Graaf GM 200 OTL power amplifier Measurements

Sidebar 3: Measurements

Following its 1/3-power, one-hour preconditioning test, the Graaf GM 200's tube cages and power transformer were too hot to touch. This is consistent with JS's findings. Since heat is the enemy of electronics, one wonders how this operating environment might affect the longevity of the tubes and the rest of the circuitry.

OTL amplifiers do not like low impedances, and although I got respectable power out of the GM 200 into 2 ohms (though at moderately high distortion), the amplifier shut down a number of times during testing, refusing to come back on until it was allowed to cool down for a few minutes. I would not expect this to be a problem in normal use.

Unless stated otherwise, the results shown are for the unbalanced configuration. The GM 200's input impedance measured 88.7k ohms (202k ohms balanced). The output impedance was 0.79 ohms at 1kHz, 0.7 ohms at 20Hz, and 0.83 ohms at 20kHz—high, but not unusually so for a tube amplifier. (It helps that there are no output transformers.) Voltage gain into 8 ohms measured 33.4dB unbalanced and an almost identical 33.2dB balanced. DC offset was 13mV in the left channel, 16mV in the right.

Signal/Noise Ratio (ref. 1W into 8 ohms) measured a fairly low 57.8dB over a 22Hz–22kHz bandwidth, unweighted, and 57.4dB over a 10Hz–500kHz bandwidth, unweighted (70.6dB, A-weighted). The Graaf is noninverting from its unbalanced inputs, and pin 3 is positive in the balanced configuration.

Fig.1 shows the GM 200's frequency response. The high- and low-frequency rolloffs are not unusual for a tube amplifier, and are no more than about –1dB up to 20kHz. The effect of our simulated load is about as expected: this amplifier should give a slightly softened sound in the high frequencies with most real loudspeaker loads. The balanced response holds up a little better at high frequencies (see fig.2). The 10kHz squarewave response in fig.3 is good for a tube amplifier. The leading edge is noticeably rounded, but the initial risetime is decent and there is no overshoot or ringing. The 1kHz response (not shown) is very good, with a fast risetime and only a slight rounding at the leading edge.

Fig.1 Graaf GM 200, frequency response in unbalanced mode at (from top to bottom at 20kHz): 1W into 8 ohms, 2W into 4 ohms, and 2.83V into simulated speaker load (0.5dB/vertical div.).

Fig.2 Graaf GM 200, frequency response (from top to bottom at 20kHz): at 2W into 8 ohms, balanced mode; 2W into 8 ohms, unbalanced mode (0.5dB/vertical div.).

Fig.3 Graaf GM 200, small-signal 10kHz squarewave into 8 ohms.

The channel separation was also respectable, at better than 80db across most of the band, with only the normal increase in crosstalk at high frequencies apparent, due to capacitive coupling between channels. The two channels are not identical, but even at its worst, the crosstalk should not be audibly significant. The THD+noise vs frequency result (fig.4) is good with 8 and 4 ohm loads. (The balanced performance, not shown, matches this within 0.1%.) It's obvious that the amplifier is beginning to complain with a 2 ohm load, however. Still, its 1kHz distortion waveform is approximately the same at all three load impedances—primarily second harmonic, as shown in fig.5.

Fig.4 Graaf GM 200, THD+noise vs frequency at (from top to bottom at 4kHz): 4W into 2 ohms; 2.83V into simulated speaker load; 2W into 4 ohms; and 1W into 8 ohms.

Fig.5 Graaf GM 200, 1kHz waveform at 5W into 4 ohms (top); distortion and noise waveform with fundamental notched out (bottom, not to scale).

The response of the GM 200 to a 50Hz input at 90W output into 4 ohms is shown in fig.6. This is not an atypical result for a tube amplifier: –39dB or about 1.1% at 100Hz, –54.8% or about 0.18% at 200Hz. At 2/3 clipping voltage into our simulated real load (not shown), the odd-harmonic THD roughly doubles—from about 0.07% to 0.18% at 250Hz—but the increase in even-order distortion is small.

Fig.6 Graaf GM 200, spectrum of 50Hz sinewave, DC–1kHz, at 90W into 4 ohms (linear frequency scale).

Fig.7 shows the Graaf's output spectrum driving a combined 19+20kHz signal—the intermodulation products resulting from an input signal consisting of an equal combination of these two frequencies—at 93W into 8 ohms. This was the closest I could get to 2/3 clipping without visible indications of clipping. The largest artifacts here are at 1kHz (–43dB or about 0.7%) and 21kHz (–47.1dB or about 0.45%), and at 18kHz and 22kHz.

Fig.7 Graaf GM 200, HF intermodulation spectrum, DC–22kHz, 19+20kHz at 93W into 4 ohms (linear frequency scale).

The 1kHz, THD+noise vs output power curves for the GM 200 are shown in fig.8. The discrete clipping levels (at 3% THD+noise here instead of the usual 1%) are shown in Table 1. While the amplifier still puts out significant power into 2 ohms, at least at 1kHz, note that the result here is the reverse of that from a typical solid-state amplifier, with steadily reduced power into lower impedances. Indeed, for the performance of an OTL amplifier at 2 ohms to match its performance at 8 ohms, it would need four times as many output tubes!

Fig.8 Graaf GM 200, distortion (%) vs output power into (from bottom to top): 8 ohms, 4 ohms, and 2 ohms.

Table 1 Graaf GM 200 Clipping (3% THD+noise at 1kHz)

Both Channels DrivenOne Channel Driven
LoadW (dBW)W (dBW)
8179.8 (22.5)182.1 (22.6)203 (23.6)
4122 (17.9)120.2 (17.8)138 (18.4)
258.3 (11.7)

The Graaf's test-bench measurements were respectable, except for the rather low S/N ratio. As with all OTL amplifiers, you should avoid loudspeakers that have unusually low impedances. And provide plenty of ventilation!—Thomas J. Norton

PO Box 570
Chazy, NY 12921
(800) 771-8279