I received my 12.5 amp switching supply today and I'm really loving it. It doesn't introduce anymore noise than my SLA did, which is to say none at all.
I was pleasantly surprised to find out the voltage can be regulated up to 14.2v instead of the 13.2v its spec'd for. With my two modded SI's playing full power it doesn't fluctuate a single millivolt.
I know 12.5amps is way more than I need, but for $24 shipped I will gladly take that extra headroom and live with it.
I was pleasantly surprised to find out the voltage can be regulated up to 14.2v instead of the 13.2v its spec'd for. With my two modded SI's playing full power it doesn't fluctuate a single millivolt.
I know 12.5amps is way more than I need, but for $24 shipped I will gladly take that extra headroom and live with it.
I guess SMPS is like everything, there are good and bad ones.
Just because it is SMPS does not mean it is bad or good.
I'm no expert, but I think that Linn and Chord use SM in their pre and power amps.
Russ Andrews is also all praise about SM, after comparing them with linear.
I am personally using a SMPS with amp3 and I can't complain of the sound and size. Of course I'm not going to start DIYing SMPS anytime soon, just buying them off the shelf is good for me.
Just because it is SMPS does not mean it is bad or good.
I'm no expert, but I think that Linn and Chord use SM in their pre and power amps.
Russ Andrews is also all praise about SM, after comparing them with linear.
I am personally using a SMPS with amp3 and I can't complain of the sound and size. Of course I'm not going to start DIYing SMPS anytime soon, just buying them off the shelf is good for me.
Some amp designs are quite a bit better than others at rejecting noise. Discretes and tubes are quite poor at it though, while the Class D are generally good. That means one can "get away" with using switchers on them, but there is no such thing as a "good" switcher that comes close to as clean as a good linear for across-the-board use.
Switchers are used for efficiency and lower cost in some higher-current applications (and sometimes a size and weight reduction). Otherwise, from a pure cleanliness and appropriateness to audio (ignoring aforementioned factors), there is no "good switcher" that is as good as any common linear (need not even be a particularly "good" linear, just a common design.
Whether one wants to weigh this against what any particular amp design will allow, whether any subjectively-audible difference would result, is another matter.
Switchers are used for efficiency and lower cost in some higher-current applications (and sometimes a size and weight reduction). Otherwise, from a pure cleanliness and appropriateness to audio (ignoring aforementioned factors), there is no "good switcher" that is as good as any common linear (need not even be a particularly "good" linear, just a common design.
Whether one wants to weigh this against what any particular amp design will allow, whether any subjectively-audible difference would result, is another matter.
! said:
I do have to take exception to this blanket statement. It simply is not true! I have worked extensively with both types and find the SMPS to be superior sonically if applied correctly. This applies to both low and high level circuits. The top designers of high end class D amps are most all using SMPS’s. I really hope you don’t think this is for increased profit as it really is far more expensive to design and implement a proper SMPS. You need to get into 1,000’s of units before your R/D costs are covered and economy of scale kicks in.
The fact is SMPS artifacts are almost all far above the audio band and cause no problem. A properly applied SMPS will have PFC which along with very good regulation is a dramatic sonic improvement. Due to far less leakage inductance causing line frequency artifacts to corrupt the grounds they are far quieter in this respect as well. Even cheap surplus units can be tweaked to sound excellent
The bottom line is that I have switched over to SM power for all my new projects because they sound better with it not because it is cheaper or more efficient.
Roger
Even if you try a SMPS with something like the AMP3 and don't like the sound you can increase the output voltage of the SMPS and drop it back down to the desired voltage using a LDO regulator with ripple rejection around the switching frequency of the SMPS. I haven't tried it yet, but I'd be willing to bet that it would sound good and be pretty compact compared to conventional linear supply. I do have an orphaned 16V IBM laptop SMPS and working AMP3 that I'd like to try this with in the near future 
sx881663 said:
They are only sonically better if you are comparing apples to oranges, say a smps with a lot of output capacitance (which of course is relatively, necessary compared to a linear) but not as much capacitance on the linear- too many people just follow reference circut which has (what, 10uF Tant or 100uF 'lytic usually?). That would make the linear higher impedance and so it wasn't a fair contest. Otherwise, any sonic differences should be noise from the SMPS distorting the audio. If you like that, it's a personal preference to deviate from clean audio.
The top designers are using SMPS for the reasons already mentioned, that Class-D suffers far fewer ill effects from the dirtier power.
"it really is far more expensive to design and implement a proper SMPS".
No, definitely not. SMPS tech is mature, there are sufficient designs out there and even if it took 300 man-hours to do one (over a linear, either way the supply type chosen has to be "created" and implemented), one has to look at the cost savings over a large transformer, heatsink, larger chasiss and potential need for more cooling- just depends on power requirement. It's a cost-savings over many units, not just one.
"The fact is SMPS artifacts are almost all far above the audio band and cause no problem." The switching frequency is in fact (usually) above audible freq. but lower-freq. noise always exists too. It's easy to hook up an amp with less PSRR and hear this noise, there is no question about it. The remaining issue is whether that noise (SNR) is enough to be audible on the target amp.
"A properly applied SMPS will have PFC which along with very good regulation is a dramatic sonic improvement."
No, PFC has nothing to do with SMPS noise, at most it is a bit of a filter for AC noise, yet has zero effect on switching artifacts themselves. Take a theoretically perfect incoming power supply and switch it, and the output will be audibly noisey with some amps- other amps merely have more immunity to it.
There is no additional corruption of ground by a linear. SMPS have all the same issues and most importantly, a relatively larger noise introduction. There is no such thing as "good regulation" that solves this, only through additional post-switcher filter circuits will there be the improvement, and THEN we would have to argue that one can implement the same post-regulation filter circuit to a linear and it'd still be cleaner.
It is known industry wide in most trades that there is no switcher as clean as a linear built with similar attention to noise filtering. Even then, the linear needs far less (if any) such additional filtering. "Sound better" is a bit of an audiophile cop-out, because it is measurably inferior to use a SMPS, with the final analysis being that it's akin to someone feeling their amp sounded better using aluminum wire and carbon resistors even though they're known to color the result if they make any difference.
Smps
!
You sound like a reasonably intelligent person. Are you expounding all this from a hearsay stand point or have you actually tried to implement one and not liked it?
The rather simple switcher I am using in my low level designs replaced a very worthy linear supply that I designed. This was a mostly discrete design with an opamp for an error signal generator. It measures less than 3 micro volts of total noise and has exemplary line and load regulation. Using small toroidal transformers it has no noticeable ground loop problems. It gave the preamp a nice clean and open sound.
Doing a couple of SM units with upgraded parts and following them with a rather large buffer cap improved my sound staging with a deeper back wall and everything more in focus. This can only mean that the noise that matters is reduced! There can be no other explanation.
You would have to ask Bruno why he chose to use SM power with the amp he designed for a high end company that wanted his best effort. I know the reason I will be using them is because an amp will always sound better with a better supply. This stands true if the “better” is regulation, noise or both. The supply must be buffered with adequate storage caps, same as for a linear. If it isn’t, yes, you will hear added and irritating noise.
As for the SMPS design being mature, well yes and no. It is not mature for powering high power amps with the wildly varying loads they cause. This would shut down or even destroy the older designs. There also is the issue of independent regulation for both rails. Add in the cost of low volume custom magnetics and it will indeed be far more expensive. I can say this from experience! It would only be cheaper if you could use some off the shelf product and even then it would be close. The economy of scale you refer to requires thousands of units being sold and this plain don’t happen with most high end companies.
PFC would seem to have no direct bearing on sound but it does! This is because of the power line distortion that conventionally powered amps cause. This will in turn mess up all the rest of your equipment. Also why conventional amps sound better when plugged into their own separate circuit. Not because of the amps sounding better but because of the rest of the equipment liking cleaner power. This also is hard won knowledge. The PFC circuit also will greatly improve a SMPS’s performance as it effectively adds another layer of regulation/isolation.
As for the end result measuring inferior to the use of a linear, this would only be true if the unit was not implemented properly and you were also measuring inaudible noise. I have had no trouble equaling or exceeding my original linear powered designs. A class D amp also will measure inferior if you don’t bandwidth limit the measurement yet you wouldn’t be here if you didn’t think they had some merit.
Roger
!
You sound like a reasonably intelligent person. Are you expounding all this from a hearsay stand point or have you actually tried to implement one and not liked it?
The rather simple switcher I am using in my low level designs replaced a very worthy linear supply that I designed. This was a mostly discrete design with an opamp for an error signal generator. It measures less than 3 micro volts of total noise and has exemplary line and load regulation. Using small toroidal transformers it has no noticeable ground loop problems. It gave the preamp a nice clean and open sound.
Doing a couple of SM units with upgraded parts and following them with a rather large buffer cap improved my sound staging with a deeper back wall and everything more in focus. This can only mean that the noise that matters is reduced! There can be no other explanation.
You would have to ask Bruno why he chose to use SM power with the amp he designed for a high end company that wanted his best effort. I know the reason I will be using them is because an amp will always sound better with a better supply. This stands true if the “better” is regulation, noise or both. The supply must be buffered with adequate storage caps, same as for a linear. If it isn’t, yes, you will hear added and irritating noise.
As for the SMPS design being mature, well yes and no. It is not mature for powering high power amps with the wildly varying loads they cause. This would shut down or even destroy the older designs. There also is the issue of independent regulation for both rails. Add in the cost of low volume custom magnetics and it will indeed be far more expensive. I can say this from experience! It would only be cheaper if you could use some off the shelf product and even then it would be close. The economy of scale you refer to requires thousands of units being sold and this plain don’t happen with most high end companies.
PFC would seem to have no direct bearing on sound but it does! This is because of the power line distortion that conventionally powered amps cause. This will in turn mess up all the rest of your equipment. Also why conventional amps sound better when plugged into their own separate circuit. Not because of the amps sounding better but because of the rest of the equipment liking cleaner power. This also is hard won knowledge. The PFC circuit also will greatly improve a SMPS’s performance as it effectively adds another layer of regulation/isolation.
As for the end result measuring inferior to the use of a linear, this would only be true if the unit was not implemented properly and you were also measuring inaudible noise. I have had no trouble equaling or exceeding my original linear powered designs. A class D amp also will measure inferior if you don’t bandwidth limit the measurement yet you wouldn’t be here if you didn’t think they had some merit.
Roger
Not just hearsay, it's a fact that they don't regulate as well. The output caps in a SMPS are an absolute necessity because it's output is horrific without them. Take same caps and add to a linear that didn't even need but ~ 10uF and it's clear that these is more noise in SMPS.
Then there must be differntiation between the needs of different amps. If an amp's local capacitance isn't sufficient for it's current swings, (which is a poor design in itself, IMO) then even more it becomes dependant on impedance of the (external) power supply. In that case and with "some" amps we could see a switcher more noisey, BUT if it had more rear-end capacitance it could reduce the amp's induced ripple. Still a linear with similar capacitance will be every bit as good in that respect with with less noise.
Engineers do often design as good as possibe- then the penny-pinchers get ahold of the budget to implement and plans get revised. Soon enough an engineer knows what kind of designs will get rejected and won't waste time doing what's theoretically best but rather best within-limits-thrust-upon-them. Even high-end gear must be profitable to manufacture, as much so as possible that a market will tolerate.
Most companies can easily recoup costs saved by implementing a switcher. If it saves $5 per unit and they sell 3000 (very low figure), what have they lost by having salaried employees design or test it?
PFC only helps within the context of having a noisey supply. A suitably sized transformer on a linear serves same goal by default.
There is no way to implement a switcher to the extent that it won't have meaurably inferior output. No matter what additional filtration you put on it's output, same additional filtration would make a linear even that much cleaner still. Linear starts out cleaner and no matter what more is done to a switcher, it's still same situation, even a PFC circuit could be put in front of a linear if you really wanted one. Linear after a smps + filter can be the best of both worlds providing there is still sufficiently low impedance output.
Remember that "audible noise" spectrum does not necessarily apply, modern amps are operating at above-audio freq. even if output is an audible frequency. Noise at higher frequency will effect their operation to a degree determined by the type & particulars of that amp. Again, it just happens that Class-D amps are more immune to noise than most (all?) older designs.
Then there must be differntiation between the needs of different amps. If an amp's local capacitance isn't sufficient for it's current swings, (which is a poor design in itself, IMO) then even more it becomes dependant on impedance of the (external) power supply. In that case and with "some" amps we could see a switcher more noisey, BUT if it had more rear-end capacitance it could reduce the amp's induced ripple. Still a linear with similar capacitance will be every bit as good in that respect with with less noise.
Engineers do often design as good as possibe- then the penny-pinchers get ahold of the budget to implement and plans get revised. Soon enough an engineer knows what kind of designs will get rejected and won't waste time doing what's theoretically best but rather best within-limits-thrust-upon-them. Even high-end gear must be profitable to manufacture, as much so as possible that a market will tolerate.
Most companies can easily recoup costs saved by implementing a switcher. If it saves $5 per unit and they sell 3000 (very low figure), what have they lost by having salaried employees design or test it?
PFC only helps within the context of having a noisey supply. A suitably sized transformer on a linear serves same goal by default.
There is no way to implement a switcher to the extent that it won't have meaurably inferior output. No matter what additional filtration you put on it's output, same additional filtration would make a linear even that much cleaner still. Linear starts out cleaner and no matter what more is done to a switcher, it's still same situation, even a PFC circuit could be put in front of a linear if you really wanted one. Linear after a smps + filter can be the best of both worlds providing there is still sufficiently low impedance output.
Remember that "audible noise" spectrum does not necessarily apply, modern amps are operating at above-audio freq. even if output is an audible frequency. Noise at higher frequency will effect their operation to a degree determined by the type & particulars of that amp. Again, it just happens that Class-D amps are more immune to noise than most (all?) older designs.
!
One more example and I will let it rest, agreeing to disagree. A customer wanted his old Bose 901 EQ upgraded. The original unit had a lot of residual noise including line freq artifacts. Upgrading most of the parts and regulating the power helped a lot but it still was too noisy. The next step was putting in a quality toroidal transformer. This also helped but still was unacceptable as it needed to be in the same case. The final solution was the input regulator was removed and a small switcher was installed feeding the circuit directly with the same filter caps left in. This solved the problem almost completely. It was now perfectly acceptable to the customer. This particular circuit is single ended and high impedance to boot. It is extremely sensitive to all types of noise. The SMPS was the only solution outside of a remote supply that would work.
Roger
One more example and I will let it rest, agreeing to disagree. A customer wanted his old Bose 901 EQ upgraded. The original unit had a lot of residual noise including line freq artifacts. Upgrading most of the parts and regulating the power helped a lot but it still was too noisy. The next step was putting in a quality toroidal transformer. This also helped but still was unacceptable as it needed to be in the same case. The final solution was the input regulator was removed and a small switcher was installed feeding the circuit directly with the same filter caps left in. This solved the problem almost completely. It was now perfectly acceptable to the customer. This particular circuit is single ended and high impedance to boot. It is extremely sensitive to all types of noise. The SMPS was the only solution outside of a remote supply that would work.
Roger
There are too few specifics to draw a conclusion about a particular application, IMO, but a complete consideration of where the noise came in and what type of filter was needed, would be necessary.
IMO, if there was noise with a linear it wasn't because it was a linear rather than a switcher, rather that whatever filtration was necessary to remove that noise, happened to be incorporated into the switcher above and beyond the bare-bones referect design of it. Same goes for a linear. Since we had a situation where the incoming power to the supply was probably too noisey, a typical line filter in front of the linear would've likely been sufficient. Apples and oranges.
IMO, if there was noise with a linear it wasn't because it was a linear rather than a switcher, rather that whatever filtration was necessary to remove that noise, happened to be incorporated into the switcher above and beyond the bare-bones referect design of it. Same goes for a linear. Since we had a situation where the incoming power to the supply was probably too noisey, a typical line filter in front of the linear would've likely been sufficient. Apples and oranges.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.