mathtracks
Active member
- Joined
- Nov 14, 2004
- Messages
- 39
What are the disadvantages/advantages of a switched power supply vs a linear power supply?Linear is better for an audio purpose.right?
Switching vs. Linear power supply
- Thread starter mathtracks
- Start date
Help Support GroupDIY Audio Forum:
BYacey
Well-known member
Linear is much cleaner due to only having to filter low frequency, and the linear action of the regulator. Switching regulators are by nature noisy, and generate lots of upper harmonics (hash) from the abrubt switching of the inductor. Doesn't seem to matter how well this is filtered, it's never as quiet as a linear.
strangeandbouncy
Well-known member
hello there!
if you are refering to power amps, switch mode power amps suck! you need seriously overated/ huge-current delivering mofo's, not flimsy "lite" crap. Current delivery equals slam.
ps try a f!!k-off power amp on your near-fields if you don't believe me! oh. and you'll lose less tweeters to boot, since your amp will clip less(clipping is effectively delivering dc to your tweeters . . . . . )
turn it up!
AndyP
ps I dont know of any serious (Analogue) outboard with a switch mode psu. Plenty of nasty digital stuff tho' - they are often blowing up in my experience, and often have wierd "digital" noise problems to boot - ho hum
if you are refering to power amps, switch mode power amps suck! you need seriously overated/ huge-current delivering mofo's, not flimsy "lite" crap. Current delivery equals slam.
ps try a f!!k-off power amp on your near-fields if you don't believe me! oh. and you'll lose less tweeters to boot, since your amp will clip less(clipping is effectively delivering dc to your tweeters . . . . . )
turn it up!
AndyP
ps I dont know of any serious (Analogue) outboard with a switch mode psu. Plenty of nasty digital stuff tho' - they are often blowing up in my experience, and often have wierd "digital" noise problems to boot - ho hum
SonsOfThunder
Well-known member
[quote author="BYacey"]Linear is much cleaner due to only having to filter low frequency, and the linear action of the regulator. Switching regulators are by nature noisy, and generate lots of upper harmonics (hash) from the abrubt switching of the inductor. Doesn't seem to matter how well this is filtered, it's never as quiet as a linear.[/quote] I hate to disagree with you Bill, but have you ever listened to a high power (>50W per channel) car amplifier? How much gain would that have and how noisy was it? How much hash did you hear? Every high power car amp has a sw power supply...there is no way around it!! I can show you a 125W per channel amp with a SNR to rated power of about 110dB.
Here is what I would say is the problem...poorly designed switching supplies! I have seen people build supplies for converting Pro EQs and processors for competition vehicles using the little switching bricks...and they are NASTY! I took an existing sw supply from another product that is made by the co I work for at my day job and was able to increase the SNR on one EQ setup by about 20dB. The trick is to get the sw freq up around 100kHz, make sure that you don't have spikes (usually caused by switching the transistors off too fast) and then use a proper 2-pole filter. This will usually make a nice quiet supply that is more efficient than a plain old linear regulated supply because the "extra" power doesn't go up as heat.
There are more and more consumer electronics with sw supplies in them. Yes, some of them are not so well done and cause interference problems, but i don't think its fair to lump them all in one basket. There are some Pro products with sw regulators in them too...one that I know of: dbx Driverack 480.
HTH...
Here is what I would say is the problem...poorly designed switching supplies! I have seen people build supplies for converting Pro EQs and processors for competition vehicles using the little switching bricks...and they are NASTY! I took an existing sw supply from another product that is made by the co I work for at my day job and was able to increase the SNR on one EQ setup by about 20dB. The trick is to get the sw freq up around 100kHz, make sure that you don't have spikes (usually caused by switching the transistors off too fast) and then use a proper 2-pole filter. This will usually make a nice quiet supply that is more efficient than a plain old linear regulated supply because the "extra" power doesn't go up as heat.
There are more and more consumer electronics with sw supplies in them. Yes, some of them are not so well done and cause interference problems, but i don't think its fair to lump them all in one basket. There are some Pro products with sw regulators in them too...one that I know of: dbx Driverack 480.
HTH...
BYacey
Well-known member
When I am speaking about noise, I am talking about low noise preamp circuits etc. I would never consider using a switching supply for an application like this. For brute force power amps and the like, where current gain is more the norm than voltage gain, switchers are more efficient and shine in this application.
dale116dot7
Well-known member
A well-designed switcher can be quiet, but it takes some effort to do it. Good layout, attention to grounding in the whole system, and properly applied filter capacitors is basically what it takes, along with a good knowledge of control loop stability, filtering, and parasitics. Even applying an off-the-shelf switcher requires some amount of work in zapping RF interference problems. A good 'scope (100 MHz or higher) and really good probes (proper RF coupling instead of the long ground lead) go a long way here.
A common thing to do these days with digital mixers and that sort of stuff is to supply the DSP core and the fader motors with a big switching power supply, and the analog and converters with either an independant linear supply, or a linear supply derived from the switcher (a post-regulator), and that cleans things up a lot.
And it can be deceiving, but a linear regulator can be noisy too - if they decide to oscillate they'll put out more noise than any switcher.
One thing about a switcher - if a feedback loop part breaks, it wil put out basically as much power as it can, frying everything in sight. If a linear regulator breaks, it is likely the series pass transistor, and if it shorts it will also blow up your board. The stressed part on a linear regulator is also its worst failure mode. A switcher's most stressed parts (diodes and power transistors) tend to cause a blown fuse on a blow-up, and your circuit just stops working.
All that being said, my Tascam DM24 has an RF problem. Adding a small capacitor in one part of the circuit fixed it, but it took me about two hours to find and fix it - and I knew what sort of problem I was looking for because that was the type of noise generated.
A common thing to do these days with digital mixers and that sort of stuff is to supply the DSP core and the fader motors with a big switching power supply, and the analog and converters with either an independant linear supply, or a linear supply derived from the switcher (a post-regulator), and that cleans things up a lot.
And it can be deceiving, but a linear regulator can be noisy too - if they decide to oscillate they'll put out more noise than any switcher.
One thing about a switcher - if a feedback loop part breaks, it wil put out basically as much power as it can, frying everything in sight. If a linear regulator breaks, it is likely the series pass transistor, and if it shorts it will also blow up your board. The stressed part on a linear regulator is also its worst failure mode. A switcher's most stressed parts (diodes and power transistors) tend to cause a blown fuse on a blow-up, and your circuit just stops working.
All that being said, my Tascam DM24 has an RF problem. Adding a small capacitor in one part of the circuit fixed it, but it took me about two hours to find and fix it - and I knew what sort of problem I was looking for because that was the type of noise generated.
mathtracks
Active member
- Joined
- Nov 14, 2004
- Messages
- 39
I have some power-one switched supplies(map80-4003) that have a +/-15v output.if i put a post linear regulator and maybe some filter caps(recommend some values please) will this be a clean enough source to run a mic pre from?i don't own an o-scope so i really can't experiment too scientifically with this suggestion.
From repairing both switchers and linears I have noted some things.
Switchers seem to have a shorter life time. When they start to go bad you need to fix them right away. Most common problem is the electro caps. LOW ESR caps are for switchers. Alot of the design is based on storing energy in the inductors and/or caps when the caps start to go bad the ESR goes up causing more self heating then if they are junk they leak etc.
For old timers, kind of the same as the flyback cap in a flyback circuit going bad if you got it when it first started going bad you could often save the flyback transformer.
If you change the caps in time you might be able to fix it: however if you let it go to long the inductors and semiconductor start to go bad from stress and a cap replacement then is a waste of time TRY to find replacement inductors for a supply that is a few years old.
FWIW I tend to recap a new PC power supply because most of the caps used are crap IMO. This is even with some name brands
A linear supply is much more repairable alot, still use the 732 (IRRC) that chip is old but good. Standard type caps etc.
I posted this in the past. I worked on a Laser power supply that used a switcher to reg most of the energy that supplied a linear at a controlled voltage and current for "fine tuning" the reg voltage and current.
Switchers seem to have a shorter life time. When they start to go bad you need to fix them right away. Most common problem is the electro caps. LOW ESR caps are for switchers. Alot of the design is based on storing energy in the inductors and/or caps when the caps start to go bad the ESR goes up causing more self heating then if they are junk they leak etc.
For old timers, kind of the same as the flyback cap in a flyback circuit going bad if you got it when it first started going bad you could often save the flyback transformer.
If you change the caps in time you might be able to fix it: however if you let it go to long the inductors and semiconductor start to go bad from stress and a cap replacement then is a waste of time TRY to find replacement inductors for a supply that is a few years old.
FWIW I tend to recap a new PC power supply because most of the caps used are crap IMO. This is even with some name brands
A linear supply is much more repairable alot, still use the 732 (IRRC) that chip is old but good. Standard type caps etc.
I posted this in the past. I worked on a Laser power supply that used a switcher to reg most of the energy that supplied a linear at a controlled voltage and current for "fine tuning" the reg voltage and current.
SonsOfThunder
Well-known member
[quote author="Gus"]A linear supply is much more repairable alot, still use the 732 (IRRC) that chip is old but good.[/quote]Gus, I think you mean the old LM723. Good chip, though I think we might have found some limitations in our application at work (300A, 14.4V supply)
Good, low ESR caps (usually rated at 105 degrees C) can be needed in a linear supply (depending on the ripple) but are definitely a neccesity in a switcher. Usually the filter is a simple LC. The filter isn't so tricky to design...having a bipolar supply helps because then you can use a common-mode choke for the L's. :thumb:
Power transistors (usually FETs) are usually the weak link in a switcher...and in my observation, more often than not, fail due to spiking (needs a bigger gate resistor) but you have to be careful not to go too big or you will have cross-conduction.
[quote author="BYacey"]For brute force power amps and the like, where current gain is more the norm than voltage gain, switchers are more efficient and shine in this application.[/quote] Bill (I promise I'm not trying to pick a fight with you :thumb: just want sw PS to get a fair shake here)...there is plenty of voltage gain going on in an audio amp that has adjustable gain to 150mV. (150mV = rated output) and this is the gain when the SNR is measured. This PS switches at about 30kHz.
FWIW, I have a design in the works for a sw PS to be used for an Altec 1567 clone (2 channels). It will switch at about 100kHz and I have confidence that it will be quiet!
HTH! Peace!
Good, low ESR caps (usually rated at 105 degrees C) can be needed in a linear supply (depending on the ripple) but are definitely a neccesity in a switcher. Usually the filter is a simple LC. The filter isn't so tricky to design...having a bipolar supply helps because then you can use a common-mode choke for the L's. :thumb:
Power transistors (usually FETs) are usually the weak link in a switcher...and in my observation, more often than not, fail due to spiking (needs a bigger gate resistor) but you have to be careful not to go too big or you will have cross-conduction.
No doubt, good caps cost $$ and those guys are operating on slim margins.
[quote author="BYacey"]For brute force power amps and the like, where current gain is more the norm than voltage gain, switchers are more efficient and shine in this application.[/quote] Bill (I promise I'm not trying to pick a fight with you :thumb: just want sw PS to get a fair shake here)...there is plenty of voltage gain going on in an audio amp that has adjustable gain to 150mV. (150mV = rated output) and this is the gain when the SNR is measured. This PS switches at about 30kHz.
FWIW, I have a design in the works for a sw PS to be used for an Altec 1567 clone (2 channels). It will switch at about 100kHz and I have confidence that it will be quiet!
HTH! Peace!
The switcher in the EQ-2NV is 150ish kHz, the single channel runs at 250K.
They are both very quiet at the inputs and outputs, but took several iterations to get there, not nearly as simple as linear regulators.
But, they also don't generate a pile of heat, fit in a small space, and don't radiate a horrendous 60hz field.
For DIY, switchers probably aren't practical, for commercial products, in my case at least, it's the only way to make the box work.
They are both very quiet at the inputs and outputs, but took several iterations to get there, not nearly as simple as linear regulators.
But, they also don't generate a pile of heat, fit in a small space, and don't radiate a horrendous 60hz field.
For DIY, switchers probably aren't practical, for commercial products, in my case at least, it's the only way to make the box work.
Flatpicker
Well-known member
[quote author="Dan Kennedy"]They are both very quiet at the inputs and outputs, but took several iterations to get there, not nearly as simple as linear regulators...[/quote]Dan,
By this I take it you designed you own switchers? Why couldn?t you use a linear supply - was size an issue? I?m not real familiar with the EQ-2NV, but had always assumed they ran on a linear PS.
By this I take it you designed you own switchers? Why couldn?t you use a linear supply - was size an issue? I?m not real familiar with the EQ-2NV, but had always assumed they ran on a linear PS.
mathtracks
Active member
- Joined
- Nov 14, 2004
- Messages
- 39
baba chom...
mathtracks
Active member
- Joined
- Nov 14, 2004
- Messages
- 39
my switchers are boat anchors?
Tim,
Size, heat and noise were all reasons I went with the switcher.
By using an off line switcher in the two channel I got rid of a big honking noisy toroid. I couldn't shield it well enough to get the noise in the audio down where I wanted it.
Linear regulators turn whatever brown out room the supply has into heat, and since the two audio channels are already dissapating 8 watts a piece, I didn't want the supply to contribute another 6 or so to the package. Just too much to get rid of without goofy heatsinks or fans.
The supply is also a good couple of square inches smaller than an equivalent linear one.
The single channel EQ uses the same power transformer as the two channel mic pre, but it doesn't work too hard, so it's hum field is relatively constrained, then the switchers just work their magic in terms of heat and size.
A heavily loaded transformer, be it EI or toroid will throw a much larger noise field than one that is loafing along, where the bulk of the magnetic action stays in the core and doesn't bother surrounding circuits.
I also stay with high frequency switchers, and use shielded inductors exclusively, so that helps keep things under conrol too.
Don't forget to synchronize your regulators...
Size, heat and noise were all reasons I went with the switcher.
By using an off line switcher in the two channel I got rid of a big honking noisy toroid. I couldn't shield it well enough to get the noise in the audio down where I wanted it.
Linear regulators turn whatever brown out room the supply has into heat, and since the two audio channels are already dissapating 8 watts a piece, I didn't want the supply to contribute another 6 or so to the package. Just too much to get rid of without goofy heatsinks or fans.
The supply is also a good couple of square inches smaller than an equivalent linear one.
The single channel EQ uses the same power transformer as the two channel mic pre, but it doesn't work too hard, so it's hum field is relatively constrained, then the switchers just work their magic in terms of heat and size.
A heavily loaded transformer, be it EI or toroid will throw a much larger noise field than one that is loafing along, where the bulk of the magnetic action stays in the core and doesn't bother surrounding circuits.
I also stay with high frequency switchers, and use shielded inductors exclusively, so that helps keep things under conrol too.
Don't forget to synchronize your regulators...
SonsOfThunder
Well-known member
Wow...thanks for sharing all that info Dan.
[quote author="Dan Kennedy"]Don't forget to synchronize your regulators...[/quote] Uhh gee, Dan...why? "Heterodyning doesn't become a problem does it?", he asked leadingly... :green:
Peace!
[quote author="Dan Kennedy"]Don't forget to synchronize your regulators...[/quote] Uhh gee, Dan...why? "Heterodyning doesn't become a problem does it?", he asked leadingly... :green:
Peace!
Flatpicker
Well-known member
Thanks for the info, Dan. Interesting solutions.
SonsOfThunder
Well-known member
DanK:
Just out of curiousity, could you tell us if you are using a single chip solution (and if so what?) or did you cook something up yerself?
Thanks!
Charlie
Just out of curiousity, could you tell us if you are using a single chip solution (and if so what?) or did you cook something up yerself?
Thanks!
Charlie
In the single channel supply I'm using a pair of LM2672-ADJ 1 amp regulators, pretty much cookbook buck regulators, but the tricks are in
keeping the noise under control. All parts with the exception of the power transformer are off the shelf, which is nice.
In the dual channel supply I use an LT1244 with custom inductors for the gate drive, transformer and output filter. The common-mode filter on the line side is a Panasonic part. It's sort of expensive with the custom wound parts but it all fits and runs clean and quiet. This is an off-line filter, which for those not familiar with switchers, means the power line ac is rectified to make about 330Vdc, which means a voltage doubler on 110, and right off the line at 220, this dc is then chopped at high speed, 150kHz, stepped down to approximately the output voltage, rectified, filtered, with a portion fed back through a linear opto-isolator back to the control chip to set the output voltage. Only one polarity of output voltage is referenced, but since the audio circuitry is class-A and very symetrical load wise, that's all I needed.
Keeping 150kHz that is whacking around at those voltages from getting into everything and out beyond is all part of the fun. I learned a lot about keeping loops short, ground planes, and all the issues about dealing with creepage distances and other fun parts of high voltage stuff.
Hetrodynes? We don' need no steenkin' hetrodynes...
keeping the noise under control. All parts with the exception of the power transformer are off the shelf, which is nice.
In the dual channel supply I use an LT1244 with custom inductors for the gate drive, transformer and output filter. The common-mode filter on the line side is a Panasonic part. It's sort of expensive with the custom wound parts but it all fits and runs clean and quiet. This is an off-line filter, which for those not familiar with switchers, means the power line ac is rectified to make about 330Vdc, which means a voltage doubler on 110, and right off the line at 220, this dc is then chopped at high speed, 150kHz, stepped down to approximately the output voltage, rectified, filtered, with a portion fed back through a linear opto-isolator back to the control chip to set the output voltage. Only one polarity of output voltage is referenced, but since the audio circuitry is class-A and very symetrical load wise, that's all I needed.
Keeping 150kHz that is whacking around at those voltages from getting into everything and out beyond is all part of the fun. I learned a lot about keeping loops short, ground planes, and all the issues about dealing with creepage distances and other fun parts of high voltage stuff.
Hetrodynes? We don' need no steenkin' hetrodynes...
fuzzy
New member
When the first dtr900s' came out they had 13 switching supplies including 5v @ 90 amps and way down in the noise floor the cutest little digital caliope going on 
The sony 3348 ran all its switchers at a multiple of Fs. The best example of switchers done well, but with heroic filtering. is the SSL9000J. All switchers but with a 300# rack of inductors and caps
kevin
The sony 3348 ran all its switchers at a multiple of Fs. The best example of switchers done well, but with heroic filtering. is the SSL9000J. All switchers but with a 300# rack of inductors and capskevin
Svart
Well-known member
My X2 console has a switching PS. it's a large box that sits under the desk, around 12"x 12" x 8". 75% of the innards are caps and chokes. :green: I have not checked the output with a scope but so far I haven't had any type of hash or noise from it at all.
:guinness:
:guinness:
