I'm fairly new to audiophile amplifiers but I've been designing power electronics professionally for years hence the following question. Has anyone used switched power supplies to generate a split supply, e.g. +-48V. I'm considering using this as the supply for a linear AB amplifier.
Of course care would have to be taken to ensure noise does not propagate into the audio stage. However, the reason I like this idea is that you don't need an enormous capacitor bank to generate a stable rock solid supply with excellent transient characteristics. Having a supply with a good transient response all comes down to the closed loop bandwidth of the supply.
Of course care would have to be taken to ensure noise does not propagate into the audio stage. However, the reason I like this idea is that you don't need an enormous capacitor bank to generate a stable rock solid supply with excellent transient characteristics. Having a supply with a good transient response all comes down to the closed loop bandwidth of the supply.
it depends.
Switching supplies (at least, general purpose, off the shelf types accessible to most DIY'ers) do not really enjoy a large delta between idle and peak load. For Class A amplifiers run within the Class A envelope at all times (SE amplifiers of most kinds), these are usually usable. Even for SE amplifiers with some output swing over the idle current, say 2x, is also by and large OK. The team I work with has developed a Class A SE amplifier with a current doubling Class A SE amplifier, and we run it off a pair of Meanwell supplies.
For push-pull amplifiers, it's probably not so good, especially at high output currents. Transient capabilities are quite poor. For Class AB amplifiers it's probably an even worse idea. That said, the Benchmark AHB2 uses a switched supply in conjunction with a Class AB output stage, and that is one of the best measuring amplifiers out there. However, it isn't as straightforward as just that. I believe there's some interaction between the power supply and the audio signal as well, and that's the next thing our team is looking at. It seems quite a tough nut to crack though.
Switching supplies (at least, general purpose, off the shelf types accessible to most DIY'ers) do not really enjoy a large delta between idle and peak load. For Class A amplifiers run within the Class A envelope at all times (SE amplifiers of most kinds), these are usually usable. Even for SE amplifiers with some output swing over the idle current, say 2x, is also by and large OK. The team I work with has developed a Class A SE amplifier with a current doubling Class A SE amplifier, and we run it off a pair of Meanwell supplies.
For push-pull amplifiers, it's probably not so good, especially at high output currents. Transient capabilities are quite poor. For Class AB amplifiers it's probably an even worse idea. That said, the Benchmark AHB2 uses a switched supply in conjunction with a Class AB output stage, and that is one of the best measuring amplifiers out there. However, it isn't as straightforward as just that. I believe there's some interaction between the power supply and the audio signal as well, and that's the next thing our team is looking at. It seems quite a tough nut to crack though.
Afaik they don't handle currents very well from + to - and vice versa. Something, imho, can only be expected when the loudspeakers are generating it's EMF. I'm not sure if low output impedance or a low damping factor circuit mitigates or worsens these factors.
For home use, it could never be a problem either way, if it is there at all. For PA use I can imagine different restrictions.
All in all you are a good judge of how great switch mode supplies can be, you can start China cheap and see what sound-wise you still need ;-)
For home use, it could never be a problem either way, if it is there at all. For PA use I can imagine different restrictions.
All in all you are a good judge of how great switch mode supplies can be, you can start China cheap and see what sound-wise you still need ;-)
Yes, that is a good point. The transfer function of a power amplifier is dependent on the load and if you have a simple PI controller for the power supply, then it responds slowly at low loads and normally at nominal loads. There are tricks of the trade to ensure a good transient response at both light and heavy loads. But it is good to know that some high end designs use this approach.
@mterbekke I don't quite follow..do you mean the loudspeaker generates a reverse current that flows back into the supply rail?
https://micro-audio.com/store/
Have a look at this company, very good quality SMPS for audio.
SMPS supplies come in many price ranges and I’ve found “you get what you pay for”. The cheapo Ali/Ebay supplies I’ve tried were not very good and noisy.
I’ve also had good experiences with Meanwell HRP and HEP series, but you need two units for a bipolar supply.
Last edited:
I agree with Vunce’s suggestion. Several of us on the forum have used them.
Another brand that are designed for audio (but some have reported difficulty in communication with the manufacturer): Connexelectronic. There are multiple options available from that vendor. I’ve had good success ordering from them and I’ve used a few different of their SMPS models. I would not hesitate to order from either company again. I’d recommend contacting Sami through the MicroAudio website, or if you order from HiFiMeDIY, make sure the model(s) you want show stock in their US warehouse.
Another brand that are designed for audio (but some have reported difficulty in communication with the manufacturer): Connexelectronic. There are multiple options available from that vendor. I’ve had good success ordering from them and I’ve used a few different of their SMPS models. I would not hesitate to order from either company again. I’d recommend contacting Sami through the MicroAudio website, or if you order from HiFiMeDIY, make sure the model(s) you want show stock in their US warehouse.
Last edited:
Yes.
It's called back EMF.
One of the things (only?) needing amps to have lowish output impedance, the whole damping factor debate thingy!
Since this current is proportional to drive current, it depends on the power you will actually use in your system.
(if it is an issue at all!)
I read about it years ago and had many other issues designing amps myself back then. This switch mode supply wasn't on the first to-do list, experimenting with topologies etc were higher and this news set it a bit lower a bit still.
But, you're absolutely right: power supplies are extremely important, that's without a doubt.
I don't foresee any real issues tbh and it seems it is either mitigated or not one at all, proof are the options provided by other responders, even though the micro audio store products optically look very solid, if you can not live with the voltage they provide (35, 40, 55) better ask them if they can e.g. be factory set to your desired 48.
Must say most standard switching supplies always come with the +_10% adjustment range, but these supplies look more like custom built.
I especially like the fact they are able to handle high cap loading. Room for experiment, sonically! Might make a big difference. Personally I doubt it would be better (sluggish isn't better in my book, but different people, different books etc😉, but depends on the loop of the switching stage etc.
You're on the right track;-)
Edit: send some sunshine over Vunce, that's what we're really waiting for here!
H
HAYK
I did try dual supply +/-36v &+/-15v&12v. It cost me $26 with regret as only +36v was regulated. Using two independent 36v is much cheaper and each is independently regulated. Nowadays you can find on Ali 36v 7A for $6 + shipment. For HIFI use you find pre AC treatment and post DC filter on Ali.
Despite their claim: "....(our SMPS) use state of the art, very efficient LLC Series Resonant Converter Topology", I'd be very careful exposing those minuscule heatsinks to constant current demands of let's say.... pure class A amplifiers.
Furthermore, exposed to AB class amplifiers, it is important that the SMPS do not change their operating regime - the switching frequency... which usually happens as the current draw increases. Detailed spec sheets (if available...) would reveal it...
I experienced both issues when I was evaluating the SMPS for use in (high-end..??) audio.
Furthermore, exposed to AB class amplifiers, it is important that the SMPS do not change their operating regime - the switching frequency... which usually happens as the current draw increases. Detailed spec sheets (if available...) would reveal it...
I experienced both issues when I was evaluating the SMPS for use in (high-end..??) audio.
Last edited:
Hi , i will add my few cents. I'm using AB amplifier with +-45v smps , cheap china from *bay , and it's unregulated, so have no noises and very efficient. Also output voltage drops less with same load , i tried to compare it with toroid . With such smps its important input capacitor capacity , as it depends voltage drop with load and 100hz ripple at output. While tested output with oscilloscope, it was hard to measure hf ripple , as 100hz ripple dominated . Also with unregulated smps you can add capacitor bank at output, just need to add some soft start. Not all topologies of smps can work like these , that one is halfbridge, constantly running with 50 percent duty cycle. If you need ready made products, then it may be harder to get what you want , but if you like to modify everything to your needs , then smps can be good choice i think.
@Extreme_Boky Do you know what topology you were using? LLC converters are highly efficient. Since they are zero voltage switching, the main loss is the conduction loss, and you just put in a good enough MOSFET. I've worked on high power LLC converters in the past that need no active cooling or heatsink whatsoever. However, the point you raise about the variable switching frequency is interesting. It is true that the LLC is a variable-frequency converter. The switching frequency over the entire operating range is typically well above 20kHz, so I'm surprised that you experienced noise, as the ripple voltage frequency is also very high. Maybe the ripple voltage magnitude was high?
I'm confused - zero voltage switching - not as described here
https://www.monolithicpower.com/understanding-llc-operation-part-i-power-switches-and-resonant-tank
The advantage in that is removal of hf energy due to switch rise and fall times. The rise and fall times of the switches need to be as fast as possible as that is where most of the losses occur. The tank circuit provides a sine wave with the hf from the rise and fall times removed 😉 probably within limits. That sine wave is then fed into a hf transformer.
Have a look at this link: https://www.ti.com/seclit/ml/slup263/slup263.pdf on page 6. It goes into the details about the operating modes of the LLC converter. Around the resonant frequency, ZVS is achieved.
Not in the usual way with say a triac.
A small dead time is needed between the consecutive transitions, both to prevent the possibility of cross-conduction and to allow time for ZVS to be achieved.
The delay can be used in other types to get around switching times that cause both parts to be conducting at the same time. Losses are still likely to be in the transition from on to off which is why that needs to be as fast as possible in all types. The faster the switching the higher the bandwidth it represents. The tank circuit removes that aspect and leaves just the fundamental, 😉 How well - pass, but that is the theory according to the other link,
The TI paper mentions the same aspect in a different way
As already mentioned, the LLC converter is operated in the vicinity of series resonance. This means that the main composite of circulating current in the resonant network is at or close to the series resonant frequency. This provides a hint that the circulating current consists mainly of a single frequency and is a pure sinusoidal current. Although this assumption is not completely accurate, it is close—especially when the square wave’s switching cycle corresponds to the series resonant frequency
That has implications even on the diode rectifiers on the output side, caps too and filtering. They show curves either side of resonance.
The rest will make interesting reading so thanks, 😉 and still more to go. Maybe switching based on the voltage actually on the tank crops up.