I assume you are correct John on the bridged output, but my friend called it a differential output and that is what caught my ear. Why is it more expensive and what advantages would it have and what contradictions would there be vs the standard approaches? I assume you would know the Spectra Sonics designs of old, but perhaps not as it was directed at the pro-audio market.
I can't imagine at least at first blush what sort of bridge mode output would not in some way be differential.
There are of course the "filterless" class D topologies which amount to using phase modulation: the two output connections to the load swing together for zero in-band output signal, resulting in large electric fields which are easy to shield, but to a good first approximation no magnetic fields. Contrast this with a typical class D stage which has the carrier blasting away swinging rail to rail at zero in-band signal, with a precise 50% "duty cycle". Unless the load is highly inductive at the switching frequency you get substantial currents, hence the need for a good output filter, typically at least an L-C. Applying feedback around that is tricky, with Putzeys invention among others making the L-C filter nearly-2Pi phase shift at the switching frequency part and parcel of the self-oscillating loop. Without some feedback the inductor in particular can be a significant source of distortion.
Typically "filterless" will use some filtering for very high frequencies, but often not much more than a ferrite bead for the L. If your antenna gain is low, that is used shortest possible speaker leads, you can achieve EM compliance without too much trouble.
There are of course the "filterless" class D topologies which amount to using phase modulation: the two output connections to the load swing together for zero in-band output signal, resulting in large electric fields which are easy to shield, but to a good first approximation no magnetic fields. Contrast this with a typical class D stage which has the carrier blasting away swinging rail to rail at zero in-band signal, with a precise 50% "duty cycle". Unless the load is highly inductive at the switching frequency you get substantial currents, hence the need for a good output filter, typically at least an L-C. Applying feedback around that is tricky, with Putzeys invention among others making the L-C filter nearly-2Pi phase shift at the switching frequency part and parcel of the self-oscillating loop. Without some feedback the inductor in particular can be a significant source of distortion.
Typically "filterless" will use some filtering for very high frequencies, but often not much more than a ferrite bead for the L. If your antenna gain is low, that is used shortest possible speaker leads, you can achieve EM compliance without too much trouble.
its more expensive because you pretty much have to build the equivalent of 2 of everything. the 2 are different things, bridged and balanced that is. you can have a bridged amp that isnt balanced (by just stacking 2 x Single ended amps together in series), but not really a balanced amp that isnt bridged =). a fully balanced amp swings both ways 'so to speak' the return current is sourced from the opposite rail, rather than ground and signal is referenced to ground through the power supply only
Last edited:
Kindhornman, I don't remember the Spectrasonics, but bridged operation is often used to double the effective output voltage. I built my first balanced bridge amp in 1969, because we needed efficient DC voltage reversal across a motor, and this is yet another advantage.
However it tends to increase the 'safe area' of bipolar transistors, by operating them at 1/2 the voltage for a given maximum power output.
I built 3 more balanced bridge designs in the '70's and '80's as prototypes. In those days, I preferred balanced bridges because they are a natural extension to the complementary differential input stage (which has two opposed outputs), and you can use the output transistors more effectively by avoiding the 2'nd breakdown region. Charles Hansen of Ayre uses the balanced bridge in order to keep the safe area region of American power Mosfets. I wish I had done that when I designed the Lineage power amplifier, where I had a lot of output failures that would have been avoided, if I had chosen the balanced bridge approach like Charles. I used to believe the spec. sheets of the American power fets, not anymore.
Today's bipolar output transistors do not necessarily need this advantage, and we can just parallel them to get reasonable performance. That is how we make Parasound amps.
However it tends to increase the 'safe area' of bipolar transistors, by operating them at 1/2 the voltage for a given maximum power output.
I built 3 more balanced bridge designs in the '70's and '80's as prototypes. In those days, I preferred balanced bridges because they are a natural extension to the complementary differential input stage (which has two opposed outputs), and you can use the output transistors more effectively by avoiding the 2'nd breakdown region. Charles Hansen of Ayre uses the balanced bridge in order to keep the safe area region of American power Mosfets. I wish I had done that when I designed the Lineage power amplifier, where I had a lot of output failures that would have been avoided, if I had chosen the balanced bridge approach like Charles. I used to believe the spec. sheets of the American power fets, not anymore.
Today's bipolar output transistors do not necessarily need this advantage, and we can just parallel them to get reasonable performance. That is how we make Parasound amps.
One option that's been explored here or near here is the correction amp side of an amplifier whose load floats, but as pointed out requires the correction one to carry all of the load current. Typically this means nearly as much increased silicon area as if it were a complete second half, as the lower voltage it will usually see doesn't help much, other than to reduce dissipation.
In an amplifier company four of us are waiting patiently to die next year (corporation-wise) the principal protagonist in the drama, who believed himself to be a tech whiz, licensed patents which he thought would provide protection for the existing design, one hatched before my involvement. This in turn was based on office action from USPTO that disallowed claims in another deeply flawed patent app. Actually the office action was pig-headedly ignorant and stupid and by and large irrelevant. And the licensed patents were essentially worthless. Perhaps my voicing of these things could have been done in a softer and more conciliatory tone, but why beat around the bush to coddle overweening negative egos?
But the licensed patents claimed the advantage of stacking stages in some sort of ill-defined modular fashion. Besides the inherent difficulties of providing appropriate drive, I pointed out that each module had to support the peak currents of some envisioned possible stack of same. And thus were burdened with this requirement even when used singly. There goes everything but the marketing story, pretty much.
And so it is, in terms of additional required silicon (or whatever semiconductor) area with bridged mode, although there are many advantages such as power supply rejection enhancement, cancellation of even-order stuff, and when your process and technology limits voltage swing, the availability of higher voltage swing and greater (~4 times) the power into a given load. I mention these things for the benefit of readers who might not know them, although of course many here very much do.
In an amplifier company four of us are waiting patiently to die next year (corporation-wise) the principal protagonist in the drama, who believed himself to be a tech whiz, licensed patents which he thought would provide protection for the existing design, one hatched before my involvement. This in turn was based on office action from USPTO that disallowed claims in another deeply flawed patent app. Actually the office action was pig-headedly ignorant and stupid and by and large irrelevant. And the licensed patents were essentially worthless. Perhaps my voicing of these things could have been done in a softer and more conciliatory tone, but why beat around the bush to coddle overweening negative egos?
But the licensed patents claimed the advantage of stacking stages in some sort of ill-defined modular fashion. Besides the inherent difficulties of providing appropriate drive, I pointed out that each module had to support the peak currents of some envisioned possible stack of same. And thus were burdened with this requirement even when used singly. There goes everything but the marketing story, pretty much.
And so it is, in terms of additional required silicon (or whatever semiconductor) area with bridged mode, although there are many advantages such as power supply rejection enhancement, cancellation of even-order stuff, and when your process and technology limits voltage swing, the availability of higher voltage swing and greater (~4 times) the power into a given load. I mention these things for the benefit of readers who might not know them, although of course many here very much do.
Difference between a $1500 stainless steel toilet bowl of a navy vessel, and a regular ceramic one.
Or a $1000 commercial airplane luggage compartment hatch in honeycomb/kevlar with adhesive foil, and a kitchen cabinet one in particle board.
Or a fuel-line on an F16 section assembly line, 100% guaranteed scratch-free mounting, plus all the official protocols & paperwork, and the exchange part of the annual car service.
Or a $1000 commercial airplane luggage compartment hatch in honeycomb/kevlar with adhesive foil, and a kitchen cabinet one in particle board.
Or a fuel-line on an F16 section assembly line, 100% guaranteed scratch-free mounting, plus all the official protocols & paperwork, and the exchange part of the annual car service.
I don't remember the LM301, but for a while Crown actually provided on request a very nice hardcopy schematic of the DC300, which I believe I still have somewhere. Why would a company give such a thing away? I don't know, and I didn't know Gerald Stanley at that time, but probably because it looked like a whole lot of trouble to attempt to copy. Perhaps it already had patent protection.
However, over the years, I embellished, in my selective recall, how elaborate the topology was. And when I ran across it a few years ago, it looked a whole lot simpler
And yes, it was famous among many for measuring well and not sounding good. I'm not sure I ever heard one, but I know they were pretty robust and employed for driving shaker tables and the ilk.
I am still waiting for a "Manufacturer's Corner" response, from Crown or at least someone at Harman, to the rather sadly lukewarm review given to the new Levinson switchmode amp in Stereophile, although I did see an LTE from someone who had acquired one, saying that he heard none of the sonic deficiencies reported in the review.
I see several advantages with bridged amp configurations.
First, power supply side, the two rails are symmetrically loaded.
Second, there is no current returns from the load trough the ground.
Common mode noises are canceled and some distortions too.
In the several tests i had done with various amps, it tend to sound better.
The only negative side i can see is the damping factor divided by two.
First, power supply side, the two rails are symmetrically loaded.
Second, there is no current returns from the load trough the ground.
Common mode noises are canceled and some distortions too.
In the several tests i had done with various amps, it tend to sound better.
The only negative side i can see is the damping factor divided by two.
- Status
- Not open for further replies.