Bridged vs Conventional Amps

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
There have been some interesting discussions about the benefits of bridged amps in other threads lately, such as the Bridged Capacitor PSU Thread. Various forms of bridging seem increasingly popular in high-end designs.

Historically I think bridging has mostly been used as a way to either save money in high power amps or to increase the flexibility of multi-channel amplifiers (i.e. allow a 2 channel amp to be used as a higher power monoblock). But the trend in many flagship high-end power amps is towards more "balanced"--i.e.bridged--output stages. Lots of companies are doing it now including Bryston, Outlaw, Emotiva and some, like Crown, have long used it for their flagship products.

If this has already been well discussed in another thread, let me know. But many of the bridged amps on the market are relatively new designs. And I haven't seen bridging presented as a cost-no-object ultimate amplifier topology by anyone. I also searched AES and the last paper on the topic was from 1984 by Sansui and it was largely in reference to using bridging instead of a transformer.

The marketing hype implies bridging is done to reduce distortion. Bryston, for example, says of their 7b and 14b: "employs a balanced-output design that reduces THD and IMD to unprecedented low values." But, from what I know, a bridged amp can only cancel even order distortion products--and that's not the dominant source of distortion. Crossover distortion typically dominates and it's mainly odd order.

A bridged amp could also, in theory at least, have a better slew rate. But slew rate, in practice, rarely causes real world distortion in modern amps. I suppose slew induced distortion *might* show up with full power sine wave testing at 20 Khz of an extremely powerful amp but even that doesn't seem very likely until you get up to obscene power levels.

In general, I would expect a bridged amp to have *more* distortion, not less. Generally distortion rises into lower impedance loads due to things like beta droop. And I would expect some of the non-linearities to be additive between the two halves of a bridged amp--i.e. perhaps thermal related distortion, phase distortions and/or mismatches between the sides.

So am I missing something as to why more and more flagship amps are bridged besides they're cheaper to build that way? Some of them, like the $8000 Bryston 14b and flagship Crowns, are supposed to be "no compromise" cost-of-little-object designs.

Perhaps once you get past a certain power level, the other benefits of a bridged design (like the broader choice of output transistors operating at half the voltage) allow it to genuinely outperform a non-bridged design? Or are the manufactures still trying to save money or just want to be different for marketing reasons?

Does anyone have any conclusive data on a bridged amp offering overall better performance than a similar conventional single-ended design?
 
In addition, I should add that most of these high-end amps, like the Brystons, likely end up connected to similarly high-end speakers that often present relatively difficult low impedance loads from lots of drivers and/or complex crossover networks, etc. To me, this makes a bridged design seem even less ideal. So why would a well respected amp manufacture like Bryston choose a bridged design for an $8000 amp?
 
Steve Dunlap said:
There have been hi end bridged amps around since at least as far back as the mid 70s. Properly done, they are NOT cheaper. In fact, they can cost almost twice as much to build.

Some of us believe the extra expense is worth the performance gain achievable. Others do not.

Yeah, the bridged Crown's for example go way back. But still, for the reasons I listed above, any sort of bridged amp would seem to be at a disadvantage. So what I'm after is how can a bridged amp outperform a single ended design? Where does the "performance gain" come from?
 
Juergen Knoop said:
headroom?

Hmmm... That's an interesting point. In a real world design, because a bridged amp works the power supply twice as hard, there would likely be more DC voltage "droop" with sustained output. That would likely give a bridged amp more measurable headroom compared to its sustained output. But power supply droop generally isn't considered a good thing in a high-end amplifier.

And, in some designs, a bridged amp is more likely to trigger its current limiting and/or suffer other losses into low impedance loads because of the effective load impedance being half of whatever is connected. So, as the load impedance drops, I would expect the single ended design to eventually have the headroom advantage all other things being equal.
 
Steve Dunlap said:
I did say properly done. That is why it can cost twice as much.

But how? No matter how high quality the halves of the bridge are, it seems to me the halves (or in the case of a grounded bridge design the "high side") will individually outperform the bridged version using the same design.

So even with an unlimited budget, if you set out to design the best amplifier in the world to drive a pair of real world speakers, why would you choose a bridged output over a single ended one?
 
2nd breakdown limited SOA in older bipolar output Q really limits I as Vsupply goes up - more than linearly so bridged @ 1/2 the supply V the safe Ipeak could be much larger and use fewer devices in total

MOSFET or thermal limited Bipolar SOA devices come out about the same in total number of output Q for given output power ( bridged can cost a little more in fixed Vdrop for the 2X bias components)

the extra small signal and driver parts usually are a minor production cost compared to mounting/heatsinking the output Qs
 
Nelson Pass said:
You can get reduced power supply noise, better common-mode noise
rejection, twice the slew rate, cancellation of 2nd harmonic and
lower (safer) voltages.

Ok, I agree with all of those except possibly common mode noise (it depends on where such noise enters into the signal path).

But, none of them AFAIK, will reduce real world distortion unless you're measuring *only* the 2nd harmonic. Given that 2nd harmonic distortion is generally much less of a problem (both subjectively and objectively) than the 3rd and higher orders, improvements there don't help the overall distortion or sound quality situation much. And I would expect the other factors I mentioned would more than offset any 2nd harmonic gains (i.e. the bridged amp having more 3rd and higher order distortion).

jcx said:
MOSFET or thermal limited Bipolar SOA devices come out about the same in total number of output Q for given output power

OK, are you saying for a given power output you need roughly the same amount of output devices in a bridged vs single ended design? I realize the transistors are in a much better part of the SOA curve in a bridged amp compared to a conventional amp of the same power. But that's offset by the peak current being twice as high, and also of course, by the need for two mirrored output stages instead of one.

I haven't done the full analysis, but I would think it comes to how much output voltage and peak current you want. For example, I'm fairly sure a 200 watt/8 ohm bridged amp would need more output devices total if you want the same 2 ohm load performance as a conventional 200 watt/8 ohm amp. For a 1000 watt amp, however, it might be the other way around.

It would seem, at least based on commercial designs, there's a crossover point at some power level where the advantages of a bridged design overcome their disadvantages. But I've not seen any hard data to back that up or identify at what power level that might be true.
 
I realized another potential advantage to the "grounded bridge" and similar topologies where the output stage power supply is not ground referenced: Such amps can have quieter grounds for the input stages of the amplifier. If the output stage power supply is floating, it avoids all the high ground currents in a conventional amp. As with SOA, this benefit is proportional to the output power of the amp.

Some also claim because the current drawn from the power supply in a bridged amp is full wave, vs half wave in a conventional amp, the bridged amp produces less induced (i.e. EMI coupled) noise in the rest of the amplifier. I'm less convinced this really helps. I'm not sure what the spectral energy differences are between the two waveforms, but I would suspect the twice higher higher currents in the bridged amp might more than offset the full wave advantage?
 
Nelson Pass said:
You can get reduced power supply noise, better common-mode noise
rejection, twice the slew rate, cancellation of 2nd harmonic and
lower (safer) voltages.

:cool:


I agree.

Also it depends on how accurate the phase splitting section of the circuit is, because if there is phase shift away from 180 degrees even at higher frequencies, there will be a related distortion. :whazzat:no transformers allowed here, eh?:whazzat:
:)

The PS has to be more robust in terms of current and uF's of capacity, but they can be lower voltage. Not a terrible difference in cost. Larger traces and thicker wires, eh, big deal.

I tend to favor the idear of using a common, balanced bridge VAS to drive each OPS instead of two separate VAS circuits. The better CMRR is a plus. In one version of my insane complex bridge amp, I drove the amp to death, deliberately, to see what would happen. One of the N-ch OPT melted short, connecting that speaker lead to the positive rail. The reaction of the DC servos that control the common mode amplifier part then attempts to correct for the CM error by trying to swing that output negative. The result was the other output phase was driven to the positive rail resulting in no DC current flow in the speaker and the circuit locked up in a steady state. This 'auto-protect' feature was an interesting observation I doubt could be achieved easily with a single end output. Hopefully sometime soon I will get some time to play around with it a bit more. I have quite a few experiments and tests I want to do yet.:cannotbe:
 
Balanced outputs vs. bridged amplifiers

Balanced outputs vs. bridged amplifiers

From Pass Literature:
"The Super-Symmetric X amplifier design exploits the symmetry of a matched balanced amplifier so that distortion and noise are cancelled at the output to the loudspeaker, then uses a small dose of a unique new form of feedback to make that symmetry more perfect.

The topology takes advantage of the character of specially matched balanced amplifiers that are cross-coupled to provide cancellation of distortion and noise. The result provides high performance with very simple linear circuits, better than previous similar efforts by an order of magnitude. It was named Super-Symmetry as an homage to particle physics, but it is popularly known as the X circuit.

Balanced amplifiers improve performance by differentially rejecting distortion and noise. To the extent that distortion and noise are identical, they vanish at the output, typically by a factor of 10 or so for matched single-ended Class A circuits.
Super-Symmetry extends this concept by using feedback only to make the distortion and noise more identical on each half of a balanced circuit, not to eliminate it as such. This gives as much as a 100:1 reduction in unwanted distortion and noise without requiring the equivalent amount of negative feedback. It is simply much easier to tweak the two halves of the circuit into identical symmetry than to eliminate all the distortion in each half of the circuit."

My ears like the sound of bipolar output Super-Symmetric amplifiers. I do not have any balanced drive music sources and use a single ended source input to generate balanced outputs.
 
Nelson Pass said:
You can get reduced power supply noise, better common-mode noise
rejection, twice the slew rate, cancellation of 2nd harmonic and
lower (safer) voltages.

:cool:
But a major disadvantage should not be unmentioned: The requirement to a small degree of deviation from the two amplifier halves is high and is easiest to realize by CSPP (Circlotron).
Therefore sometimes I observed a lack by quality of power amplifiers in the upper frequency range, when they switched from stereo to mono. I experienced this by certain, individual devices of the NAD power amp 2600 and 2400. The reason was mainly not selected output power devices from Toshiba - so I think
 
Hi,

In theory you can get 4 x the power into the same load, that's if the power supply is beefy enough of course. So that in itself is an attractive proposition.

This also means the voltage rails can be lower, with the benefit of reduced thermal stress, and cooling requirements, and lower voltage output devices.

I think a correctly balanced input and output is hard to beat.
 
Here's a fun circuit which is grounded bridge:

http://media.qscaudio.com/pdfs/discontinued_products/PL9.0PFC.pdf

schematic:
http://www.qscaudio.com/support/library/schems/Discontinued/PowerLight Series/pl9.0PFC.pdf

4500W / channel, 2 ohms :bigeyes:

The 4 step class H would have needed eight power supply rails and six rail switches if it had been a non-bridged amplifier. Of course, now the power supplies for the two channels need to be separate, but having one channel working in case of PSU failure is probably worth it in this application.
 
One specific benefit of bridging comes with true class-A operation of the halves. Then the total current draw from the supply is constant, the amp cannot modulate the supply with load currents. Still the supply should be low ripple and low and flat impedance of course. Since there are no load currents in the GND, a single supply can be used and a reference GND derived from that in a way that it optimzes PSRR for the individual halves right to start with.

Another point would be, for low/zero FB designs, that one can trade some even harmonics to favor low amounts of odd ones, as the former are cancelled by the bridge.

- Klaus
 
"My ears like the sound of bipolar output Super-Symmetric amplifiers. I do not have any balanced drive music sources and use a single ended source input to generate balanced outputs."

Hadley 622 (60's vintage):

Hadley622gif.gif
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.