In my linux pc-based dsp system for crossover and EQ of active loudspeakers, I use a pc to send 6 channels of processed audio over hdmi, with an Atlona AT-HD570 to convert all channels to analog before amplification. Recently I discovered that this setup can distort badly at moderately high signal levels, in a way that produces large transient subsonic signals which, when amplified, could permanently damage loudspeaker drivers. My measurements point to a flaw in the HD570.
Sine Wave Response
I connected my pc’s hdmi output to the Atlona HD570, and output a full-scale 400Hz sine wave on one channel while capturing the HD570’s analog output. The first 250ms of output looks like this:
Strangely, the output has a 15ms rise time, and the sine wave rides on top of a large subsonic transient that takes more than 250ms to die out (the black curve is the same analog output, but low-passed at 20Hz). It seems the HD570 has an under-damped low-frequency resonance in its transient response, at least when it’s driven to high levels.
Attenuating the input by 6dB excites this transient somewhat less but it’s still really bad:
The steady-state level is only 3dB (not 6) down from the the full-range case so there is some compression/limiting going on, though it doesn’t look like clipping.
At -12dBFS the transient is much reduced in amplitude, and the steady-state level drops 6dB as it should:
But a 20ms subsonic transient response is still evident; it doesn’t really disappear until I drop the input level below -15dBFS.
Response to a Pulsed Sine
To test whether the distortion originates in the pc or in the HD570 itself, I made a test signal consisting of a full-scale 400Hz sine wave pulsed 4x per second with 10% duty cycle; it looks like this:
I captured the analog output of the HD570 while feeding it this test signal over hdmi from the pc. I then did the same, but feeding the signal from a Sony BDP-BX120 media player instead of the pc.
With the pc as source the HD570’s steady-state response to the pulsed tone looks like this:
Every pulse has a slow rise time, and initiates a subsonic transient that doesn’t damp out before the next pulse. Exactly the same thing happens with the Sony player as the source:
(the vertical scale is different because I changed the pre-amp level on my recording device).
With identical results from independent sources, I’m led to believe the distortion is caused by a serious flaw (feature?) in the HD570.
Update (18/8/2016): The engineers at Atlona have confirmed this issue on other units. No word on a possible fix. I recommend strongly against using the HD570 for any input signals above -15dBFS.