|
Post by jimbo on Jan 23, 2021 10:00:57 GMT
Our friends at audiosciencereview think not??
|
|
|
Post by jimbo on Jan 23, 2021 10:15:27 GMT
|
|
|
Post by sq225917 on Jan 23, 2021 10:54:12 GMT
The straight answer is that there's no reason why a smps can't be as good as a linear supply. There's lots of reasons why they aren't as good though, usually cost cutting and poor design.
Some older gear just isn't designed to work well with smps, but the gear that is can be as good as it gets.
That said, a good linear supply is usually a safer bet.
The smsl400 uses dual smps to supply the whole dac, the gustard a18 uses linear, both measure about as good as it gets.
Beware generalisations.
|
|
|
Post by misterc on Jan 23, 2021 11:53:06 GMT
This is very straightforward this is about size/space and cost.
SMPS use much higher frequencies to generate the power from 50Khz+ (That is from 50 thousand cycles per second)as opposed to a linear PSU which uses either 50/60Hz which is why you need a larger transformer and storage/smoothing caps for the rectified dc.
A SMPS can generate much higher power density than a linear one, smaller footprint and costs. However they are noisier without question, its just the way the work by switching FET's on and off hard, this generates a lot of RF, power droop and output ripple which needs to be addressed.
Linear power supplies are more costly, use more space, however are much quieter, far lass drop under load are more suitable for low noise and quieter circuit requirements, for bigger power ratings that SWPS are more suitable.
We use bench test power supplies up top 30 amps DC in linear and between 30 and 300 amps in SMPS for our design work, Kikusui manufacturer some of the very best power supplies and loads for labs, educational and research establishments.
When designing smps a great many more variables are required to complete the task correctly, with Control Loop Response Analysis, Power device analysis, Power Device Losses, Safe Operating Area, B-H Hysteresis Curves, Control Loop Analysis, Line Power Analysis.
Also remember that SMPS operate at much higher voltages commonly over 340Vdc, generally have smaller smoothing caps due to the switching frequency. Where as say a 500Va tx 12v 10amp supply would require at least 100Kuf of capacitance to keep the reliable constant full bridge rectified DC output.
Then when you have finished your smps you will need to integrated into your power distribution network (all smps in quality electronics are controlled by PDN) these parameters to be taken in to consideration are Digital Multi-phase Power Management ICs (PMIC) Voltage Regulator Modules (VRM) Point-of-load (POL) switching regulators Low-dropout (LDO) regulators Half-bridge DC-DC converters Unregulated DC supplies Embedded system power rail debug and analysis Embedded system power delivery network (PDN) validation.
With the following traits needing to be addressed Per-cycle parameters, Per-cycle Waveforms, Ripple, Ringing, Overshoot, Droop Noise, Settling Time.
Some of the SMPS chips and control systems with the latest dc<>dc converters are capable of of producing ripple under low lower DUT conditions of around 30uv and some of the new IOT devices are required to operate at extremely low 'off power states' around 8/9 picoamps at stand by. But these are at very low voltages and current demands.
A while to go yet before I fit smps to any of my audio deigns, my non audio deigns are a different matter!
However I have no question that a correctly designed linear psu with deliver far better sonic results than a smps currently.
|
|
|
Post by stevew on Jan 23, 2021 13:26:50 GMT
Thanks for that Tony. Always good to hear a solid technical or scientific reason for things as opposed to ‘opinions’. Never regretted dropping a linear power supply in when I could, and it always seemed much better to me. Mind you, usually replacing cheapo power supplies.. but still.
|
|
|
Post by sq225917 on Jan 23, 2021 14:45:29 GMT
Quite how would a linear supply improve on the performance of the mola mola dac for example?
|
|
|
Post by jimbo on Jan 23, 2021 16:18:33 GMT
Quite how would a linear supply improve on the performance of the mola mola dac for example? Or DAVE which contains a SMPS?
|
|
|
Post by stevew on Jan 23, 2021 16:35:44 GMT
Not sure I understand Surely in both those instances the manufacturer has made the decision. The only choice you have then is what iec 240 v power cord you want to put in. I was surprised myself that Lyngdorf have decided on switching power for their new amps....unlike the linear one in my old Tdai 2200... but I’m told they sound pretty good.
|
|
|
Post by misterc on Jan 23, 2021 17:04:41 GMT
Folks If you have the space and budget you can achieve a great deal with a linear psu. With Chord electronics Jimbo the the very first Dac64's (The Dave just a pimped up version with 1 million taps LOL) had a small torriod tx in that clam shell case (the first 50 or so I believe back in 2002?) it did sound much less Chord like and more enjoyable than the later SMPS version, I know I still have one. Due to the case size (and Budget) a smps was required. Over the years I have built linear psu's for Dac64, (one using a empty Dac64 case we machined out to fit, customer was fastidious for matching casework)significant improvement reported by the customers who chose that route. Interesting that Chord have reported to be 'uninterested' in warranty claims if one of their products had an after-market PSU fitted (Possibly MCRU could fill the blanks in there). Also they approached a well known industry figure to design a linear power supply for them last year. The Hugo TT responds very well to a linear power supply, although it is rated at 4amps @ 15 volts, (if using the inbuilt pre amp)I fact it transforms them into quite listenable devices. The mola mola dac Simon, no thank you, I would rather listen to a naim nait3 having listened to one not so long ago., Seriously its ultra fi at best. Interesting how the ginger whinger stopped recommending them and then decided a £600 dac was better?, not sure I agree with that sentiment to be fair. Also peppering a design with LT3045's in parallel (if you require more than 500ma current) doesn't automatically give you low noise figures, it depends on the supplying rails, load, board layout, veeres, proximity to other delicate i/c' etc. As with all quality designs implementation is the key. In fact we do not use any of the LT3042/45 in our designs, they are other ways of skinning a cat I am happy to perform a full spread of measurements on our R&S UPV a dac before and after a linear psu is fitted but also on a RT spectrum analyser to look at those 'pesky' above 200Khz FFT's just to get a fuller picture we have real time bandwidth capabilities of 300Mhz. You might be surprised. Also If I was just making quick ripple noise measurements with a DMM I would be using one of these as a bear minimum.
Though fort long term dc power supply work these are superb, the act as a source/sink and scope and data logger plus arb (arbitrator wave form generator) generator so you can programme your own start up current/ recovery plots/droop recovery etc. Tie into the dedicated PC analysis software and it seriously cuts your design time by a significant amount.
|
|
|
Post by sq225917 on Jan 23, 2021 19:18:52 GMT
The lt3045 is a curious beast, great numbers, when dropping more than 2v.....
While it's great to have test equipment and to be able to measure milifarts at megahertz, where's the wealth of evidence to show the audibility of the same in the output?
|
|
|
Post by jimbo on Jan 23, 2021 20:26:17 GMT
Folks If you have the space and budget you can achieve a great deal with a linear psu. With Chord electronics Jimbo the the very first Dac64's (The Dave just a pimped up version with 1 million taps LOL) had a small torriod tx in that clam shell case (the first 50 or so I believe back in 2002?) it did sound much less Chord like and more enjoyable than the later SMPS version, I know I still have one. Due to the case size (and Budget) a smps was required. Over the years I have built linear psu's for Dac64, (one using a empty Dac64 case we machined out to fit, customer was fastidious for matching casework)significant improvement reported by the customers who chose that route. Interesting that Chord have reported to be 'uninterested' in warranty claims if one of their products had an after-market PSU fitted (Possibly MCRU could fill the blanks in there). Also they approached a well known industry figure to design a linear power supply for them last year. The Hugo TT responds very well to a linear power supply, although it is rated at 4amps @ 15 volts, (if using the inbuilt pre amp)I fact it transforms them into quite listenable devices. The mola mola dac Simon, no thank you, I would rather listen to a naim nait3 having listened to one not so long ago., Seriously its ultra fi at best. Interesting how the ginger whinger stopped recommending them and then decided a £600 dac was better?, not sure I agree with that sentiment to be fair. Also peppering a design with LT3045's in parallel (if you require more than 500ma current) doesn't automatically give you low noise figures, it depends on the supplying rails, load, board layout, veeres, proximity to other delicate i/c' etc. As with all quality designs implementation is the key. In fact we do not use any of the LT3042/45 in our designs, they are other ways of skinning a cat I am happy to perform a full spread of measurements on our R&S UPV a dac before and after a linear psu is fitted but also on a RT spectrum analyser to look at those 'pesky' above 200Khz FFT's just to get a fuller picture we have real time bandwidth capabilities of 300Mhz. You might be surprised. Also If I was just making quick ripple noise measurements with a DMM I would be using one of these as a bear minimum.
Though fort long term dc power supply work these are superb, the act as a source/sink and scope and data logger plus arb (arbitrator wave form generator) generator so you can programme your own start up current/ recovery plots/droop recovery etc. Tie into the dedicated PC analysis software and it seriously cuts your design time by a significant amount.
I have heard a few folk say they liked the DAC64 and actually preferred it to later Chord DACs such as the Qutest.
|
|