Columns Retired Columns & Blogs |
Melos SHA-1 headphone amplifier Measurements
Sidebar 2: Measurements
Footnote 1: Excellent cans, these; I intend to review them in the near future.—John Atkinson
I bought myself a Melos headphone amplifier after hearing it drive Sennheiser HD560 Ovations (footnote 1) to perfection at the 1991 Summer CES. I was in need of something of quality to do monitoring with while making live recordings, and it was only when I found myself temporarily without a preamplifier capable of driving long cables that I tried it as a line stage. To say that I was astonished at how sweet yet neutral this inexpensive unit sounded in this application would be an understatement! Perhaps instruments and voices are a trace more forward in the soundstage than they are with my McCormack Line Drive Deluxe, but the music came through without a trace of hardness, grain, or all those other audiophile afflictions.
Oh, well. Corey gets paid to wax poetical. My job is more mundane: maybe some pointers to its sonic excellence will emerge in the measured fine print.
The SHA-1's gain with the volume control wide open was almost to specification at 18.8dB, 100mV in at 1kHz giving 870mV out. When the SHA-1 is used as a preamp, this will be a little on the high side for CD sources. With Wilson WATTs/Puppies driven by Audio Research Classic 120s, I ended up using the volume control between 8 and 11 o'clock, for example. Surprisingly for a design with, I believe, a single voltage gain stage, the SHA-1 doesn't invert polarity from either output. The measured input impedance at 1kHz was high, at 91k ohms, while the output impedance was 2.7 ohms from the RCA jacks or 0.7 ohms from the headphone socket, these figures again taken at 1kHz. Unweighted audio-band noise was also to spec, at better than -84dB referenced to a 1V output level (footnote 2). The volume-control tracking was excellent, within 0.1dB from full up to 9 o'clock, though the right-hand channel had a constant +0.3dB gain error compared with the left.
The frequency response for 50mV in and the volume control set to its maximum was flat from 10Hz to 70kHz, with then a rolloff to -3dB at 200kHz, as can be seen from fig.1. What also can be seen from this graph, however, is that the amp's ultrasonic behavior depends on the volume-control position. As the control is reduced below the 1 o'clock position, which is where the response is maximally flat, an increasingly large peak between 100 and 200kHz appears, reaching +5dB (left) and +2dB (right) at 200kHz with the volume control set to 9 o'clock. While the audio-band changes resulting from this behavior are less than +0.25dB at 20kHz, it is still inelegant engineering.
Fig.1 Melos SHA-1, frequency response from headphone output with volume control set to (from bottom to top at 50kHz): maximum, 1:00, 12:00, and 9:00 (0.5dB/vertical div., right channel dashed, load impedance 100k ohms in parallel with Sennheiser HD 560 headphones).
Channel separation was adequate (fig.2), though much better in one direction than the other. More important was the limited separation between the two pairs of inputs. The top trace in fig.3 shows the capacitive leakage from the driven input into the other when the latter was unterminated (very much the worst case), while the bottom is the leakage with the undriven input shorted to ground (the best case). In real life, the leakage from, say, an FM tuner hooked up to one input while a CD is playing through the other will depend on the relative levels and output impedances of the source components. Such leakage can be eliminated, of course, by reducing the unwanted source's output level to zero. But if you turn off the other source, you increase the possibility of RF pickup, which will then leak into the Melos's other input, with the possibility of sonic degradation. To get the best from the SHA-1, leave all sources turned on all the time or fit shorting RCA plugs into the unused pair of inputs.
Fig.2 Melos SHA-1, Inter-channel crosstalk at 100mV input, volume control at maximum (10dB/vertical div.).
Fig.3 Melos SHA-1, Inter-input crosstalk at 100mV input, volume control at maximum, with undriven input open (top) and shorted to ground (bottom) (10dB/vertical div.).
The SHA-1 seems to be intrinsically linear, as indicated by the plot of THD+noise vs frequency (fig.4, which was taken with the volume control wide open, a 100mV input level, and the amp loaded with a pair of Sennheiser HD560 headphones), though the rise in the measurement above the audio band presumably ties in with the circuit's propensity for peakiness a decade or so higher in frequency. And what distortion there is at audio frequencies consists of the subjectively benign low orders, primarily second and third, as can be seen in fig.5, which shows the harmonic structure up to 1kHz resulting from 100mV of 50Hz input and the amplifier driving the Sennheiser cans. Both second and third harmonics are 76dB down from the level of the fundamental. Though traces of AC mains-related components are visible at 120Hz, 180Hz, and 240Hz, all of these are 88dB down or more, and innocuous.
Fig.4 Melos SHA-1, THD+noise (%) vs frequency for 100mV input, headphone output, volume control set to maximum, Sennheiser HD 560 load (right channel dashed).
Fig.5 Melos SHA-1, spectrum of 50Hz sinewave, DC-1kHz, from headphone output at 870mV into Sennheiser HD 560 headphones (linear frequency scale).
Footnote 1: Excellent cans, these; I intend to review them in the near future.—John Atkinson
Footnote 2: My SHA-1 developed a weird problem soon after I bought it, with very low-frequency oscillations appearing on the output—the sight of the Puppy woofers very slowly pumping in and out was quite alarming, even though no damage occurred. I sent it back to Melos to be fixed; in the nine months since then, there has been nary a hint of a problem. This was despite the unit being left on all the time, apart from a two-week vacation when Laura and I went to the UK to see Nigel Mansell demolish the Formula One competition in front of his home crowd at Silverstone. Motor-racing hacks may refer to "Red Five" as "The Whiner," but he is a perfectionist and 1992 is his Championship year; he's entitled to bitch and moan if he wishes.—John Atkinson
- Log in or register to post comments