[stv014]

Let me know if any other test configurations would be useful. Should I do one with the line input connected by a regular cable?

 
I think these configurations might be worth testing:
- headphone load with no resistors at all, only regular cables
- resistor load with regular cables, maybe also with input resistors if those make any notable difference in RMAA

 

[stv014]

Here are a few graphs created from the loopback recordings posted so far:


The THD vs. frequency graph looks interesting in the 1.1 Vrms/0.2 ohms case, as the distortion varies greatly with the frequency, perhaps depending on the impedance/phase of the load.
 
THD vs. level at 60, 1000, and 15000 Hz:



These assume that 0 dBFS output is 2.2 Vrms on the load. Without the resistors, the distortion always hits a wall at about 1-1.2 Vrms, but the exact threshold varies with the frequency. Interestingly, the distortion also appears with the resistors at 1 kHz at 2 Vrms (double voltage). The high noise level measured at 15 kHz is because of noise shaping in the DAC and ADC.
 
Edit: fixed THD vs. level graphs

 

[stv014]

More graphs:
 

This is the waveform of the 1 kHz tone at slightly less than the maximum level of 2.2 Vrms without output resistors (top), and the difference from the undistorted original sample (bottom).
 

Frequency response from MLS, with channel balance adjustment. It is less affected by distortion than the one generated from sine sweeps, and has a wider frequency range.

 

[stv014]

  Frankly if there is oscillation it's probably just defective.

 
I am not saying it is necessarily oscillation, but it does look like there is a problem with the Magni, or at least with this particular unit that may indeed be defective in some way. It does not seem to be simple voltage or current limiting, as the exact level where the distortion begins depends on the load, and a higher impedance load allows for both higher voltage and current, at least if the measurement setup worked correctly. 1 Vrms is well below the specifications of the Magni, and it is only about 45 mA peak current into a ~32 ohms load. From the above waveform displays, it appears the distortion occurs only below a certain negative level, which depends on the load.
 
It is still not clear why adding serial resistors to the line input makes any significant difference.

 

[MindsMirror]

Thank you for that analysis. I contacted Schiit support and linked them here. They said it looks like there is a problem with my amp, so I've sent it to them for repair. I'll post an update once I get it back.
 
It looks like on the frequency sweep and the 60Hz test you can see (where there is not the problematic distortion) that the measured distortion is higher on the tests with the resistor on the amp's output, because of the distortion coming from the headphone, and that distortion is not present at the output of the amp when it has a high damping factor. Would the high damping factor also result in the driver producing less physical/acoustic distortion, assuming you have an amp that is working correctly? In the measurements you linked to earlier of the DT880, the THD measured by the microphone with 10 and 110 Ohms look almost the same.

 

[stv014]

It looks like on the frequency sweep and the 60Hz test you can see (where there is not the problematic distortion) that the measured distortion is higher on the tests with the resistor on the amp's output, because of the distortion coming from the headphone, and that distortion is not present at the output of the amp when it has a high damping factor. Would the high damping factor also result in the driver producing less physical/acoustic distortion, assuming you have an amp that is working correctly? In the measurements you linked to earlier of the DT880, the THD measured by the microphone with 10 and 110 Ohms look almost the same.

 
It did make a difference of about 2 dB to bass distortion, which is not a major increase, but the damping with the extra serial resistance was also only 1.4 times (= 3 dB) worse:

 
By the way, it seems the THD vs. level tests also have some distortion from the sound card (Sound Blaster Z ?), as when they reach 0 dBFS, there is always some clipping on the negative peaks, even with the output resistors, regardless of whether the voltage on the load is 1.5 or 2.2 Vrms. But this would not explain the distortion at well below 0 dBFS.

 

[Greenears]

  Thank you for that analysis. I contacted Schiit support and linked them here. They said it looks like there is a problem with my amp, so I've sent it to them for repair. I'll post an update once I get it back.
 
It looks like on the frequency sweep and the 60Hz test you can see (where there is not the problematic distortion) that the measured distortion is higher on the tests with the resistor on the amp's output, because of the distortion coming from the headphone, and that distortion is not present at the output of the amp when it has a high damping factor. Would the high damping factor also result in the driver producing less physical/acoustic distortion, assuming you have an amp that is working correctly? In the measurements you linked to earlier of the DT880, the THD measured by the microphone with 10 and 110 Ohms look almost the same.

 
I'm glad you solved your defective amp problem.
 
The answer to your question about damping factor is "no".  Damping is just a fancy term for resistor in this context.  If series resistance was advantageous they could just put it in the amplifier box.  The ideal voltage mode amplifier has zero output impedance or negligibly small amount (0.12 ohn of the Magna is just perfect).  Within its operating range of Vmax and Imax, it gives the load whatever current is asked for and holds the voltage steady with negligible background distortion through 20 kHz.  That's it.  The magic happens in the transducer, and it is much easier to design a transducer to a known source than to start to introduce additional stuff which makes it all go to mush.

 

[Sonic Defender]

Thanks to all in this thread. While my literacy in these matters is un-measurably small, I find such threads very interesting to read. I will be interested to see if Schiit confirms the problem and if so, do they offer an explanation in keeping with the technical nature of this thread. 

 

[davidsh]

Thanks to all in this thread. While my literacy in these matters is un-measurably small, I find such threads very interesting to read. I will be interested to see if Schiit confirms the problem and if so, do they offer an explanation in keeping with the technical nature of this thread. 

Agreed

 

[MindsMirror]

The amp is on its way back, they didn't find any problem with it.
 

 
By the way, it seems the THD vs. level tests also have some distortion from the sound card (Sound Blaster Z ?), as when they reach 0 dBFS, there is always some clipping on the negative peaks, even with the output resistors, regardless of whether the voltage on the load is 1.5 or 2.2 Vrms. But this would not explain the distortion at well below 0 dBFS.

That is correct. That clipping at 0dBFS is there if I measure straight from the line out to the line in, and it goes away when I decrease the volume slightly.