I would think any "real" RF easily just jumps over any solid state switches (back-to-back MOSFETS etc) via their "off" capacitance, wouldn't it?
Isolation at DC/mains/audio frequencies should be close to perfect, though.
Thought I saw "bipolar" when scanning it, but guess not. Probably mosfets then.
He does say that the Rds(on) < 1 Ohm.
I think I can answer this now. My best guess:
1) The switching frequency is double the freq of the mains voltage (100/120 Hz) at the zero crossings of which the rectifier output would be measured and compared against the reference value.
2) If the above result returns a deficit, the boost portion (which I guess is something like I've drawn in post #42) would be inserted into the circuit to bring up the voltage, while surplus (or normal) conditions would have the boost portion bypassed, thereby maintaining status quo. In other words, the boost portion seems to operate during the mains cycles whose zero crossing instances returned voltage deficits at the rectifier output, sort of "integral cycle control".
In order to visualise the above, you may replace the 90V & 30V capacitors in post #42 with absolute sinusoidal voltages of similar magnitudes. The additive state (both caps add) gives 120V cycles to compensate for heavy load current draw, while the bypass state (only 90V cap used) would be used under normal conditions.
Also possible is a subtractive state (60V in the example) that could be effectively used during quiet passages (of the audio) to reduce amplifier quiescent losses. Unfortunately, the said Magtech regulator white paper does not appear to include this state.
Nevertheless, I still don't understand how the unidirectional rectifier output manages to make zero crossings ! Maybe near-zero crossings or resonant switches used.
It is very easy to amplify a sample of the secondary until you have a square wave and use the rising and falling edges as 0-xing points.
Then it is just a comparator to compare the output voltage to the set voltage (or look at the error voltage of the regulator loop) and if there is an error, close the switch on the next 0-xing.
The circuitry is trivial, the point is, is it a good thing, or just a gimmick?
And it certainly is not non-switching - you can close the switch at a 0-xing, but current only starts to flow if the high secondary peak rises above the regulated voltage, and that constitutes a switching action. Just like a diode; it is always 'closed' but only conducts (switches on) when the sine wave amplitude gets close to the peak.
Jan
jan.didden said:
Yes, Sir, but those would be the zero crossings of the AC, not that of the rectifier output, as given by the white paper:
"What about the output from the rectifiers? This is pulsating DC. While the peak of each pulse is at high voltage and power, the voltage at the end and beginning of each pulse is at -- ZERO!"
Since the rectifier output does not touch zero for a C-filter (even with load), it's quite possible that the filter is LC in which case it would be sufficient that the zero crossings of the AC be used instead, like you've mentioned.
jan.didden said:
Yes, but in order to cut switching losses, isn't it enough that ZVS alone be performed ? A zero switching loss (power being product of current and voltage) is obtained even when (only) one of the two quantities is made zero, isn't it ? It's usually either ZVS or ZCS.
jan.didden said:
Well, in my opinion, the switching frequency (120 Hz) is much lower than those used in huge thyristor choppers, and therefore, savings due to ZVS do not matter much, especially at voltages used by audio amplifiers, as switching losses are roughly proportional to the frequency and the square of the supply voltage (both insignificant in this case). Further, you may already be aware that raising the bus voltage is the main idea behind the Class G/H amplifiers.
Nevertheless, Mr. Sanders appears to be quite senior to me and I really don't want to take a swipe at his invention. I think he's just doing business and I would probably do something similar if I were in his place. Also, I must say that I'm not aware of the requirements of electrostatic loudspeaker systems (in general), maybe they're different, after all.
BTW, the 'topping-up' would take place before the rectifier, if I understood correctly, so this process causes no ripple.
Jan
Jan
Got it, but for ZVS to save switching losses, it's necessary that the "switch" turns on while the voltage blocked by it is zero. However, since the capacitor back is already charged by the low voltage rectifier, the ZVS instants for the "switch" would no longer coincide with those of the AC on the winding.
Yes, I see what you mean. My hunch is that he has a diode in series with the switch, otherwise the reverse bias on the BJT switch at switch-on could kill it.
I just fired off an email to Roger, lets see what he replies.
Jan
I just fired off an email to Roger, lets see what he replies.
Jan
Well, if it's indeed before the rectifier, then the "switch" needs to be an AC switch (back to back) in which case, the turn-off must happen at exactly zero current (not zero voltage), in order to avoid flyback (and possible avalanche of switch) due to the secondary leakage inductance of the transformer. However, there has been no mention of anything such as that.
Besides, from the white paper,
"digital control circuitry is used to monitor the rectifiers' wave form and cause the transistors to switch states"...
I happened to note the plural form in bold, that made me (and possibly also MarcelvdG) think of two separate rectifiers doing the job.
Looks like we have to wait for Mr. Sanders' reply.
Besides, from the white paper,
"digital control circuitry is used to monitor the rectifiers' wave form and cause the transistors to switch states"...
I happened to note the plural form in bold, that made me (and possibly also MarcelvdG) think of two separate rectifiers doing the job.
Looks like we have to wait for Mr. Sanders' reply.
To me, the apostrophe has been misplaced. We're taught the British form here and I don't know how other forms differ.
rectifier's => of the rectifier (singular)
rectifiers' => of the rectifiers (plural)
🙂
rectifier's => of the rectifier (singular)
rectifiers' => of the rectifiers (plural)
🙂
Name from the past. I remember his first ESL in Speaker Builder. 🙂
You can switch a lot of amps with a HexaFet or TrenchFet and not need much more than some board foil to cool. Zero crossing switching is nothing new. As Jan says, Roger is not a fool, but he is not telling the full truth here either. One thought is the boost supply is significantly higher voltage than the rail and he is relying on the TC not to exceed rail. That 2% is very hard to swallow though.
FWIW:
If you know the load very very closely, as within a couple percent, you can use a very small circuit to vary the saturation of the transformer core. However, that would not work for an amplifier as the load varies. Trying to remember back 40 years, we had a 60A 5V supply for a digital system that used a 2N2222 transistor and an op-amp. Just saying, extremely efficient regulation is possible.
You can switch a lot of amps with a HexaFet or TrenchFet and not need much more than some board foil to cool. Zero crossing switching is nothing new. As Jan says, Roger is not a fool, but he is not telling the full truth here either. One thought is the boost supply is significantly higher voltage than the rail and he is relying on the TC not to exceed rail. That 2% is very hard to swallow though.
FWIW:
If you know the load very very closely, as within a couple percent, you can use a very small circuit to vary the saturation of the transformer core. However, that would not work for an amplifier as the load varies. Trying to remember back 40 years, we had a 60A 5V supply for a digital system that used a 2N2222 transistor and an op-amp. Just saying, extremely efficient regulation is possible.
In advertising speak, 'no switching noise' and 'no dissipation' normally mean 'very little switching noise' and 'not a lot of dissipation' ;-)
Jan
Jan
I liked the comment by Richard Feynman to the basic effect: If you don't like the physics in this universe, go find another one because that is how they are.
Hi, I'm new here. Also, although I've been an electronics tech for over 40 years and a hobbyist for over 50, I'm only a part-time sort-of audiophile and I haven't designed and/or built a lot of audio equipment.
That said, I feel there's something that's largely missing from the discussion so far. Mr. Sanders has a cool idea, and I seem to remember his name from audio magazines of the past, so he has some cred. Nevertheless, I can't help thinking that this solution misses the real problem. (Apologies in advance for the length of this post).
Yes, the voltage in this scheme is "regulated". But in my understanding and experience, the primary goal of regulating the supplies of audio amplifiers is to reduce and linearize the impedance of the supplies, such that the AF currents circulating in them don't add distortion and intermodulation components. Keeping a rock-steady voltage is almost a by-product of this. (And in an amplifier topology that uses lots of negative feedback, it must be said that the amp itself acts to some extent as its own supply regulator). In a well-designed amplifier with good PSRR, (pre- or power), minor supply voltage variations may not have much if any negative impact. And throwing boatloads of reservoir capacitance at the problem can go a long way toward solving it, as long as things like inrush currents, diode switching noise, and signal modulation within the diodes themselves, are taken proper care of.
I think the real problem in audio amplifiers is that power supplies inadvertently become active, non-benign parts of the signal path. As far as I can tell, Mr. Sanders' design doesn't address this issue, beyond perhaps reducing IM distortion on signal peaks. Then again, a few hundred more millifarads of rail capacitance might accomplish the same end. But oh, my poor aching diodes!
One final note: in a no-holds-barred amplifier design, Mr. Sanders' topology just might make a killer input stage for a stupidly expen$ive high-power low-impedance regulator.
Just my two cents' worth, and again, apologies for my long-windedness. I also apologize if I've offended anyone.
That said, I feel there's something that's largely missing from the discussion so far. Mr. Sanders has a cool idea, and I seem to remember his name from audio magazines of the past, so he has some cred. Nevertheless, I can't help thinking that this solution misses the real problem. (Apologies in advance for the length of this post).
Yes, the voltage in this scheme is "regulated". But in my understanding and experience, the primary goal of regulating the supplies of audio amplifiers is to reduce and linearize the impedance of the supplies, such that the AF currents circulating in them don't add distortion and intermodulation components. Keeping a rock-steady voltage is almost a by-product of this. (And in an amplifier topology that uses lots of negative feedback, it must be said that the amp itself acts to some extent as its own supply regulator). In a well-designed amplifier with good PSRR, (pre- or power), minor supply voltage variations may not have much if any negative impact. And throwing boatloads of reservoir capacitance at the problem can go a long way toward solving it, as long as things like inrush currents, diode switching noise, and signal modulation within the diodes themselves, are taken proper care of.
I think the real problem in audio amplifiers is that power supplies inadvertently become active, non-benign parts of the signal path. As far as I can tell, Mr. Sanders' design doesn't address this issue, beyond perhaps reducing IM distortion on signal peaks. Then again, a few hundred more millifarads of rail capacitance might accomplish the same end. But oh, my poor aching diodes!
One final note: in a no-holds-barred amplifier design, Mr. Sanders' topology just might make a killer input stage for a stupidly expen$ive high-power low-impedance regulator.
Just my two cents' worth, and again, apologies for my long-windedness. I also apologize if I've offended anyone.