Laboratory test report
When Newport Test Labs started testing the Lavardin IT’s power output, the lab techs quickly discovered this would not be possible using the standard design, because the 2-amp fuses on the internal PCBs blew whenever the amplifier approached rated power. After advice from Lavardin, the fuses were temporarily replaced with a larger, 3.15A fuses, even though Lavardin warned that these, too, ‘ may also blow during power measurements.’ When questioned about the use of the low-value fuse, the company said that it was chosen because this 2-amp rating: ‘ provides a real protection when playing music and has never made any problem to any customer for the past 19 years.’ The company was at pains to point out that if this fuse has to be replaced, it should be with a standard transparent glass fast-blow 2-amp fuse, stating specifically that ‘ neither ceramic nor esoteric high-end fuses should ever be used.’
Once the higher-value fuses were fitted, the Lavardin IT’s power output was able to be measured at three different frequencies and three different loads and you can see the surprising results of this testing in the power output table. I say ‘surprising’ because the results are exactly—and I mean ‘exactly’—what I’d expect to see from a valve amplifier, not from a solid-state amplifier. In fact I’ve never seen this type of result from any solid-state amplifier. As you can see, power output essentially remained the same irrespective of the load impedance, so the power output into 2Ω loads was almost identical to that into 8Ω loads. Another peculiarity was that the power output at high frequencies when driving 8Ω loads was less than half the unit’s rated power, so the amplifier was only able to deliver 22-watts at 20kHz into 8Ω loads, whereas at 1kHz it was able to deliver 52-watts per channel into 8Ω with only one channel driven, and 46-watts per channel into 8Ω with both channels driven. This, too, is very valve-like behaviour. When Newport Test Labs examined the both-channels-driven into 8Ω high-fre- quency behaviour of the Lavardin IT more closely, the lab discovered that it will deliver 45-watts at all frequencies from 1kHz up to 10kHz, after which output power drops with increasing frequency, to 37-watts at 15kHz and then to 22-watts at 20kHz, as shown on the accompanying table.
When driving 4Ω loads the amplifier was able to deliver almost the same power output at 20kHz (45-watts) as it could at 1kHz (52-watts), while when driving 2Ω loads, it delivered slightly more power at 20kHz (50-watts) than it could at 1kHz, where it was only able to deliver 47-watts per channel.
I can’t explain these results, because they make it appear like the Lavardin is not working like a standard voltage-source amplifier, but somewhat akin to a unity-coupled design.
Looking at the distortion spectrograms, I can say that harmonic distortion is very low, and at an output of 1-watt into 8 , at least, is completely atypical of distortion spectra I’ve seen from any other amplifier when driven into the same load at the same output. What’s unique is that there’s no second harmonic distortion at all, and almost no third harmonic distortion either. The two obvious distortion components are the fifth and seventh harmonics, each at around at –95dB (0.0017%). All other harmonic distortion components are more than 110dB down (0.0003%). Driven into a 4 load at one watt, distortion is still low, but the distribution of the distortion components is more typical, with a second harmonic at –110dB (0.0003%), a third at –97dB (0.0014%) and all other harmonics at around—or more than—110dB down (0.0003%).
When the Lavardin IT is delivering 52-watts into 8 it exhibits a more familiar distribution of harmonic distortion components, with a dominant second-harmonic component at –95dB (0.0017%), a third harmonic at –98dB (0.0012%), and fourth and fifth harmonics both at –110dB (0.0003%), after which all higher-order harmonics are hovering around –120dB (0.0001%). At 57-watts into 4 (Graph 4) distribution of distortion components is similar to that into the 8 load, but distortion has actually reduced a little in level compared to the 8 result. That said, overall distortion at these output levels is still very, very low, with Newport Test Labs recording an overall THD+N figure of just 0.001%, ‘way below the threshold of audibility.
Intermodulation distortion (Graph 5) of the Lavardin IT was also exceedingly low. Either side of the two test tones at 19kHz and 20kHz (1:1) used for this test, you can see the adjacent unwanted sidebands are both close to 105dB (0.0005%) down, and the other sidebands around 105–110dB down. Significantly, the unwanted regenerated difference frequency (at 1kHz) is around 102dB down (0.0007%).
The frequency response of the Lavardin IT, as measured by Newport Test Labs, was certainly flat across the audio band, both when driving a standard non-inductive 8 resistor (black trace) and when driving a load that simulates that of a two-way bass reflex loudspeaker (red trace), but the response rolls off quite quickly at high frequencies, so it’s 1dB down at 19kHz when using the simulated speaker load, and 1dB down at 18kHz when using the 8 laboratory test load. Both traces are 3dB down at 33kHz. At low frequencies, the frequency response of the Lavardin IT was around 1dB down at 4Hz and 3dB down at 2Hz.
Channel separation was excellent, with Newport Test Labs measuring it as being 119dB at 16Hz, 97dB at 1kHz and 72dB at 20kHz. All are good results. Interchannel phase errors were also exceptionally low.
The signal-to-noise figures measured by Newport Test Labs were also excellent. Referenced to a 1-watt output, the lab measured the signalto-noise ratio of the Lavardin IT at 85dB A-weighted, improving to 101dB A-weighted when referenced to rated output. However, most of the noise was outside the audio band, as you can see from the noise floors on Graphs 1–5, which are at around –120dB when referenced to 1-watt and down at –140dB when referenced to 50-watts. Most of the low-frequency noise is mains-related, as you can see from the left hand side of each graph.
I was surprised that Newport Test Labs measured the Lavardin’s output impedance as being 0.09 (at 1kHz), not because of this isn’t a good figure, resulting in a damping factor that’s high enough to guarantee control over any loudspeaker you connect to it, but because I was expecting it to be much higher, not least because Lavardin itself says that it’s 8 in its specifications. If it had been 8 , I would have taken a guess at at least one of the secrets of this amplifier’s design, but because it was measured at only 0.09 , I am still left in the dark.
Square wave performance was good, but does show the bandwidth limitations of the design quite dramatically, even on the 100Hz square wave, where the tilt in the top shows the low-frequency restriction, and there’s even a little rounding at the top of the leading edge showing the high-frequency limitation. On the plus side, no phase shifting is evident at low frequencies. The 1kHz square wave is rounded but otherwise fine, while the 10kHz square wave is so rounded it’s beginning to look like a triangular wave. When loaded with a highly capacitive load, the Lavardin’s performance was highly unusual. There is some ringing, which is to be expected, and it’s nicely restricted and nicely controlled, but it seems to be somewhat elongated, which is certainly atypical: not the same as you’d get from either a solid-state or a valve amplifier. As you’ve no doubt already grasped, the Lavardin IT returned a set of seemingly contradictory results during Newport Test Labs’ tests. Its power output is lower than specified almost across the board, and it sags inexplicably at high frequencies into 8 loads, but not into 4 and 2 loads. Distortion is exceptionally low (both THD and IMD), and the signal-to-noise ratios are high. And although the frequency response rolls off earlier than is usual for a solid-state amplifier, it starts rolling off at such a high frequency, and so gradually, that it would not be perceptible as a ‘roll-off’ per se. The Lavardin IT’s performance on Newport Test Lab’s bench was so unusual that I can only guess that when he designed the Lavardin IT, designer Gérard Perrot deliberately set out to build an amplifier with the frequency response of a classic British solid-state Class-AB amplifier from the 80s, the super-low distortion of modern Class-D amplifier, and the output-stage characteristics of a single-ended triode valve amplifier. If this was indeed his intention, he certainly succeeded! Steve Holding