Even though it's irrelevant for normal use, amplifiers do sell better when they can continuously deliver a ridiculously high output power. This supply concept helps to achieve that. When an amplifier can only deliver a high power for a very short time and the manufacturer specifies that as its output power, the manufacturer will be accused of unrealistic specs.
For the record, I have a two times 20 W amplifier with unregulated supply and have no intention of changing that. I do think this is an interesting concept, though.
For the record, I have a two times 20 W amplifier with unregulated supply and have no intention of changing that. I do think this is an interesting concept, though.
If you would remove that 'regulator' and run the amp on the max supply voltage, you would have higher peak power.
The supply would sag under high load but then it would only be down to the 'regulated' voltage of this sytem.
I don't see what you gain by it.
Jan
The supply would sag under high load but then it would only be down to the 'regulated' voltage of this sytem.
I don't see what you gain by it.
Jan
You would have to design your amplifier for higher peak voltages and currents then. For a given peak voltage and current handling of the amplifier, a regulated supply helps to increase the continuous output power rating.
Only my wife is 100% efficient. The rest, not so much. I operate at about 50% efficiency. I have a friend who endowed care of the sloths at the NY Zoological Garden (formerly Bronx Zoo).
The best I have seen are the controlled slew-rate controlled SMPS from ADI which go linear for a portion of the curve and burn some energy. These are easily implemented in DIY.
Ain't no free lunches. TANSTAFAL.
The best I have seen are the controlled slew-rate controlled SMPS from ADI which go linear for a portion of the curve and burn some energy. These are easily implemented in DIY.
Ain't no free lunches. TANSTAFAL.
Also, the correcting (series) portion of the supply seems to run at low switching frequency, thereby limiting control bandwidth, resulting in poorer transient performance when compared to an SMPS running above 20kHz (highest audio frequency).
The max power supply would also cause higher quiescent power dissipation, assuming a linear amplifier.
Last edited:
Or the supply for lower voltage. Continuous power is only of interest for Stereophile. ;-)
Which is a valid reason I suppose.
@jackinnj: wasn't it TANSTAFL?
Jan
Would error correction that increased PSRR substantially couldn’t as a ‘regulator’. No series pass element losses at all
That's just audiophile misinformation!
When designing a decent power amplifier you also design a decent power supply for it.
You know your wanted output power, you do the calculations to get to the voltages and current ratings, then you make sure that the power supply can handle that maximum continuous current without too much drop or voltage ripple.
A decent rated power transformer would drop only a few percent of it's rated voltage under heavy load ( because that's how things are done ).
The same with filter caps, a decent amount of filtering capacity would make sure that under the maximum continuous load there is low enough voltage ripple.
Thus the amplifier is able to deliver both high transients and high enough continuous power to the load.
Anything else is as i said just marketing crap, and i dare anyone to prove me wrong!
It's simply stupidly wasteful to have 2 different high power supplies for the same amplifier.
You don't have a steady line voltage? No problem, design a decent switching power supply with an active PFC which makes line voltage variations irrelevant ( you need PFC anyway for high power ratings ).
That solves another problem, because there is no need to complicate things by designing a regulated main converter ( very complicated and very expensive for high power ), you already have a quasi-regulated supply due to the PFC having a regulated HV output.
That is doing things smart, not wasteful.
According to the white paper, they have common filter capacitors and chances are that they just use different secondary taps of the same power transformer, so basically they then made an automatic transformer tap selector. Using separate transformers would indeed be a waste.
I've learned long ago to ignore a lot of so-called "miracle solutions" that so-called experts have dreamed up.
Once you've understood the actual fundamentals of electronics (physics, theory, etc), it's designs, and any of its possible and effective improvements (within reason), anything else is hearsay and chalked up as snake oil.
Once you've understood the actual fundamentals of electronics (physics, theory, etc), it's designs, and any of its possible and effective improvements (within reason), anything else is hearsay and chalked up as snake oil.
So every 10 ms the control circuit decides if the boost is needed or not. Sounds like the rails will have a lot of 100 or 120 Hz modulation
Good luck to passing EN 61000-3-2 and EN 61000-3-3 with this technique
Good luck to passing EN 61000-3-2 and EN 61000-3-3 with this technique
"There ain't no such thing as a free lunch" -- the motto of the University of Chicago Economics Department
This is not a regulator in my book. It's a stepped DC servo updating in a 100/120Hz time grid, letting all the line and load ripple pass through.
@MarcelvdG : Spot on with the switched xformer taps, that was my first thought as well after I read through that bloaty white paper.
@MarcelvdG : Spot on with the switched xformer taps, that was my first thought as well after I read through that bloaty white paper.
I think you can also doubt the non-switching aspect. Even if the 'top up' supply pass element (a BJT apparently) is switched on at a zero crossing, it will not conduct until the input voltage from the higher xformer tap rises above the voltage on the reservoir cap. At that point the BJT effectively switches on, just like a diode would. Definitely will have switching noise!
There most probably is a diode in series with the switch BJT anyway to prevent reverse bias I guess.
Jan
There most probably is a diode in series with the switch BJT anyway to prevent reverse bias I guess.
Jan
I think I read something about separate diode bridges in the white paper. No idea why they use a BJT as the tap switch or where rayma found that they do.
It may be possible to set up a resonance-induced zero crossing before that moment, which is possibly what the control circuitry does.
I think there's a thread here, several years old, talking about a near-perfect-isolation-from-ACmains power supply that used switched capacitors. Standard transformer + rectifier + filter caps gives a raw and ripply DC, WITH ACmains noise for the portion of the mains waveform where the rectifiers are ON. But there is additional circuitry downstream of the raw and ripply DC supply.
On clock phase1, switches1 close -- they connect the raw DC to a giant reservoir capacitor. On clock phase2, switches1 open and switches2 close. Switches2 connect the reservoir capacitor to a second reservoir capacitor. Second cap feeds a conventional linear regulator which is engineered to have EXCELLENT line rejection at the clock phase1/phase2 frequency. Voila, there is never a direct current path from mains to regulator, and HF hash + RFI on the mains are never manifested on the regulator output. And of course you, the circuit designer, control the waveshape (rise time, tail-off, nonoverlap interval, etc) of the phase1 phase2 clocks, and can tailor them for minimum fCLK injection onto the reservoir capacitors.
It doesn't claim zero losses. It doesn't claim perfect efficiency. It does claim near-perfect isolation from HF/RF on the mains. And if I remember correctly, it actually did work.
_
On clock phase1, switches1 close -- they connect the raw DC to a giant reservoir capacitor. On clock phase2, switches1 open and switches2 close. Switches2 connect the reservoir capacitor to a second reservoir capacitor. Second cap feeds a conventional linear regulator which is engineered to have EXCELLENT line rejection at the clock phase1/phase2 frequency. Voila, there is never a direct current path from mains to regulator, and HF hash + RFI on the mains are never manifested on the regulator output. And of course you, the circuit designer, control the waveshape (rise time, tail-off, nonoverlap interval, etc) of the phase1 phase2 clocks, and can tailor them for minimum fCLK injection onto the reservoir capacitors.
It doesn't claim zero losses. It doesn't claim perfect efficiency. It does claim near-perfect isolation from HF/RF on the mains. And if I remember correctly, it actually did work.
_
Attachments
