My purpose is to capture relative differences.

And your measurements/graphs still fail in that regard. You can use a potato to record headphones and if the results are bad in an "absolute" sense they won't tell you much in the way of relative differences. The fact that the only variable being switched out is the headphone doesn't change this.

DT990 are bright and bassy to my ears. I'm don't read personalaudio.ru so I cannot say much. Are their measurements smoothed in any way? Sometime smoothing will eliminate dips. Also, sometimes dips arise from interactions between coupler and headphone - wave cancellations. This phenomena is difficult to eliminate and varies among measurement rigs.

As for which to trust: You already know the answer. It's relative differences within the same framework that matter. Comparing measurements taken using different rigs is a really bad idea. As Jude says in his HF profile, different measurement rigs, different results. You can see a picture of my first iteration measurement rig if you click on Jude's profile. I'm now using a damper solid coupler with hair, ear simulators now for my V2 iteration measurements.

Any attempt to correct or compensate the measurements here to another system will result in catastrophic error and disaster. World might possibly end with floods, earthquakes, etc.

DT990 (old 600 Ohm) ARE bassy and trebly to my ears and my (yet based on other types of measurements) rig as well.
Of course my measurements too are incorrect in an absolute sense but are useful for me in a relative sense to my other measurements.
They may be useful for others as well though.
Only one person on this planet seems to have an absolute accurate (proprietary) rig but he won't be measuring the DT990 for you. 



Non-smoothed and somewhat 'corrected' so that a horizontal line would sound 'flat' to my ears.
The test rig only set me back a few bucks so don't complain about inaccuracies please.

The plot basically is very similar (when smoothed) to the personal audio.ru plots.
The (extremely smoothed) plot below is from the mentioned Russian site:
http://personalaudio.ru/raa/otchety/naushniki/beyerdynamic-dt-990-600-ohm/




Since this thread is about the HD650 I'll also post what my rig 'hears' of it..... for reference sake.



For what it's worth... IMO you can't EQ the DT990 so it'll sound like an HD650. You may be able to get it closer tonally but they will still sound different.
At least that's my experience (own and EQ'ed both pretty flat). The HD650 is pretty easy to EQ really flat, the DT990 is nigh-on impossible without steep filters.

@Solderdude, @marvey
I expressed myself wrong by talking about emulating HD650 sound. Forget about it. I just wanted to point out that by correcting dt990 according to specific measurements(take a look at headphoneinfo which is significantly different in some regards) you get a cloudy-lifeless sound.
This shouldn't be happening, realizing the fact that fidelity is something objective, very specific and measurable(regardless of how humans perceive sounds) and fidelity involves flat FR. Even if we couldn't simulate hd650's sound(which is the most reasonable for certain reasons) there should have been an improvement.

This shouldn't be happening, realizing the fact that fidelity is something objective, very specific and measurable(regardless of how humans perceive sounds) and fidelity involves flat FR. Even if we couldn't simulate hd650's sound(which is the most reasonable for certain reasons) there should have been an improvement.

What specific EQ settings did you use?
Maybe you prefer the dt990 with a bit of brightness?
Maybe the dt990 driver has a tendency toward being bright and correcting it does not yield good results? Like trying to make someone who is gay straight?
Maybe your dt880 measures slightly differently than here? strong possibility here.
Maybe your subjective experience was influenced by the measurements that you took?
Maybe you have no or little idea what you are doing?

I dunno, lots of possibilities.

As an aside: Sorry, fidelity is not measurable. I'm on my my 22nd or 31st speaker project / variation now. I can tell you that fidelity is not measurable. Even if you take the trinity of FR, non-linear distortion, and CSD, that trinity only tells you 50% of what is going on.

Finally, if you want easy numbers lining up and elegant solutions that work out well, I suggest a hobby in pure mathematics below the university level. Otherwise your whining and bitching about stuff not making sense is getting annoying. I simply don't have the answers. Lots of shit doesn't make sense, even with regular ol' speaker measurements, lots of shit doesn't make sense; but you figure it out over time.