Relative Power Supplies

It's actually quite unusual to see a dual-rail SNAPS2. There's an unused regular in the SNAPS2, and when you hook it up the performance jumps significantly. It's still different from a Flat-Cap, though. I had posted an account of my perceptions on the old forum. Here's a repost of that message:

I recently upgraded my system at home from 3.5/Hi/102/NAPSC/Flat/140 to CDX/82/2*Hi/250. This freed up the 3.5, Flat-Cap, and 102 for resale. Since my office system is also partially Naim, for a while I was running with the 3.5/Flat/32-5/110, then Cambridge D500/32-5/Flat/110, then finally D500/32-5/110.

I didn't cry too much when the 3.5 left, as I knew it was "too good" for my office system (mostly used for background listening). Besides, my D500 is a pretty acceptable player. When it was the Flat-Cap's turn to go, however, it was a much harder transition, and the lower cost of the unit made me question my decision to sell it.

Once the Flat-Cap was removed, the sound from the 32-5/110 seemed to shrink to a pinched version of its former self. I should point out that with the Flat-Cap gone it was primarily dynamics and scale that suffered. After a time I was able to adjust to the change, and even to think it sounded "good" again.

Regardless, I could still recall how much better it had sounded. Then I learned of a SNAPS2 for sale, and after laboring over the decision I decided to take the plunge and buy it. Since the cost was less than my Flat-Cap's resent sale price, the decision was a little easier for me.

While I was at it, I decided to have the unit converted to dual-rail mode. The SNAPS2 has two regulators, one of which isn’t used. With a little rewiring and soldering, plus the addition of a 5-pin socket, it can provide two independent 24V supplies.

Once the SNAPS2 was plugged into the system, my initial impressions were that there wasn’t much difference. The scale was a little better, but nothing compared to the losses when the Flat-Cap was removed. However, the unit was very cold (temperature-wise), so I let it warm up for about 24 hours before doing further testing.

The second day showed a much different presentation. I listened to the system with the SNAPS2 in place, then yanked it. While removing the Flat-Cap make the whole signal shrink, it didn’t really affect tonal qualities all that much. In contrast, removing the SNAPS2 robbed not only size and dynamics, but also tonal realism: the bass wasn’t quite as full, the cymbals seemed tinnier, and vocals seemed hazier.

Reintroducing the SNAPS2 added a sense of natural realism that was quite enthralling. There also seems to be a much increased sense of PR&T; the music had a palpable driving quality that wasn’t there without the SNAPS2.

Therefore, I prefer the dual-mode SNAPS2 to the Flat-Cap. Although I think the Flat-Cap made it a little bigger and it was a touch quieter (i.e. “blacker”), the current performance with the SNAPS2 seems more “real”. BTW, the office setup is now Cambridge Audio D500, NAC32-5, SNAPS2, NAP110, NACA5, ended by Royd Merlins.

Catch you later!

-=> Mike Hanson <=-

Smilies do not a forum make.

I just had a Snaps recapped and dual-railed by NANA, and they said that was standard when they got one for service. Presumably, since they test by listening, they should have an idea how it sounds compared to a Flatcap.

OTOH, the Snaps sounded dreadful when it arrived, in many ways worse than powering from the 140. It took two days to sound even decent, and a week to sound good. So if they didn't warm it up/break it in, they wouldn't have a decent grasp of how it sounded compared to a broken in FlatCap.

--Eri

Eric,

I have currently 2 flatcaps and have opened them up several times. I have seen the innards of a hicap in a magazine once but never the snaps, so I can't comment on the latter.

The flatcap's circuitry is pretty simple, and a simplified version of the hicap's. The toroidal trafo used in the flatcap is the same 120VA one as used in the nait3/nap90. In fact it was even laballed Nait3/Nap90! Then comes a single bridge-rectifier for both rails, followed by a single 10,000uF BHC aerovox electrolytic cap. 2 voltage dividers follow, one for each 24v supply rail, made up of 3 resistors each. And also for each supply rail an LM317 regulator. That's basically it.

The hicap has a (much) bigger toroid... dunno how big though. Someone once told me 500VA, but I'm not sure. There are 2 separate secondary windings, one for each supply rail. Circuitry is like the flatcap's but now the 2 rails are entirely separated. So therefore 2 bridge rectifiers, one for each rail, then 2 electrolyt. caps, again one for each rail. Then the rest is like in the flatcap.

There are 2 small blue things next to the voltage dividers in my flatcap. They look like electrolytic capacitors to me as well.... if they are, they are probably connected in parallel with the voltage divider to snub out hf noise. This capacitance would then be multiplied with the beta value of the LM317 to make a "super-capacitor" as how some books put it.

Hope I answered your question... partly!!

cheers!
Steven

Eric,

although yours wasn't run-in I've no doubt that they will have heard some other SNAPS that they've modified in the past.

Cheers, Martin

I don't know electronic components very well, but I believe the Snaps has the transformer from the 110 (200 vA??), with the same single capacitor and regulation.

--Eric

Stephen,

I think your reading of the circuit is misunderstood.

The LM317's are three-terminal linear voltage regulators. The voltage dividers form part of the voltage setting for the LM317's, and the little blue devices are tantalum capacitors. There is usually one on the regulator output for improved transient performance and stability, and, as is the case with most adjustable regulators, one on the adjustment terminal ('voltage divider').

This second tantalum provides increased ripple rejection from the regulator.

Sometimes there is an input capacitor, but this is often not necessary when it is close to the main supply smoothing capacitors (the large can style caps).

There are numerous differences between versions, in terms of higher capacity, and improved noise performance as a result of device selection. I wouldn't be suprised if bandwidth was also a factor here, also, but obviously do not know the full gamut of tests that are performed on a *cap. It is just logical to assume that the above factors could have significant audible effects if addressed.

Interestingly, for those of you who have built their own PSU's, I've found, as a result of FFT spectral analysis of an LM317 regulator output (thanks to Martin Carrington for the article that instigated this avenue of investigation), that certain capacitor values can cause noise peaks in the audio bandwidth under certain conditions, some 15 dB above the regulator noise floor.

I'm sure it is just such detail that Naim attends to when designing their products, and why it is difficult to emulate this performance with homebrew kit.

If anyone wants details please email me.

Andy.

Andrew L. Weekes
alweekes@audiophile.com

Andrew,

thanks for the correction. Always good to clear up a misconception.

cheers,
Steven