I have plenty of switching power supply units. It seems that they have noise issues. But my computer is silent and can produce good sound. And so do newer commercial DVD players. So what is the real problem here?
Recently I took out a switching power supply which is also used to power up computer monitor. I guess the voltage and current is suitable for at least providing B+ for a tube preamp. Anybody know if this is possible or not?
Recently I took out a switching power supply which is also used to power up computer monitor. I guess the voltage and current is suitable for at least providing B+ for a tube preamp. Anybody know if this is possible or not?
Jay,
SMPS and so called Iron PSU's all get built for a cost which means there either good ok or rubbish, that said I have a pre-amp which runs from an smps and it is very clean. I think it all depends what your preference is and what you can get.
There is no reason a well designed SMPS can't be as good as a conventional PSU. Having said that I wouldn't say that computer smps's are always good examples of state of the art electronics.
But I have heard of people modifying them to supply other voltage/current then what they were designed to.
I think it really becomes a personal taste aspect and what your comfortable wit.
Robert
SMPS and so called Iron PSU's all get built for a cost which means there either good ok or rubbish, that said I have a pre-amp which runs from an smps and it is very clean. I think it all depends what your preference is and what you can get.
There is no reason a well designed SMPS can't be as good as a conventional PSU. Having said that I wouldn't say that computer smps's are always good examples of state of the art electronics.
But I have heard of people modifying them to supply other voltage/current then what they were designed to.
I think it really becomes a personal taste aspect and what your comfortable wit.
Robert
I did some R&D on SMPS for automotive tube audio use and found it as good as a SS bench linear supply.
The only think to remember is careful shielding and proper bypassing. I double-shielded (with gap to avoid a ground-loop), looped the wire a couple of times through an RFI toroid and finally out through feedthrough capacitors.
The only think to remember is careful shielding and proper bypassing. I double-shielded (with gap to avoid a ground-loop), looped the wire a couple of times through an RFI toroid and finally out through feedthrough capacitors.
.do u have 'audiophile' ears?30khz is inaudible to humans!
rascal101 said:
This is a bit of a sweeping statement, aka generalisation.
SMPSs, like all PSUs, are built to spec. Most digital systems can tolerate a certain amount of noise at 50kHz etc, but not so for audio. So, audio SMPSs are heavily filtered and are run at non-audio frequencies. I have built many many SMPSs, a couple of which I use for valve audio use. These are not hack designs, but are high-end systems.
Basically, for the average person, its far easier to build a quiet non-SMPS that an SMPS system - SMPSs are tricky to get right. However, if you do get it right, they are cheap, small, light, highly efficient and very effective. Noise is barely measurable on a good system.
When I use an SMPS for audio use, I use good filters on both input and output e.g. MuRata BNP002-03 and encase the SMPS in a shielding box (as used for small RF projects - you can also use a box made from soldered pieces of single or double sided PCB - another Ham Radio trick).
Nick
I design SMPS for a living and yes if you use the standard run of the mill designs used in the last 20 years, you will not get good performance in audio...
I have designed switchers that switch as high as 2 Mhz and and produce little noise...
ALso , the use of quasi-resonant/resonant SMPS circuits are very low with EMI ....
There is a place for SMPS in audio, I would recomend switching above 450kHz and also using a resonant topology...
Chris
I have designed switchers that switch as high as 2 Mhz and and produce little noise...
ALso , the use of quasi-resonant/resonant SMPS circuits are very low with EMI ....
There is a place for SMPS in audio, I would recomend switching above 450kHz and also using a resonant topology...
Chris
Hi Chris,
I guess you are familiar with the Vicor units, which I think are quasi-resonant (or zero-current switching). I have used a few (the 150W 2nd generation units) in audio designs with good results, but that may have been luck
.
What is your opinion on the Vicors for audio use? They also have a RAM (Ripple Attenuation Module) available to further decrease output noise by some 20dB, but the final price gets steep so I am not sure it would be worthwhile.
Jan Didden
I guess you are familiar with the Vicor units, which I think are quasi-resonant (or zero-current switching). I have used a few (the 150W 2nd generation units) in audio designs with good results, but that may have been luck
.What is your opinion on the Vicors for audio use? They also have a RAM (Ripple Attenuation Module) available to further decrease output noise by some 20dB, but the final price gets steep so I am not sure it would be worthwhile.
Jan Didden
Chris, further to Jan's post, can you name a few brands of switchers you think give good performance for audio use? Would much appreciate your opinion. I've been looking on the internet, but cannot seem to latch onto proper figures of merit to guide my search.
Any opinions on Tamura switchers? For instance, this one.
Any opinions on Tamura switchers? For instance, this one.
I currently design semiconductor ATE for a living, and due to space and power consumption constraints mostly use SMPS for my raw dc supplies. Simple external filters using a single common mode choke, a couple of inductors and some low ESR capacitors can knock down conducted noise by several orders of magnitude or more.
For suitable chokes and inductors take a look at http://www.coilcraft.com
You can also use high performance linears after the SMPS for somewhat improved noise performance and to define a low source impedance at audio frequencies. (Note most linears do not have enough loop bandwidth to effectively remove noise above 100kHz. Linear Tech has several specified at 100kHz that do work well.)
Noise levels of much less than 50uVrms over a 100kHz bw are achievable with some care using the above techniques.
Shielding to reduce electrostatic coupling is also very important. Look at some of the other posts for more on this.
ATE applications often require measurement accuracies of 10ppm or better, including the noise contribution. No audio device is even close.
Kevin
Edit: Link is correct URL but does not seem to work correctly in firefox, copy and paste into your browser address bar if the link is misdirected. Might be some problem with the webserver as well.
For suitable chokes and inductors take a look at http://www.coilcraft.com
You can also use high performance linears after the SMPS for somewhat improved noise performance and to define a low source impedance at audio frequencies. (Note most linears do not have enough loop bandwidth to effectively remove noise above 100kHz. Linear Tech has several specified at 100kHz that do work well.)
Noise levels of much less than 50uVrms over a 100kHz bw are achievable with some care using the above techniques.
Shielding to reduce electrostatic coupling is also very important. Look at some of the other posts for more on this.
ATE applications often require measurement accuracies of 10ppm or better, including the noise contribution. No audio device is even close.
Kevin
Edit: Link is correct URL but does not seem to work correctly in firefox, copy and paste into your browser address bar if the link is misdirected. Might be some problem with the webserver as well.
kevinkr said:
Great post. I once designed a preamp for a capacitor microphone (Bruel & Kjaer) which was portable, so necessitated the use of SMPS to get all the required voltages. An ordinary flyback was used, the real stuff is in careful control of all current paths, and proper magnetic shielding - very much like what you describe. It proved much more troublesome to provide amp noise performance that was lower than the SMPS injected noise. Anyone who knows what these microphones are capable of, knows what I mean.
I didn't make a sweeping statement when I said that SMPS do not have a home in high end audio. This is due to my experience working with switchers for the past 8 years. Ripple, & Noise and conducted/radiated emi (not audible yes but imagine the noise from the switcher coupling to the audio electronics - you need very good shielding).
Have also done auditions on vintage amplifiers using switchers.
Have also done auditions on vintage amplifiers using switchers.
I'm not sure ripple and EMI are the most important factors for judging the sonic worth of a power supply. I might speculate that a power supply's ability to track changes in load current without generating spurious voltages in the process + a power supply's ability to reject mains grunge might be more important. I'm running a pair of Charlize amps via a cheap $25 switcher, and it sounds very good. Makes me wonder what a good switcher could do.
The highly rated and lovely sounding Emm Labs digital gear uses switchers.
The highly rated and lovely sounding Emm Labs digital gear uses switchers.
I believe you are referring to transient response. SMPS should have better transient response than linear power supplies.
Speaking of ripple and noise I still have to see ripple and noise figures in the 1mV range for SMPS.
Speaking of sound, this is purely subjective. If you don't hear it then SMPS is good for you.
Speaking of ripple and noise I still have to see ripple and noise figures in the 1mV range for SMPS.
Speaking of sound, this is purely subjective. If you don't hear it then SMPS is good for you.
Sir Serengetiplains,
Transient response refers to the response of the SMPS PSU to load changes eg from 10% load to 50% load or from 90% load to 40% load. Typically SMPS companies design power supplies for a step load of 50%. Power supply must maintain regulation and get back to nominal output within a specified settling time.
Transient response is a measure of the responsiveness of the feedback loop and also gives you an idea of the phase margin - stability.
What we do to minimize noise from the FET switch is minimize the loop area, use a star connection for the ground, avoid low voltage PCB tracks passing through high voltage tracks, capacitor layout at edge of tracks to ensure filter action, use copper shield at transformer etc. Despite this you still have EMI (albeit passing CISPR or FCC requirements).
To offset this for audio application is to isolate SMPS from the audio electronics itself and to use a lot of low esr e-caps to minimize ripple & noise.
Despite this I still do not want any SMPS for audio application. The effort and time required to equal the quietness of a linear power supply is not worth it. It is quite possible but it will very very difficult using an ordinary SMPS topology (eg flyback, forward, 2 FET forward - you would have to go half bridge or full bridge).
Transient response refers to the response of the SMPS PSU to load changes eg from 10% load to 50% load or from 90% load to 40% load. Typically SMPS companies design power supplies for a step load of 50%. Power supply must maintain regulation and get back to nominal output within a specified settling time.
Transient response is a measure of the responsiveness of the feedback loop and also gives you an idea of the phase margin - stability.
What we do to minimize noise from the FET switch is minimize the loop area, use a star connection for the ground, avoid low voltage PCB tracks passing through high voltage tracks, capacitor layout at edge of tracks to ensure filter action, use copper shield at transformer etc. Despite this you still have EMI (albeit passing CISPR or FCC requirements).
To offset this for audio application is to isolate SMPS from the audio electronics itself and to use a lot of low esr e-caps to minimize ripple & noise.
Despite this I still do not want any SMPS for audio application. The effort and time required to equal the quietness of a linear power supply is not worth it. It is quite possible but it will very very difficult using an ordinary SMPS topology (eg flyback, forward, 2 FET forward - you would have to go half bridge or full bridge).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.