[MindsMirror]

Can anyone explain the following phenomena?
 
These tests were done using my Magni amp with my SR125e headphone as a load. It was measured by my SBZ mic input using a Y splitter to measure the electrical signal directly on the headphone plug. The green "no resistor" line is with the splitter connected directly to the amp, the white "33 ohm resistor" line is with a 33Ohm resistor between the splitter and the amp, simulating a 33Ohm output impedance. The -1dB test tone as measured from the splitter is 0.5VRMS in all cases.
 
THD test

No resistor = 3.737%
33 ohm resistor = 0.018%
 
IMD test

No resistor = 0.045%
33 ohm resistor = 2.827%
 
The distortion of the 1KHz tone in clearly audible with the headphones sitting next to me in the no resistor test. In the resistor test, with the headphones at the same loudness there is no audible distortion of the 1KHz tone. I can't hear the 60Hz tone or its distortion without the headphones on my head, and it would be too loud to do that. None of the graphs are shifted up or down, it's a little odd that the noise floor is so different in the first graph.
 
Can anyone explain why the damping factor has this effect on the distortion?

 

[dprimary]

How can you be listening to the headphones, while measuring one channel with the 33 ohm resistor and one without? What is a SBZ? What it seems that you are measuring is the response of the amp into what ever the impedance of the SBZ is and then again with the SBZ plus 33 Ohms. With the headphones in series as well?

 

[MindsMirror]

  How can you be listening to the headphones, while measuring one channel with the 33 ohm resistor and one without? What is a SBZ? What it seems that you are measuring is the response of the amp into what ever the impedance of the SBZ is and then again with the SBZ plus 33 Ohms. With the headphones in series as well?

The two lines are not different channels, they are different tests plotted on the same graph. SBZ = Sound blaster Z sound card. The SBZ microphone input is high impedance and doesn't effect the response in any of these tests as far as I can tell. The headphone is always in parallel with the SBZ's mic input, so I am measuring the response at the headphone in both. In the first test that is also in parallel with the amp. In the resistor test, the resistor is in series between the amp and the headphone.

 

[dprimary]

I think this is what you are trying to do
 
https://cdn.shopify.com/s/files/1/0321/7609/files/Headphone_Amplifier_Performance_-_Part_2.pdf?1361
 
It is hard to tell what effect the input impedance of the sound card might have I don't see it listed in the specs microphone inputs tend to fairly low. Still the headphones are only 32 ohm, not that I can find an impedance curve for them.
 
Then again look at the the impedance of an 8 Ohm speaker

 
 
 
 
In your case it seems reducing the damping reduces the distortion. It should be the opposite. 

 

[stv014]

I think there is some problem either with your measurement setup or the Magni, because distortion products at only 30 dB down from the fundamental (> 3% THD) are definitely not normal when an amplifier is not clipping. Maybe the issue is related to the microphone input of your sound card (by the way, why not use a line input instead ?), which was probably designed for lower input levels. Although if the levels are matched, then that does not explain why there is so high distortion only without the resistor. Another explanation I can think of is that the stability of the Magni might be marginal, and the combined capacitance from the two cables and the microphone input cause it to oscillate, while the serial resistor isolates the capacitance and stops the oscillation. I have seen reports of people "receiving" radio stations with the Magni, and the effect depended on the headphone cable (e.g. it only happened with an extension cord).
 
For comparison, this older post shows the results of a 10 vs. 110 ohms output impedance test on a DT880-250, which make more sense. It includes both electrical and acoustic measurements.
 
By the way, are you sure that the level was really only 0.5 Vrms ? If you used a DMM to measure it, especially one of the cheaper models, then it is possible that at 1 kHz the real voltage was significantly higher than that. Some multimeters are only accurate when measuring AC voltage at 50-60 Hz, and roll off anything much higher than that. This can be tested easily by generating a tone at a constant level, but at different frequencies, and measuring the unloaded output voltage from the DAC/sound card with the DMM.

 

[MindsMirror]

I used the mic input because with the line input at 100% the measured signal was low, the mic input amplified it more. I don't think the measurement setup has anything to do with it though. The distortion on the 1KHz tone is clearly audible without any measurement, with the headphones plugged directly into the amp. I didn't think it was a problem with the amp though. With the 33 ohm resistor as the load instead of the headphone, I measure low distortion on the amp (>0.01%) in THD and IMD tests, at significantly higher output power.
 
Yes, it was only about 0.5V. The SBZ line output is 2.2V, I set the amp to unity gain and turned the windows volume to -12dB. It measured 0.4V on my DMM, I have tested it previously as you said with different frequencies at constant amplitude and it doesn't roll off much at 1KHz.

 

[stv014]

  I didn't think it was a problem with the amp though. With the 33 ohm resistor as the load instead of the headphone, I measure low distortion on the amp (>0.01%) in THD and IMD tests, at significantly higher output power.

 
Well, if there is clearly audible distortion at 1 kHz with only the headphone as the load, but not with resistors of comparable resistance, and neither with serial resistors at matched SPL, then that does suggest there is a problem somewhere. On these InnerFidelity measurements, the SR125 has only about 0.3% distortion at 1 kHz at 100 dB SPL (0.329 Vrms, a few dB lower than your test level, but that does not explain the >10x difference), and that is with a low impedance amplifier. What happens if you try different serial resistors, like 4.7, 10, 15, etc. ohms ? Does the distortion suddenly drop at a certain resistance ?

 

[MindsMirror]

   
Well, if there is clearly audible distortion at 1 kHz with only the headphone as the load, but not with resistors of comparable resistance, and neither with serial resistors at matched SPL, then that does suggest there is a problem somewhere. On these InnerFidelity measurements, the SR125 has only about 0.3% distortion at 1 kHz at 100 dB SPL (0.329 Vrms, a few dB lower than your test level, but that does not explain the >10x difference), and that is with a low impedance amplifier. What happens if you try different serial resistors, like 4.7, 10, 15, etc. ohms ? Does the distortion suddenly drop at a certain resistance ?

The distortion does start very suddenly, right at the level that I tested. So I guess it's possible that the one on Innerfidelity would have done the same. All I have are 5x33 ohm resistors, but I did test the magni at 6Vrms into ~19 Ohms (33 // (33 + (33//33//33)), I was trying to test around 33 ohms at a high voltage without clipping the line input, which only goes up to about 2V) and if I remember correctly the distortion was still bellow 0.1%.

 

[stv014]

  The distortion does start very suddenly, right at the level that I tested. So I guess it's possible that the one on Innerfidelity would have done the same. All I have are 5x33 ohm resistors, but I did test the magni at 6Vrms into ~19 Ohms (33 // (33 + (33//33//33)), I was trying to test around 33 ohms at a high voltage without clipping the line input, which only goes up to about 2V) and if I remember correctly the distortion was still bellow 0.1%.

 
You can still test with 16.5 ohms serial resistors then, or several different values if only one channel is measured.
 
It could be possible that the distortion of the headphone increases so much from a relatively minor increase in voltage, but I am not really convinced, as you reported it does not happen with the serial resistors and matched level on the drivers, and without the resistors, the very high distortion can also be measured on the amplifier output, which presumably has < 1 ohm impedance. The fact that increasing the serial resistance decreases the distortion by a large factor suggests the source of the distortion is the amplifier. So, I still suspect it might be a problem with the Magni.

 

[MindsMirror]

Okay, I retract my statement about the measurement not affecting anything. With everything plugged directly into the splitter, no serial resistance, the distortion starts at about 0.4V. Unplugging the SBZ mic/line input removes the audible distortion, which then starts again by increasing the level to ~1V. It makes sense to me that the headphones would produce audible distortion at that level. Is there just some interaction going on between the SBZ input and the headphone that is causing them to distort more, and adding serial resistance to the amplifier somehow reduces that?

 

[stv014]

  Is there just some interaction going on between the SBZ input and the headphone that is causing them to distort more, and adding serial resistance to the amplifier somehow reduces that?

 
With the added serial resistance, the amplifier has less control over the voltage on the headphones, so if there is any kind of interaction with the microphone input, then the resistors should in theory make it worse. On the other hand, if the microphone input interacts with the amplifier, for example by making it (more) unstable, then it would make more sense that the resistors fix the problem.
 
By the way, is the microphone input muted on the sound card ?
 
Also, it may be worth trying the test also with the line input. Even if the level is low, 0.5 Vrms relative to 2 is -12 dBFS, so amplifying it in software should not be a major issue. The microphone input is often noisier than the line input anyway.

 

[stv014]

Originally Posted by dprimary /img/forum/go_quote.gif
 
It is hard to tell what effect the input impedance of the sound card might have I don't see it listed in the specs microphone inputs tend to fairly low.

 
They also often output a few V of DC voltage to power the microphone, but I do not think that is responsible for the problem.

 

[MindsMirror]

 
By the way, is the microphone input muted on the sound card ?

I'm not sure what you mean. The distortion disappears when I unplug the mic input from the tested circuit, I didn't mute it.
 
The line input produces identical results, I hear and measure the distortion starting at the same point as with the mic input, and the distortion stops when it is disconnected.
 
I don't know if this reasoning is valid at all, but I was thinking that maybe the high damping factor controls the driver too tightly, and by trying to control the distortion generated by the driver, it just makes it worse. With the series resistance added, the distortion in the driver is not being over corrected by the amp. Is that what you mean by the amp being unstable?

 

[stv014]

  I'm not sure what you mean.

 
I meant the microphone input is not audible on the output of the sound card, otherwise there could be an unwanted feedback effect.
 

Originally Posted by MindsMirror /img/forum/go_quote.gif
 
The line input produces identical results, I hear and measure the distortion starting at the same point as with the mic input, and the distortion stops when it is disconnected.

 
The two inputs probably have very similar electrical properties then. Although it is not impossible that they are in fact the same, and the microphone mode simply applies some digital gain, and maybe enables the "phantom power" DC offset; a solution like that would not be very surprising on a cheap sound card.
 

I don't know if this reasoning is valid at all, but I was thinking that maybe the high damping factor controls the driver too tightly, and by trying to control the distortion generated by the driver, it just makes it worse.

 
That does not explain why the sound card line input, which should have at least 100 times higher impedance than the headphones, makes such a huge difference. Also, electrical damping is limited by the DC resistance of the voice coil itself, so the difference in damping between 0 and 33 ohms source impedance with 33 ohms headphones is in fact only a factor of 2, while you measured an increase of distortion by orders of magnitude. Not to mention higher damping should generally reduce distortion. Finally, if the Magni has less than 0.1 ohms output impedance as advertised, then for 3.7% distortion to be fed back onto its output, the distortion signal from the driver would have to be well above 100%.
 
From the information available so far, there are not many explanations I can think of that make much sense:
- your Magni may be broken in some way
- the Magni becomes unstable/oscillates in the RF range because of capacitive loading from the cables and sound card input, or the sound card loopback interferes with its performance in some other way (ground loops, etc.)
- there is an unknown problem with your measurement setup, but it is difficult to figure out what it is without having access to the hardware; there are a number of things that can go wrong in an RMAA loopback measurement, these are discussed in detail on a certain blog that is not allowed to be linked here
 
There are still a few thing that may be worth trying, however. What happens if you use 16.5 ohms serial resistors (2x33 ohms in parallel), or if you add the 33 ohms resistors to the sound card input instead (as close to the amplifier as possible) ?

 

[Steve Eddy]

Couldn't resist.

se