[amplifierguru]

Hi Dan,

You're good at skirting questions.

There's a very good reason audio amplifiers use unregulated power supplies en masse - headroom (and associated dynamics)!

Cheers,
Greg

[Dan Banquer]

Yes to some extent, and you pay a price for headroom as you do for regulation. If you look at most speakers, the rooms they are in and the max spl at the loudspeaker generally tolerated then massive amounts of dynamic headroom are not generally required. For some of us there is a real difference between the demands of sound reinforcement and the demands of the home or project studio.
BTW: I have noted that more of the pro audio high power switching amps that are used for sound reinforcement are using regulated supplies. Beat that with a stick!
                   d.b.

[amplifierguru]

Hi dan,

Consider a 50W/ch amp using a common supply and traffo of a size comensurate with it's needs - regulation will be sufficiently poor that under REAL MUSIC conditions this amplifier, which might scrape in it's both channel rating with sinewaves, might produce peaks of 200W /ch before clipping on average musical programme.
 
If the same amplifier were to have interposed a regulated supply, it would have to be set below the lowest mains sag, full 4 ohm load sag plus allow for regulator operating drop, the amp would be lucky to produce 35Wrms and 70W on peaks - a shrinkage of output by 5dB!!

Way to go? I don't think so, but hey, the PCB would be a walk in the park and require no special talent. Not to mention wasteful of energy, power and resources. And cost twice as much through added complexity.

Cheers,
Greg

[NealH]

Approach your preamp concept from four tiers, most to least important (though they all are important).

Active signal passing components
Passive signal passing components
Layout
Power supplies
 
The active signal passing components, tube or solid state,  should get the most attention.  This is where the signal has the best chance of being "colored" so to speak.   The first rule of audio amplifier design - whatever  can be done with op-amps, can be done much better with discrete devices (transistors or tubes).  There is really no reason outside designer laziness ( or incompetence) to be using op-amp devices in the signal passing section of a preamp.  The designer has so much more control over the signal and amplifier parameters when using discrete componentry.  Hip op-amps today are there for cost savings - not ultimate fidelity.  Use the op-amps in the voltage regulators, not the signal circuitry.  (of course all in my opinion)

Passive components in the signal chain are the second most important sector of the circuitry.  Less is better here and, for mostly obvious reasons.  Passive components can induce non-linearities which show up as vagueness, veil, limited dynamics, etc.  Using high quality components obviously helps mitigate these non-linearities but, using less means less non-linearities to begin with.  

There is nothing worse than crosstalk.  Layout is very important again for the obvious reasons.  Poor layout leads to crosstalk and susceptibility to EMI.   It only makes sense that you don't want one amplifier channel or supply affecting or influencing another.

Use the old fashion regulated power supplies and this area will likely need  little further attention.  "Pi" and double inverted "L"filters, chokes...all componentry employed to build quality regulated supplies in the years past.  You can build them cheaper today with op-amps, but not necessarily better.  No need to use any hip technology on the supplies - just go with the tried and true.  Those supplies of yesterday had no trouble meeting the class B FCC conducted emissions.  Supplies of today may well need filtering added.

[Dan Banquer]

Hi dan,

Consider a 50W/ch amp using a common supply and traffo of a size comensurate with it's needs - regulation will be sufficiently poor that under REAL MUSIC conditions this amplifier, which might scrape in it's both channel rating with sinewaves, might produce peaks of 200W /ch before clipping on average musical programme.
 
If the same amplifier were to have interposed a regulated supply, it would have to be set below the lowest mains sag, full 4 ohm load sag plus allow for regulator operating dr ...

Your statement above is at odds with your numbers. A regulated power supply for a 50 watt amp must be able to have the rails set so the amp has the voltage across it to do the 50 watts rms into 8 ohms and if the amp is up to the task, 100 watts into 4 ohms. A lot of this has to do with the unregulated portion of the supply and how it performs under the designed for load and also the AC line voltage at high line and low line.
I can see from your statement above that you haven't really thought much about PS regulation and what it entails for an amplifer. It can be rather involved, and definetly not a trivial design exercise.
          d.b.

[_scotty_]

Dan can you please expound a little design wise on what is required for a regulated supply to do the job correctly. I would think that the supply should have symetrical rise and fall times equal to or less than the amplifier it is to power as well as a dynamic impedance that is as low or lower than the passive supply it replaces.
Scotty

[amplifierguru]

Hi Dan,

You appear to not have read my quoted statement on the way to pursuing your agenda. I said "If the same amplifier were to have interposed a regulated supply" means just that. I started with an amplifier then added the reg supply to highlight the tradeoff it brings.

The result is a drop in useable power to typ 1/3 of the original and complete loss of headroom, as well as double the complexity as the supplies are now as complex as the amps, subsequently more than doubling the cost with heatsinking, etc..  


So comparing: original amp or a 1/3 the useable power, no headroom and twice the price version with reg supplies.

My approach is to de-sensitize the amp board, layout and topology to reduce the Class AB modulation/feedthrough that inevitably occurs on unreg supplies achieving MOST of the advantage of reg supplies without the downsides.  Now, if you wish to discuss a different scenario where the unreg part of the supply is resized up to accomodate a reg supply to fund 50Wrms/ch, the headroom is still lost! Still 3dB less practical power than the original. Still not a match, practically as the original has better dynamics.

I must admit that reg supplies are great for spec'ing amps though as they give the best figures irrespective. Doug Self found this when he reg'd his amp design and THD dropped from 0.04% to 0.006%, from memory. Improved PSRR topology, PCB layout and lead dress would have likely achieved the same result.

For value, power and dynamics unreg supplies are the winner.

Cheers,
Greg

[Dan Banquer]

"Improved PSRR topology, PCB layout and lead dress would have likely achieved the same result. "
If you really believe that, then you really haven't understood anything I have said about gain bandwidth vs. supply voltage. You've also missed/ignored a fair amount of theory. Suit yourself.
  Have a nice day;
               d.b.

[Dan Banquer]

Dan can you please expound a little design wise on what is required for a regulated supply to do the job correctly. I would think that the supply should have symetrical rise and fall times equal to or less than the amplifier it is to power as well as a dynamic impedance that is as low or lower than the passive supply it replaces.
Scotty

I am not sure I understand your comment about symmetrical rise and fall times, as the output of the regulator is DC. If you can expand on that Scotty I would appreciate it.
Most Solid State regulators use feedback, and the typical output impedance of most of the linear feedback regulators is a few milliohms.
Hope this helps;
                 d.b.

[amplifierguru]

Hi Dan,

I haven't ignored anything, just turned the BS filter on!
The techno-babble you quote just didn't make it through.

Enjoy your day.

Cheers,
greg

[_scotty_]

Assuming that the regulator is a perfect source of DC is a simplistic viewpoint.
It may sag under load and be slow to recover,they can exhibit ringing,ballistic
overshoot and undershoot, and all of this performance can change over the audible bandwidth. The impedance measured in milliOhms can vary with frequency and vary under dynamic signal conditions.
Wouldn't this varying source behavior and impedance have some impact on the performance of the circuit it is supplying power to or are we assuming that all regulated DC sources are behaving as perfect batteries.
Scotty

[amplifierguru]

Hi Scotty,

I would have to say in the big picture there are real issues of significant consequence such as the Class AB half-wave spray of harmonics generated in the output stage and superimposed on the power supply impedance and it's subsequent entry into the signal path as a pollutant through the PSRR of the design, PCB proximity interactions and lead dress. These can have a huge effect on the quality of the result.

Regulation of the power supply is one, in my view inelegant, solution to some of this, at the expense of GREATLY increased cost, loss of power and headroom and poor resource usage.

The types of effects you describe w.r.t reg supplies would be minor to negligible and bordering on a witch-hunt IMHO. I have designed commercial amps with both reg and unreg supplies and I prefer unreg and finessed design/layout as it delivers the goods without impost.

Cheers,
Greg

[Dan Banquer]

Assuming that the regulator is a perfect source of DC is a simplistic viewpoint.
It may sag under load and be slow to recover,they can exhibit ringing,ballistic
overshoot and undershoot, and all of this performance can change over the audible bandwidth. The impedance measured in milliOhms can vary with frequency and vary under dynamic signal conditions.
Wouldn't this varying source behavior and impedance have some impact on the performance of the circuit it is supplying power to or are we assuming that all regulated DC sources are behaving as perfect batteries.
Scotty

Yes they can do that and unregulated supplies exhibit similar characteristics. The extent a regulated supply will do that will depend on  how the feedback loop is compensated in the supply. I have found in test that the supply I am using for the power amp is compensated well enough to minimize these artifacts. In test I will monitor the supply line vs the output signal. What appears to predominate the measurement on the supply line is the signal coming out of the power amp. This is very normal, and at higher levels that signal will be across the main supply caps in the unregulated section of the supply. At first this may seem to be a fault, but when you start moving higher voltages and currents this is what happens.
I have also noted that the supply rail even under square wave testing at high power varies less than 400mV. Just a further note: the supplies in the LNPA 150 are designed for 20 amps max. The fuses between the supply and the amp will limit the supply to about 15 amps peak for 5 milliseconds. Output impedance of the supply even with the amp running a 20 KHz sine wave at full power into 2 ohms is still in the low milliohms.

The LNPA 150 power supply regulator is a variation of the good old op amp regulator. The basics of this come out of one of my textbooks. The op amp regulator is also the basis for the Walt Jung super regulator.
Does that answer your question?
               d.b.

[_scotty_]

Thank-you Dan, your answer addresses most of my concerns.
It looks like you and Greg have a difference in viewpoint on the audible effects of the real-world behavior of regulators. Having heard the effects first hand of the difference between a conventional 3 pin regulator and shunt regulation in my preamp I have to say that I share your opinion that regulator behavior does matter and can significantly impact the sound of the circuit it supplies power to.
Scotty

[Dan Banquer]

Thank-you Dan, your answer addresses most of my concerns.
It looks like you and Greg have a difference in viewpoint on the audible effects of the real-world behavior of regulators. Having heard the effects first hand of the difference between a conventional 3 pin regulator and shunt regulation in my preamp I have to say that I share your opinion that regulator behavior does matter and can significantly impact the sound of the circuit it supplies power to.
Scotty

I have never used a shunt regulator........ yet. A couple of years ago I was doing a prototype linestage/headphone amp. I used the LM317's and LM337's with the adjust cap. This cap reduces power supply ripple and noise, mainly because it reduces the "bandwidth" of the regulator. I observed that the "bass slam" into the headphones was not as good as it should be, and found that the square wave response at the "knee" of the square wave had a bit too much of an exponential curve. Did I need a faster regulator to keep up so to speak?
Well, I could have done that, but instead I decided to put 220uf caps on the supply line very close to the headphone amp, in fact right at the collectors of the output transistors  of the headphone amp. This worked just fine and the bass slam without added noise worked just fine. The scope confirmed this when re-examinig the square wave.
So the moral of the story is: When using a low bandwith/low noise regulator, make sure you have enough distributed capacitance where you need it on the board.
Hope this helps;
             d.b.

[amplifierguru]

Hi Scotty,

At no time did I say differences between regulator types were not audible. Regs can make a substantial difference to the sound both in low level circuits (preamps, crossovers, etc..), and power amplifiers - in fact I use discrete 80dB EF BJT regs in some of my low level products.

However in power amplifiers there is a huge impost against using full regulation as the problems they alleviate can be fixed more elegantly without the impost of significantly lower power, no headroom, complexity, $$$ extra cost and excessive resource use.

Simple and refined works for me.  

Cheers,
Greg