Thursday, April 25, 2013

ASoNAudiophiles - February 2013

ASoN Meeting –  February 3, 2013

Ultra Low Distortion Amplifier – Mark 3

An Audiophile-quality DIY Amplifier Project from Silicon Chip Magazine


Introduction:

As a kid, I was interested in all things electrical, mechanical and then electronic. Yes, folks, I was a kid once – now, I’m a big kid.

(Contrary to what my offspring think, I did not hatch from an egg at age 40 – they think that’s why I just don’t understand their generation.)

Without the benefits (???!!!) of personal computers and the internet, one of my main sources of electronics information and education was the Australian magazine Electronics Australia. The high point of every month was when a new issue appeared at the newsagent. (No, I wasn’t a nerd nor a geek – they hadn’t been invented yet.)

Each issue contained lots of interesting articles, and best of all, projects to build, pocketmoney permitting. Parts were expensive then!

One of the main writers and designers was a chap called Leo Simpson. And one of my first “serious” amplifier products was the “3 plus 3” – a solid state amplifier with a massive 3 watts per channel – designed by Leo.

I also built a model train controller, power supplies, and a signal injector probe using the very first integrated circuit, with – count them – 6 transistors! Or was it 4?

Turn the clock forward 50 years, and Leo Simpson is now publisher and editor of Silicon Chip magazine. Their latest amplifier is 200W per channel, designed by talented staffmember Nicholas Vinen, featuring less than 0.0006% harmonic distortion, and claimed to be the lowest distortion amplifier design ever published in such a magazine (if not ever). In fact, this distortion is so low that it is almost impossible to measure, as it rivals the inherent distortion in the test instruments themselves. And it is also almost impossible to separate the distortion from the noise, so in fact it’s probably even lower.

So, how does this translate to actual perception, to actual listening?

I contacted Silicon Chip and asked if we could borrow the amplifier, and also if Nicholas or Leo would like to do a presentation.

I was thrilled, and felt quite honoured, when Leo agreed.


ULD3 amplifier Specifications:

Officially, the ULD3 is a power amplifier, but it is also used as part of an integrated amplifier design, featuring the Silicon Chip Audiophile Pre-amp. It was the integrated amplifier that we had for the meeting.




Output Power: 200 watts RMS into 4Ω; 135 watts RMS into 8Ω
Frequency response: +0, -0.3dB (8Ω); +0, -1.0dB (4Ω) – 10Hz-20kHz)

Rated Harmonic Distortion (8Ω): <0.004% 20Hz-20kHz, typically 0.0006%
Rated Harmonic Distortion (4Ω): <0.007% 20Hz-20kHz, typically 0.0006%
Signal-to-Noise Ratio: 123dB unweighted with respect to 135W into 8Ω (22Hz to 22kHz)
Damping Factor: ~180 with respect to 8Ω at 1kHz & below
Stability: unconditionally stable with any nominal speaker load ≥ 4Ω

Equipment:

For comparison, ASoN member John Sawley provided his Odyssey Stratos power amplifier, rated at 150W per channel.

Speakers were the club’s KEF Q900 floorstanders.

Sources were:

  • CAL CD player, with its Philips parallel DAC, still favoured by many over the current generation of Delta-Sigma DAC’s.
  • Oppo 95 upgraded by Joe Rasmussen.
  • Sony DVD player DVP-NS300 (vintage 2000) feeding my Behringer SRC2496 DAC/sample rate converter (a very useful box, by the way – ask Google) – used at the start, as a backup, as the CAL was late arriving, and the modded Oppo takes up to half an hour for the clock to stabilize.


Listening and Music:


1)    Female opera singer Natalie Desai – Composer Sans Saens – La Rose
EMI Vocalisers - Oppo

2)    Male opera – The Mikado highlights – A Wondering Minstrel I

We used the Sony, as the Oppo would not play this disc properly, and the CAL had not yet arrived.

The mids were edgy and hard, which was put down to a bad recording, and the fact that the KEF’s are very revealing in this range. Some felt the KEF’s were ringing, but I doubt it. Although I remember previous meetings at Haberfield where the highs through the KEF’s would set your teeth on edge, I put this down to interaction between the KEF’s wide dispersion (directional characteristics) and the room. I have used them extensively at home, and while they are “a touch bright”, that is all.

3)    Opera – male and female – Mozart K441 - “Das Bandel”
Philips sampler – Best of the Best


Leo Simpson’s Presentation:

And now, here comes the moment you’ve all been waiting for…. (and there it goes).

Leo Simpson began his presentation, which was a rare insight into amplifier development over the years, and some “tricks and treats” that he was prepared to share with us. I can tell you that much of this valuable information is not to be found in textbooks and engineering papers, and is far from obvious.

Leo spoke about bipolar transistors versus FETs – he felt the former were superior (for power amplifier output stages), a sentiment I have heard from most leading solid state designers. He also talked about distortion mechanisms in Class B push-pull amplifiers, and the role the power supply – and layout! -  can play, being just as important as crossover distortion (which occurs as the signal swings through zero, and one of the output transistors “hands over” to the other).

Leo pointed out that at this level of performance, even the inductance of the output (emitter) resistors is a significant factor, and how they were able to compensate for it.

Finally, he touched on the preamp design, which uses the LM833 opamp. He claimed that after substantial evaluation and testing of many devices, this was their choice, giving best performance versus cost (“bang for buck”).

I should add that although this is a DIY amp from a hobbyist magazine, it employs some of the latest research in its design. Nick has also painstakingly modeled the circuits using simulation software such as SPiCe, to optimize the design and check for sensitivity to component variations. And then the prototypes(s) were measured using an Audio Precision test set – the system of choice in many top audio labs.

(Simulation software such as SPiCe allows a circuit to be entered into a computer, and its behaviour to be modeled and tested in great detail, and very accurately, without having to pick up a soldering iron! The software provides a suite of virtual test instruments, such as oscilloscopes, multimeters, spectrum analysers, and so on. Such programs can even factor in the effects of layout, stray coupling between components, the material from which the circuit board is made, and more. Thousands of man-hours – sorry, person-hours – can be saved, and measurements can be performed that are almost impossible on the bench e.g inside an integrated circuit. What’s more, these test instruments don’t load or disturb the circuit under test, as real instruments do.)

Unfortunately, at this point, Leo had to leave, so yours truly became DJ, playing some more music, and then swapping to the Odyssey.

More Music and Comparison:

Because the Odyssey is a power amp only, we used a Lightspeed attenuator as volume control.

Tracks played were:

1)    Walt Disney’s Lion King – a selection of tracks featuring some outstanding male and female vocal, IMNSHO (in my not so humble opinion); and the fight scene, with very loud and dynamic drumming, brass and percussion.
2)    Stevie Ray Vaughan – Riviera – a delicate electric guitar track
3)    K D Laing – Hallelujah – beautiful female voice, nicely recorded piano
4)    Carly Simon – In My Dreams – as above, but Carly’s voice readily reveals any tendency toward sibilance.
5)    Tommy Emmanual with
6)    Josh Groban – When It Rained – one of the best and most versatile male voices, with piano and orchestra

Of course, I committed one of the most frequent and cardinal sins – I did not match the volume settings in the comparison.  Frankly, the logistics of swapping systems, powering up and down, CD’s with very different recording levels, and limited time, made this extremely difficult.  

But why is this such a problem?

The issue is that a change in volume changes perceived quality, and tonal balance.
For example, at lower levels, we are more sensitive to mid-range frequencies, and bass and treble become much softer relative to these. Conversely, as we increase volume, bass and treble become more prominent.

Early amplifiers – typically those that had tone controls -  usually had a loudness switch. In the more basic designs, these provided additional, fixed boost to treble and bass. In the better amplifiers, they were linked to a control on the back of the volume control, such that when the volume was turned down, they provided progressively more boost to treble and bass.

This listening characteristic is described in the so-called Fletcher-Munson curves, which show the ear’s varying frequency response at different levels.

Summing up, a volume change of even one decibel or less – barely perceptible – can be heard as a significant change in tonal quality or balance.

Usually the louder amplifier or speaker will be preferred.

In this comparison, I didn’t think that there was an outright winner, but I would describe the Odyssey as having “more weight”. Bass seemed more extended and powerful.

But the amp’s owner, while agreeing, added “there’s really not much in it”, and then stated that he would seriously like to build himself a ULD3! Clearly, he was very familiar with the much-more-expensive (but still considered to offer great value) Odyssey, so I consider this high praise indeed.

The difference may have been due to:
  • volume settings – as above;
  • the use of the Lightspeed attenuator with the Odyssey may have made it more dependent on the characteristics of the interconnects (cables), due to the higher impedances;
  • we heard the Odyssey after the ULD3, and second listening tests exhibit a statistical preference;
  • and, of course, placebo effect - bigger amps (seem to) sound bigger.

(Of course, there’s heaps to discuss/debate here.)


Conclusion:

No matter how you look at it, for a kit costing just $750, this is one hell of an amplifier! You could spend ten times that on a finished product that might be quite inferior. Sure, you’ll have to put in a lot of (hopefully enjoyable) hours building it – provided you have the skills - but the result is worth it on many levels. Perhaps we might run a DIY SIG to help interested members if there is demand.


A very special thankyou (and congratulations) to Leo Simpson and Nicholas Vinen! Now, when can we have the DAC you showed me?????


Morris Swift