I've been listening to my newly-received DT1990 for the past couple of hours on both my Asgard 2 and my also newly-received A20. I've also given my DT880/600 and my HD598 a try on both amps. Here are some initial impressions:
BUILD:
Beyerdynamic has always had great build quality, and the DT1990 continues that tradition. High quality plastics in the ear cups, beefy metal gimbals, a nicely padded leather headband, and velour pads with memory foam. They've improved their headband adjuster mechanism over the older DTxx0 series so that each click of adjustment gives a solid tactile feedback. My only minor complaint is regarding the swivel joint where the gimbals hold the ear cups -- it's a pretty tight joint with quite a bit of friction, so the ear cups don't swivel very freely. I swiveled the ear cups back and forth for a while to try to work the joints in and they started to loosen up -- so I suspect this will improve with age.
COMFORT:
About on par with the older DTxx0 series. I think I like the ear pads on my DT880 better -- the DT1990 memory foam pads are a little thicker and don't breathe quite as well, so heat tends to build up a little more. However, I like the headband on the DT1990 better than the DT880 -- it's got softer padding, so while the DT880 can bother the crown of my head a bit after an hour or so, I have no such problem with the DT1990.
SOUND:
To be clear, these impressions are for the "balanced" ear pads that came pre-installed. Overall, I find the sound signature to be pretty neutral, with the bass standing forward in the mix a little bit over the mids and treble. It's a brighter sound than my HD598, but darker and meatier than the DT880.
Holy bass, where did that come from?! Knowing my DT880, and having owned the DT990 and T1 in the past, I would not have expected this kind of bass from a Beyerdynamic. Very robust, forward in the mix, fast and impactful. It's restrained enough to keep these headphones in the category of "neutral" tonal balance, but it provides a strong foundation and excellent dynamic impact that lends to an exciting listening experience.
The mids are a pleasant surprise -- they have more body (i.e. lower mids) than the DT880 and are probably a little bit more forward in the mix as well. Male vocals in particular sound a little chestier; female vocals are largely similar between the two models. Overall I find the DT880's midrange presentation is particularly suited for acoustic and classical music, whereas the DT1990 seems to be able to carry all genres with equal aplomb.
The treble is typical of Beyerdynamic -- clean, clear, and well-extended. However, it is toned down just a bit when compared to the DT880, which is something I think a lot of Head-Fi'ers would prefer. I've always found my DT880 to be borderline sibilant -- it's riding that edge where it's almost too bright for me, but not quite. The DT1990 takes a nice step back from that edge. People who have tried Beyers in the past and found them to be a bit too treble-happy might want to give the DT1990 an audition.
Soundstage and imaging is similar to the DT880, which is to say it is very good. Though Beyers don't present the biggest soundstage, it is nicely-sized and seems spherical in shape (e.g. equal width, depth, and height), which lends to a natural spatial presentation. Imaging is very sharp -- it's easy to pinpoint the location of various sounds and instruments.
A20 vs. ASGARD 2:
My impression is that these two amps are technically very similar (e.g. in terms of noise floor, detail retrieval, dynamic range, etc.), and that any differences in tonal balance mostly arise from the difference in output impedance between the two. The Asgard 2 is the more conventional amp with a roughly 2 ohm output impedance, whereas the A20 has a 100 ohm output impedance. The effect of this impedance difference will depend entirely upon what headphones you are using.
First, a couple of notes about the A20. Continuing the Beyer tradition, build quality is excellent. The aluminum chassis is well-finished and very attractive, and the amp is surprisingly heavy (about the same weight as the Asgard 2, despite the Asgard having a beefier transformer). The volume knob is very smooth. While running, the A20 is just a little bit warm to the touch (contrast against the Asgard, which can double as a space heater or a hot plate). I think any worries about the A20 being under-powered are completely unfounded, at least if you're using dynamic driver headphones -- with the HD598 I could barely turn the volume knob past 9:00, maybe 10:30 for the DT1990, and 12:00 for the 600-ohm DT880.
With the DT880 and DT1990, the difference between the two amps is slight but noticeable. The tonal balance is largely the same, but I'd say the A20 gives both headphones just a touch more bass -- so it comes down to a matter of preference which is better. I'd probably lean toward the A20 for these headphones, particularly for the DT880, which can benefit from a little extra bass IMO.
With the HD598, the Asgard 2 is the very clear winner. On the A20, the HD598 has significantly more bass, and it starts to bleed into the midrange and muddies up the sound. The HD598 needs an amp with low output impedance, plain and simple.
AN ASIDE -- OR, WARNING: SCIENCE AHEAD!
The effects of amplifier output impedance on headphone frequency response can be modeled with reasonable accuracy. Dynamic drivers (headphones or speakers) can be modeled with a simple passive circuit:
Here RE1 is the electrical resistance of the voice coil, LE1 is the electrical inductance of the voice coil, RM1 is the mechanical damping of the driver suspension, CM1 is the mechanical spring force of the driver suspension, and LM1 is the mechanical moving mass of the cone, suspension, and air load. RM1 also incorporates friction and other losses. You can pick values for these parameters to try to match the measured impedance curve for a headphone.
When the headphone is connected to an amplifier, the amplifier's output impedance will be in series with the driver circuit model. Since the driver's impedance changes with frequency, the amount of power that will be delivered to the driver will also be dependent on frequency. This power difference can be converted into SPL changes.
I picked parameter values to match InnerFidelity's impedance curve for the DT880, and then modeled the effects of varying amplifier output impedance on the frequency response of the headphone. The plot below demonstrates the effects -- the black line corresponds to a 0 ohm amp output impedance (similar to Asgard 2), whereas the purple line corresponds to a 100 ohm output impedance (similar to the A20). As you can see, the differences in frequency response are fairly minimal -- just a fraction of a dB of increased output in the bass and treble regions. Some might argue that these differences are not even audibly noticeable, but I definitely heard a small difference when listening, probably a little more than is predicted by this model.
I also did the same computation for the HD598, giving the plot below. You can see the A20's 100-ohm output impedance results in a nearly 6 dB increase in bass response, which is most definitely noticeable. Subjectively, the effect was not a good one.
I'd do the computation for the DT1990, but InnerFidelity hasn't measured these headphones yet and I'm too lazy to do it myself