More food for thought...
Without knowing the output impedance (OI) of sources other reviewers used to perform their DIY-measurements and listening tests, let's take a look at one of the most popular amps for the HD800 / HD800S: the HDVD800.
According to
this post (and others I've found), Sennheiser deliberately designed the HDVD800 with an OI of 43 ohms, to warm the HD800 up a bit. Thing is, even though nominal impedance of the HD800 / HD800S is 300 ohms, their
impedance curve has actually a massive bump in bass around 100Hz, so driving them from a higher OI source will make them sound slightly warmer and perhaps a bit easier to listen to.
However, higher OI also means lower electrical damping, and as a result, slightly less control of the driver and slightly more distortion. I haven't found any measurements for the HD800 or HD800S in that regard, but
this post demonstrates the effect nicely for the 250 ohms DT880 Pro.
So, considering that the damping factor of a HDVD800 driving the HD800 / HD800S is about 330 times lower than that of your Audio Precision APx1701, the listener may hear the headphones as both slightly bassier and less controlled than what your measurements would suggest. Moreover, even a small difference in THD between the HD800 and HD800S (provided there is any in the first place) might increase too, with higher output impedance.
As I said, just food for thought... particularly for those thinking the matter is straightforward and prematurely conclude that the emperor has no clothes.
@james444, you are correct that using an amp with low output impedance is going to be more ideal for measuring headphones, to minimize any chance the amp will affect the measurements. While we mostly use the APx1701, we also sometimes use the Neve RNHP (which also has low output impedance). We try to remember to always show (and link to) the instruments we're using for a given measurement, as it's certainly relevant. We have two entirely different headphone measurement fixtures in three possible configurations, so it's important information to know when you're looking at our measurements.
While I very much enjoy several tube amps for listening with the HD800 and HD800S, they would not be reasonable candidates for the purpose of measuring the headphones, for obvious reasons. And before using any device other than the APx1701 to drive headphones or earphones for measurement (like the Neve RNHP, which we use sometimes), we first measure those devices on theAPx555 to assure they're up to the task of both driving the headphone
and staying out of the way (and to set levels).
As much as is reasonably possible, the idea is to measure only the device under test (in this case, in this thread,
headphones), with the instruments being quiet enough to stay out of the way, and also precise enough to get the full measure of the DUT (device under test). By any measure, I think most would agree that both the Sennheiser HD800 and HD800S are high-precision, low-distortion devices. As such, you need instruments of even greater precision and lower distortion than these high-precision headphones to attempt to get their full and proper measure. To that end, we're using the best audio analyzer currently available and the some of the best measurement microphones and ear simulators currently available.
Given that these are also
acoustical measurements, you also want to eliminate as much environmental influence (ambient noise, vibration, etc.) from the setup. If there is no attempt to isolate the headphone and measurement microphones from environmental influence,
it will reflect in the measurements. At our office, then, we do also attempt to isolate using a laboratory-grade acoustic and vibration isolation enclosure:
This is something we should definitely examine in greater detail in the future, but, for now, read Dan Foley's article in
audioXpress that discusses the importance of precision in audio measurements
: “I Can Hear It. Why Can’t I Measure It?”
We've been discussing harmonic distortion and total harmonic distortion (THD). Let's look at Audio Precision's definition of THD in their APx500 manual:
Audio Precision in the APx500 manual said:
THD stands for Total Harmonic Distortion, a measure of all the harmonic distortion products in the DUT’s output, with the fundamental stimulus tone removed, and without consideration of noise. THD measurements are usually made with band-limiting or weighting filters. True THD measurements can be valuable but are rare (although easily done in the Audio Precision APx analyzers), since with most techniques it is difficult to measure the distortion products without also measuring the noise. See THD+N.
Here's another interesting read at AP.com about THD:
More about THD+N and THD
Now let's take a look at one of the first FFT measurements I posted in this thread (click on it to enlarge):
If THD is the measure of all the harmonic distortion products in the DUT's output, let's look at that in this FFT and the distortion products shown. In this measurement, the headphones measured were measured enclosed in the isolation enclosure shown in the above photo. Again, the fundamental in that FFT is a 40 Hz sine wave. H4 is 80 dBSPL below the fundamental, and H5 is more than 80 dBSPL below. H6 is about 93 dBSPL below the fundamental. While perhaps I'm oversimplifying here, I think it makes a relevant point: I've been asked repeatedly about comparative measurements from
DIY headphone measurements rigs (that probably generate most of the headphone measurements you find on the web), but I just do not feel they are precise enough to give us the answers we're looking for in this discussion -- a discussion of the harmonic distortion of two very high-precision headphones. Are these
DIY headphone measurement rigs precise enough to selectively measure these two headphones' true harmonic distortion amidst instrument noise (see Dan's article above)
and environmental noise? If you're hemming and hawing about the answer, let me answer this for you: No.
Tyll also acknowledged the difficulty of properly measuring distortion:
Tyll at InnerFidelity said:
...Maybe the most important thing to know about my THD+noise measurements is that they're the most likely plot to be erroneous. Because it's a very sensitive acoustic measurement and I don't have a laboratory grade isolation chamber, if a large truck goes down my street during a THD+noise measurement you'll probably see it in the plot as a small rise. Also, if water is running in my house or if the refrigerator kicks on, these might produce enough noise to increase the recorded THD+noise...
Even with the use of a lab-grade acoustic and vibration isolation enclosure, we turn off our two HVAC systems when measuring. If we see environmental influence in a measurement, we either notate it or discard it entirely. If while a measurement is running, if we hear a loud car, truck, or airplane -- or even if someone flushes a toilet in the office next door -- we stop the measurement and wait. As Tyll mentioned, trying to properly measure a headphone's harmonic distortion means having to make a very sensitive acoustic measurement. And the more precise your device under test, the more precise your measurement setup needs to be. And these two headphones are precise.