[serengetiplains]

I think I have found the articles you referenced online.
http://www.scribd.com/doc/2610442/Capacitor-Sound
What is interesting is that he found that bypassing electrolytic capacitors with film shunt capacitors gave no useful reduction in distortion and the lowest distortion electrolytic power supply caps were a pair series connected bi-polar capacitors.
Fascinating reading.
Once again there appears to no measurable reason for an observed sonic improvement, it figures.
For what it's worth I have also heard improvements from bypassing electrolytic power supply caps with film types.
Scotty

That be the series, Scotty.

There's a part of me that prefers tube amplifiers for the sole reason they can be built entirely without electrolytics.

[serengetiplains]

Speaking of (almost) electrolyticless amplifiers, ahh:

http://purityaudiodesign.com/ultra%20gt.htm

[kc8apf]

Back to the comparator for a moment.  That chip performs probably the most important function in the amp, and as such would be the most important source of nonlinearity and distortion.  A little jitter on that delicate oscillator and ...

Based on:

If you look at section 8 of the data sheet it is for control and monitoring:
The microprocessor has three main functions: to provide an interface for controlling the amplifier, to
monitor the supply voltages in order to prevent spurious operation during power up/down and to
detect error conditions.

the PIC is not acting as a comparator nor is it part of the analog path at all.  It only handles supervisory functions.  It should be fed from a separate digital supply and use a dedicated digital ground.  Extra bypassing won't hurt anything but it shouldn't change the sound in any way.

[serengetiplains]

Thank you, kc.  Perhaps the comparator is discrete, in which case I won't be tampering with it.

[serengetiplains]

I have bridged mode up and running.      It's too early to say anything much about the sound, but it does sound different, and I think I like it.

[Regnad]

So, you appear to be the first to assemble a bridged setup!  Yay!!!

Are you using 2 SMPS for the 2 NC400's?   150 nf cap?

Thanks a LOT for any info on this and your listening impressions.

[serengetiplains]

Yes, up and running!  I'm using a single smps.  My reasoning is that psu noise cancellation is probably best with a single supply as, all other things being equal (they never are), I want the same noise signal on both amps to allow summing of such on the output.  But with two Ncore modules drawing from a single supply, that supply will be forced to operate at higher currents, therefore creating more noise to that extent.  Two smps might sound better, but at the volume levels at which I normally listen, I don't think that would be the case.

In any event, bridging sounds to me to be an improvement.  The effect I heard, on initial impression, is subtractive---as in, hey, something's missing.  What's missing is not detail, not power, not resolution; it must be noise.  Bridge mode brings greater subtlety, better definition, more obvious power.  I like what I hear, but these are impressions based on only a short period of listening to non-bridged mode.  I've only had these modules a few days, so my aural memory of non-bridged mode is not very strong.  Let's put it this way: I don't hear anything I don't like that I feel has been added by operating the Ncores in bridge mode.  I would be able to identify something like that.

Re the cap, I'm using what I had quickly on hand, which happened to be a 0.068uF cap.  I'll experiment later with a larger cap, toward the 0.15uF value Bruno recommends.

[Regnad]

I'm enjoying the single monos very much but, with 81dB speakers, it's hard not to think about more power.   

Is there somewhere where the single/dual SMPS for bridging is discussed, recommended?

Thanks!

[serengetiplains]

I certainly don't need more power myself, though I do hear a benefit of having more power in reserve.  The music emerges more relaxed, if you will. 

I don't have a clear understanding of the topology of this amplifier yet, but it seems to me that bridging should give greater power supply noise reduction.  By power supply noise, I mean anything entering, riding on or produced by the main output or any of the power rails---whether RFI/EMI picked up from god knows where, voltage variations emitted by the switching elements of the supply, variations created by current draws passing from a positive-impedance, dielectric absorption voltages from the ugly output and supply electrolytics, various resonances and instabilities---all of which exist dynamically in a complex state of constant movement and flux (simplistic sine-wave analyses don't come close to modeling this stuff).  I'm speaking here of the entire gamut of AC existing on the power rails, which should be pure DC. Subject to limited feedback reduction, that noise will pass 1:1 through the Ncore FETs into your speakers into the air into your ear. 

This is what I call power supply noise.  It is always present and always affects any component operating where- and however.  This noise, because largely dynamical, cannot be heard by placing one's ear at a tweeter.  It's a kind of dynamical noise that veils the signal.

Subject to having a perfect DC supply that remains perfectly DC always (hence zero-impedance and absolutely linear, with perfectly ideal components), not to mention having a perfect amplifier, the only way to reduce this noise beyond reductions wrought by feedback or component quality is to have that noise cancel itself by operating differentially.  It seems to me that bridging offers better differential noise cancellation than a mono amp can attain.  One of the reasons this amp sounds so good stock is the considerable attention Bruno gave to differential operation. From what I can tell, the amp operates differentially from input to out.

Fwiw, bridging sounds quieter to me, more relaxed, less edgy in a kind of edge I associate with noise.  I want to listen a little more, but merely my listening today has convinced me I will go with bridging using a single supply.  I may someday hook up another supply to compare single- vs. dual-supply, but theoretically, one wants the very same noise signal on the + and – legs of the speaker terminal.  That voltage will produce no signal in the driver.  A single supply theoretically gives a closer approximation to the same noise signal on the output.  Because the drivers are operated push-pull, that noise will cancel.  What remains to produce sound is (more) the signal.

Also remember that current modulations in each of the supplies---the output, gate and buffer supplies---will produce noise as I've briefly described above that will intermodulate in complex ways with the operation of those supplies one to another.  Reducing noise in any of the supplies will reduce that intermodulation, leading to cleaner DC rails for all active components in the amp.

[*Scotty*]

serengetiplains, If I understand your most recent post, you are saying that Bruno advocates using a .15uF cap for a bypass of the positive and negative power supply rails on the module,correct ?
If so it should be possible to find a very small form factor capacitor per Cyril Bateman's recommendations for a very low inductance, low distortion cap. In as much as the cap shouldn't have to withstand more than 150volts, even a .15uF cap should be small enough that you could shorten the leads to the point that the lead inductance is at least the same or less than the intrinsic inductance that cap has due to its type of construction. It might be possible to actually have the cap function as usable bypass for reducing HF and RF noise if the total self inductance can be kept low enough. What a concept. Does Bruno have anything more specific for a recommendation for this cap other than just the value. I am fishing for a brand or film type recommendation.
Scotty

[TomS]

Scotty,

They recommend the small cap hot output to hot output when bridging.

http://www.audiocircle.com/index.php?topic=106827.msg1102629#msg1102629

Tom

[serengetiplains]

Thank you, Tom.  That's how I've connected my Ncores in bridge mode (though currently using a 0.068uF cap).

Somewhat OT, for those wanting a good bypass capacitor for cheap, try these:

http://www.ebay.com/itm/0-5uF-250V-0-5-K71-7-Polystyrene-capacitors-Lot-of-50-NOS-/251060469200?pt=LH_DefaultDomain_0&hash=item3a745ebdd0

These capacitors (and this comment pertains only to the 0.5uF varieties) measure incredibly.  I have a capacitance meter that measures DF to 6 decimal places.  You will notice from Bateman's article that DF roughly correlates with DA which strongly correlates with level of capacitor distortion.  The above caps typically measure 0.000005 DF.  I've measured quite a few at 0.000000 DF---off the scale!  A good teflon typically measures 0.000010 to 0.000020, sometimes less.  Good polypropylenes typically measure 0.0000150 to 0.0000300.  Mylars, which are used as the output filter caps in the Ncore, measure 10 to 20X this.  Mylars are also polar (not polarized), so distort worse for that reason also.  I'll be bypassing these caps in the Ncores (I'd rip them out if I could do so without ruining the amp).

[*Scotty*]

Something that may be poorly understood about switching power amps like the Ncore. The Class D amplifier pulls current from the power supply at the switching frequency not at the audio frequency present at the analogue input. The higher the power the amplifier supplies to the load the higher the current demanded from the power supply. When power supply delivers these higher current levels to amplifier this is exactly the same as turning up the amount of watts emitted by a RF transmitter at its antenna.
 The power supply wiring is the transmitter antenna. This is the predominant noise in the box and its loudness is a function of the power delivered to the loudspeaker. This is also true for a conventional amplifier, the difference being that the frequency varies as a function of the audio input.  The power supply wiring can also radiate HF noise and the power supply can also ring at high current levels.
The secondary source of noise is the SMPS switching frequency. It would be very interesting to look at the voltage output of the SMPS on an oscilloscope. This would show any and all noise contaminating the DC power supplied by the SMPS.
Scotty

[mgalusha]

The secondary source of noise is the SMPS switching frequency. It would be very interesting to look at the voltage output of the SMPS on an oscilloscope. This would show any and all noise contaminating the DC power supplied by the SMPS.

If I can scare up some time I'll poke at mine with the scope and take some photos. I don't have a DSO, so it's camera on the tripod for oscilloscope images. But I do like my analog Tek scope, 400Mhz, so it sees HF noise pretty well.

[serengetiplains]

Is it possible you can compare the noise on the + and – rails?  If that noise is not a mirror image one rail to the other, bridged operation will I think of necessity give better PSRR.

[*Scotty*]

Thanks Mike, any data is good. I have very little experience with SMPS. I get how one can see a 60Hz or 120Hz charging frequency show up and add to raising the noise floor via hum. What is unclear to me is how the SMPS switching frequency gets suppressed to the same degree. It would seem as though you would need to incorporate a high current L or TT filter to suppress the switching frequency adequately.
Scotty

[serengetiplains]

Also Mike, can you measure power supply noise when playing music through the Ncore?  How the overall noise changes under dynamic conditions would I think be telling.  Also telling, I think, is whether noise signatures on the + and – rails change relative to one another in dynamic conditions.  That might be a difficult task with a camera, but you might be able to see such differences, if there are any, and report what you see.  If you can and if you have the time ... Cheers

[serengetiplains]

Something that may be poorly understood about switching power amps like the Ncore. The Class D amplifier pulls current from the power supply at the switching frequency not at the audio frequency present at the analogue input. The higher the power the amplifier supplies to the load the higher the current demanded from the power supply. When power supply delivers these higher current levels to amplifier this is exactly the same as turning up the amount of watts emitted by a RF transmitter at its antenna.
 The power supply wiring is the transmitter antenna. This is the predominant noise in the box and its loudness is a function of the power delivered to the loudspeaker. This is also true for a conventional amplifier, the difference being that the frequency varies as a function of the audio input.  The power supply wiring can also radiate HF noise and the power supply can also ring at high current levels.
The secondary source of noise is the SMPS switching frequency.

Yes, let's call that current a pulse current in the shape of a PWM signal.  Tell me if I'm off here with what I see.  That pulse current will create voltage fluctuations given the non-zero impedance of the supply, and those fluctuations will more or less mimic the PWM timing of the pulse currents.  It seems to me those voltage fluctuations would themselves approximate square waves (a very distorted, possibly more analogue version of which, surely).  As I understand them, they are but the music signal plus an appreciably high level of harmonics.  An audio signal is therefore travelling inside of those currents, albeit a signal with a nasty amount of noise.  Those harmonics are themselves recycled via the output filter, allowing the music signal to emerge from the amplifier output.  I'm not disagreeing with anything you're saying above, but am wondering if I'm picturing this supply activity properly.

Re radiation, is it positive impedance on any part of the supply rail that causes radiation?  Why does a circuit trace radiate, Scotty?

[*Scotty*]

You have asked a question without a very simple answer. In the case of the ClassD amplifier you have the transmitter carrier frequency which is the switching frequency and the amplitude modulation of the carrier which is done by the audio signal fed into the amplifier. It's an AM transmitter and depending on the effectiveness of the output filter, the speaker cables can be the transmitters' antenna. I could easily pick up the output of my TA 3020Tripath amplifier on a hand held AM transistor radio when I held it near the speaker wires.
 To put it simply the higher the frequency you try put into a wire and have it go from point A to point B the harder it is to keep inside the wire. If the frequency is high enough it won't really be in the wire at all, it will exist entirely as a field around the wire at this point and a lot of energy will be radiated into space.
 For example, the radar transmitter of a warship is located somewhere in the bowels of the ship below decks. The radar mast is located in the superstructure of the ship. The transmitter is connected to the antenna with highly polished copper tubing that cannot have any dents or dimples in it or you will have a short. The copper tubing is acting as a highly efficient wave guide for the radar signal that you wish to propagate from the antenna topside. The signal exists as a field that is contained in the tubing.  Weird huh?
For the amplifier, the shorter the traces are, the less effective they are at either radiating noise or picking up noise that is present in the environment. You also have a much easier time building a high speed circuit if you keep the circuit traces short and the board small. The Ncore module is a textbook example of how to do it right. If the power supply wiring is kept as short as possible in the Ncore the ability to radiate noise is curtailed because the antenna is smaller.
Hopefully this simplistic explanation makes some kind of sense, I don't write textbooks for a living.
Scotty

[serengetiplains]

That's an excellent explanation, Scotty.  It works for me.  And this stuff truly is weird sh*t.  It reminds me of an anecdote regarding Einstein.  Einstein was sitting at his desk in Princeton when an excited grad student burst into his office explaining his excitement over some new collider and how we will know more blah de blah.  Einstein looked up at him and said, "I just want to know what an electron is."

Yes, Bruno's board is fantastic.  It's the smallest board I've seen, which helps everything including allowing a tighter, more precise feedback operation.