CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => Headphone Measurements => Topic started by: stv014 on September 16, 2013, 02:31:02 PM
-
Tyll's plots of the newer DT880 indicate around 0.2% at 1kHz which will consist of D2 mainly and makes the measured 0.3% plausible.
If you refer to the InnerFidelity PDF files, note that they include THD+N graphs, not THD, and at 90 dB SPL the result is often dominated by noise. That is why you can so often see lower THD+N at 100 dB than at 90 dB. On the other hand, HeadRoom shows actual FFT plots generated from the same data as the InnerFidelity PDF files, and the test signal is a 500 Hz tone at (I think) 90 dB SPL. On those graphs, D2 is typically at about -70 dB, which agrees with my 0.05% figure. For comparison, here is also a graph of my DT880 Pro at 500 Hz (recorded with an ECM8000 microphone):
(http://i39.tinypic.com/2yva3j6.jpg)
This shows a D2 at -68 dB at -10 dBV (about 90 dB SPL), so it is fairly similar to the HeadRoom graphs.
-
You sure about that? 0.05% is very very low and more typical of non-chip amps.
I do not think it is unrealistic at midrange frequencies and not too high SPL. I have just shown it above, although the accuracy of my not particularly high end test gear might be in doubt. However, the HeadRoom graph shows very similar low order harmonics.
-
Isn't that -58db instead of -68 for D2?
-
Isn't that -58db instead of -68 for D2?
It is -58 dB for the 0 dBV (loud) test tone, which is red and orange on the graph. For the -10 dBV signal (blue), D2 only reaches about -68 dB, and D3 is at -78 dB; this admittedly might not be easy to see, I should have created two separate graphs to avoid the confusion. These levels were chosen to roughly correspond to 90 and 100 dB SPL, based on the InnerFidelity measurement of -10.49 dBV voltage required for 90 dB SPL.
-
The InnerFidelity graph for the 880-32 shows THD+noise at about 0.3% at 1kHz. The Headroom FFT plot at 500Hz doesn't look like the above one.
-
It is -58 dB for the 0 dBV (loud) test tone, which is red and orange on the graph. For the -10 dBV signal (blue), D2 only reaches about -68 dB, and D3 is at -78 dB; this admittedly might not be easy to see, I should have created two separate graphs to avoid the confusion. These levels were chosen to roughly correspond to 90 and 100 dB SPL, based on the InnerFidelity measurement of -10.49 dBV voltage required for 90 dB SPL.
Ah got it. The red sort of blended in with the purple. Eyes getting bad.
-
Tyll's plots show -53dB at 500Hz
Purrin's plots show -53dB at 500Hz IF they were referenced to 0dB in the plots they too would show -68dB
STV's plots show -68dB at 500Hz
Science is fun but measuring headphones isn't as easy as measuring electronics.
It all stands or falls with references and accuracy of the gear.
10dB is a LOT though. :D
-
Different headphones though. The 880-2003 may have some very different characteristics.
-
The InnerFidelity graph for the 880-32 shows THD+noise at about 0.3% at 1kHz. The Headroom FFT plot at 500Hz doesn't look like the above one.
As noted above, the InnerFidelity graphs include noise, which makes them unreliable for low distortion levels at low SPL. At HeadRoom, the 32 Ω graph shows similar D3 as mine (note that only the blue traces are relevant here, the orange/red ones are at higher SPL, and obviously show more distortion), and better D2; that can be down to sample variation, or better microphone quality. At very low distortion levels, the ECM8000 does become a bottleneck.
-
Different headphones though. The 880-2003 may have some very different characteristics.
That is true, I noted that possibility already in the first post. One might also wonder if the fact that I have the "pro" version makes a difference, although it is allegedly identical other than having a more clamping headband, which could affect bass performance.
-
Tyll's plots show -53dB at 500Hz
Actually, no (I do not know how to post this link correctly because of the characters it includes, you need to copy and paste it):
http://graphs.headphone.com/graphCompare.php?graphType=1&graphID[]=953&graphID[]=963&graphID[]=2751&scale=30
Taking into account the significant sample variation, these are roughly similar on average to my graph for D2 and D3 (the high order distortion is even more random). That is, assuming that they were made at 90 dB SPL.
-
Let me get an FFT. The distortion measurements above were actually combobulated from centered impulse responses which were in turn combobulated from frequency sweeps. This sometimes yields different results.
-
purrin, so far as subjective performance goes... do you like these way more than the newer dt880s or do those do other things better?
-
purrin, so far as subjective performance goes... do you like these way more than the newer dt880s or do those do other things better?
Probably and likely - I don't have new pair on hand to compare. Mid-treble brightness, but very manageable with EQ. Definitely much smoother overall sound. Doesn't seem to "stab" as much or at all, but again, bright sounding. I think the newer ones seem to extend lower.
-
Here's a 500Hz tone at ~90db. (-10db corresponds to 90db SPL on the graph - at least I hope and the mic / overall system hasn't gone off calibration). The signal is being pushed harder here than the combined FR/distortion graph shown in the second post - since those sweeps were taken at 90db SPL/A white noise, which meant 500Hz would be ~87db.
-
500Hz @85db.
-
Purrin's plots show -45dB (0.56%) for D2.
What's also clear is that D3 and D4 are noticeably lower in amplitude (favourable) while in the HR plot D2, D3 and D4 have about the same amplitude (for 250Ohm version)
In your (stv014) plots D3 is much lower in amplitude and D4 drowns in noise where it doesn't in HR plots.
The link points to headroom not Tyll's plots..
When I mentioned Tyll's plots I actually mean Tyll's (Innerfidelity) plots ;):
http://www.innerfidelity.com/images/BeyerdynamicDT880250ohm.pdf
http://www.innerfidelity.com/images/BeyerdynamicDT880600ohm.pdf
The strange HR plots (equal amplitude of harmonics) makes me question that plot.
Can't measure it myself as I do not own a 2003 DT880 any more (I did own the 2003 version for a while)
Either the headphones differ considerably or something else is wrong,
In any case DT880 doesn't sound bad and I feel it is better to listen to a headphone than measure it. :)p17
-
The link points to headroom not Tyll's plots..
If I recall correctly, HeadRoom actually use (or at least have used) Tyll's gear for the measurements.
The high amount of high order harmonics is probably not normal, it could be bad quality control, or some unknown problem. It is not present for the 32 Ω and balanced 250 Ω versions, but it is shown for the regular 250 and 600 Ω models. It seems to be fairly random.
-
I had a look at a couple of other headphone measurements, and there seems to be a general trend of higher distortion than at InnerFidelity (for example, the HD598 here is shown to have almost as much bass distortion as on Tyll's graph at 100 dB SPL) and HeadRoom. I wonder if there is a difference in how the SPL is calibrated ?
-
Dunno, I haven't calibrated my mic/CLIO system in several years. It could be off or maybe I broke it after some recent high SPL testing.
BTW, I was using A-weighted SPL for the HD598 measurements. The HD598 is fairly even-keen and has a very broad bump centered around the upper bass. A-weighted, SPL is likely to be 3-5db louder for such a headphone than non-weighted SPL which I believe Tyll uses. For trebly headphones, such as the SRH1840, the case should be opposite, which it is (my SRH1840 harmonic distortion numbers are slightly lower than IF.)
I'm likely to discontinue use of A-weighed SPL and line up 1kHz at 87db going forward for the harmonic distortion measurements. The SPL/A calibration is just too screwy.
I think the HeadRoom measurement of the DT 880s are messed up with those roughly equal magnitude high harmonics popping up at every 500Hz interval. That data was probably taken when Tyll was still at Headroom as I remember seeing the same data (but with older visualization several years ago.) We measurers all learn quite a bit over time and refine our methods (it's always tough to implement refinements if we want our measurements to be comparable and consistent with each other.) I know Tyll made a clean break when he went off on his own with IF and his current measurements are different from HeadRoom legacy data.
BTW I think your DT-880 measurements are messed up too. You are getting way too low second harmonic for the DT880 driver or any dynamic driver. In my years of experience, I don't recall seeing any driver have a second harmonic at -68db at any reasonable volume. That's nuts. I know Tyll's IF measurement ~0.3% @1kHz is THD+noise, but THD+noise is typically only a little be higher than THD by itself, not a magnitude higher. Subtract the noise, and the THD is likely to be ~0.15 to 0.25%, with the second order distortion being similar to THD (since second order dominates), not your 0.04%.
-
BTW I think your DT-880 measurements are messed up too.
Well, that is not impossible, maybe the microphone "corrects" the distortion by having second order distortion with inverse phase relative to the headphone drivers. At high SPL and high frequency, the measurement definitely does become microphone limited (if I move the microphone away from the drivers, so that it does not receive as high SPL, the distortion drops significantly). However, the result seems to be fairly consistent with changing frequency, level (graphs will be shown soon), and even with different headphones I still get low distortion.
One thing I cannot guarantee is that the SPL is correct. I simply used Tyll's sensitivity measurements as a reference, so that I only needed to measure voltage, which I can do accurately enough.
You are getting way too low second harmonic for the DT880 driver or any dynamic driver. In my years of experience, I don't recall seeing any driver have a second harmonic at -68db at any reasonable volume.
It is actually fairly consistent with the HeadRoom distortion measurements, at least as far as the 2nd harmonic is concerned. For some Sennheiser headphones, HeadRoom in fact shows even lower distortion.
I know Tyll's IF measurement ~0.3% @1kHz is THD+noise, but THD+noise is typically only a little be higher than THD by itself, not a magnitude higher.
0.3% THD+N at 90 dB SPL is 39.5 dB distortion + noise SPL. Without A-weighting (and THD+N measurements are usually unweighted), it can plausibly be entirely noise. Especially if the possibility of inaccurate SPL (due to the headphone not having perfectly flat response) is taken into account, so it may even be a few dB less than 90.
As an example, the THD+N (unweighted 20 Hz to 20 kHz) of the same tone that I plotted earlier to show the -68 dB distortion is about 1%. Of course, that was recorded with a crappy Behringer measurement microphone in a "normal" room, but even with better equipment it is not easy to get the noise at 90 dB SPL to under 0.1% without weighting.
The fact that the InnerFidelity graphs so often show less THD+N at 100 dB SPL than at 90 dB also makes it likely that the lower SPL measurement is actually dominated by noise.
Subtract the noise, and the THD is likely to be ~0.15 to 0.25%, with the second order distortion being similar to THD (since second order dominates), not your 0.04%.
Of course, if there is actual high THD, then adding the noise will not increase it as much. But the point is that even the noise alone can already produce a relatively high THD+N, so it masks any really low THD. Even for your example of 0.25->0.3% increase, 0.166% noise is needed.
-
One thing I cannot guarantee is that the SPL is correct. I simply used Tyll's sensitivity measurements as a reference, so that I only needed to measure voltage, which I can do accurately enough.
Maybe that is part of the problem. There is sensitivity variance in same make/model headphones. Consider Innerfidelity Beyer 70 2 sets of measurements which hit 90 dB SPL at 0.099 Vrms (0.03 mW), and 0.073 Vrms (0.02 mW). I think that's about 1.76 dB, but 2 headphones is a small sample space to get a feel for the variations. Furthermore, some headphone models might have more or less variation.
It is actually fairly consistent with the HeadRoom distortion measurements, at least as far as the 2nd harmonic is concerned. For some Sennheiser headphones, HeadRoom in fact shows even lower distortion.
I know Tyll's IF measurement ~0.3% @1kHz is THD+noise, but THD+noise is typically only a little be higher than THD by itself, not a magnitude higher.
I think HeadRoom displays the distortion contribution of a single 500 Hz signal, while I think Innerfidelity (Tyll's newer approach), Marv and some others use sinusoidal sweeps to derive THD+N. In these cases THD+N might not just be the contribution of a 500 Hz tone (missing the 250 Hz, 125 Hz and so on contributions which may or may not be negligible). Here is a hopefully useful Linky (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.33.1614&rep=rep1&type=pdf).
Cheers.
-
I have created some THD vs. frequency graphs:
No smoothing (http://i39.tinypic.com/nvzbrr.jpg)
Smoothed version (still affected by noise) (http://i43.tinypic.com/2rps9rq.jpg)
Frequency response (http://i43.tinypic.com/2mfkajb.jpg)
A DT770 measurement is also included for comparison. Note that this headphone is slightly (2-3 dB) more sensitive, so the SPL is higher, even though the voltage of the sweeps was the same.
The frequency range is limited to 22-6000 Hz. Treble is not shown, because I do not have suitable equipment for making usable headphone measurements in that range. Also, there is in fact less bass roll-off, partly because the microphone adds some roll-off too (a few dB at 20 Hz IIRC), and partly due to the lack of a good seal (in the best case, I get up to +6 dB bass boost on the 770). Distortion and FR graphs were generated from the same sweeps, but the former have even lower usable frequency range, because above 2 kHz D3 already falls outside the relatively flat 6 kHz bandwidth. THD products up to the 3rd or 5th harmonic are included, depending on the SPL. Distortion was calculated as total power of harmonics / total power of fundamental + harmonics (so it can never be higher than 100%).
Edit: here is also an older measurement of the same DT880 (http://i43.tinypic.com/110erh1.jpg). This shows somewhat more distortion, and some microphone placement related problems in the >1 kHz range. The difference could be caused by a number of factors like positioning/seal, pad wear, amplifier (it had much higher output impedance then), or even burn-in. :)p17
-
Maybe that is part of the problem. There is sensitivity variance in same make/model headphones. Consider Innerfidelity Beyer 70 2 sets of measurements which hit 90 dB SPL at 0.099 Vrms (0.03 mW), and 0.073 Vrms (0.02 mW). I think that's about 1.76 dB, but 2 headphones is a small sample space to get a feel for the variations. Furthermore, some headphone models might have more or less variation.
Yes, that is quite possible. Although my -10 dBV level assumes 94 dB/mW sensitivity, which is already 2 dB worse than the manufacturer's specification. But it could still be quieter by a couple of dB.
I think HeadRoom displays the distortion contribution of a single 500 Hz signal, while I think Innerfidelity (Tyll's newer approach), Marv and some others use sinusoidal sweeps to derive THD+N. In these cases THD+N might not just be the contribution of a 500 Hz tone (missing the 250 Hz, 125 Hz and so on contributions which may or may not be negligible). Here is a hopefully useful Linky (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.33.1614&rep=rep1&type=pdf).
If it is an actual sine sweep (only one frequency playing at any time), like what I have used, then it should not really make a difference. For multi-tone tests where multiple frequencies are played at the same time, I think the frequencies are chosen such that the distortion products will not overlap (which might not be easy once IMD is also taken into account).
-
Yes. The sweep seems logarithmic and more than likely will and should not overlap. Besides the sensitivity variations and SPL settings I'm not sure. Maybe the number of FFT bins and/or windowing. Could also be rig differences like you said.
The thing that bothers me about noise (THD+N vs THD) being an issue though is that in Marv's plots, D3 and D4 are very low at 1 kHz... unless this noise is affecting even harmonics only at 1 kHz which brings it back harmonic distortion instead of noise floor.
-
The thing that bothers me about noise (THD+N vs THD) being an issue though is that in Marv's plots, D3 and D4 are very low at 1 kHz... unless this noise is affecting even harmonics only at 1 kHz which brings it back harmonic distortion instead of noise floor.
I was referring only to Tyll's graphs regarding THD+N vs. THD. Obviously, when an FFT plot shows the actual distortion, then it is not just noise. I do not doubt that Marv's plots really show high distortion. On the other hand, a typical Tyll's plot like this HD650 one
(http://i39.tinypic.com/71r9zn.png)
shows low THD+N at 100 dB SPL (only about 0.1% in the midrange), and the graph becomes noticeably "fuzzy" when the THD+N level is low. This makes one suspect it is in fact mostly noise above 300 Hz, and at the presumably less demanding 90 dB SPL the driver actually has less than 0.1% THD.
-
I really do agree with Marv's remark about the drivers being different.
The DT880 2003 might have very different drivers (and thus measurements ?) as newer versions as one expects progress over the years and while bringing out different looking models over the years.
For all we know ALL measurements are (very close to) being correct and we are talking of different headphones with the only real common factor of both having 'Beyerdynamic DT880' on it.
THD will always show values (slightly) higher than the highest harmonic peaks and they will always be FAR above the noise limit of headphone amplifiers but might very well be near those of microphone amps.