Why are Switching power supplies so uncommon in power amps?
Hello !
please excuse my trivial question
But SPS have a lot of benefits
if we do not want to talk about efficiency at least the weight
And still power amps powered by SPS are a very small part of the totality
I thought that only very expensive SPS were good enough to be used in a audio hifi power amp
I knew of Halcro, Linn Klimax range, Lab Gruppen ... all extremely good but also not precisely cheap
Now I read that Mackie has a line of amps with SPS that are not expensive and sound good. They are traditional class AB amps but with SPS.
To end all the ramblings why these SPS are so rare in amps ?
Do they have any drawback sound wise ?
Actually I am asking because I would like to try one of these products
https://www.coldamp.com/store/products/audio-power-supplies
in an old amp I have at hand. Just for curiosity
Thanks a lot for any info
Kind regards,
gino
Hello !
please excuse my trivial question
But SPS have a lot of benefits
if we do not want to talk about efficiency at least the weight
And still power amps powered by SPS are a very small part of the totality
I thought that only very expensive SPS were good enough to be used in a audio hifi power amp
I knew of Halcro, Linn Klimax range, Lab Gruppen ... all extremely good but also not precisely cheap
Now I read that Mackie has a line of amps with SPS that are not expensive and sound good. They are traditional class AB amps but with SPS.
To end all the ramblings why these SPS are so rare in amps ?
Do they have any drawback sound wise ?
Actually I am asking because I would like to try one of these products
https://www.coldamp.com/store/products/audio-power-supplies
in an old amp I have at hand. Just for curiosity
Thanks a lot for any info
Kind regards,
gino
Last edited:
Looked at the SPS30 datasheet but it doesn't have a schematic or mention a topology, so here's a general answer.
I'm aware of three basic issues, though those more familiar than I with SMPS designs may have some corrections. 1) Typical home audio listening levels are on the order of 10mW output power, at which an SMPS is not cost effective, 2) an SMPS sized for the 100W or so most class AB amplifiers target will typically be in DCM for the few watts or milliwants an amp draws at quiescence, and 3) filtering switching noise off of the rails of a class B or D load requires the filter corner above the load bandwidth but sufficiently far below the switching frequency that meaningful rejection occurs. Since either switching frequency or pulse width goes to zero in most topologies and feedback implementations as the load goes to zero effective passive filtering of the supply becomes impossible since the switching fundamental and its harmoncs collapse into the audio band. The usual solution is to follow the switcher with an LDO but this is difficult at power due to limited LDO availability, need to maintain headroom on the linear regulator for good performance, and restores a fair bit of the cost and thermal complexity one hoped to avoid with a switchmode supply. With an amp appropriately sized for home audio SPLs it's often cheaper to pay for linear heatsinks than switchmode parts and magnetics though design inertia plays a part in this as well.
I'm aware of two classes of solutions to the above. One is a self oscillating supply---which is just applying Hypex/Philip's UCD class D amplifier topology or similar to output a DC rail rather than a audio signal---and has the rather useful property that the switching frequency varies with the output voltage rather than the load, making it considerably easier to keep the switching frequency above the audio band. Being a UCD patent licensee this is probably how Hypex implements their SMPS modules but I'm not aware of other implementations. The other is quasiresonant or resonant (mainly series LLC) topologies, wherein the magnetics remain in continuous conduction and switching noise/EMI is reduced. These topologies have only recently seen much adoption (my impression mainly in the past couple years) and implementations such as Connextronics' supplies still have CLC pi filters on their outputs.
I'm aware of three basic issues, though those more familiar than I with SMPS designs may have some corrections. 1) Typical home audio listening levels are on the order of 10mW output power, at which an SMPS is not cost effective, 2) an SMPS sized for the 100W or so most class AB amplifiers target will typically be in DCM for the few watts or milliwants an amp draws at quiescence, and 3) filtering switching noise off of the rails of a class B or D load requires the filter corner above the load bandwidth but sufficiently far below the switching frequency that meaningful rejection occurs. Since either switching frequency or pulse width goes to zero in most topologies and feedback implementations as the load goes to zero effective passive filtering of the supply becomes impossible since the switching fundamental and its harmoncs collapse into the audio band. The usual solution is to follow the switcher with an LDO but this is difficult at power due to limited LDO availability, need to maintain headroom on the linear regulator for good performance, and restores a fair bit of the cost and thermal complexity one hoped to avoid with a switchmode supply. With an amp appropriately sized for home audio SPLs it's often cheaper to pay for linear heatsinks than switchmode parts and magnetics though design inertia plays a part in this as well.
I'm aware of two classes of solutions to the above. One is a self oscillating supply---which is just applying Hypex/Philip's UCD class D amplifier topology or similar to output a DC rail rather than a audio signal---and has the rather useful property that the switching frequency varies with the output voltage rather than the load, making it considerably easier to keep the switching frequency above the audio band. Being a UCD patent licensee this is probably how Hypex implements their SMPS modules but I'm not aware of other implementations. The other is quasiresonant or resonant (mainly series LLC) topologies, wherein the magnetics remain in continuous conduction and switching noise/EMI is reduced. These topologies have only recently seen much adoption (my impression mainly in the past couple years) and implementations such as Connextronics' supplies still have CLC pi filters on their outputs.
I'm inclining to agree with @didiet78, it is probably more of a marketting choice, a good and stable SMPS is not easy to design, especially regulated ones witch for audio aplications must have a verry good EA compensation, so a verry good transient response where it would be the most needed, another reason could be the popular believe that switchers with theyr noise have real bad effects on audio amplification, more complex filterring solutions have to be considered and it would probably in all, cost maybe more than a simple 50-60Hz power transformer.
Evidently what i sayd is a subjective personal ideea, you all are in title to dissagree
Evidently what i sayd is a subjective personal ideea, you all are in title to dissagree
If you try to transport, setup and use 5kWatt power amplifiers, as Sound Engineers do when producing music shows in large dance halls etc, a 5kWatt power amplifier with a SMPS is less than 1/4 of the weight compared to a standard transformer coupled power supply weighing in at almost 1cwt, (50kG in new money). Most of us old timers have back pain issues!
Thank you very much ! I understand
actually i got my back pain already just we much less power
But ok ... i understand that if a system is not so common there must be drawbacks
Clearly when the amps are a lot and must be moved from one location to another the weight is really an issue for the professionals
no more doubt on SPS amps for home purposes
Kindest regards,
gino
I agree in general terms with twest, but also I would add a hard misconcept about SMPS´s As I said in other posts, I have made one of them without trouble from about 12 years ago,
The problem expressed by this guy is true, but it is easily (almost for me) to compensate the loop for a too low current, and also I added some fixed loads for the supply not to enter in "hiccup mode" (Quickly start/stop switching as the load current becomes negligible). In my case, not only 6 power amplifiers are powered from a UC3842+STP7N60+TL431+BYT80 +-14V +12V +5V total 80W flyback topology, also the radio receiver (TDA7211+TDA1220B) and some TL082's in the audio path, and two early (very old) CD players and the control logic are powered from it, maintaining the minimum load necessary to not enter in hiccup mode. It is fixed 50KHz. This is the aluminium box at the last photo. No AM radio interference of appreciable high levels.
The problem expressed by this guy is true, but it is easily (almost for me) to compensate the loop for a too low current, and also I added some fixed loads for the supply not to enter in "hiccup mode" (Quickly start/stop switching as the load current becomes negligible). In my case, not only 6 power amplifiers are powered from a UC3842+STP7N60+TL431+BYT80 +-14V +12V +5V total 80W flyback topology, also the radio receiver (TDA7211+TDA1220B) and some TL082's in the audio path, and two early (very old) CD players and the control logic are powered from it, maintaining the minimum load necessary to not enter in hiccup mode. It is fixed 50KHz. This is the aluminium box at the last photo. No AM radio interference of appreciable high levels.
Attachments
Osvaldo, do you have a write up on your compensation scheme and minimal load levels you could link? Generally I find a base load of 5 to 10% of max load is required and making the control loop more sophisticated to more gracefully support lighter loads still leads to switching frequenies close enough to audio band they're difficult to LC filter effectively. I've managed higher than 50kHz at minimum load in buckers, though not by a whole lot---60kHz or so.
Having a quick look at ebay or elsewhere online SMPS tend to be expensive.
It seems only top end amps have SMPS at the moment.
SMPS can be disasterous when they go wrong. I have seen mine blow up both mosfets, the SMPS chip and the SMPS power supply components. And there is much more to go wrong in a SMPS.
Behringer do a £140 1000w amplifier with SMPS for about £140.
It seems only top end amps have SMPS at the moment.
SMPS can be disasterous when they go wrong. I have seen mine blow up both mosfets, the SMPS chip and the SMPS power supply components. And there is much more to go wrong in a SMPS.
Behringer do a £140 1000w amplifier with SMPS for about £140.
Perfect, I would say.
I have the project in paper, and haven't scanner, but with time I can draw in LTS or Express Schematic.
Actually not only Behringer but also Mackie has a line of SPS powered amps
Mackie - FRS Series Amplifiers
It seems that good SPS are becoming cheaper
Better to wait and see the feedback from users anyway
Maybe they will be not reliable
Regards,
gino
I think it would be hard to find a brand of professional amplifier that did not have some SMPS powered models. They may still make mostly linear supplied ones, but they all see the market as needing light weight models too, and that means SMPS. Consumer audio is probably a different story. COnsumers will feel that the amp that is 20 pounds heavier is of "better quality." They won;t see the light weight of an SMPS as an advantage.
I am talking about power amps, but increasing numbers of guitar amps are using them too. Rack mount processing gear is full of them, as well as many audio mixers. I remember the Yamaha SPX90 effects processor had an SMPS, and that was 20 some years ago.
Peavey has a whole line of them, the DPS1600, 1600 watts and weighs 7 pounds for example. I remember some Crest models with switchers on each channel, and that was before Peavey bought the company.
Any audio engineer can also design a linear power supply. To put an SMPS into an amp would require the company hire SMPS engineers or outsource the job. That right there is enough to cause some companies to go without.
It probably doesn't count, but I spent some time in the coin operated amusements field, and our arcade machines were running on SMPS over 30 years ago already. They were digital logic boards mainly, but they did have audio. SO I guess it almost relates.
I am talking about power amps, but increasing numbers of guitar amps are using them too. Rack mount processing gear is full of them, as well as many audio mixers. I remember the Yamaha SPX90 effects processor had an SMPS, and that was 20 some years ago.
Peavey has a whole line of them, the DPS1600, 1600 watts and weighs 7 pounds for example. I remember some Crest models with switchers on each channel, and that was before Peavey bought the company.
Any audio engineer can also design a linear power supply. To put an SMPS into an amp would require the company hire SMPS engineers or outsource the job. That right there is enough to cause some companies to go without.
It probably doesn't count, but I spent some time in the coin operated amusements field, and our arcade machines were running on SMPS over 30 years ago already. They were digital logic boards mainly, but they did have audio. SO I guess it almost relates.
There is quite a bit of maths involved with SMPS design.
Look up the datasheet for the irs27951.
I was given a spread sheet for the IRS27951 where I just input my output voltage and current and it gave me the transformer and control component values.
If you pm lorylaci he has the original copy.
Mine has been bodged to suit my mosfets and transformer core.
A 27951 thread is here:
http://www.diyaudio.com/forums/power-supplies/184487-irs27951-irs27952.html
A lot of people do not use smps due to old stereotypes of the technology. The truth is that smps are at a point of being on par with linear ps when they are designed correctly. I am actually putting together a smps powered jlh amp at the moment and I am already impressed by the sound I am getting on my test speaker.
One project I want to try soon is a smps powered tube amp.
One project I want to try soon is a smps powered tube amp.
Properly filtered and shielded, it is just a box that puts out voltages. And like ANY power supply it must be up to the task of providing enough current and stand up to the vagaries of the load. It can make +400vDC, -70vDC, and 6vDC just as easily as making +5, +12, and whatever that we see every day.
I think a lot of consumers have antiquated views of SMPS, but I think the engineering community is well aware of its advantages and capabilites.
I think a lot of consumers have antiquated views of SMPS, but I think the engineering community is well aware of its advantages and capabilites.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.