Conrad-Johnson MF2500 power amplifier Measurements

Sidebar 3: Measurements

While the Conrad-Johnson MF2500 will drive 4 ohm loads with music, the output specification is given as 240Wpc (23.8dB/W) for 20Hz–20kHz, at less than 1% of IM or harmonic distortion, both channels driven into 8 ohms. Don't be concerned at that high-looking figure of 1% distortion—in practice, the figures are much less than this. What this figure reflects is C-J's disdain of slavish conformance to the artificially low distortion figures adopted by many companies without regard to sound quality.

Tested with the UK's 242V, 50Hz power supply, the MF2500 reached 309W into 8 ohms, one channel driven, and 295Wpc, both channels driven. It wasn't shy at 4 ohms, and the shared power supply was man enough for 451Wpc into 4 ohms. This, only 1.3dB below the 8 ohm single-channel test result, indicates good regulation and load-driving ability.

Playing music-related (IHF toneburst) signals, the MF2500 pushed 24.9dBW into 8 ohms (315W) and 24.77dBW into 4 ohms (600W), but began to current-limit a little into 2 ohm loads with 23.9dBW (976W) available. This relative indifference to loading provides good verification of the easygoing muscularity identified in the listening trials. The maximum peak current capability was approximately ±35A—good enough for all but the most severe speaker loads, and certainly those nominally rated at 4 or 8 ohms.

Power bandwidth was fine, at full power rated –0.45dB at 20Hz and –0.3dB at 20kHz—very solid results. Distortion at rated power was more than satisfactory: 0.15% at 20Hz, 0.14% at 1kHz (–57.1dB), and 0.4% at 20kHz. While looking at the bandwidth, I happened to check the distortion at 1W, 150kHz, where readings of several percent are common, and where the switching rate of the output devices dominates the issue. The test result showed the MF2500's fine internal behavior: just 0.5% THD, which is impressive.

In case you're worried that the modest result for 20kHz might have consequences for in-band high-frequency intermodulation products, don't. Tested with 19+20kHz IM signals at full power and 8 ohm loading, the MF2500 gave a difference tone well down at 0.0375%, or –68.5dB (see fig.1). Note that the second-order difference product was a little higher than the 1kHz difference tone, so the overall IM result was not quite as good as it might at first appear. At significantly lower powers the linearity cleared up quickly, and the excellent intermodulation result at 1W output (fig.2) was typical.

Fig.1 Conrad-Johnson MF2500, HF intermodulation spectrum, DC–22kHz, 19+20kHz at 250W into 8 ohms (linear frequency scale).

Fig.2 Conrad-Johnson MF2500, HF intermodulation spectrum, DC–22kHz, 19+20kHz at 1W into 8 ohms (linear frequency scale).

While this is a bipolar amplifier run at modest class-A/B bias levels, its harmonic distortion signature remained of low order over a wide range of levels and frequencies. Thus the 1W harmonic spectrum for a non–supply-related frequency of 177Hz showed an almost tubelike signature (fig.3): second harmonic was predominant at a moderate 0.1%, third harmonic at 0.015%, with an almost monotonic harmonic decrement thereafter. Hum components were quite low, at –80dB or lower relative to the 1W output level.

Fig.3 Conrad-Johnson MF2500, spectrum of 177Hz sinewave, DC–2kHz, at 1W into 8 ohms (linear frequency scale).

With the MF2500 fired up with a stressful two-thirds of rated level into a 4 ohm load, I analyzed the spectrum for supply harmonic breakthrough (fig.4). Once again, that aurally "kind" monotonic distortion spectrum is seen, with hum components held to negligible levels. Those early stage regulators must be doing their job well. Short-circuit trials resulted in blowing the rail-protection fuses but nothing else.

Fig.4 Conrad-Johnson MF2500, spectrum of 37.5Hz sinewave, DC–2kHz, at 400W into 4 ohms (linear frequency scale).

The output impedance was satisfactorily low, typically 0.12 ohm at low and mid frequencies. This rose to 0.32 ohm by 20kHz, but remained of no great significance. The input was easy to drive, its impedance nominally 105k ohms in parallel with 120pF, which will be suitable for tube, FET, and solid-state preamplifiers. Input sensitivity was about average: 1.9V for full level and 124mV for 1W, not quite sensitive enough for the amplifier to be used with a passive controller unless the associated CD player offers more than the standard 2V maximum output.

Frequency response for the precision –0.5dB points was rather better than 2Hz–21kHz; ie, very slightly dulled at the edge of the audible range, and a bit worse than specified. The rate of rolloff above that point was quite gradual; eg, to —dB at 140kHz.

Channel balance was tolerably close at 0.120dB midband, while DC offset at the output was also quite satisfactory: 38mV left, 25mV right, and remaining quite stable. Of more interest perhaps was the rather ordinary result for channel separation, beginning at 83.6dB for the 20Hz bass range, drifting down to 49.3dB by midband 1kHz frequency, and falling further to 26.3dB by 20kHz.

These results were not special, and I was concerned that if the crosstalk product emanated via the common power supply, then it could well be dominated by harmonic distortion from the other channel. As it happened, this crosstalk was innocuous, purely signal-related and probably arising at the high-impedance input section. A direct measurement of the "silent" channel while the other was driven to 100W into 8 ohms gave a fine distortion result of better than 0.2%. No worries here.

The UK's 50Hz power supply caused some mild mechanical hum from the transformer at certain times of day—the peak demands of TV sets in the evening, for example. On electrical noise, for IHF 0dBW, unweighted I measured a fine 77.2dB rising to 89dB A-weighted. Relative to full power, the unweighted reading was 101dB (112 excluding hum) and 112.6dB A-weighted—very satisfactory results.

Finally, the MF2500 was analyzed for stability (fig.5). It proved unconditionally stable; even a 2µF load and direct, wide-bandwidth squarewave drive resulted in a clean result, with little overshoot and negligible ringing. This tidy behavior further reinforces my good opinion of its load tolerance.—Martin Colloms

Fig.5 Conrad-Johnson MF2500, small-signal 1kHz squarewave into 8 ohms in parallel with 2µF.

Conrad-Johnson Design
2733 Merrilee Drive
Fairfax, VA 22031
(703) 698-8581