Link Sotlights
The Pinnacle of The Electrostatic Sound

Monday, October 26, 2015

EAN SpecCheck!
Benchtest Measurement:
Mytek Manhattan D/A Converter

by Bascom King
BHK Laboratories

  The Mytek Manhattan is a DAC/line preamp/headphone amp with an array of digital and analog I/O. Measurements were made at 120V AC line voltage with both channels being driven except when otherwise noted. Measurements made on left channel through the AES/EBU inputs and balanced analog outputs unless otherwise noted.
  I measured frequency response with a new technique using noise stimulus as it makes it easier to see the cutoff characteristics of digital filters is shown in Fig. 1 for the PCM sharp filters at sample rates of 44.1, 96.0, & 192.0 kHz. In Fig. 1A, the responses are shown for the slow filter choice. One consequence of using the slow filter at a sample rate of 44.1K is the aliasing products that result for a 19.1 kHz test tone at 0 dBFS. This is shown in Fig. 2, first with the sharp filter, and then in Fig. 2A with the slow filter. When the sample rate is changed up to 96.0 kHz with the slow filter, things behave more like the Fig. 2 regarding the lack of aliasing products. These tests were done with in the analog bypass mode.
  Regarding the analog bypass mode vs. using the analog volume control attenuator, I measured THD+N of a 1 kHz signal as a function of decreasing digital input level in a 22 kHz measurement bandwidth. In Fig. 3, this shows much lower distortion and ultimate noise level is for the bypass mode vs. the analog volume control mode which was set for 0 dB for the test. When measuring how THD+N varies with frequency, it was found that it was quite constant over the audio band which is a good measurement attribute. This is shown in Fig. 4 for both and bypass and analog modes. When tested in the digital volume control mode, the performance was essentially the same as the analog bypass mode.

  In the case of the bypass mode, the difference is about 20 dB for a 3-plus bit increase for 19-plus bit resolution for 24-bit data and 1-plus bit increase for the analog volume control mode — yielding 17-bit resolution.

  An interesting test for determining the effective bits of a DAC's resolution is to plot a spectrum of DAC output noise in the audio band with a 16 bit “digital off” signal and then compare the same spectrum with a 24 bit signal. One assumes that the noise level of the 16 bit signal is say 96 dB and then we see how many “6 dB” increments are contained in the difference. One bit for every 6 dB of noise reduction in the two curves. This is shown in Fig. 5 for the bypass mode and Fig. 5A for the analog volume control mode. In the case of the bypass mode, the difference is about 20 dB for a 3 plus bit increase for 19-plus bit resolution for 24-bit data and 1-plus bit increase for the analog volume control mode — yielding 17 bit resolution.
  Channel separation vs. frequency (not shown) for both directions and for bypass and analog attenuator modes was 95 dB or better from 20 Hz to 10 kHz rising to about 90 dB at 20 kHz. When checking the analog in – analog out characteristics, I found the general input/output linearity and difference between analog volume control and bypass mode distortion to be similar to the D/A behavior. Frequency response was very wide and flat from 10 Hz to 200 kHz.
  Looking at the headphone amplifier briefly, I found that I could only get output from the headphone jacks with the digital attenuator mode selected. I presume this is normal behavior for this unit. With a 32 or 16 Ohm load, the power output at the onset of clipping was about 2W.

  Finally, dynamic range and signal to noise ratios with A weighting at sample rates of 44.1, 95.0, and 192.0 kHz were all about the same and for both channels at 103.5 dB for the analog volume control mode and 113.3 for the analog bypass mode. Signal to noise ratios in the bypass mode in a 22-22 kHz bandwidth were about 110.5 dB and wideband, about 99 dB. In the analog volume control mode, the corresponding numbers were 100.4 dB and 86.6 dB. Note these signal to noise numbers in the 22 – 22 kHz bandwidth relate to and are similar to the lower input levels in Fig. 3.
      Fig. 1  Sharp Filter Noise: 44.1k, 96k, 192k

Fig. 1A  Slow Filter Noise: 44.1k, 96k, 192k

          Fig. 2 Sharp Filter Alias Effect Noise

 Fig. 2A Slow Filter Alias Effect Noise

Fig. 3 Bypass vs. Analog Volume THD+N/Input

     Fig. 4 Bypass/Volume Control: THD+N/Freq.

  Fig. 5 Analog Bypass Noise Spectrum: 16 vs. 24 bit

        Fig 5A Analog Volume Noise Spectrum 16 vs. 24 bit

No comments: