|TEST REPORT 1965 Akai M-8|
From The Tape Recorder, March 1966
LOOKING back, I find that I reviewed the M-6 in November 1962, and the M-7 in June 1964. They have all used valve amplifiers and the same basic layout and styling, but the M-7 and the present M-8 use the exclusive 'cross-field' biasing system. All these recorders have a continuously variable playback equalisation, which can be used as a form of tone control.
Only the record/play heads are switched, the bias and erase heads are moved up and down to provide the various recording facilities. In the stereo model they are aligned with the record heads, and on mono the heads are moved so that the top head or the bottom head are off the tape. This novel system avoids any problems of oscillator load variations, which would need to be compensated if the heads were switched.
The twin monitor speakers are mounted on the top of the cabinet and stereo monitoring is only available to the operator near the recorder. At greater listening distances the stereo effect from these speakers is negligible and widely spaced wide-range external speakers must be used for normal stereo listening.
A 'sound-on-sound' button has been added so that one amplifier is set to play and the other to record with appropriate cross connections for track-to-track transfer with added recording. The long-term speed stability was checked by strobe and standard frequency test tapes and found to be within limits of ± 2% at all speeds at the beginning and end of 7 in. reels.
The short term speed stability was tested in the usual way by recording a 3 Kc/s test tone at each speed and then playing it back via a sensitive frequency discriminator circuit to a wide range pen recorder. The fluttergrams of fig. 1 show that wow and flutter are satisfactorily low at tape speeds of 3¾ i/s and 1 i/s with no sleeve fitted to the capstan (traces C and D), but that capstan wow at 7 - 8 c/s and 3 - 3½ c/s were unpleasantly obvious when the capstan sleeve was fitted and the recorded and play speed deviations happened to be in step (lower traces of A and B). For short periods the capstan wows cancelled to give the smoother traces (top A and B). It should be mentioned that the capstan sleeve on this particular machine was a very loose drop-on fit and that the effect would not be so bad on a correctly fitting sleeve.
Next, test tapes recorded to known levels, corresponding to surface induction characteristics of 70, 140 and 280µS time-constant, were played and the tone controls set for the best line output response. The arrows on fig. 2 show the position of the tone control knobs at each speed. Noise and hum were 33dB below test tape level with no tape running through the machine at 7½ i/s, and 28dB below test tape level at 3¾ and 7½ i/s.
The tape supplied was then loaded on the machine and test recordings were made to find which VU reading corresponded to test tape level. This was found to be -6dB. Thus true peak recording level (12dB above test tape level) would be above the full-scale reading of the VU-meters, and 0dB corresponds to a level 6dB below true peak recording level. This is standard practice on professional VU-meters to allow for the inertia of the meter movements on normal programme material.
Test tones were now recorded at -6dB level and the playback responses measured at line output to give the curves of fig. 3. It will be seen that the tone control positions are similar to those shown in fig. 2 indicating that the recorded time-constants are close to those shown against these responses.
Overload tests at 500 c/s showed that waveform distortion was low up to recorded levels 4 to 5dB above full-scale deflection of the VU-meters. This corresponds to levels 13 to 14dB above test tape level and proves that bias is optimum for minimum distortion. Further CRO tests showed that amplifier overload was some 6 to 8dB above that of the tape. This ensures that the tape is the first part of the system to overload, even with other lower sensitivity tapes.
Peak recording level was erased on the machine and the wide band signal-to-noise ratios were 45dB at 7½ i/s and 41dB at 3¾ i/s.
Finally, 25 one-third-octave bands of white noise were recorded at 7½ i/s and the sound output of the loudspeakers measured with a calibrated microphone on playback. The acoustic responses of fig. 4 show the expected fall in low note response below 200 c/s due to the small cabinet volume and indicate that the response is reasonably level in front of the machine (lower dotted curve).
The free air response of one of the microphones supplied with the recorder was measured in a white noise sound field to give the response of fig. 5. The impedance is high at 50K ohms to suit the valve input stage. Polar response is omni-directional so the spaced microphone technique must be used for stereo recording. Listening tests confirm that this response is indeed smooth and wide range and that the microphone is a worthy partner to the M-8 recorder.
The promised very wide frequency response at low tape speeds with the cross-field method of recording has still not materialised, but there is some evidence that higher recording levels are possible with low distortion and with adequate frequency responses for the three speeds. It was thought that the slight roll off at very short wavelengths might be due to high frequency overload, but tests at very low recording levels did not show any startling extension of frequency response.
Personally I would choose low distortion and a good signal-to- noise ratio any day, and this is not because my own hearing may be deteriorating above 10 Kc/s, but because I am somewhat cynical about unlimited extension of frequency response until all other distortions and inter-modulation products have been reduced to a very low level indeed.
Cross-field bias is only one step in this direction.