Power supplies, philosphy, and diminishing returns

I agree with all you say but do think the leap to 52 head unit is a large one, after all you're using 13/15ths of your SC by then.

You say: 'You get most from upgrading the pre-amp power supply rather than the head unit if you are setting off down the upgrade path. Ergo, Naims philosophy is correct, surprise, surprise.' That's our philosophy, I'm not sure its Naim's. If you have a 32.5 or 72 Hi Cap their advice would usually be to buy an 82 head unit not a Supercap.

With hindsight, my upgrade path would certainly be 32.5/HC > 32.5/SC > 52/SC.

Alex

Steve

You are correct in your statements, and the PSU's are fundamental to the performance of most audio equipment.

As one of the more vocal contributors to some of the other threads, I do have to make one thing very clear, which has got somewhat lost along the way.

I've never criticised a Supercap, either in sonic terms or with regard to value for money, when utilising most of its capability.

My only comments relate to its use with devices like my Prefix (or the 72) when Naim could, if they so chose, manufacture a simpler PSU with the same performance.

I know it provides a large upgrade in both the above scenarios, but knowing the same result could be acheived for less money is a personal irritation, for cost and space reasons.

If ever I wish to upgrade significantly beyond a Prefix, I'd seriously look at the Tom Evans Groove, which sounds fabulous and turns Naims PSU philosophy on its head. It's expensive but costs less than the Supercap my Prefix would need.

It offers fantastic results from a small box, containing the mains transformers and some very high quality regulation that is several orders of magnitude better than Naim's.

Naim may well disagree with my sonic estimation of the Groove though, I only have to please myself although there's a few others that agree.

Andy.

quote:

---

Do you think that's likely to be the case with a Naim product aimed at an arguably specialist application?

---

How many people would like a cheaper, smaller Supercap for their preamp / phono stage?

I bet the demand is greater than you might at first suspect, it's going to be many years before I can justify the expense of a 52 to utilise one, and I'm sure there's others in the same league as me. In the meantime they'd all like to upgrade economically, I'm sure.

You have to bear in mind though that the abscence of a product at this point drives the market for the mid / hi-end equipment, so it's unlikely to happen.

Naim have also stated previously they are not interested in producing such a product, since they disagree with some of their customers over the merits of a PSU upgrade vs. a preamp upgrade. Naim would recommend an 82 / HC over a 72 / SC.

Andy.

Andy.

Ta, Steve

quote:

---

In preparation for my Alex inspired orgy

---

I've met him - he hasn't had that effect on me

Andy.

Getting two out of those three are pretty easy. The trick is to have them all. I'm sure there are other crucial design parameters (such as bandwidth, low radiation etc... but maybe not quite as fundamental as those above).

It wasn't that long ago that the Hicap was the ne plus ultra of naim-sourced power supplies. The Supercap offers vastly better performance and is not entirely out of place even when coupled to a 42.5 preamp (I have done just this in the sixpack system). But as good as the Supercap is.. it is designed, sourced and built to a price point. As the market allows prices of the true high end gear to soar to house-level expenses, these true Cost-Be-Damned components will see the light of both the day and of my music room. The biggest dilemma is what to call it... since the Supercap is already a superlative moniker!

Ron
Dum spiro audio
Dum audio vivo

quote:

---

Originally posted by naimless!:
Stallion,
I note you have a lot of experience with the higher end naim kit and enjoy reading your posts. Assuming we are all seeking this top end nirvana, I just wondered if there was a limit as to how far you should go depending on the type of music listened to....what I mean is, on well produced stuff there is no limit to how far you would go because the sound just gets more and more real....so lets get all the powers supplies we can afford right? but what about the other 90% of which is commercially produced...in particular indie type material where the naim kit is so revealing and with all the blemishes within the recording, you have to ask is it worth it?

---

That is a question I have asked myself recently. I think a lot of people upgrade because they think all of the music will sound better (at least, that is what I used to think). But I think the only things that sound even better are well recorded pieces. I've had an entry level Naim system for several years and will probably not upgrade past a 72/Hicap. It just seems that I will get even more frustrated with the music I listen to because it might sound better if I upgraded just one more time. But, as usual, it is a two edged sword. Either some of the music sounds really good and some really bad or, by using crappier stuff, it sounds middle of the road all the time. I think the only think that could possibly improve the sound is if the recording engineers and producers actually started to give a shit.

Mike

Can you connect 2 hi's to a 72? Don't think so, but would like to know for sure.

Happy testing

Lobo - im on it. Thanks for the tip. Off to do some listening...

Steve

quote:

---

In my neo-layman understanding of electronica, the key to a great power supply are
1. Stable DC supply (their raison d'existance)
2. Low output impedence.
3. High rejection of spuria and perturbances and noise.

---

I think there's more to it in the context of a Naim setup, as some of the connection schemes idealise some factors at the expense of others.

From a technical perspective the ideal DC PSU has the following characteristics: -

No AC content (noise),
Stable DC voltage,
Infinite line rejection (removal of 'raw' DC supply noise),
Zero output impedance.

For audio these parameters need to be maintained, as a minimum, over the entire bandwidth of the amplifier / device being powered.

Naims supplies optimise earthing at the expense of output impedance, since the preamps are stuck on the end of a long cable with impedance in the milliohms range, several orders of magnitude greater than the best regulators.

Theoretically line rejection should eliminate 'raw' DC supply quality issues, but I suggest that the finite values available from standard circuits leaves room for improvement, in terms of absolute rejection and bandwidth over which it's maintained.

I also suggest the spectral content of the noise on the regulator output is possibly more important than it's absolute level, since none of Naim's supplies, when compared to the state of the art, are that quiet.

I'm very interested in products like the Tom Evans Groove, that are able to achieve excellent results with much more elegant PSU's.

I shall discover more soon, since the regulator I'm now working on offers something like 40-50dB better line rejection than an LM317, about 1000 times less noise and lower output impedance than a 'ZEROHM' surface mount link or a very short piece of wire.

Even more importantly these parameters are maintained over a bandwidth 100 times greater than a monolithic regulator.

Since it's discrete, I can vary any of the parameters above and determine their effects, should be fun (if you like that sort of thing)!

Andy.

So...first thing to say is that it was a very, very close call. The 72/super has great grip and PRaT when compared to the 82/Hi. Not easy to ignore. Instruments have real presence and demand to be noticed. In contrast, the 82/Hi had less glare and fizz (which is something the 82 suffers from sans supercap in my experience). Perhaps a little less authority? What I just could not ignore though was the additional information getting through with the 82. The 82 just plain resolves more than the 72. This is most noticeable in the sense of space portrayed around vocals and instruments – theres lots more of it. Lots. I can live without this when the equation involves shelving out 2K extra because the PRaT of a basic 72/Hi is so good but in this context, when the price differential is much closer, Id go for the 82. Of course this wouldn’t apply if you were certain of a move to 52 in the future, in which case it would be economical to go straight for the supercap. It might also not apply if you were an utter confirmed flat earther – the 82/Hi is better hi-fi but the 72/Super has flat-earth qualities in spades. Your choice.

I did try the 72/Hi, which reaffirmed my impressions of last night – that the supercap is a significant upgrade when compared to the head unit (both in cost and benefit). I also tried the 52 finally. It’s the best, just...at a very hefty price.

Steve

regards,

dave

Steve B

quote:

---

the 82/Hi is better hi-fi but the 72/Super has flat-earth qualities in spades. Your choice.

---

So, to restate my upgrade path 42.5 or 32.5 or 72 Hi-Cap to pre/Supercap to 52/SC. Leave out the 82 if you know you're going to a 52. If you're not going to a 52 eventually then its time to look at the competition from DNM and Dynavector to name two.

BTW These PSU discussions seem always to show that the 'Super Hi-Cap' would be much appreciated and would make Naim some money.

Alex

quote:

---

Originally posted by Alex S.:

BTW These PSU discussions seem always to show that the 'Super Hi-Cap' would be much appreciated _and_ would make Naim some money.

Alex

---

Maybe Naim can call it "UltraCap"...

Alex

Regards,

Alex

PS How could I possibly count sheep on the Naim forum?

quote:

---

Originally posted by ebirah:

So...first thing to say is that it was a very, very close call. The 72/super has great grip and PRaT when compared to the 82/Hi. Not easy to ignore. Instruments have real presence and demand to be noticed. In contrast, the 82/Hi had less glare and fizz (which is something the 82 suffers from sans supercap in my experience). Perhaps a little less authority? What I just could not ignore though was the additional information getting through with the 82. The 82 just plain resolves more than the 72.
Steve

---

From a purely Vinyl perspective would the 72/HC to Prefix/72/SC not make more sense than to a 82/HC. Anyone tried this?

quote:

---

Originally posted by Mr Tibbs.:
Welcome to a fellow NI resident.
Can't be two IM's in this wee world!

Mr Tibbs.

---

Can confirm that, to the best of my knowledge, IM is truly unique!

quote:

---

Any chance of a circuit diagram when you're finished?

---

As a little teaser, here's a quick discrete circuit I knocked up (using surface mount on veroboard!). It's based on an article that appeared in American magazine 'The Audio Amateur' in the 80's, with one very important modification by me, that reduces AC gain of the feedback loop, simultaneously lowering noise, reducing output impedance and increasing line rejection (which can be about 10-20dB better than an LM317, if built carefully).

This is proper discrete too, none of your namby-pamby op-amps here

It gives an indication of the performance that can be acheived, there's a dramatic reduction in HF noise above 400Hz or so, but owing to the simple circuit used it's not possible to filter the reference (which is acually a sub-surface 'buried' zener offering the lowest self-generated noise I know of). The filtering is brute force, caps a across the reference diode, and therefore is only effective at the point the capacitors impedance is small relative to the reference.

The up-and-coming op-amp based version allow the reference to be filtered to acheive that same noise performance across the entire audio band and above. Line rejection is 40-50dB better than a 317, noise is <<10nV / root Hz (against approx. 200 for a 317) and output impedance is measured in microohms!

Andy.

P.S. VERY IMPORTANT - If you wish to try this circuit you'll need an output crowbar and an input fuse or similar to ensure your precious kit isn't damaged in the event of circuit failure.

All measurements are relative, not absolute, the discrete reg. hits my current measurement noise floor at HF, the peak at HF is the cheap soundcard, not the circuit.

Naim acheive a similar HF noise reduction using an RC network that feeds most circuits, but I wonder what a PSU rail that acheives the same performance but much lower impedance will sound like. I need to build another one to find out!

Andy.

Have you built this and measured it or have you just done the CAD simulations? I would be very interested to see two plots: a plot of open-loop gain and phase and a plot of output Z, both from dc to 1MHz with your loading assumption.

It's possible to do this, but it's a more complex circuit. Bootstrapping the reference bias from the low impedance, quiet, output works very well and is elegant. One can always make circuits like this better, but since reducing noise is the primary goal, a simple circuit like this cannot acheive it, due to the inability to filter the reference.

Output impedance is not in any way primarily determined by this area of the circuit, so it's a question of balance.

2. Be aware of resonance effects when caps are placed in parallel at the output, especially with the inductance of electrolytics.

Well aware of this, there's only one output cap, but one has to be careful of it's type and the connected circuitry.

3. The Q1 circuit looks a little weird to me. What is it doing that a constant current source could not do?

It is a constant current source, biased using an LED. In comparison to a FET source it allows lower dropout, and by freak of nature offers almost perfect thermal stability. Multiple transistor circuits again add complexity, which isn't desirable in this circuit, bearing in mind it's inherent performance limits.

4. The feedback loop compares a fraction of the output V with the zener across a pn junction. This is not as linear as using a matched LTP.

Works well though - again simplicity rules in this specific case.

The FFT plot is noise of an actual prototype.

The big advantage of a truly discrete circuit such as this is a large bandwidth, without the inherent rolloffs present in op-amp based circuits. This results in interesting characteristics, most parameters are flat and consistent until well outside the audio band.

I'll sim the data your interested in, when I get a minute, but the output impedance of the original circuit (without C4) was approx. 2mOhms, ruler flat from 20Hz to 200kHz. This impedance is lower than a 'Zeroohm' surface mount link. For comparison an LM317 offers similar impedance below 1kHz, rising to about 0.2 ohms @ 40kHz and above.

The original SPICE sim is unbelievably accurate too, emulating the measured performance very accurately.

I've used different devices, and added C4 which lowers noise and output impedance significantly, whilst improving line rejection.

It's intended solely as an indication of the noise levels acheivable, doing a fully discrete circuit (diff. amp ) that competes with the op-amp I intend to use at both AC and DC is a ridiculous amount of work, though tempting, long term.

It's taken Tom Evans 9 months!

Andy.

P.S. Gain / phase (open loop) attached

[This message was edited by Andrew L. Weekes on SATURDAY 12 January 2002 at 23:09.]