Hello!
I hear people complaining about noise and spurs from SMPS and I also hear people rave about Class-D amps and this is confusing me.
I know from experience that SMPS can be really noisy (bad designs!) but it must be much easier to design a good low noise SMPS than a low noise, low spur, low distortion, wideband etc. class-D audio amp???
To me SMPS+analog amp sounds lika a better idea than trafo+class-D.
(SMPS+Class-D with syncronized switching frequencies is probably the future, but we are not there yet)
I have just compared a LM3886 amp with the Zap Pulse 2.2se (with regulated PS at +/-40V) and I do prefere the LM3886. Am I doing something wrong?? What are your experiences with different combination of analog/digital PS and amps???
I hear people complaining about noise and spurs from SMPS and I also hear people rave about Class-D amps and this is confusing me.
I know from experience that SMPS can be really noisy (bad designs!) but it must be much easier to design a good low noise SMPS than a low noise, low spur, low distortion, wideband etc. class-D audio amp???
To me SMPS+analog amp sounds lika a better idea than trafo+class-D.
(SMPS+Class-D with syncronized switching frequencies is probably the future, but we are not there yet)
I have just compared a LM3886 amp with the Zap Pulse 2.2se (with regulated PS at +/-40V) and I do prefere the LM3886. Am I doing something wrong?? What are your experiences with different combination of analog/digital PS and amps???
It is difficult to get a good and quiet design using both Class-D amplifiers and SMPS. However, it is possible. We (coldamp) have been using and selling this combination with no problem, obtaining better sound quality than with linear PSUs, if everything is laid out carefully and cabling is properly done.
The future is, no doubt, SMPS + Class-D. Nobody said it was the easier way, though.
The future is, no doubt, SMPS + Class-D. Nobody said it was the easier way, though.
I've done a lot of testing and listening to class-d with battery, linear and switching power supplies. As ssanmor indicates above, much depends on the implementation.
Class-D amp by their very nature kick a lot of noise back onto the power rails. Usually much more noise than any well designed power supply. How well the power supply deals with the amp noise goes a long way toward quality.
There are fans of battery supplies, because batteries are the ultimate quiet supplies, right? Wrong. Batteries are very quiet until you turn on the amp, then the noise floor rises considerable. It seems that the battery can't cope well with the RF noise generated by Class-D amps. A good linear or switch mode supply can.
As many of you know, SMPS will have a lot of HF noise, much more than a good linear supply or battery, but it will tend to have less 50/60 Hz than a linear supply. But even that noise is far below the noise floor of the Class-D amp. Once the amp is running the noise spectrums are quite different from supply to supply.
Leaving the noise issue behind, how well does the power supply regulate and how does that regulation sound? That is were a SMPS often takes the upper hand. I don't know why (yet) but they do seem to sound better than other supplies, at least with Class-D. A good linear supply can also sound very nice. And of course even a battery supply could be regulated and filtered, if you wanted to.
Where does that leave us? In limbo, sort of. I would not use a SMPS for a sensitive or high gain circuit such as a preamp, but in a Class-D amp most SM noise will be drowned by the noise of the amp. And the SMPS noise can be filtered out to a large degree. They are not as bad as you might think, and in fact are quite good when used properly.
Class-D amp by their very nature kick a lot of noise back onto the power rails. Usually much more noise than any well designed power supply. How well the power supply deals with the amp noise goes a long way toward quality.
There are fans of battery supplies, because batteries are the ultimate quiet supplies, right? Wrong. Batteries are very quiet until you turn on the amp, then the noise floor rises considerable. It seems that the battery can't cope well with the RF noise generated by Class-D amps. A good linear or switch mode supply can.
As many of you know, SMPS will have a lot of HF noise, much more than a good linear supply or battery, but it will tend to have less 50/60 Hz than a linear supply. But even that noise is far below the noise floor of the Class-D amp. Once the amp is running the noise spectrums are quite different from supply to supply.
Leaving the noise issue behind, how well does the power supply regulate and how does that regulation sound? That is were a SMPS often takes the upper hand. I don't know why (yet) but they do seem to sound better than other supplies, at least with Class-D. A good linear supply can also sound very nice. And of course even a battery supply could be regulated and filtered, if you wanted to.
Where does that leave us? In limbo, sort of. I would not use a SMPS for a sensitive or high gain circuit such as a preamp, but in a Class-D amp most SM noise will be drowned by the noise of the amp. And the SMPS noise can be filtered out to a large degree. They are not as bad as you might think, and in fact are quite good when used properly.
Only a couple of additions to Panomaniac comments, that has done a very good job in explaining why SMPS can sound better with Class-D:
Usually, switching frequencies of Class-D amplifiers are much higher than the switching frequency of SMPS's. The only problem that may arise are "beat" frequencies between these frequencies that may fall within the audio band, producing subtle high pitched (and even variable frequency) noises. However, practice has demonstrated that it is quite easy to filter out the rails noise so this effect disappears, so it is no longer an issue with good SMPS implementations.
About regulation: linear power supplies are almost never regulated, and this is specially true as power increases, due to efficiency and heat dissipaton issues. On the other hand, it is easy to make regulated SMPS, that will hold rails voltage up to full power with no problem. This translates in a deeper and more powerfull bass, mainly, although it has sonic impacts in the rest of the audio band. Further, if you add a respectable output capacitance (most SMPS manufacturers fail to use it due to increased costs and add a few hundred microF per rail arguing that the refresh rate is much higher than in a linear PSU), you get the best from both worlds: efficiency, excellent load regulation and high energy reservoir.
As a curiosity, we have been manufacturing both linear and switching power supplies, and almost all customers that have used both of them with our Class-D amplifiers (and others), agree in that they get a better overall sound when using SMPS...
Usually, switching frequencies of Class-D amplifiers are much higher than the switching frequency of SMPS's. The only problem that may arise are "beat" frequencies between these frequencies that may fall within the audio band, producing subtle high pitched (and even variable frequency) noises. However, practice has demonstrated that it is quite easy to filter out the rails noise so this effect disappears, so it is no longer an issue with good SMPS implementations.
About regulation: linear power supplies are almost never regulated, and this is specially true as power increases, due to efficiency and heat dissipaton issues. On the other hand, it is easy to make regulated SMPS, that will hold rails voltage up to full power with no problem. This translates in a deeper and more powerfull bass, mainly, although it has sonic impacts in the rest of the audio band. Further, if you add a respectable output capacitance (most SMPS manufacturers fail to use it due to increased costs and add a few hundred microF per rail arguing that the refresh rate is much higher than in a linear PSU), you get the best from both worlds: efficiency, excellent load regulation and high energy reservoir.
As a curiosity, we have been manufacturing both linear and switching power supplies, and almost all customers that have used both of them with our Class-D amplifiers (and others), agree in that they get a better overall sound when using SMPS...
And again - I agree with all above.
I didn't mention in my post that I was running tests on the low power T-Amps for which it's easy to build a linear regulated supply. Larger, more powerful amps are rarely regulated, as ssanmor mentions.
I'm also a fan of a large capacitance reserve, no matter what the supply. There are a lot of sonic advantages. The disadvantages seem to be cost and weight.
Power supply design and implementation is very important in an amplifier. Because the amp is a power supply, basically.
I didn't mention in my post that I was running tests on the low power T-Amps for which it's easy to build a linear regulated supply. Larger, more powerful amps are rarely regulated, as ssanmor mentions.
I'm also a fan of a large capacitance reserve, no matter what the supply. There are a lot of sonic advantages. The disadvantages seem to be cost and weight.
Power supply design and implementation is very important in an amplifier. Because the amp is a power supply, basically.
Thanks for your answers and explanations.
Well, it seems like it is already possible to make a good SMPS+Class-D amp but I still think SMPS+analog amp is also a very good solution, especially if you use chipamps (LM3886/TDA7293) for an active speaker. I do not think that I would like to use a SMPS with 6 Class-D amps in the same box but SMPS+6 chipamps should be no problem (cheap and small size : ) Has anyone tried???
The advantages I see with SMPS is good regulation at high currents and high efficiency. The smaller caps needed is probably good because smaller caps will have a larger bandwidth (resonance at a higher frequency) and better quallity (at least for the same price).
The high frequency regulation will also be good at taking care of the back generated current (Back EMF???) from the speaker.
Maybe a SMPS has a better isolation to 220V/GND because of the MUCH smaller transformer...and that is always good.
I also think that in a good SMPS the LF noise (<20k) can be very low and that the hf noise (switching frequency etc) might not be a big problem because it will not end up in the audio band (this is not really true..there will always be some second order effects that will transform the hf noise to lf noise....)
Too bad that it is almost impossibe to find data/measurements on the noise and output impedance vs frequency and output current.
ps smps can also be very good in cd players and preamps. One exampel is the Linn CD12.
Well, it seems like it is already possible to make a good SMPS+Class-D amp but I still think SMPS+analog amp is also a very good solution, especially if you use chipamps (LM3886/TDA7293) for an active speaker. I do not think that I would like to use a SMPS with 6 Class-D amps in the same box but SMPS+6 chipamps should be no problem (cheap and small size : ) Has anyone tried???
The advantages I see with SMPS is good regulation at high currents and high efficiency. The smaller caps needed is probably good because smaller caps will have a larger bandwidth (resonance at a higher frequency) and better quallity (at least for the same price).
The high frequency regulation will also be good at taking care of the back generated current (Back EMF???) from the speaker.
Maybe a SMPS has a better isolation to 220V/GND because of the MUCH smaller transformer...and that is always good.
I also think that in a good SMPS the LF noise (<20k) can be very low and that the hf noise (switching frequency etc) might not be a big problem because it will not end up in the audio band (this is not really true..there will always be some second order effects that will transform the hf noise to lf noise....)
Too bad that it is almost impossibe to find data/measurements on the noise and output impedance vs frequency and output current.
ps smps can also be very good in cd players and preamps. One exampel is the Linn CD12.
Have a look at:
http://www.coldamp.com/opencms/opencms/coldamp/en/aplicaciones/
This one was used for a classic music concert hall. You will understand that this is one of the most sound-quality demanding applications you may find.
Pierre said:I'd like to see a high-quality battery supply feeding a 400W amplifier (directly)
You're right. I forgot that there are still people out there, who need 400W to get A sound out of the speakers.
Or the other way around: I didn't consider that people are tight to an amplifier which theoretically - due to construction reasons - is able to supply even 1000Ws (showing up with +/- 48V rails or similar), from which 10W (1%) or less are used in average!
(BTW - I own Icepower 1000ASP modules by myself -- I know what I am talking about - I gave up on the mains power mess. >> For subwoofers only!)
Cheers
ssanmor:
Cool, really cool!
soundcheck:
"I'd like to see a SMPS beating a high-quality battery-supply."
Have you tested a high quallity SMPS vs. a high-quality battery-supply?
panomaniac tried and he liked SMPS better but as always it is about the implementation. SMPS can be amazingly good.....but usually there is more noise than DC (sorry, now I'm evil!)
darw82
"You have to synchronize the SMPS and class D amp to a master clock."
Yepp, that is probably a good idea! Did you have to do that ssanmor??
I didn't want to start a PS war, they can all be very good! I was just confused why people were so afraid about noise from SMPS and not from Class-D amps.
Cool, really cool!
soundcheck:
"I'd like to see a SMPS beating a high-quality battery-supply."
Have you tested a high quallity SMPS vs. a high-quality battery-supply?
panomaniac tried and he liked SMPS better but as always it is about the implementation. SMPS can be amazingly good.....but usually there is more noise than DC (sorry, now I'm evil!)
darw82
"You have to synchronize the SMPS and class D amp to a master clock."
Yepp, that is probably a good idea! Did you have to do that ssanmor??
I didn't want to start a PS war
, they can all be very good! I was just confused why people were so afraid about noise from SMPS and not from Class-D amps.
soundcheck said:
Not true. A battery supply can be and should be filtered and maybe even regulated. Batteries are not perfect, but some people seem to think they are. Batteries work well for DC - when you put a variable load on them, they don't perform as well. So a filter can help.
As I said above:
panomaniac said:
Hello,
At first, we tried to synchronize our Class-D modules (BP4078) together, and also with our SPS80 / SPS30 supplies.
After a lot of experimentation we found that it was not necessary with a proper layout, so we have stopped recommending or doing so. If the amplifier and SMPS is properly designed, you don't need to sync them and you won't have any noises produced by beating or so.
I agree that batteries are nice for preamps or headphone amps, but nobody will even think about feeding a medium-powered amplifier with batteries for obvious practical reasons. Don't know if many batteries will have the dynamic response and output impedance that our switching power supplies reach, anyway, and you can be sure that it translates to sonics.
At first, we tried to synchronize our Class-D modules (BP4078) together, and also with our SPS80 / SPS30 supplies.
After a lot of experimentation we found that it was not necessary with a proper layout, so we have stopped recommending or doing so. If the amplifier and SMPS is properly designed, you don't need to sync them and you won't have any noises produced by beating or so.
I agree that batteries are nice for preamps or headphone amps, but nobody will even think about feeding a medium-powered amplifier with batteries for obvious practical reasons. Don't know if many batteries will have the dynamic response and output impedance that our switching power supplies reach, anyway, and you can be sure that it translates to sonics.
ssanmor said:
My battery (Northstar 90) has an impedance of 2mR ESR. Try to beat that with your SMPS! It delivers 4500A short circuit current.
You can dream of such a power reservoir delivered through the mains. My T-Amp runs almost a week on this battery.
I do agree though, that smaller batteries (+-10 Ah) do have an issue with its dynamic response, that's why I changed to the big ones. However, the small ones still deliver a more quiet background than SMPSes.
One big downside, especially of small batteries, they degrade (increase of ESR, which lowers the dynamic response) over time, since they are reloaded much more often than bigger batteries.
This will not happen with an SMPS.
You'll also have an issue if you need higher voltages. I'd regard
24V as the maximum, which in the end requires a high efficieny speaker and a special amp.
If you look from the invest side:
If you run after value-for-the-money: SMPS is your choice
If cost-is-no-object and you wanna push it to the limits: A big filtered battery supply is IMO something one should think of.
Cheers
PFC
IMO an active PFC based switcher works very well indeed for power amplifiers.
Some companies use unregulated switchers in their amplifiers - the worst of both worlds - higher output impedance, and the noise to contend with.....
Of course switchers have their 'power envelope' that you cant really get around, but a well designed switcher for audio can rock....
my 2 cents......
IMO an active PFC based switcher works very well indeed for power amplifiers.
Some companies use unregulated switchers in their amplifiers - the worst of both worlds - higher output impedance, and the noise to contend with.....
Of course switchers have their 'power envelope' that you cant really get around, but a well designed switcher for audio can rock....
my 2 cents......
soundcheck said:
Ahhh.... But at what frequency? DC? That's the point. My testing seems to indicate that battery ESR rises pretty fast with frequency. One would imagine that different batteries would be quite different in that respect.
FWIW, I started listening tests with a big (200 AH, IIRC) marine battery. It just didn't seem any better than a good linear or SMPS. It should have been, but I couldn't hear it.
Now here is where I'm going to look like a complete idiot.
I did extensive spectrum analysis of different power supplies, SLA battery, linear regulated, SMPS in both the audio and radio bands. Tested the power rails and the outputs at idle and under load of several amplifiers. That sounds simple, but it wasn't, it took days.I wanted to publish my findings, but all the measurement files are now corrupt! arrrggggghhhhh! And the spectrum analyzer was not mine; I had to give it back.
Maybe I'll start over some day, or try to fix the files. The findings were certainly interesting.
But hey, I'm willing to do battery listening tests again, because everyone raves about them so. Maybe I was wrong. It's certainly worth another try.
Pierre said:I'd like to see a high-quality battery supply feeding a 400W amplifier (directly)
Hi Pierre,
Have a look HERE
This beasts have a Ri of approximately 10 mOhm. If you short them with a screwdriver the screwdriver will vanish.
Only disadvantages: Cost! And you’ll need a complicated battery management system (BMS).
Cheers
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.