Given this supply with too high a voltage and too low a current, I suggest that if you want to use it, you try a capacitance multiplier circuit. It will increase momentary energy stored from the transformer, decrease rippple, provide higher current (depending on what the drop is, lower being even better). If you had an LC stage inbetween the SMPS and capacitance multipler, even better. If you wanted to set a *regulated* voltage via the multiplier circuit by including a zener diode, that would work too. However, unlike (in a basic config, the bare transformer and bridge rectifier plus smoothing cap) an unregulated PSU, trying to stretch a SMPS beyond it's current capability can have ill effects. It might shut off. It might fold back to a lower voltage, which is fine if only a moderate amount and if behind a filtration stage,
For example,
http://www.tcaas.btinternet.co.uk/jlhcapmult.htm
I would not aim for least voltage drop from a cap multiplier though. As others have mentioned, the PSU is more than a little light on current so IMO you should shoot for a more significant voltage reduction. Perhaps 30V is more suitable but to be frank I would think about re-ebaying the PSU and going with a more conventional method. I've built capacitance multiplier regulation boards for a gainclone and the board was as large as the gainclone itself, at some point you have to pause and think "it is going to be better or just take a lot longer". Capacitance multipliers are not necessarily a bad thing, but better seen as an improvement on an already adequate supply, rather than a way to try to stretch a supply with only a fraction of the needed current to a level beyond their design.
However, my last few words opened a whole 'nuther can of worms. Given time to reverse engineer the SMPS, it is possible to tweak it to get a lower voltage but higher current at that lower voltage (within limits, don't try to halve the voltage or anything near that). For example if you hacked a +-42V PSu to output only +-32V, in "most" cases it would stay stable and have a little higher current. To "some" extent, a SMPS is VA limited, if you dropped the voltage 20% you may increase current by 20% (if it's well engineered, you have to check on current ratings for cerain parts after the transformer like the diodes, but frankly if they are that close to their margins already, it's not a SMPS you'd want to use long term for an amp). 20% is not much, but suppose you got closer to your target voltage, say that target was about 32V and you lowered the supply to 36V by hacking it. A capacitance multplier might do pretty well in that scenario.
I don't really mean to encourage it though, this is more of a stuck-on-deserted-island-pnly-have-one-PSU, perspective. Ideally you just want a transformer with lower voltage and more current, then decide how elaborate the PSU is after the rectification and filtration.
For example,
http://www.tcaas.btinternet.co.uk/jlhcapmult.htm
I would not aim for least voltage drop from a cap multiplier though. As others have mentioned, the PSU is more than a little light on current so IMO you should shoot for a more significant voltage reduction. Perhaps 30V is more suitable but to be frank I would think about re-ebaying the PSU and going with a more conventional method. I've built capacitance multiplier regulation boards for a gainclone and the board was as large as the gainclone itself, at some point you have to pause and think "it is going to be better or just take a lot longer". Capacitance multipliers are not necessarily a bad thing, but better seen as an improvement on an already adequate supply, rather than a way to try to stretch a supply with only a fraction of the needed current to a level beyond their design.
However, my last few words opened a whole 'nuther can of worms. Given time to reverse engineer the SMPS, it is possible to tweak it to get a lower voltage but higher current at that lower voltage (within limits, don't try to halve the voltage or anything near that). For example if you hacked a +-42V PSu to output only +-32V, in "most" cases it would stay stable and have a little higher current. To "some" extent, a SMPS is VA limited, if you dropped the voltage 20% you may increase current by 20% (if it's well engineered, you have to check on current ratings for cerain parts after the transformer like the diodes, but frankly if they are that close to their margins already, it's not a SMPS you'd want to use long term for an amp). 20% is not much, but suppose you got closer to your target voltage, say that target was about 32V and you lowered the supply to 36V by hacking it. A capacitance multplier might do pretty well in that scenario.
I don't really mean to encourage it though, this is more of a stuck-on-deserted-island-pnly-have-one-PSU, perspective. Ideally you just want a transformer with lower voltage and more current, then decide how elaborate the PSU is after the rectification and filtration.
It's probably not that hard to convert the voltage.
I've done it with old computer power supplies. There will probably be a feedback voltage drawn from a voltage divider on the output going back to the PWM controller. They work by increasing or decreasing the pulse width to match the feedback voltage to a reference voltage.
Usually if you're only trying to drop the voltage, all you have to do is change one of the resistors in the divider such that the output voltage of the divider will be correct when the amp is putting out the voltage you want.
Going higher in voltage is tougher, because you usually need to modify some overvoltage protection circuits as well (usually just replace a zener)
I'm not a total expert on it, but I've redone the regulation to give me a solid single 12V output on a few ATX supplies. Also upped one to 16V to replace a missing supply for a laptop I bought for cheap in an auction.
It's definitely doable if you're willing to figure out how they work.
And as ! mentioned, they do tend to be capped by the VA rating of the transformer, assuming there's enough margin on the output diodes and chokes, so you may be able to eek out a little more current as well.
I'm not saying it's a good idea, but if you can't find a good home for the supplies, it may do a halfass working supply. You certainly won't get full power out of it.
You may find it shuts down when a heavy bass note hits, etc.
If it's too cheaply built, you may just blow out the choppers from all the stress.
But I think there's probably still a lot of unrecognized potential in switching power supplies in audio stuff.
They've gotten a bad name in the past, but possibly just due to bad designs. When transistors were first invented, they built some pretty wretched amps with them, but these days people have figured them out pretty well.
-Nick
I've done it with old computer power supplies. There will probably be a feedback voltage drawn from a voltage divider on the output going back to the PWM controller. They work by increasing or decreasing the pulse width to match the feedback voltage to a reference voltage.
Usually if you're only trying to drop the voltage, all you have to do is change one of the resistors in the divider such that the output voltage of the divider will be correct when the amp is putting out the voltage you want.
Going higher in voltage is tougher, because you usually need to modify some overvoltage protection circuits as well (usually just replace a zener)
I'm not a total expert on it, but I've redone the regulation to give me a solid single 12V output on a few ATX supplies. Also upped one to 16V to replace a missing supply for a laptop I bought for cheap in an auction.
It's definitely doable if you're willing to figure out how they work.
And as ! mentioned, they do tend to be capped by the VA rating of the transformer, assuming there's enough margin on the output diodes and chokes, so you may be able to eek out a little more current as well.
I'm not saying it's a good idea, but if you can't find a good home for the supplies, it may do a halfass working supply. You certainly won't get full power out of it.
You may find it shuts down when a heavy bass note hits, etc.
If it's too cheaply built, you may just blow out the choppers from all the stress.
But I think there's probably still a lot of unrecognized potential in switching power supplies in audio stuff.
They've gotten a bad name in the past, but possibly just due to bad designs. When transistors were first invented, they built some pretty wretched amps with them, but these days people have figured them out pretty well.
-Nick
Has anyone tried the techique used in this article?
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
Looks promising.
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
Looks promising.
This is pretty much false. The advantages of a SMPS are (sometimes) lower weight, sometimes smaller size, usually greater efficiency, sometimes lower cost, often lower weight.
What a SMPS practically never does, is offer the same clean output. The primary difference we are all interested in, in a audiophile forum, is not typically size or weight or efficiency, it is that given practically any (every) design, the PSRR factor is far far worst with a higher noise frequency necessarily present in even very well filtered SMPS. Above all else, high frequency noise in the supply is a target for elimination!
SMPS have no place in audiophile land, they are always a concession that goes against fidelity. If you "like" the coloration caused by their noisey output, good for you, but there are so so many different ways one can degrade output and still say they "like" it, that we can still chaulk it up to a subjective impression that should not be assumed as a goal.
The FACT is, a SMPS cannot perform as well. There's no "good design" that changes this, unless you're only arbitrarily contrasting with a "bad" non-switching design which is not a fair comparison.
! said:
This is a pretty closed minded view of things. Nothing I stated was "pretty much false" I just stated that I think there's potential in them and that their problems were
possibly just because of bad designs.
I for one am willing to experiment.
A smaller, lighter, more efficient supply seems like a good idea to me, and I don't think that the problems are insurmountable. If you can explain to me scientifically that I'm wrong, my theory will change to align with your "FACT", but I don't think you can.
If this kind of thinking isn't allowed in your "audiophile land" I'll just pick up my power supply and go hang out elsewhere. (I can do that, It's much lighter than your big transformers)
Maybe science land. Scientists are all about learning new things, and changing theories as new information dictates.
Back to my previous analogy. If there were internet forums back when people first started playing with transistors, I'm sure someone would be saying the same things about them, and yet today the general consensus seems to be that they're a pretty useful technology.
-Nick
alleycat's post above is interesting and then some ... Yes, this works. 
alleycat: " Has anyone tried the techique used in this article?
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
Looks promising. ..."
alleycat: " Has anyone tried the techique used in this article?
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
Looks promising. ..."
pocketman said:
I'm working on number 3 at the moment, though I still haven't finished the chassis for #2 yet. I'm far from being an expert at it, but I'll certainly be willing to lend a hand if I can.
#1 Is one of Peter Daniels' kits. Definitely an easy way to get started for a beginner.
#2 is a 6 channel which I'm using for tri-amping. I finally got around to picking up the caps and hooking up the last 2 channels a few days ago. So far, so good.
#3 will be a bridged parallel LM3886 bass(guitar) amp, though it's probably still a while off. I'm working on the PCB layout right now.
-Nick
If you prefer the size or weight more, that's certainly your choice to make. It's a choice of those instead of higher quality audio. I already explained why, but you seem to have ignored the fact which goes against what you wrote below.
Actually my view is scientific, the scientific method does show the noisier output in high frequency range directly conflicts with far worse PSRR at HF of audio amps.
You seem to be suggesting switching PSU are "new" or something. There is good reason why it isn't common on good equipment and it is nnot that nobody tried, it's that it doesn't work as well. One of the goals should always be matching the qualities of the supply to the requirements of the load. If you have other requirements for your aamp that supercede the audio output quality it would be seen as a departure from our core goals, except for the generalized goal to DIY and experiment.
You might actually prefer the distorted sound produced by using a switching supply. Some people also prefer the sound of distortion caused by tube amps. The difference is there's a scientific backing for the latter having even order harmonics, while the former tends to just introduce an ever increasing random error rate as the audio frequency rises. You could do a lot of work engineering addt'l output filtration for a switching supply to (mostly) overcome this, but as you do so, the cost, size, weight increase which goes against the point of using one, and the addt'l complexity rise makes it a shorter lived amp. To some a ~ 10 year lifespan might be acceptible but generally it is accepted a switching supply has a shorter lifespan and if an amp is good, it should be timeless, worthy of listening to 30 years from now.
There is one other good reason to use SMPS, to make it more energy efficient, though it wouldl only follow that the amp is then some derivative of class D instead of A / B.
I expect your power supply will likely need some new caps within 10 years too. That's really the most likely part to fail in a switching supply as well. Sure, the output transistors may fry when the caps do give out, but I'm willing to spend the extra couple bucks when the time comes, if I ever let them get that bad.! said:
I may switch to class D at some point, it makes a lot of sense. I don't know enough to make a very good class D amp right at the moment, though. Chipamps are cheap, easy, and still reasonably efficient.
Not every design needs to be on the extremes, and people may have different design criteria. A big part of the reason my current amp will have SMPS is for weight and size, so I'm simply choosing a different balance.
Maybe some people are more concerned with efficiency. Obviously SMPS don't make awful sounding amps, or people wouldn't say they sound better than their linear supplies. I'm not saying they sound better either, but there many criteria involved, and it's easy to get stuck on one. Perhaps the linear supply has some downfalls which you haven't considered which contribute to these people liking the switching supplies more.
Just because the supply has more high frequency noise on it, and people like the sound of it doesn't mean they like the noise. There could be other factors involved.
Why do people listen to vinyl? Do you think they enjoy the clicks and pops?
Personally I think it's all a bunch of crap. I doubt I could hear the difference. Nor could a large percentage of the people who claim to, given a double blind test.
-Nick
Arx said:
No, this was specifically why I mentioned caps. The lower frequency in a non-switching design allows much longer life from same quality caps.
Chipamps are only as efficient as you choose to make them. You might do class A/B to some extent but so you could with a non-chipamp, though class A isn't out of the question if a modest output power.
You can DIY a class D if you want. The sound quality is something many will argue about but I think most people will agree they're good enough now to not only be reserved for driving a sub anymore. There are some nice premade kits at 41Hz.com to serve as an introductory project.
Fair enough, if all amps were the same it would be incredibly boring?
This is the crucial bit, that they are liking the change it makes to a lower fidelity sound. Everyone is entitled to have their own preference, but there are so many ways to change an amp and IMO picking a suboptimal PSU type is not the answer unless they have the other criteria as you mentioned.
If they say it sounds better, I think we do have to concede they are liking the noise... since it is PSU noise and that modulates the output sound even too fast for an amp with a feedback loop to correct to a high degree.
I think it's because that's the medium it came on, the medium on which they first heard song "X" and they became accustomed to it sounding like that and want to recreate the same sound. However, I think no so many people do listen to vinyl anymore, many who formerly did have digitized their collection.
switching power supply & lm4780
I really am new to this building amps thing. I built a 3w tube amp as a project in school, which was cool, but even then it did not compete with the detail of the discrete amp that I bought. The reason I am building one now is, I wanted to see how these chip amps compare to the amps that one can purchase.
I need either a 4 or 6 channel amp to run my dedicated 2 channel system which uses a digital crossover. Many of the amps out there are over $1000 and a 6 channel chip amp should come up at half of that.
The switching power supply intrigued me it was cheap to purchase and efficient and supposedly does not have a 60Hz hum, which I thought would be pluses.
At this time I cannot make the amp work, however I need to check if its the power supply or the amp. I do not really know how I am going to do this yet , as I do not have test equipment.
I have been interested in audio for the past 30 years and I have been able to hear the differences of many of the technologies that have come (and gone). I eventually think that digital amps are the future, its just I am not able to afford a Millennium amp.
I really am new to this building amps thing. I built a 3w tube amp as a project in school, which was cool, but even then it did not compete with the detail of the discrete amp that I bought. The reason I am building one now is, I wanted to see how these chip amps compare to the amps that one can purchase.
I need either a 4 or 6 channel amp to run my dedicated 2 channel system which uses a digital crossover. Many of the amps out there are over $1000 and a 6 channel chip amp should come up at half of that.
The switching power supply intrigued me it was cheap to purchase and efficient and supposedly does not have a 60Hz hum, which I thought would be pluses.
At this time I cannot make the amp work, however I need to check if its the power supply or the amp. I do not really know how I am going to do this yet , as I do not have test equipment.
I have been interested in audio for the past 30 years and I have been able to hear the differences of many of the technologies that have come (and gone). I eventually think that digital amps are the future, its just I am not able to afford a Millennium amp.
Interesting thread. As a tube guy for years, I've followed the chip amp movement because it appears to me to be quiet similar to the resurgence of tube equipment. People got tired of being told what was and was not accepted and branched out into side endeavors.
Chip amps seem to be the low parts-count answer to ss amps with buckets of components trying to control every real and imagined aspect of audio signal generation.
Now I'm reading things like "the proper way" and "the correct way" being applied to a DIY crowd, concerning something which seems to have gotten very little attention or research.
In my work, I repair commercial optical laser imaging equipment. As switchers have gotten better, they have been replacing linear supplies. At this point, I can't think of a single linear supply left in any of my equipment. All have been replaced by switchers, including critical laser and optics control circuits.
That must say something for the quality of newer switchers.
Another bonus is that personal liability should be mitigated by using an off-the-shelf ps for power generation. Not an issue for some, but an issue for others, I imagine.
Good luck.
Bob
Chip amps seem to be the low parts-count answer to ss amps with buckets of components trying to control every real and imagined aspect of audio signal generation.
Now I'm reading things like "the proper way" and "the correct way" being applied to a DIY crowd, concerning something which seems to have gotten very little attention or research.
In my work, I repair commercial optical laser imaging equipment. As switchers have gotten better, they have been replacing linear supplies. At this point, I can't think of a single linear supply left in any of my equipment. All have been replaced by switchers, including critical laser and optics control circuits.
That must say something for the quality of newer switchers.
Another bonus is that personal liability should be mitigated by using an off-the-shelf ps for power generation. Not an issue for some, but an issue for others, I imagine.
Good luck.
Bob
It could say more about the idea that we want analog audio improved, a deviation from what can already be bought at reasonable price. Some circuits aren't as vulnerable to PSU noise contamination as others. Build a switcher with output as clean as a linear and you're then off to a good start, although this thought is a bit ironic considering the unregulated PSU often used in gainclone type projects and yet chipamp PSRR goes down as noise frequency goes up.
"quality" is sometimes an arbitrary or subjective concept, and for some it is to approach an ideal power source from the perspective of audio reproduction even to the extent of having to build it ourselves, mitigated by the time, space and expense in doing so. Certainly due to these latter factors some people make a reasonable choice in buying a PSU instead of following the popular DIY topologies proposed in these forums.
If we ignore that difference in switching PSU noise, might we just ignore building the amp too and buy a ready-made stereo? Some feel that is the right answer but for others... here we are talking about it.
Last edited:
Sure, SMPS generate high frequency noise, but how much? A modified Cosel SMPS measured 4 mV PSU ripple with no load and 6 mV ripple at 35 W load in this thread: http://www.diyaudio.com/forums/chip-amps/75987-best-smps-gainclone-3.html#post1850799
That doesn't look too bad to me. 6 mV on a 24 V rail is somewhere around -72 dB. Allow for the PSRR of the amp itself, and I'd be surprised if the ripple is audible.
How much ripple would you expect from a well-built linear PSU at light loading? A couple hundred mV? Rod Elliott reported even bigger numbers in his article on linear PSU design: Elliott Sound Products - Linear Power Supply Design
That doesn't look too bad to me. 6 mV on a 24 V rail is somewhere around -72 dB. Allow for the PSRR of the amp itself, and I'd be surprised if the ripple is audible.
How much ripple would you expect from a well-built linear PSU at light loading? A couple hundred mV? Rod Elliott reported even bigger numbers in his article on linear PSU design: Elliott Sound Products - Linear Power Supply Design
True, tweaked switching PSU can be within the tolerance someone would accept.
Did you mean to link the page you did? It does not have linear PSU on it, only unregulated. Certainly linear PSU are far far cleaner than 100mV ripple even if the most basic barebones types.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.