:1 advantage for the A/B amp. Nothing new here Bruno already addressed this issue in one of the early threads for the UcD’s, unfortunately not much can be done about it yet. It will have to wait for better parts.
Roger
I guess it depends on what amplifier you would be talking about but just by looking at a published efficiency vs output power curve I found here (go down to figure 6 http://www.bashaudio.com/technologies.htm) you can obviously see that class d would still have better efficiency even at low power.
Here's some more good reading.
http://www.audiodesignline.com/howto/portableaudio/189602183
Class ab amps aren't used in low power battery powered applications because they usually dissipate more power than class d; even at idle.
Here's some more good reading.
http://www.audiodesignline.com/howto/portableaudio/189602183
Class ab amps aren't used in low power battery powered applications because they usually dissipate more power than class d; even at idle.
so if i dont care abt battery the A/B will be an better choice, right?
I found some cheap chinese PRO amplifier with nice specs and high power, but really dont know if they are good, btw, i can found wich output trnsistors it use, this can help me to found if are good amps?
I found some cheap chinese PRO amplifier with nice specs and high power, but really dont know if they are good, btw, i can found wich output trnsistors it use, this can help me to found if are good amps?
samsagaz said:
No
My guess is that you'd likely be way better off with even chip-amps. Pro amps aren't built for their sonics, but to be reliable. Being cheap doesn't help & commonly published specs are largely meaningless.
dave
Actually I don't see low power specification here. These test were made with full drive, and that curve can't be read at 100 mW. And if it were as you said, still you can't generalize. There are many ClassD wich dissipates more then a comparable ClassAB at room volume. (These amps are tipically very big.)
It depends on your requirements. Power output, cooling, etc...
Do you want to build it completely, or you want to buy modules? If you can buy, UcD180s are very good (however a litle bit expensive). If you want to build, then forget about discrete ClassD, unless you have oscilloscope and many experience.
Pafi said:
well, i have an osc but not too much experience so will be hard do it 100%myself, btw, i was thinking in modules.
I really want to spend the same money that i will use for an commercial unit but do it myself. I really cant use Class A because i want at least 200W per channel. So need to use A/B or D. I think that D is easy to build (with the kits), but i dont care abt it, i want nice sound, with specs similar to commercial ones.
i need at least 5 channels (can use diferent case for each one). maybe can build 7 channels to byamp the front stage.
The speakers that i will use will be a pair of Polk LSi 15 (at least to start, later i can do it myself other colums, i have 3 Ribbon Tweeters, but cant spend money in woofers now, i want to purchase the LSi and some processor (DTS, DD...).
About the power, i need to build an nice amp, i dont want to change it in a few years, i want to use it for looooonnggg time, i want to upgra processor, maybe speakers but not amps 🙂
Thanks for all your help guys.
there may be a distinction between how power is dissipated in a class A/B amp vs. a Class D amp.
In a class A/B amp, the dissipation takes place mostly in the output devices. because they are essentially voltage-controlled variable resistors.
You can view a class D amp as a capacitor being alternately connected to either the positive rail or the negative rail, through an inductor whose sole purpose is to limit the current charging up or discharging the capacitor. It is the charging and discharging that consumes energy, but in the power supply, not in the output devices.
The amount of the current I guess depends on the size of the inductor / capacitor, rail voltage and oscillation frequency: the larger the inductor, or the smaller the capacitor, or the lower the rail voltage, or the faster the frequency, the lower the current.
In a class A/B amp, the dissipation takes place mostly in the output devices. because they are essentially voltage-controlled variable resistors.
You can view a class D amp as a capacitor being alternately connected to either the positive rail or the negative rail, through an inductor whose sole purpose is to limit the current charging up or discharging the capacitor. It is the charging and discharging that consumes energy, but in the power supply, not in the output devices.
The amount of the current I guess depends on the size of the inductor / capacitor, rail voltage and oscillation frequency: the larger the inductor, or the smaller the capacitor, or the lower the rail voltage, or the faster the frequency, the lower the current.
the way I view it is that the capacitor is like a bucket and the inductor is like a valve. It gets filled up by the positive rail, and the valve controls how fast the bucket gets filled up. and then the bucket empties itself to the negative rail, through the valve again.
The capacitor is essentially shorting the positive rail to the negative rail, indirectly and alternately.
The capacitor is essentially shorting the positive rail to the negative rail, indirectly and alternately.
Pafi said:
What do you mean you don't see low power specs? You can read the graph down to about 1W, where their efficiencies are about equal. Why would you bother calculating efficiency below a watt for a power amp? The most power that will be wasted below 1W output will clearly be less than a watt.
For most speaker systems today, I would consider "room volume" to be at least 1W output for lower listening levels. Maybe more like 2-4W for more reasonable/adequate listening levels.
Another way to think about class d operation is just a high power PWM signal being fed to a low pass filter.
- Status
- Not open for further replies.