advertisement


The best regulator I've built so far

Yes, I am aware of this (even if actually is DC with little AC current, it's class A circuitry after all).

But there is the filter capacitor there, which provide all the AC current above 125 Hz. Let's say, above 1 kHz. If the filter works, actually the PS does not see higher frequency currents. Its bandwidth is not an issue.

I know that this reasoning is not correct, still I am not able to find the bug.
 
Naim preamps use first order filter with 27R and 47uF, at 20KHz it gives a ratio of about 1:160, or 8db attenuation. This regulator uses a second order filter with a much higher ratio providing an attenuation of over 100db. In addition, due to the ESL of the electrolytic capacitors used by Naim they simply don't filter at all at higher frequencies.

You may ask what do we care about higher frequencies. Well I can tell that in this application it makes a big difference. I have experimented with many types of capacitors and only capacitors that work up to the MHz provide these results. You can clearly hear the difference.
 
Many thanks, this can be a possible explanation. Qualitatively I understand it, but the real numbers still let me dubious.

A ratio of 160 is 44dB, which actually is a significant attenuation in any case, also because the noise of a linear PS is low to begin with.

Moreover, Naim preamps have much higher internal noise(the well known hiss, audible in most speakers).

I do not know the exact capacitor type Naim uses, but are they so inductive to have problems at 20 kHz? If so, why Naim did not place a cheap film resistor in parallel to correct for this? We are speaking of very expensive equipment, and a design that has remained basically the same and been continuously improved for 20 years (or more).

Difficult for me to believe that a so high regarded stuff is poorly designed, or deliberately designed to be PS sensitive (to sell additional PSUs?).

Things must be far more subtle.
 
You're obviously correct regarding the db calculation, I forgot the log10...

Anyhow 44db is still considered poor for a regulator (the ALWSR is more than 100db), especially that the 317 on the Hicap has also poor rejection at high frequencies.

On the NAP102 they use cheapo Samwha capacitors (earlier preamps used superior SLCE capacitors). Don't know why, the bottom line is that they are a successful company.

I will not be surprised at all if they read this forum, and know how their amplifiers could be improved, but they probably have their own successful business driven reasoning.

I must admit though, that I was once very sceptical about the possibility to improve their designs, until I tried... ;)
 
nicpom said:
Difficult for me to believe that a so high regarded stuff is poorly designed, or deliberately designed to be PS sensitive (to sell additional PSUs?).

Things must be far more subtle.
Click to expand...
I think they are!

One thing is that - unlike many designs - the basic Naim line / gain stage circuit is not especially sensitive to PSU impedance, because its loaded with current sources but it does also have fairly poor PSU rejection, something like 18-22dB at worst estimate. So a very clean supply helps no end, a clean, low impedance supply is even better - and the easy brute-force way to do that is simple RC decoupling, because the R value can be tolerated easily in terms of audible performance. It's doubly effective because HF noise passes straight through 3-pin regs.

The poor intrinsic PSU rejection is no reflection the performance of the basic 5-transistor gain stages though; it's rather elegant in terms of excellent basic linearity, bandwidth and output drive. The PSu susceptibility then gets treated in a separate stage would be one way of viewing the PSU. A perfectly valid modular approach really, and one that happens to suit a business model really well.


PS. Most of Naims hiss comes from the slightly excessive pre/power gain structure BTW - lots of gain in the front end (preamp) simply exaggerates the contribution here. There's nothing inherently noisy about the gain stage or power amplifier designs, quite the opposite - drop the preamp gain (as described elsewhere on this board) and the hiss drops into the floor too.
 
Linearity of what? How do you define the linearity of a DC PS?
Click to expand...
Not picking on you nicpom - that is a really good question.

Yes, linearity in terms of output impedance vs frequency; or better IMHO output impedance vs. output current vs. frequency. If you are relying on the PSU to keep operating conditions 'clean' for the powered circuit (any audio circuit, not just Naim I would add) these things really matter to make the AC components on the supply 'dissappear' as far as possible.

It's quite easy to bent a bent transfer function in an active regulation circuit - and hear it as an artifact in the audio! Very low ESR 10-22uF caps on the output of an LM317 is an easy and popular way to try this at home...
 
Lots of very interesting comments. Many thanks to all.

About hiss, I read an old Julian Vereker post where he explained that Naim preamps high gain was originally set for phono signals. Nowadays the majority of sources are line level, and many speakers have high efficiency, so maybe they should consider to change philosophy.

Now the Naim approach is a bit clearer to me: they use RC decoupling because that particular preamp circuit can manage the increased power rail impedance. Then, they add stiffer and stiffer external PSUs, but always based on 3-pin regulators. In any case the RC filter seems to me a key component and should very carefully optimized, and made with the best components.

BUT, if a different PSU is used, an ALWSR or the advanced designs described here, should the RC-decoupling be removed? It seems to me that in this case its presence is useless or even harmful. If you have wide bandwidth, low impedance, and low noise, a direct connection should be the best.
 
If a different PSU is used, an ALWSR or the advanced designs described here, should the RC-decoupling be removed? It seems to me that in this case its presence is useless or even harmful. If you have wide bandwidth, low impedance, and low noise, a direct connection should be the best.
Click to expand...
You can remove it if the reg is very close to the circuit served; but if the 'better' regs remain offboard and some distance away, it can be a good idea to retain it - to ensure stability in the regulator. I agree with reducing supply impedance whenever possible though.

You can also play with the RC values to suit other pruposes. Using one local reg to supply several stages it could be preferable to keep a touch of RC filtering simply to separate the stages effectively from each other at HF. Another reason to keep a small series R value, even with a local reg, is to allow the use of very low ESR caps (like large oscons) for decoupling, when these would upset the regulator without; as little as 0R33 works for this. There are lots of iterations to play with :)
 
OK, so in practice:

- Naim solution is a sort of compromise of many technical aspects (cable length, PSU characteristics, grounding philosophy...), plus economical ones. Not necessary the best possible compromise, just one which demonstrated to be successful.

- If you change one component, i.e. the PSU, then you could (should?) revise all the others for the new best compromise. The RC filter is only one, surely.

- The more you change, the more you need to revise to obtain a balanced system. At the end of the day, you probably will obtain a completely different component. The point is that also the sound is completely different.

A question for all the tweakers: all these heavily modified preamps and sources, with local regulators, complex grounding, switching DC-DC converters etc. are still sounding like a Naim unit? Or are they closer to Linn, or Quad, or...? Are you obtaining an improved Naim or something else in a Naim case?

Regarding this, some months ago I made a test: as I have a source (CEC belt-drive player) with a variable output, I connected it directly to my NAP200. Previously I had asked Naim if this was safe, and they answered that it was perfectly possible, but the sound would be different.

In fact, the sound was completely different. Much more tightness, more detail, also better soundstage. Not Naim at all, however. It reminded me some american stuff, Spectral, Krell or so.

Just a thought, not a critic to anyone.
 
I think that's a very good summary: if you change the operating bandwidth(s), psu arrangements or components you are by definition moving away from Naim's own balance.
 
Probably, there are some limited modifications that can be done still maintaining the sonic presentation.

For example, the effect of a NAPSC can be obtained with a cheaper PSU. After all, it is only for control circuitry.

On the other hand, signal involved PSUs, bandwidth management and grounding are key NAIM features and I definitely think that modifying them you really are making another thing.

It is equivalent to buy another brand product.

Tweaking can be fun, and far cheaper, I admit.

My excuses to Teddy for having inserted such pseudo-philosophical comments into his highly technical thread.
 
My excuses to Teddy for having inserted such pseudo-philosophical comments into his highly technical thread.
Click to expand...

IMO they are very valid comments.

Basically, no matter what you alter you will shift away from the original sound to some extent (that's the point of modding though!). Then again, the 'naim sound' has itself changed considerably over the years, with the new kit having a quite different sound to the old 'chrome bumper' stuff.

Mr Tibbs
 
Not quite true Mr Tibbs. If we start with e.g. a 102, change the component values to those of an 82 or 52, and add a naim style power supply for each amp section, we will be going towards the higher end pre-amps won't we? Ok, so there are still differences in switching arrangments and the layout of the PCBs, but we will still be within what I would call the 'Naim design envelope'.

But I would agree that adding ALWSRs or Avondales supplies, or changing the circuit topology, we then start straying out of that envelope.
 
Not quite true Mr Tibbs. If we start with e.g. a 102, change the component values to those of an 82 or 52, and add a naim style power supply for each amp section, we will be going towards the higher end pre-amps won't we? Ok, so there are still differences in switching arrangments and the layout of the PCBs, but we will still be within what I would call the 'Naim design envelope'.
Click to expand...

yeah but no but... ;)

Mr Tibbs
 
After this truly illuminating discussion, I would have a question for Teddy.

You are in the best position to make a comparison between standard Naim PSUs, the ALWSR alternative and this last design.

Just to help persons like me to understand which is the personal best choice, as we obviusly cannot go to a shop and listen to them, could you summarize the sonic differences? As a Naim user, I am interested in PRAT, but also in refinement, transparency, bass control, and soundstage.

And are all these things different using these PSUs in a source and in a preamp?

It is clearly your opinion, but I highly value it, and so many others I am sure.
 
As the name of this thread suggests, this one is the best. When I say best, I mean in terms of being as close as possible to the real thing, that is, if you close your eyes and listen to a well recorded record, you can imagine yourself sitting in a concert.

The HiCap has the highest coloration of all, though some may find it pleasant to the ear. Its sound is very distinguishable. Voices may sound warmer than they really are, and details are lost. Comparing to the other regulators there’s some emphasis in the mid-bass region. As I said, I can understand that some people may like it with specific types of music (vocals, saxophone, etc).
I must say though, that Naim managed to get extraordinary results from a basic regulator (LM317), any time I tried to build a PSU based on the HiCap circuit I didn't even get close to it.

The ALWSR is a big step forward, suddenly you hear lost details, and clarity is improved. The results while using ALWSR are still somehow dependent on the transformer and reservoir capacitors, that is, they may sound differently in different configurations. Adding the VBE in front of them improve things significantly, both in the aspect of transformer/capacitor independency, and in the final result.

This regulator is the best in all aspects, it sounds the most natural of all, and it improves PRAT, clarity, bass is tight and full at the same time. The difference is not night and day comparing to a VBE-ed ALWSR, but there is a difference. As with any regulator, the results with this regulator depend on the component choice (capacitors, transistors).

Unlike preamp modifications, PSU modifications maintain the so called Naim sound signature. I think that the effect is more like going up with the Naim model. The Naim sound signature starts to change when you change the internal amplifier circuits. IMO, however, the higher you go on the Naim model, the more the sound signature disappears and converges with other vendors’ high end material; they just reach there via a different path… In other words, ultimately there shouldn’t be any difference between two ideal systems coming from different vendors; they should all sound the same, like the reality. :)
 
Never heard a standard HiCap in my system, but I have one of your VBE-ed ALWSR, not the latest version, and must say that when I connected it to may Nait 5 the difference was impressive.

Then, when I bought a NAC202/NAP200, the improvement was audible but less than I expected. The law of diminishing returns in action, I suppose.

In fact, listening to my source directly connected to Stax earphones, I can hear nuances that got lost if I connect the earphones to the tape out of the preamp, so audible improvements in the preamp section are possible in my system. And I am really interested in your recent developments, especially if, as you say, thay do not alter the fundamental Naim signature.

However, if it is true that all high-end system tend to converge, I begin wondering if the Naim path is a sensible one, economically speaking.
 
The Stax comment is interesting. For the last year I have been listening exclusively with headphones via a Stax 4040 / 006tII setup directly from a CDX or LP12/Linto. Was wondering if going through a pre would offer any advantage as I believe that the 006t has essentially a passive attenuator on the input as a volume control. Will have to dig out a preamp and try, but sounds like it's best not to use a pre.
 
One question, guys.

For the Fetlington, which 2SK117 is more suitable? The 2SK117-GR or the 2SK117-BL? Or the rated IDss does not matter in this application?

Thanks.
 
You must log in or register to reply here.


advertisement


Back
Top