Laboratory Test Report

2023-04-17 - Steve Holding

Readers interested in a full technical appraisal of the performance of the OAD Ultrafidelity Vajra power amplifier should continue on and read the LABORATORY TEST REPORT published on the following pages. Readers should note that the results mentioned in the report, tabulated in performance charts and/ or displayed using graphs and/or photographs should be construed as applying only to the specific sample tested.

Newport Test Labs measured the power output of the Vajra using an eight-ohm non-inductive laboratory test load as 180 watts per channel at all three frequencies tested — 20Hz, 1kHz and 20kHz, which matched OAD’s specification exactly. When using a four-ohm non-inductive laboratory test load, Newport Test Labs measured power output at exactly 360 watts per channel at 20Hz, 1kHz and 20kHz. These results are tabulated below and shown graphically as a 3D bar chart on the next page.

When the lab switched from single-channel testing to driving both channels simultaneously, the lab measured output at 162 watts per channel into eight ohms for all three frequencies tested, and 272 watts per channel into four ohms for all three frequencies tested. These results are also tabulated below and shown graphically as a 3D bar chart on the following page.

The di¤erence between the two sets of results could be because of the increased demand placed on the Vajra’s power supply, or that Newport Test Labs did not maintain the mains voltage at 240 volts, which it has ceased doing now that the Australian standard is 230 volts, and the voltage supplied at the mains is always slightly higher than this. (If you’re one of the many who thinks it’s 240V, you’re mistaken: back in 1983 Standards Australia adopted a 20-year plan to convert Australia from the nominal 240 volts grid to a nominal 230 volts grid in order to align with European Standards.) Or it could be a combination of both. Either way, it’s of no practical significance whatsoever, because the di¤erences amount to only 0.4dB in output for the eight-ohm results and 1.2dB for the four-ohm results, both di¤erences that are far too small to be audible.

The Vajra has a very extended frequency response, with Newport Test Labs reporting that under its test conditions, it extended from 1Hz to 148kHz –1dB, and from 1Hz to 290kHz –3dB. Across the audio band the response was even flatter, as you can see from Graph 7, where the response into a non-inductive eight-ohm test load (the black trace) is just –0.1dB at 5Hz and 0dB at 20Hz, and just 0.05dB down at 20kHz and 0.18dB down at 40kHz. This means that when normalised, the audio-band frequency response is 20Hz–20kHz ±0.025dB.

The red trace on Graph 7 shows the Vajra’s frequency response when it’s driving a test load that simulates a load presented by a typical two-way sealed enclosure loudspeaker. Newport Test Labs uses the same test load as Stereophile magazine, a circuit that was originally developed by Ken Kantor (NHT) and subsequently modified by John Atkinson by the addition of a Zobel

network (a resistor and capacitor in serPieosw) the input terminals. If you’re interested in what the circuit looks like and the load it presents, you can find both here: https://tinyurl.com/Kantor1z

As you can see, the Vajra’s response into this simulated loudspeaker load tracks its response into a resistive load almost perfectly out to 4kHz, after which it drops to 0.1dB below it out to 30kHz, and after which again it is slightly higher out to the graphing limit. This slight drop is too small to be audible. The normalised response with this load is still 20Hz – 20kHz ±0.04dB.

Channel separation was outstandingly good at 1kHz, with Newport Test Labs reporting a measurement of 113dB at this frequency. Separation diminished at the frequency extremes, as you can see for yourself from the tabulated chart, but results of 71dB at 20Hz and 71dB at 20kHz are still figures that will guarantee perfect stereo imaging and no audible bleed from one channel to the other. Channel balance was a brilliantly good 0.016dB at 1kHz, demonstrating both excellent design and quality control.

Interchannel phase accuracy was also outstandingly good, with the OAD Vajra returning errors of just 0.07 degrees at 20Hz, 0.05 degrees at 1kHz, and 1.05 degrees at 20kHz.

Newport Test Labs measured distortion at 1kHz at an output of one watt into an eight-ohm noninductive load, the result of which test is shown signal at 1kHz is at the left of the graph. You can see that the related harmonic distortion components are all more than 100dB down, so overall distortion is less than 0.001%.

As for the individual harmonic components, the second harmonic (at 2kHz) is around –102dB (0.00079%), and the third (at 3kHz) is at –104dB (0.00063%). Even though these levels are too low to be audible, they are benign in that they’re ‘good sounding’ distortions due to being harmonically euphonious with the fundamental (that is, they’re the octave and the fifth). The fourth and fifth harmonics are down at –118dB (0.00012%). As you can see, other harmonic and non-harmonic distortions are visible above the noise floor higher up in frequency, but all are more than 125dB down, and so would contribute less than 0.00005% to the overall THD (which, as you can see from the tabulated results, was measured as being 0.007%).

On Graph 1, the noise floor is very, very low, at around 140dB down referenced to just a onewatt output, except at low frequencies, where some power supply noise is visible at the extreme left of the graph. It appears that there is some 50Hz signal that is 90dB down (0.00316%), after which the harmonics are mostly more than 100dB down (0.001%).

Reducing the load impedance whilst still maintaining the output level at one watt resulted in the performance shown in Graph 2. You can see that the second harmonic has remained at –102dB (0.00079%) but the third harmonic has increased a little in level to be –92dB (0.00251%), where it would still be inaudible. The lower impedance has also caused the odd fifth, seventh and ninth harmonics to increase in level, all to around –105dB (0.00056%). Even higher in frequency, some distortion components are now above –120dB (0.0001%).

Graph 3 shows the performance of the Vajra when it’s delivering 20 watts of power into a noninductive eight-ohm load with a 1kHz test signal. The most obvious di¡erence is the lack of noise inmtehtea-ocuhtaprtu.cto—mnoise between 3kHz and 20kHz is all more than 140dB down, which is an amazingly good result. (Note that this means noise at each frequency, not overall noise!) Note, too, that the noise associated with the power supply (mains hum and harmonics) has also dropped, so that the highest peaks are now at around –105dB, though this is of course to be expected, given that the top of the graph (0dBFS) is now 20 watts (12.6 volts).

You can also see on Graph 3 that the second harmonic is at –100dB (0.001%), the third and fourth harmonics are both at around –110dB (0.00031%), and the fifth and seventh harmonics are both at around –105dB (0.00056%), with the sixth at –120dB (0.0001%) and the eighth at –130dB (0.00003%). An excellent performance.

When the load impedance was halved to four omhmetas,-cuhsainrtg.coam1kHz test signal at 20 watts, the result was as shown in Graph 4. The noise floor across most of the audio band has remained much the same as with an eight-ohm load (a very slight increase) and the power supply noise has also increased very, very slightly. The primary di¡erence is that most of the harmonic distortion components have increased in level with all of the odd-order components out to 9kHz sitting at around –100dB (0.001%).

Intermodulation distortion (CCIF twin-tone, using 19kHz and 20kHz test signals of equal level) into eight ohms is shown for a one-watt output in Graph 5, and for a 20-watt output in Graph 6. At one watt, you can see an excellent result, with only two significant signals either side of the test tones, with the 18kHz signal at –92dB (0.00251%)

Overall, a superb performance... a quiet, high-powered, ultra-linear and lowdistortion amplifier

and the 20kHz signal at –88dB (0.00398%). The other high-frequency sidebands are more than 110dB down (0.00031%). There is a difference signal at 1kHz, but it’s sitting way down in level, at –98dB (0.00125%).

At the higher power level of 20 watts, the 18kHz and 20kHz sidebands have increased in level to –80dB (0.01%), which is still commendably low. Although there are more high-frequency sidebands, their level is still very low, with all except one being more than 110dB down (0.00031%) and the exception being at –105dB (0.00056%). The regenerated 1kHz signal is slightly higher in level, but still more than adequately low at –92dB (0.00251%).

Newport Test Labs measured the signalto-noise ratios of the OAD Vajra referenced to a one-watt output as being 85dB unweighted and 92dB IHF A-weighted. Referenced to rated output (the more usual measurement), the lab reported signal-to-noise ratios of 97dB unweighted and 105dB A-weighted.

Given the flat and extended frequency response of the OAD Vajra, we were expecting to

see very well-formed square waves resulting from the square wave testing, and our expectations were met in spades.

As you can see for yourself, the 100Hz square wave is textbook — the perfect square wave, looking as though it came straight from Newport Test Labs’ square wave generator, rather than via the Vajra.

The 1kHz square wave is also a perfect representation, as you can see. Both the verticals and the horizontals track the relevant graticules of the oscilloscope. An outstanding performance yet again.

The OAD’s performance with a 10kHz square wave was exceptionally good, to the point of very likely being the best we’ve ever seen from any amplifier. The rise time is exceptionally fast, with only the very slightest curve at the top of the leading edge. A superb performance.

The Vajra also excelled itself when driving the very difficult and complex load presented by an eight-ohm resistor in parallel with a 2µF capacitor, which is the fourth oscillogram shown in the series. The amplifier is not only completely stable into this demanding load, the overshoot is less than one-quarter amplitude and the output settles completely in just two cycles. Yet again, superb performance!

Input sensitivities were very close to those specified by OAD, with Newport Test Labs reporting that using the unbalanced inputs, you’d need 152mV to result in an output of one watt into eight ohms, and 1.93Volts to deliver rated output. Via the balanced inputs, you’d need 304mV to deliver a one-watt output, and 3.86 volts to deliver rated output. This puts overall gain at 25.6dB, a fraction less than the specification of 27dB.

Output impedance was measured as 0.3 ohms at 1kHz, which gives a damping factor of 266 — somewhat less than the “greater than 500” claimed by OAD, but at the same time far, far more than the DF of 50 that Floyd E. Toole says “is more than sufficient to properly control the back-EMF from any loudspeaker”.

The OAD Vajra will draw just under 80 watts from your mains power supply whenever it’s switched on, even if you’re not using it to listen to music (and a bit over when you are). If you’re playing at its maximum output, it will be pulling a shade under 500 watts from your mains.

Overall, a superb performance from the OAD Ultrafidelity Vajra. It is a quiet, high-powered, ultra-linear and low-distortion amplifier.

SPECIFICATIONS

Frequency Response:

0Hz–100kHz (–3dB)

Power Output (8Ω):

180 watts per channel

Power Output (4Ω):

360 watts per channel

Gain: 27dB

Input Sensitivity:

4V/2V (Balanced/Unbalanced)

Input Impedance: