Where the upshot on performance is enviable, the circuit itself is quite complex. Unlike a standard push-pull array where the phase splitter is a triode, here the job is handled by two 6BM8 pentodes plus a constant-current sink ECC88/6922 triode for their common cathode. For this the supply voltage needs to be at 600V so large voltage swings are handled with low distortion. The input to the phase splitter is DC-coupled to the first stage's output. This means the entire stage rides at 90VDC, another reason for the high supply voltage. From the splitter stage, the output stage is driven through a cathode follower for which Andreas uses the triode section of the 6BM8. This tube is commonly referred to as an audio triode with a beam pentode. The already mentioned bias adjustment on the back of the amplifier sets the negative grid bias of the four cathode follower triodes to provide the desired bias voltage for the four KT power tubes.

In the ECC88/6922 first stage, there are two inputs. One is at the tube grid where global feedback gets applied, the other at the triode's cathode where the input signal from the second half of the double triode gets injected. This makes for a cathode follower with a common cathode that has a constant-current sink. Andreas chose the ECC82/6922 for their stability under global feedback which is an integral part of his circuit. All of it depends on a high-quality stable power supply. Here the power transformer sports ten secondaries as well as taps for 110, 120, 220, 230 and 240V - yet another complex transformer. The power supply also feeds an HT regulator for the screen grids of the output stage; and one for the phase splitter. For the latter to work, Andreas reached deep. The regulator needs to deliver 600V and to run off a +700V supply. Here a 6C19π becomes series pass element to place a variable element in series with the load. By changing the resistance of the series element, the voltage drop across it can be varied to ensure the voltage across the load remains constant.

Another HT regulator is used for the input stage. This is tricky because there is a 120V output and no less than a 600V input from the previous regulator. Andreas committed a 6BM8 as another series pass element to overcome the large voltage difference. The 6.3VDC heater supply is regulated with transistors just as are the three negative bias supplies for the signal stages. Finally, there is one negative bias supply for the output stage. All in all, this amp is complex. The designer encourages experiments with the output tubes but caution should be taken when rolling any other tubes. Replacing the 6BM8 with an ECL82 for example can end up badly as the 6DBM8 is able to handle the 600V whilst the ECL82 is only designed for 300V.

By now the familiar DHL courier was back at our doorstep with the same crate but clear markings on it like fragile and this side up. Had these done the trick? We'd soon find out when the amp was unpacked again and lifted into position in the listening room. NVO had now shipped the amp with the 22 tubes plus some spares in their original boxes or individually wrapped in bubble foil. See below. Each tube was clearly marked so seating it in the right socket was easy, the more so as we had received a clear tube layout sheet.

Once all tubes were seated, we let the amp which was cold from transit adapt to our room's temperature and humidity. The next day was final check/bias day. Once connected to the 240V coming from the Dutch grid and powered up without a load, we went over the innards and listened for funny sounds. All switches were solid, all tubes glowed nicely and the amp proved mechanically completely quiet. After ample warm-up, we took the multi meter and measured the bias settings, having to adjust only two out of the four tubes and then only marginally. Andreas provided guidelines for the bias of the KT-120 to prevent under bias that might induce a fuller sound but early distortion; or over bias that might cause a thin cold sterile sound.

Time for first tunes. With lots of power on board but completely unknown start-up behaviour, we designated our standard-efficiency Sounddeco Alpha F3 loudspeakers as a first match [above]. Our blunt rationale was that if anything went bad, its drivers would be easier and less expensive to replace than our widebanders. After switching off, letting the amp cool down a little and connecting the speakers and input from the La Rosita Beta streamer, we finally switched the amp on with a load to see the red power LED change to green after about a minute, accompanied by a faint boof as the circuit went online. NVO conveniently include a generic Seki remote for volume and mute. When the amp arrived for the first time, we'd already noticed something slightly peculiar, a bit like French books if you will. French books print the titles on their spines 180° out of sync with most other countries. When going over a bookshelf, we are used to tilting our heads right. With French books, you cock your head to the left; Gallic idiosyncrasy no doubt. With the NVO ISA-2, the volume knob sat on the left, the input selector on the right. To our recollection, most amplifiers do it the other way. Did this indicate that Andreas was left-handed?