[Joe Rasmussen]

Ah, Churchianity, I love it.  We have it in Oz too, and it's BIG BUSINESS, preaching that wealth and ostentation is God's reward to a hard working, entrepreneurial flock...

I think Man is always looking for ideas to which he can hitch his star, and for which often he will fight to the death.  The most recent has been the failed 70 year social experiment which crashed in 1989 - a great concept, but not the answer...

Has anybody observed the latest Messiah? It's called the Free Market and that will fix everything, bring prosperity and good will to all. Just beware that you don't have oil in your back yard or oil off your nearest beach.

Re GNFB, has anybody actually tried the opposite? Now there is a heretical thought. Oh no, it's not as crazy as it sounds. It's been done and yes, it works, in surprising ways. But use like you would salt, sparingly. Wouldn't use it in conjunction with negative loops as that would be counter productive.

I think of GNFB and NFB in general as suppressing an uprising and rebellion. You suppress it down hard and it just causes more nasties to pop up elsewhere and just gets more angry and aggressive as ever. It doesn't fix the problem unless you look at what caused the problem in the first place.

More on a technical level, NFB has so many unpredictable inter-actions with stuff like slew rate and vanishing headroom with increased bandwidth. I use your grey description often. In fact I used it here, read under the heading Overview: http://www.audiorevelation.com/cre/product_info.php?cPath=33&products_id=191

So what about PFB, or if "negative" matches the "back" in feedback, then what about positive feedforward? PFF? Many topologies and certainly active devices have an inertia problem (my choice of words) as discussed by Hawksford (both Allen Wright and Terry Demol thinks of him highly). That is that any initial movement from a static point, that movement is likely to be non-linear or have a significant non-linear component. Positive feedforward, PFF, may in fact, as kind of intimated by Hawksford, be an error correction. While he was mainly concerned with bi-polar action, I can testify that it is not just limited to that. So I am suggestion that make we make our choices to make an amplifier topology behave in a consistent linear manner... and then add a little... tastefully according to the ear??? Kind of a gentle nudge forward. Think of a car, the hardest job it has is to get of the line from a standing start, or stuttering start. A little auxiliary shove could help.

I am currently listening to a tube amplifier that has it switch-able in and out. It is quite audible. Low level details, image specificity and stability, clarity of soundstage and soundspace, all noticeably affected. Elson Silva has tried it too and liked it. His words, it make the amp sound "bigger". I know exactly what he means. The sound opens up. I also use it with great effect in I/V conversion in digital players and DACs. There are more to be tried...

Unlike NFB which has to be used in significant amounts to even be justified in the first place, PFF is used in tiny amounts and is unlikely to affect headroom, bandwidth or indeed stability margin. In a stable amp it seems perfectly safe at <2dB and in some cases <1dB is enough.

Now I will adopt a new middle name, maybe Arius? Getit?   

Cheers, Joe

[AKSA]

Hi Peter,

An extremely elegant, simple, well set up NAKSA70!!  Congratulations....  I am thrilled you have it up and running, and I'm delighted by your report card!  I hasten to add that the NAKSA Peter owns was built by ME ME ME, as an example of how he should in turn build his ten.

I ruefully admit that I believe Peter's NAKSAs are actually better built than mine.......

Joe, GREAT post, yes, I've been looking hard at PFF but you did not mention that it's a tough design problem from a stability standpoint!

Kieran,

Thanks for the nice post.  I am very leery of judging anyone about religion;  it's literally a sacrosanct subject.

Cheers,

Hugh

[gaetan8888]

Hello

I did remember that there was an article about non-feeback amps, I've search and found it, I did not read it all for now, but it seem an interesting article. Here is the link;

http://stereophile.com/reference/70/index.html

Bye

Gaetan

[AKSA]

Very good article, Gaetan, with the usual polarised response from the dress circle.....

When experts don't agree, you have an empirical playground, where camps conflict and accusations are frequent.

What really stands out to me is this comment:  'Negative feedback reduces low order distorton but creates, through intermodulation, a spray of higher order artefacts, albeit at very low amplitude'.

The issue is, can these high order products be heard, and to what extent might they affect the subjective listening experience?

Hugh

[gaetan8888]

Hello Hugh

I think that in sound an music there is some references signal in the hf region that the brain use, with those signal the ear/brain can reconstruct the 3 dimentions, life and ambiance of the sound and music and the personality of a musical instrument, and for the music there is a special section in the brain to give the sense of beat and music.

So wen there is some spray of higher order artefacts, it mask the references signal and the brain lose the 3 dimentions, life and ambiance of the music. But maby there is other thing, but nobody did a serious and neutral physiological and psycological and acoustic study to find out more about that an the relation with the electroacoustic (amps, speakers, etc...)

The ear and brain are a marvel, we can situate a sound at 360 degree horisontaly and verticaly with precision.

Bye

Gaetan

[ginger]

Hugh,
Will give you a buzz in the next day or so to order a NAKSA. I've got a 55N+ on the bench at the moment. The owner shorted a speaker connection and now it has about 1V offset and bad hum in one channel. Should have that fixed shortly.

As you know I am an electronic design eng. in the day job.

My two favourite engineering "wisdoms" are:

1) An expert is someone who knows more and more about less and less untill he knows absolutely everything about nothing.

2) In any design project the first 1/2 of the job takes 90% of the time and the second 1/2 of the job takes the other 90% of the time.

Cheers,
Ian

[jkeny]

Here's my favourite related quote -
"An expert is someone who has made all the mistakes that can be made, but in a very narrow field" - Niels Bohr

[jkeny]

Progress Report:
I got my Naksa pcb from Hugh just the other day - flown half way around the world in about a week, fantastic - beautifully made & soldered. So compact - a work of art.

I have been planning my construction.
- ordered a 300VA 30-0-30 toroid trafo
- I have a nicely suitable donor case which just fits neatly - it's a Samson Servo 550 stereo studio amplifier of 275 watts per channel. The heatsink is a perfect size for the Naksa pcb & the rest of the case will neatly hold the toroid. (The existing beefy EI trafo is 54-0-54 may well suit a 100W Naksa?). The dual mono pots will be used to eventually control a lightspeed vol. control. The balanced inputs will be changed to SE inputs but that's all that really needs changing. As I suggested to Hugh, this might be a good second-hand buy for Naksa builders - it may be less than the cost of a heatsink alone never mind the case. (I picked up 3 of these at my local recycling centre )
- Once the existing pcb is removed (good parts source - 6*SC5200 & 6*SA1943), I will be tapping the heatsink for the Naksa board - hopefully today

[AKSA]

Gaetan,

I have just written this for the NAKSA Instruction, detailing distortion and how it is perceived for audio amps.  I think it pretty much agrees with your comment!

NAKSA Measured Performance

Measured peformance for audio amplifiers is a highly contentious issue, with often acrimonious debate between the objectivists, who say human senses are inconsistent, and the subjectivists, who insist that ‘how it sounds’ is the only valid test.  Like so much in technology, and indeed life, the reality probably lies somewhere between the two extremes.
Distortion measurements are based on a single sinusoidal test tone, usually at 1KHz.  This input test bears little resemblance to music, of course, which is polytonal and much more susceptible to intermodulation effects between the tones.  A single tone is a very easy test for an audio amplifier, liable to give flattering results.
Aspen believes that the human hearing system, and this includes the brain which processes the output from the ears, is very acute indeed at discerning odd order harmonics.  In fact, even order harmonics, the second, fourth, sixth, etc, are perceived to be musical, with many scales, notably the Western Major scale, emphasising and exploiting even order harmonics.  Scales such as the pentatonic (the major scale of China and Mongolia)  and the slendro (the Javanese Gamelan scale) are disharmonious, and tend not to rely on even order harmonics.  These scales sound markedly different to the usual major scales, and are, by Western standards, an ‘acquired taste’.  Aspen believes it is these subtleties which underlie our apprehension of a musical amplifier or otherwise, and therefore the nature and quantity of the distortion artefacts from audio amplifiers bears directly on human aural perception.
The vast majority of modern solid state amplifiers produce a harmonic spectrum where, after the often prominent second harmonic (musical), the odd order artefacts dominate, usually 5-10dB above the evens.  This results in the oft described ‘sterile, mechanical’ rendition.  There are also other important factors, such as the length of the harmonic spectrum and the decay of notes, but essentially high feedback amplifiers produce a long string of artefacts at very low levels which continue until at least the 15th harmonic and often further.  The higher order odd artefacts are vanishingly low, but there is some evidence that the ear can detect these tiny sounds and register them subliminally as ‘non-musical’.  The listener is strangely discomforted by such a presentation, and listener fatigue results, negating any ‘engagement’ (emotional connection) between the listener and the recording.
The NAKSA was designed specifically not for vanishingly low distortion, but for even harmonics predominating over odd.  This is achieved by single ended circuitry wherever possible, both global and interstage feedback, and an asymmetrical output stage, which offers the efficiency of Class AB operation with the distortion profile of a single ended amplifier. 
The following measurements were taken by Dr Graham Huon of Huon Labs, Waverley, Victoria on 8th July 2010 using the Audiomatica program Clio, a PC-based high accuracy distortion analyser.
70W per channel NAKSA power amp:    Results at 14.14Vrms, 20Vp, 25W into 8R at 1KHz:

H2   -72dB   H3   -79dB   H4   -79.5dB
H5   -84dB   H6   -83dB   H7 Unmeasureable
H8   -86dB   H9  -87dB   H10  -85dB

This sums to a THD of 0.0404%, 91% of it H2, H3 and H4.  At 1W output, 2.83Vrms, THD is 0.008%, of which H2, H3 and H4 comprise 0.005%.  The profile of this distortion is monotonically decreasing.
These are good results, particularly at very low powers where the crossover event of Class AB amplifiers normally greatly increases distortion.  These figures give some credibility to the notion that a good sounding amplifier will have an impressive distortion spectrum.

Ian (Ginger),

The expert gets a pretty bad rap - but I agree with those figures, they seem about right.  Everything always takes about twice as long as you plan, and delivers results half as good as you expected.

John,

That's a great idea, a very clever way of buying a nice, elegant case.  That Sampson is a bargain from a recycling center, I must check mine out - not that I need any more electrical equipment in my workshop, hell, I can hardly get in as it stands......

Thanks for the pictures, we will watch with rapt fascination just how you add the NAKSA to this case!!

Cheers,

Hugh

[jkeny]

My recycling centre has reaped some interesting finds - along with the 3 Samson studio amplifiers, I also got a fully working Rogers Cadet III integrated amplifier. Of course they are not too keen in my removing these items so a certain amount of stealth is required - recycling, how are you! I haven't paid them a visit in a long while - I always think what useful equipment has gone through there that I missed & is destined for the dump?

Thanks Hugh, I believe this will fit like a glove - as I said only small bits need changing to accommodate the Naksa - the EI trafo replaced with toroid, the balance-in inputs changed to SE. That's it really.

[AKSA]

John,

How goes it?  Have you installed your NAKSA without problems and got it going?

Heck, I'm as keen as you are.....   

Hugh

[jkeny]

Not yet, Hugh, yesterday was busier than I thought & I didn't get a chance to buy the M3 drill taps. I will be doing so today & will post pics & report.

[hybride]

I will be doing so today & will post pics & report.

Review review!!  (Hi John!, are still playing with the ESS9022?)

[jkeny]

Review review!!  (Hi John!, are still playing with the ESS9022?)

Hi Hans,
Just back from town with M3 taps & hand tool. Got to eat, relax & then address this tonight.

Yes, the ES9022 is still a great DAC - did you power yours up & have a listen?

[AKSA]

John,

Well fed?  Warm and Comfortable?  Calm? 

It seems we have international interest here, rivalling the Gulf Oil Spill......  I hope the birthing goes nicely, John!

Brad,

Have you received your NAKSA yet?  It's been about ten days..... 

Hugh

[jkeny]

Ok, just down from my work room with an update. Tapped the heatsink & got as far as mounting the Naksa board on the heatsink & back panel. Tomorrow I'll connect the xfmr, vol pots & outputs - should be listening then.

Couple of problems - all my own making - so progress was slow: Drilling & tapping went surprisingly easy - this is the first time I have done this!. One worry though, the holes are pretty close to existing tapped holes & mica washer sits over these pre-existing holes - cuts down the heat dissipation efficiency somewhat but hopefully it won't be a problem.

All my spacing washers fell out when I tipped the pcb to offer it up to the heatsink - these are awkward to put back in between the output devices & the underside of the pcb. Hugh, it might be an idea to somehow stick these in place when you are assembling the board.
 

Pics attached:
Heatsink Tapped





Naksa mounted on back panel




Naksa mounted on Heatsink




Samson Case Full & Empty

[AKSA]

Hi John,

Couple of points I'd make here from past experience!

#1  If you drill and tap any hole for mounting pcbs, always countersink the hole after drilling and before tapping.  This removes any chance of burrs breaking through the mica washer at fitting.
#2  Tapping is best achieved into cast or extruded Al with kero, meths, WD40, or some other cutting fluid.  Just a few drops on the tap will make all the difference.
#3  After this operation, go over the mating surfaces with wet/dry sandpaper grade 250 to remove any burrs or sharp peaks.
#4  Thoroughly clean after the tapping and sanding operation.  Allow to dry.
#5  When fitting semis to flat surfaces with mica washers, first grease up the washer ONLY, BOTH sides.  Then, carefully position the washer on the heatsink, press hard with the finger tip to adhere it to the heatsink.  Then turn the heatsink so it is vertical, and move the NAKSA pcb towards the heatsink in the vertical position with bolts and washers held horizontally in place by gravity.  First gently screw in the leftmost bolt, then the right most bolt.  This will locate the board correctly over its holes on the sink.  Then, using a very small jeweller's screwdriver, engage the driver stud, an 8mm bolt, and screw it in four turns.  Then guide the bolts for the inner semis, once in, turn the assembly so the pcb is now uppermost and horizontal, and screw down the four bolts carefully until they are firm.  Recheck that the mica washers are square, properly aligned with the longest dimension of the heatsink.
#6  Now, using a beefy philips head with a shaft diameter not exceeding 4.5mm, screw down the driver fairly tight.  Not too tight, this is only Al, mind!
#7  You can now tighten down the four semis, ensuring that washers lie towards the center of the board and do not foul the earth track on the outside of the pcb.


Once this is all done, inspect the semis between the pcb and the heatsink to ensure they sit flush against their washers and heatsink, and double check all bolts are tight, but not so tight they risk pulling out the thread.  Holding the pcb vertical at assembly will prevent the washers falling out;  sticking is not required in this situation.  Using the above approach you will ensure good thermal contact and electrical integrity, with not too much heatsink paste spilling out!

Hugh

[jkeny]

Thanks, Hugh, I'll go over my work tomorrow - I did check that there were no burrs or edges - everything felt smooth to the touch - didn't have 250 grade sandpaper.

I take what you're saying about the washers - still if they fall out for any reason, they are a bugger to get back in & lined  up.

[tinears]

Hugh,
I have found your comments generally very interesting, esp
in relation to the harmonic distortion profiles. Thank
you for your thoughts, it's been an education in
itself in many ways.
 
Suppose you had a choice of suppressing the
3rd harmonic of the Naksa amp down by another say,
5db while increasing the 4th harmonic by the same
amount, would that be preferable (all other harmonic
levels being the same)?

[jkeny]

Not much progress, I'm afraid - well it is Friday  & I'm now having a glass or two of wine which I find is not conducive to soldering or planning layout.

Anyway a pic - I've wired the outputs - both speaker outs are on the right side  of the case so one lead goes across the board which probably isn't optimal.

I have orientated the Naksa so as the inputs are on top - I'll eventually put lightspeed attenuators next to the inputs but at the moment the Balanced inputs are on the board on the left - I'll change these to SE inputs.

The PS toroid I'll probably mount just in front of the power section of the Naksa

Any/all comments appreciated (as long as they're clean  )