Eagle-eyed readers blessed with -grade memories know that I've used GRAAF's flagship pre-amp off and on since reviewing it five years ago. The sonic performance of the GM13.5B made it one of my references for a tube pre-amp with balanced output, its appeal enhanced by cool looks, user-friendliness, utterly faultless construction* and every facility I need. So you'd expect a Mk II version to be something I'd welcome with glee. Which I would. Only the 13.5B II a Mk II version of the GM13.5B.

It's an entirely new pre-amp.

Oh, it may look similar, with the same frontal aspect and knob positions and sloped cut-out section on its upper left hand edge and peek-a-boo valve window and large glass plate on top. And you'd expect the slight change in the black finish, somehow more modern, less-glossy, more sophisticated. But side-by-side comparisons will only keep you believing that the GRAAFs are father-and-son, when the only other things in common between the original and the new are the model name and therefore the amount of gain: 13.5dB.

OK, OK, so they're both line level, all-tube pre-amps. But there it ends. Where the first had a single power transformer, the new has two. The original used one large motherboard; GM13.5B II contains five separate PCBs. The first used a half-dozen 2C51 twin triodes; these have been replaced by six 6922 twin-triodes. In the original, all outputs were parallel; now they're fully independent. All of the internal parts - caps, resistors and the like - have been changed for better alternatives as part of Ing. Mariani's ongoing research into component sound.

But the main reason for regarding the GM13.5B II as a different beast entirely are its novel passive/active modes. The aforementioned eagle-eyed types will by now have also noted that the Mk I's deep power-on push-button has been replaced by a flush, er, button. Well-spotted: it's a sexy piezo device like those seen on certain Audio Analogue models, with no travel and therefore it's a party trick. You just graze it and, hey, presto! The unit switches on. A green LED flashes above it for about a minute until all is stabilised. What's that got to do with passive/active? Nothing. Only that there's an identical piezo button on the other side of the volume control to choose the modes, thus setting the level of the pre-amp's gain between 13.5dB and 0dB.

Ordinarily, this would just mean feeding the input signal straight to the attenuator, no additional gain being provided by the preamplifier stages. And sceptics would ask, so why pay for all that other circuitry if you should find yourself preferring the lower-gain mode all the time? Remember: this is not the first pre-amp to offer a by-pass of the gain section. But the answer is simple. Passive pre-amps of the straight-through variety, while squeaky-clean and detailed, usually lack dynamics and sweetness; invariably, users with switchable pre-amps find themselves using both, not either/or, because different conditions dictate different solutions. But in the GM13.5B II, the 0dB line-level input signal still passes through the valves, so it still sounds like a tube pre-amp and you're able to determine your preference by having to assess only how dynamic character changes in using 0dB vs 13.5dB; the actual sound character is identical.

If we're being accurate, then the 0dB stage isn't really 'passive' at all because the signal is travelling through all of the pre-amp's active circuitry, only with no increase in level. It had not actually bypassed the workings between input and volume control. But I suppose the term appeared in the product's description because it would make identifying the two levels of performance somehow easier in the minds of consumers, much in the way that we no longer bother to distinguish between efficiency and sensitivity. And, also if we're being precise, you could say that 'passive' apply because the dictionary definition reads 'not acted upon', and in this case, that could refer to the absence of gain being applied. Or as my suddenly-aware-of-babes son just told me, when we were watching one of his fave shows (the rather bouyant ), in a scene where Xena is possessed and not acting herself, 'If she looks like Xena and she sounds like Xena and she kills like Xena, then she's Xena.' So I shut up.

Offering a switch to choose between a 0dB and a +13.5dB signal sounds like a long-winded way of - effectively - giving you a switchable volume control which operates at two levels: high and low. But note that there's a bubbling-under-the-surface cult which believes that one of the problems we've been having with line sources, especially digital ones, is that they're way too loud. One highly-respected guru recently told me that the best thing I or any other listener could do would be to pad down the output of all our CD players and DACs by a volt or so and it would improve things immeasurably. And it's in situations like this where the 0dB/13.5dB option becomes useful. Not that it's ever that clear, of course, because I never did find a hard-and-fast rule as to when I preferred either. As with all things, it turned out to be component-, cable- and music-sensitive.

With all the extras crammed in, I should add for any wishing to simply slot a GM13.5B II in place of a GM13.5B Mk I that the earlier unit's 410x320x105mm (WDH) no longer apply: the Mk II is 420mm deep, the extra 100mm needed to accommodate all the added goodies. Also altered is the rear-panel complement to match the new array of inputs and outputs. The original's front-panel controls consisted of rotaries choosing playback and record-out for tape, a balanced source and four RCA line inputs. Now the line reads Aux 1-6, with Aux 1 and Aux 2 being fully-balanced inputs. At the extreme right is the volume control and below it the piezo on/off switch. So now the back's gold-plated socketry includes two sets of XLR balanced inputs and gilded phonos for the remaining four pairs of line inputs, two sets of XLRs for output and two sets of single-ended outputs, PLUS the unusual feature of a mono-only single-ended output AND a dedicated subwoofer output. This will prove invaluable when auditioning subs in future in a non-A/V format.

To assess this baby in both all-balanced and all-single ended mode, I relied on the Krell KAV300cd CD player and the (reissued) McIntosh MC275 power amplifier, both of which provide single-ended balanced connection. I also used 'em mixed, with the source balanced and the amp single-ended, and vice versa, and there remains no doubt in my mind that balanced-throughout is the way to go if you can. Forget the absolute quietness aspect, because few of us use nasty cables or long-enough runs to make it an issue. What appeared to me to be painfully obvious were gains in two key areas, balanced operation providing (1) greater dynamic behaviour, including both speed from extreme soft to extreme loud and the distance between them, and (2) an overall sense of control and cohesiveness.

At no point did I prefer single-ended operation, but I must also add one other suspicion: that balanced operation appears to be more important between pre-amp and power amp than between pre amp and source component. One day, maybe more work will be done in this area, but it's academic because all preamps with balanced outputs seem to offer at least one balanced input, and there are enough source components and power amps out there with balanced operation, only a few of which are balanced-only. So, if you've got it, use it. And I suspect further that the only reason makers of balanced components even bother with single-ended option is to retain universality.

Also used for the single-ended usage were the Unison Research Smart 845 monoblocks and the California Audio Labs Tempest II SE CD player. Speakers throughout the sessions were Sonus Faber Guarneris, wired with A.R.T. cable, while my balanced interconnects included Mandrake and Kimber Select. For AC wiring, the GRAAF was plugged into the Kimber-wired ring, the CD player in the Transparent-wired ring and the McIntosh was powered by the Siltech ring. All sounded better than the standard-wired circuit.

First, it's gotta be said: the new one stomps the old one. I can't name a single area where the GM13.5B II didn't improve on the old, attesting to the worth of a five-year gestation period. Conversely, the GM13.5B of the old variety survived five years unchanged, which demonstrates just how brilliant it was and is. But here's what's on offer, whether we're talking balanced or single-ended.

For my money, the GM13.5B II finally makes real the absolute promise of an all-tube pre-amp with solid state virtues. It's as quiet and precise and controlled as ANY high-end solid-state pre-amp I've used, but with an added kiss of warmth you just don't get with an absence of valves. It sounds big and palpable and open, it's a real Palladium of a pre-amp creating a vast arena with no apparent walls beyond those in the recording. And it is blindingly quick. Digging out a handful of recent bluegrass and C&W items with ridiculously fast - and I mean heavy-metal-fast - plucking of exceedingly metallic-sounding instruments, the GRAAF provided both the speed and the correct ringing and twanging, with decays and full stops as clear-cut as the real thing. Better still, it keeps them all separates, quite a feat when you have a stage full of rednecks trying to out- each other.

But tubes are about warmth, and it was here that I got to savour the true worth of the GM13.5B II. Alison Krauss' voice - as unspoiled an instrument as was ever heard, so sweet it makes Dolly Parton sound like Bessie Smith - issued from those Italian sculptings with the lightness of down, while at the same time her backing band's instrumenst retained all the kick and weight they possess. A dazzling, glorious balancing act, truly a mix of strength and finesse and power, all at the same time.

But back to this unit's almost-USP: the 0dB setting. I actually found myself changing it from recording to recording, but I'm sure that I would eventually discover circumstances and partnering components where either 0dB or 13.5dB were preferred for the majority of the time. But that presupposes a normal life, in which a system is left unchanged for long periods. After two months in reviewer mode, I was still at a stage where the system sat unmolested for no more than two sessions in a row.

Whatever, I fell in love with the GM13.5B II. And I swooned when I found that (because the lira has dropped, not doubt because it's linked to the repulsive Euro) the price is only £3950 - just £200 more than the original! To hammer into you just what a bargain this makes it, let me tell you with hand on heart, wallet and Butcher Sleeve that its only rivals in my book are the Nagra PL-P and the Audio Research Reference One. And we all know what they cost.