Integrated Amp Reviews

Sidebar 3: Measurements

In line with Krell's design practice of recent years the KAV-300i is a low-feedback design, which means that we shouldn't expect extravagant results for damping factor or its more practical equivalent, output impedance. Nor is exceptionally low distortion likely. Nevertheless, Krell amplifiers aim to deliver clean, neutral power to a range of loudspeaker loads and while this base-level design isn't rated for 1 or 2 ohm duty, it does crack the whip powerfully into 4 and 8 ohm loadings.

Noting that my test sample was a European model with a nominal 230V, 50Hz supply specification, I powered it from the typical UK voltage of 242V, a well-within-tolerance 5% lift. Output readings for this unregulated supply amplifier will therefore have benefited by around 8%, depending on transformer quality and its regulation.

With that proviso, this is undoubtedly a powerful amplifier, especially considering that it is a compact, integrated type. Into a nominal 8 ohm load, single-channel driven, it delivered close on 200W continuous (23dBW) from 20Hz to 20kHz. The good 20Hz figure was a surprise in view of the modest power supply. The peak-program output approached 250Wpc—this is some little powerhouse!

A more demanding power test comprised firing up both channels with a tougher 4 ohm loading. Under these conditions, the output level fell by 2dB, rather more than is seen for the big Krell power amplifiers. Nevertheless, the KAV-300i was still putting out 250Wpc into 4 ohms; this is actually a greater amount of power than the power transformer's specified 400VA. This seeming impossibility is explained when the transformer regulation is taken into account. If you are prepared to push a transformer really hard, it will go on delivering more power, but at poorer regulation. For this design's total continuous output of 500W, the transformer must actually be supplying something like 750W.

Both this behavior and the relatively small internal heatsinks indicate that the KAV-300i will not be the amplifier of choice for continuous, flat-out duty into 4 ohms or lower loadings (footnote 1). This isn't a Krell dreadnought, but it is more powerful than many other amplifiers, particularly in its dynamic capability. There's certainly no shortage of peak current. That 250W peak program, 8 ohms, grew to 400W peak into 4 ohms with 620W peak available into 2 ohms. These figures are backed by ample peak current, measured at approximately ±45A.

It may be compact but there's nothing downsized about the very good load tolerance shown by the '300i. The output impedance measured a moderate 0.3 ohms—this somewhat dependent on the amplifier's thermal history, and higher than the specification—and was virtually constant over the audio frequency range. This is equivalent to an 8 ohm damping factor of 27. Mild changes in bass damping and impact may be noticed when comparing this amplifier with bigger amplifiers of lower output impedance. With source impedances below 0.2 ohms, however, these differences tend toward inaudibility.

DC offset was respectably low at the speaker terminals, 10mV left channel and 8mV right channel. I inadvertently tested the protection with momentary full drive into a short circuit. This abuse blew the 20mm supply fuse in the 3-pin IEC power input socket. Conveniently there was a spare already clipped in the little inbuilt sliding drawer.

I made distortion measurements both after a modest warm-up and after some full-power operation had heated up the amplifier, to an estimated 60 degreesC on the internal heatsinks. Bias levels had clearly altered at the higher temperature, and midband 1W distortion had increased from 0.035% (-70dB) to 0.07% (-63dB). This behavior is partly related to the low negative feedback, but neither distortion figure would be considered audible.

At full power, the harmonic distortion level was typically 0.07% at low and mid frequencies, rising to 0.56% by 20kHz. Again, I don't consider this rise to be aurally significant, and at lower powers, the 20kHz distortion level improved, to -57dB (0.14%) at 1W into 8 ohms, for example. The twin-tone intermodulation results—these are more important in subjective terms as a measure of high-frequency quality—were good, the 1kHz difference tone lying at -70dB (0.033%) at full power (fig.1), rising to -67dB at 1W into 8 ohms. However, some higher-order products can be seen clustered around the 19kHz and 20kHz fundamental tones, which might just take the edge off upper treble clarity.

Fig.1 Krell KAV-300i, HF intermodulation spectrum, DC-25kHz, 19+20kHz at 150W into 8 ohms (linear frequency scale).

Fig.2 shows the distortion harmonic spectrum for a 200Hz tone at 1W into 8 ohms (0dBW). This shows odd-order harmonics to predominate, mainly the third and fifth, with a rapid decay in level for higher orders. The second harmonic spectrum (fig.3) concerns the breakthrough of power supply ripple under a stressed power condition with a low-frequency tone (37Hz). The amplifier was driven to two-thirds the rated power into 4 ohms. The dotted vertical markers in this graph are aligned with the harmonics of the UK's 50Hz supply. Thus spectral lines that don't coincide with the markers are simply due to distortion of the 37Hz fundamental. There is one 100Hz supply harmonic present, but this is at a mild -80dB relative to full level.

Fig.2 Krell KAV-300i, spectrum of 200Hz sinewave, DC-2kHz, at 1W into 8 ohms (linear frequency scale). Note that the third harmonic at 150Hz is the highest in level.

Fig.3 Krell KAV-300i, spectrum of 37Hz sinewave, DC-200Hz, at 200W into 4 ohms (linear frequency scale).

Squarewave testing for stability showed a clean risetime, barely affected by adverse loading, even a pure 2µF of capacitance. This is one stable, tolerant amplifier.

Though the KAV-300i is capacitor-coupled at the input, the frequency response (fig.4) was flat and wide. I measured +0dB, -0.5dB, from 8Hz to 31kHz, while the benchmark -3dB or "half-power" points were at 0.6Hz and 90kHz. Channel separation was fine, if neither state-of-the-art nor to modern CD or digital standards. I recorded crosstalk at -68dB at 20Hz, -86dB at 2kHz, and -70dB by 20kHz. (These figures were somewhat dependant on input termination and whether I was using the balanced or single-ended inputs.)

Fig.4 Krell KAV-300i, frequency response at 1W into 8 ohms (5dB/vertical div.).

The KAV-300i was quiet both physically and electrically, the latter measuring better than -110dB for background noise relative to full power. Channel balance was better than ±0.015dB at high volume-control settings, with ±0.02dB accuracy even at a -60dB setting!

I won't elaborate too much on the behavior of the 256-step volume control except to point out that there was a deep -93dB of maximum attenuation while the next few settings ran -72, -68, -66, -62, -60, -58, -55dB etc., and that above -50dB, a more critical region, the resolution improved to 1dB or better. I checked a stack of settings from 0dB (full level) to -20dB and noted a fine resolution of either 0.5 or 0.75dB according to the position in the code. There was an almost irrelevant 0.2dB of backlash according to whether you were ascending or descending the volume scale. The backlash increased below -60dB, which initially confused me. The microprocessor appears to respond too rapidly to volume setting inputs—to access fine-level steps, a very quick action on the remote handset button is needed. I feel the internal volume-setting rate could be usefully slowed, though it should be noted that I only noticed this behavior on the test bench.

The '300i's input impedance was confirmed as being a desirably high 210k ohms, while the input also showed a wide dynamic range, accepting up to 9 volts before the onset of mild distortion. At full volume, 55mV input gave the IHF-standard 1W output, while 580mV produced full output into 8 ohms, This is medium sensitivity and is well-matched to modern source components.

The preamplifier section's output impedance was moderate, at 120 ohms. The preamp wasn't as powerful as a separate, fully specified Krell preamplifier, but it will drive a range of normal amplifier loads, putting out up to 8.5V of level. It performed equally well for both balanced and normal inputs.—Martin Colloms

---

Footnote 1: An IHF 1/3-power preconditioning run was out the question, a fact confirmed by Dan D'Agostino.—Martin Colloms