DIY Power Supplies

I cannot say that building your own copy will give superior performance to a Naim supply (I haven't done it) but I am confident that you will certainly get 90% of the way there at 10% of the cost. Not a bad deal, really.

However, you have to know what you are doing - following design guides from Maplin won't cut it in high-end audio. Attention to detail is crucial, especially to grounding and interconnects and component choices. This is what you are paying for when you buy Naim. But it is a heavy price.

BA

Please be very carefull on this.

A couple of years ago, I visited Tanais Foxes house in order to listen to his system which was a modified LP12 (Ninja) stacked on a pile of Mana.

The sysytem (Naim based)sounded good, in fact superb, and then TF switched some standard Naim Hicaps for one of his modified "mycaps" which he named Epona.

Tanais has a hobby of modifying PSU's to suit his own ears and they sell for half the price of the equivalent Naim product under the Epona label.

Neither my wife nor I liked the sound of these modified "mycaps", in fact we thought they were edgy and over hyped in sound. Eventually, Tanais and I struck a deal, I purchased one of the Naim Hicaps in standard form from him. There was no way I could have lived with the sound that came from the Eponas.

TF is an extremely intelligent man who has a good working knowledge of electronics, is heavily into music and I have a lot of respect for him.

However, IHMO it is impossible for one man to produce a component which sounds better to most people, in competition with a company such as Naim who have spent millions in developing and perfecting the product. You may be one of those who prefers the sound of a modified PSU, but the odds are agasinst it.

If cash is an issue.......purchase a second hand Hicap. Its much safer and you can always get it serviced and brought up to new standard by Naim for about £90.00.

You obviously like the Naim sound so carry on buying Naim.

Regards

Mick

Problem is, batteries tend to be quite noisy, relative to a properly-designed rectified AC supply, and you might still need a voltage regulator downstream, thus cranking your initial voltage requirements up a notch.

After all the effort you'd have put in to the project, you'd probably have ended up with a FLATCAP 2anyway, with a lower-noise spec.

Dave Dever, NANA

As I said, I would have been too chicken to try even if I knew what I was doing but it appears the evidence against is overwhelming anyway.

I'll wait for the Hicap!

Nice to hear from you.

Please be honest....when you were listening to Mr Pogs system at 3.00am.......had you consumed a few drinks. That can make a lot of difference to ones overall impression.

Also I believe that this system is a Stealth Amp on a pile of Mana.......phase 50 or whatever, so is a true compaison possible. Could you accurately differentiate and critique a PSU, be it Epona or Naim. It would just be lost in the system. You were listening to a very unusual set up and normal comparison is unrealistic.

Regards

Mick

Here's my view, and it's largely in accordance with BAM's except I've actually done it.

It is quite easy to design a supply that will provide a significant improvement from not having one. It is not as cheap an option as one might suppose though, particularly when s/h equipment is considered.

Aesthetically it's often difficult to achieve an appearance one is happy with at a sensible price.

Implementation and detail are more important than specific circuits, although I don't like batteries, they are noisier than people expect. once you start to draw current from them they exhibit more AC content than expected.

My own PSU, using a conventional rectified, regulated design exhibits very low AC content across the entire audio band, but it takes some work, and suitable measuring equipment to acheive this. Largest noise component is the 2nd harmonic of the mains frequency, and double regualtion has brought this down considerably.

'Improved' measured performance doesn't always result in better sound, and it's frustrating to try and understand why.

My own Diy-cap sounds great (although I've only compared it to my Flatcap 2), but it's a LOT easier to buy a second hand SNAPS (from £70) and recap it yourself. It's less work, cheaper, looks better, and still sounds great.

Do it if you're confident and enjoy the work or are doing it for self-education, otherwise take the easy option!

My 2 pence worth.

Andy.

I've just looked at the link you posted, I've talked to Hon and was the one that recommended he buy a SNAPS. Unless you do the earthing correctly you will not acheive the full benefit of an external supply.

I'm sure he won't mind me saying (since he admitted it himself) the build quality of his is not in the same league as mine.

Adding the cost of buying new SNAICS for a DIY supply makes the whole thing much less economic than you may at first think.

Andy.

Arye

The Hi-Cap supplies 24V DC. You can't patent that, if you want to make your own then go for it. I would think the biggest problem is getting the earthing done correctly and not mess ing up the signal by passing it through the power supply. That's probably why it's so tough to get right.

An issue on batteries that also means they are not really suited to hifi applications is that they generally show quite high internal resistance. This impies that you can not expect a constant voltage supply which is what you are after.

Matthew

Arye

If you know electronics, and can find suitable parts, it's actually a very enlightening project. (And a valuable one if you're on a budget, too). I'd say it might give you even more respect for what Naim does. Those who say, "Naim's making a killing on their power supplies" are totally wrong. The devil is in the details, and that's where the genuine Naim product wins over any homebuilt version.

M

PS Ayre: I think it is making a big assumption to imply that no forum members can design as well as Naim engineers. Remember, Naim is a small company in Salisbury with £6m pa turnover: it has a limited development budget.

quote:

---

How do you go about choosing your replacement caps for s/h Naim psus?

---

Depends upon application.

For reservoir / smoothing it's usually based on an examination of what is being replaced (most of these caps are easily identifiable) and then a decision to replace with same, or uprate slightly (given that replacement devices are often smaller) to reduce ripple pre-regulation. It's usually something from the BHC Aerovox range, of suitable rating.

Transformer rating / heating (I squared R) losses need to be considered (increasing caps increases transformer heating), along with rectifier ratings. You'd have to go fairly mad for it to be a big issue though.

Tants are usually replaced with AVX Kyocera parts where available, although I've used others in PSU's with good results. Changing ESR radically can have untoward results on regulators (actually increasing noise amplitude under certain conditions).

I've had superb results using LM317AT's on my SNAPS, seems a big improvement on the original reg's, although papaer noise spec. is no better.

I'm just building a stand-alone regulator noise test fixture - I'll let you know what results I get from different types.

I'm also building a low-noise discrete regulator for very low noise performance, it'll be interesting to see how that turns out, sonically.

Andy.

Imagine you were able to draw a step current from the supply, what would its output voltage waveform look like? Would it be a perfect mimmic of the current or would it ring or distort? Alternatively, if you drew an ac current from the output across a wide frequency range what would the voltage amplitude and phase response look like?

I suggest that an ideal psu would look like a perfect voltage source in series with a perfect resistor. If the series impedance is not resistive or it contains resonances this may affect the performance of the device being powered. Note that resonances may be caused by the output capacitance of the supply in series with the inductance of the dc cable that goes to the device being powered unless there is adequate damping resistance.

Incidentally, these measurements are applicable to the output of a power amplifier as well. Things get very complicated once there are "active" circuits in the system such as electronic regulators. The regulators contain feedback systems to simulate low impedance and they contain sources of non-linearities.

Personally, I'm not a big fan of remote psus. I think it inherently better to control the power source as close to the circuit being powered as possible. I also think that each amplification stage in a circuit should, in general, have its psus isolated from the previous and next stages to reduce interactions. All this costs more money, of course, but aims to keep the audio signal as uncorrupted as possible. I wonder why Naim don't do this: have a pre-amp with integral psus.

BAM

I agree with regard to transient response, although in the case of Naim preamps the circuits are class A with extensive use of current sources, which means that transient loads to the PSU correlated with the audio signals are much reduced when compared to class B stages. There are other 'inputs' to consider though.

As for 'perfect' supplies, I'm inclined to agree again. My own experiements with active noise shunts proved them to be a very effective and efficient means of reducing noise to inaudible levels. Particularly at the frequencies below which the decoupling is effective. The sonic results were not pleasant though, probably due to the interaction between the circuit being fed, and what is, in effect, another amplifier / active circuit. I need to revisit this one since it was so effective I'm not inclined to leave it unsolved.

As for capacitance and inductance of cables, the following circuitry is part of the design, so if done properly it shouldn't present a problem. The output inductance of the regulator is possibly more significant. I've proved (with FFT analysis of my own PSU) that the effective inductance of the regulator can easily be made to resonate with the output capacitor, the Q of which is determined by the ESR of the capacitor, hence my original comment w.r.t. changing ESR radically. The load current and a host of other parameters all have an effect on the effective regulator output inductance.

On interactions between circuits, in general I'd agree, but in specific circumstances (particularly PA's with over-all feedback) the PSU has to be carefully considered as an amplifier input. My own simulations show that whilst one can improve PSRR by increasing regulation, the seperation of different areas of the circuitry, which have their own overall feedback can bring about problems in terms of stability margins. PSRR may be better, but if the reduced amplitude output signal is long-decay ringing, which has little direct correlation to the input signal then the results are unlikely to be good.

Naim do, I believe, have pre-amps with internal PSU's, the higher end pre's have local regulation (in fact even the 112 has, I think). But placing noisy AC components in close proximity to sensitive wide bandwidth circuitry is not clever. One at least needs to feed a supply with low AC content into the preamp, which is what a *cap does.

The earthing is another significant factor that should not be under-estimated. Naim are the only company to my knowledge that pursue the concept of an absolute system reference to such a degree. I'm certain that this is a significant factor in the end result.

Let's face it, we'd all like to think we know better, but Naim's engineers aren't stupid, and would have considered exactly the same things. This implies to me that there is a good reason for doing things the way they do. My own experience has shown this is the case. It's easy to incorporate the ideas used in higher end components into the lower end ones, but this is part of the compromise a component has to have at a given market point.

Andy.

My memory may fail me but I thought the pre-amp gain stages were very simple common emitters with resistive loads and capacitive coupling due to their single-ended psu. Perhaps the circuits have changed. In any case there will always be some coupling to the psu either directly through resitors or via current source capacitance. Hopefully this is very small in which case, as you imply, it relaxes the demands on the psu.

"The load current and a host of other parameters all have an effect on the effective regulator output inductance."

Nasty semiconductors!

"the seperation of different areas of the circuitry, which have their own overall feedback can bring about problems in terms of stability margins."

Err...perhaps. I'd argue that solving stability problemns via psu feedback is not the right thing to do. The feedback signal will pick up stuff in the psu lines that you really don't want in the signal path. This is especially true in the amps themselves although I notice that Naim choose not to isolate the LTP and v gain stages from the psu (not sure about the NAP500).

"But placing noisy AC components in close proximity to sensitive wide bandwidth circuitry is not clever"

The transformer and rect don't need to be that close.

"One at least needs to feed a supply with low AC content into the preamp, which is what a *cap does."

What about the series impedance of the cable? It's one thing to feed ac into your preamp but it is also one thing to give the pre-amp a floppy impedance.

"The earthing is another significant factor that should not be under-estimated."

Absolutely.

"This implies to me that there is a good reason for doing things the way they do."

I'm sure there is.

"but this is part of the compromise a component has to have at a given market point."

I'm sure it is. Naim engineers have other priorities as well as sound quality. The home designer just has to worry about sound quality...oh, and his/her warranty!

quote:

---

I'm sure it is. Naim engineers have other priorities as well as sound quality. The home designer just has to worry about sound quality

---

bam,

Whatever other priorities Naim have, it would make no sense for them to champion separate power supplies in the face of the rest of the market if they didn't provide a noticeable sound quality benefit.

One only has to see the benefit of moving the power supply further from the preamp to understand why they do this.

cheers, Martin

Nothing remotely like a simple common emitter circuit has ever been used to my knowledge. I'm not prepared to discuss Naim's circuit details here though.

"Err...perhaps. I'd argue that solving stability problemns via psu feedback is not the right thing to do. The feedback signal will pick up stuff in the psu lines that you really don't want in the signal path. This is especially true in the amps themselves although I notice that Naim choose not to isolate the LTP and v gain stages from the psu (not sure about the NAP500)."

I wasn't particularly clear hear, but I've been modelling the effect of running the output / driver combination from one supply, and running the rest of the PA stages from a stiff regulated supply, since this is easier to implement, cost effectively into existing space / PSU limitations. It can have some pretty odd effects though if one starts to feed signals (i.e. noise spikes) into the unregulated output stage. This is probably why Naim go the whole hog on 250 / 135 and regualted everything.

"The transformer and rect don't need to be that close."

True enough, but there's still going to be some residual AC content that can be removed with greater finesse by a regulator than a dustbin full of electrolytics!

"What about the series impedance of the cable? It's one thing to feed ac into your preamp but it is also one thing to give the pre-amp a floppy impedance."

You assume that this increased impedance is a hindrance, not a benefit - I believe you're wrong in that assumption, particulary bearing in mind the actual circuits used. The DC impedance of the cable is an irrelevance compared to the impedance of the circuit it feeds.

Andy

You are joking, right?

So email me traderbam@yahoo.co.uk

"It can have some pretty odd effects though if one starts to feed signals (i.e. noise spikes) into the unregulated output stage."

Send me the design and I'll be happy to help you figure out what the problem is.

BAM

quote:

---

Martin: "Whatever other priorities Naim have, it would make no sense for them to champion separate power supplies in the face of the rest of the market if they didn't provide a noticeable sound quality benefit."

BAM: You are joking, right?

---

Rather than glib know-it-all comments, it would be much more constructive and useful to have your reasons for why you think you know better than the Naim after their 25 years of product development and improvement, bearing in mind the obvious commercial disadvantage of having to have an ever increasing number of boxes and cables as one progresses through the upgrade path

If I may butt in [as someone whose knowledge of electronics is currently zero and which never reached much beyond A level physics]:

Andy Weekes said:

quote:

---

...but this is part of the compromise a component has to have at a given market point.

---

To which BAM replied:

quote:

---

I'm sure it is. Naim engineers have other priorities as well as sound quality. The home designer just has to worry about sound quality...oh, and his/her warranty!

---

It is not implicit in BAM's statement that Naim's engineers are inferior to BAM or anyone else. Nor did BAM say that external power supplies don't give a significant improvement in sound quality.

As far as I can see, BAM was simply recognising that Naim [and, for that matter, every other hi-fi company] has to take into account many things other than sound quality in making commercially succesful products - which the home designer does not.

Everyone - please apologise to each other at once.

Ermmm... Sorry - getting a bit carried away - I shouldn't have said that.

--Jeremy

"the obvious commercial disadvantage of having to have an ever increasing number of boxes and cables as one progresses through the upgrade path"

I cannot disagree more.

I submit this is a MAJOR COMMERCIAL ADVANTAGE for Naim. I submit one of Naim's key differentiators from competitors is the DIY nature of their systems - the allure of the tweak and add-on. The attraction to the customer of being able to craft their own system step by step. It's fun! Even rewiring one's house is part of the fun. More components, more revenue.

Is this wrong? Not really. It's business. It's a targetted strategy. It works pretty well for a particular market segment. Although it goes against the grain I'm not prepared to assume that "25 years of experience" has only gone into the pursuit of sound quality. Just look at Sony - they have even more experience.

BAM