Here are some numbers for you that will show why driving balanced headphones from the DAC1's line outputs will result in distorted frequency response. This says nothing about how a balanced amp -- with very low output impedance designed for driving phones -- will sound driving your Senn 650s.
To keep the numbers concrete, let the output voltage from the amp be a constant 10 volts at various frequencies. I've picked points at the inflection points of the Senn 650's impedance curves: 340 ohms @ 20Hz (340R@20Hz) (at the low end), 490R@80Hz (at the resonant peak), 340R@600Hz (near the end of the peak), 300R@1kHz (where it's "flat"), 300R@6kHz (where it stops being flat), and 360R@20kHz (where it's been rising).
Now, an ideal amp has 0 ohms output impedance. Here are the voltages, currents, and powers that would be delivered to the phones at each of the above frequencies.
Code:
Code:
[left]Ideal output impedance: 0 frequency impedance voltage current power variation ref. 1kHz (db) 20 340 10.000 29mA 0.29 -0.5 80 490 10.000 20mA 0.20 -2.1 600 340 10.000 29mA 0.29 -0.5 1000 300 10.000 33mA 0.33 0.0 6000 300 10.000 33mA 0.33 0.0 20000 360 10.000 28mA 0.28 -0.8[/left]
The variation with reference to the power at 1kHz looks pretty bad. That is, these phones consume 2.1db less power at 80Hz than 1kHz, but give relatively flat response (ie, constant SPL). How come? These power differences are combined with the mechanical behavior of the phones and give the 650s their more-or-less flat response. Thank the designers for this.
Now let's look at these values if the amp's output impedance is 0.11 ohms, the upper limit for the output impedance shown in the DAC1PRE manual.
Code:
Code:
[left]HPA2 output impedance: 0.11 ohms (less than 0.11 per DAC1PRE manual) frequency impedance voltage current power diff from ideal 20 340 9.997 29mA 0.29 0.0 80 490 9.998 20mA 0.20 0.0 600 340 9.997 29mA 0.29 0.0 1000 300 9.996 33mA 0.33 0.0 6000 300 9.996 33mA 0.33 0.0 20000 360 9.997 28mA 0.28 0.0[/left]
The power delivered to the phones is identical to the ideal case because there are negligible losses in the 0.11 ohm output impedance.
Now for the fun part. Here are the 60 ohm XLR outputs driving Senn 650s. The power delivered to the phones is very different from the ideal case, being roughly 2db down over the entire range. This power is lost in the 60 ohm resistor. But there is much more of importance in the last column!
Code:
Code:
[left]XLR output impedance: 60 frequency impedance voltage current power diff from variation ideal ref. ideal 1kHz 20 340 8.500 21mA 0.18 -2.1 -1.6 80 490 8.909 16mA 0.14 -1.5 0.6 600 340 8.500 21mA 0.18 -2.1 -2.3 1000 300 8.333 23mA 0.19 -2.4 -2.4 6000 300 8.333 23mA 0.19 -2.4 -2.4 20000 360 8.571 20mA 0.17 -2.0 -1.6[/left]
The last column compares the variations in power (2nd last column) to the last column for the ideal case. Look at the variation now! Whereas the power at 80Hz should be 2.1db lower than the power at 1kHz (ideal case) for flat response, the power at 80Hz is 3.0 db higher (0.6 - (-2.4)). That's a huge bass boost of 3db SPL. Similarly, the power at 20kHz should be 0.8db lower than at 1kHz, but is now 0.8dB (-2.4 - (-1.6)) higher, nearly 1 db of boost at 20kHz. You might not be able to hear 1db boost at 20kHz, but you will certainly hear 0.5db at 10kHz.
In general, the frequency response will become shaped like the impedance curve as the output impedance of the amp increases. It's just a matter of scale as to how big the deviation from flat becomes.
To complete the picture, look at the XLR line outputs driving a preamp with a fixed 20,000 ohm input impedance:
Code:
Code:
[left]XLR output impedance: 60 frequency impedance voltage current power 20 20000 9.97 0.49mA 0.005 80 20000 9.97 0.49mA 0.005 600 20000 9.97 0.49mA 0.005 1000 20000 9.97 0.49mA 0.005 6000 20000 9.97 0.49mA 0.005 20000 20000 9.97 0.49mA 0.005[/left]
The last two columns don't really matter, since the DAC1 isn't being called upon to deliver any current or power. All that matters is the voltage, and it is unchanged from frequency to frequency because of the constant load impedance.
Hope this helps!
- Eric