I know folks are probably more interested in ‘unveiled’ impressions, but I figured I’d give a Susvara 2024 updated take with new data. I’ve been spending time with it again.
Frequency Response
The most important metric
B&K 5128:
Raw
Calibrated
This is a generally solid result for open-back front sealed planar magnetic headphones. So with these types of headphones you shouldn’t expect to see a bass shelf, unless there’s a filter added (like what you see on the Expanse for example).
A few things to note with the Susvara’s tonality. There’s a slight forward character to the lower mids, followed by a dip around 2khz. This is typical for HiFiMAN headphones. What’s not typical for the high end ones though is that the Susvara is actually quite a bit less bright than say the Arya or HE-1000 series headphones. Those have notably more ear gain.
But this is just one head, here it is on another head.
GRAS:
vs Harman 2018
vs DF Tilted
This uses the calculated DFHRTF for the GRAS KB5000, tilted by 10dB.
vs “High Res Harman”
The Harman target is highly smoothed, so this applies the same preference filters from 2018 to the high res, unsmoothed DFHRTF for the GRAS KB5000 (calculated).
Note the dip from 9-11khz that’s actually part of the measurement rig’s DFHRTF, but it’s missing from Harman OE 2018. In the past we would’ve typically said “ignore the 9khz dip” for headphones that exhibited it, but here we can actually see those fine-grained features.
It is still slightly different from Harman OE 2018 though since the in-room baseline isn’t exactly identical to the DFHRTF for ear gain, but it’s very close.
For those wanting to still use the GRAS data and Harman 2018, but have an updated read of it, this might be helpful. I’ve been playing with EQing to the treble here and comparing it to the EQ results I get from the 5128. Ultimately it’s just different ‘heads’, but this is still interesting.
Harmonic Distortion
SPL calibrated
Ah yes, the controversial one that doesn’t really matter. I think some saw a post where the Susvara was ‘disqualified’ due to SPL limitations, and you’d see that in harmonic distortion. Let’s see if that’s the case.
But it should be noted that Harmonic distortion isn’t audible even at far lower volumes unless things are truly awful. So none of this actually matters since in this case harmonic distortion under normal circumstances is far below audible. I just wanted to push it to the edge.
95dB
105dB
114dB
Turns out things are still below 2% harmonic distortion, even at 114dB.
How to read this? Harmonic distortion products are multiples of the fundamental, so a 1khz tone would have a 2nd order product at 2khz. This just plots each product for a full sweep rather than a singular tone.
Typically it’s the higher order stuff that you have a reason to worry about, so in this case that would be the green, pink and teal lines, since those are significantly further away from the fundamental and more likely to be audible (outside the masking window). But… with that said, for the Susvara these are still below 0.2% at 114dB.
I was also unable to detect any clipping, even when playing infrasonic tones (below 20hz). This can occur when there’s an SPL limit, like in the case of some of the open-back Focal headphones that hit their driver excursion limit around 107dB, but there’s none of that here.
As usual, be careful with how loud you listen… but the Susvara will have no problem making you go deaf without any audible distortion products in your music if that’s your aim.
Now let’s see if there are any issues with resonances and if there’s a reason to care about time domain information.
Excess Group Delay:
Shockingly not.
EQ
So despite this headphone having a somewhat modal response (the wibbles), it’s actually a solid candidate for EQ.
Now as mentioned, I’ve been experimenting a bit with the GRAS KEMAR DFHRTF and its fine-grained features to see how that tracks on my head. So I actually have two EQ profiles for this headphone, and it could be that each works for different people.
GRAS EQ
I made this one to more closely have things fit with the treble features of the GRAS. So you’ll notice above that the GRAS data shows some excess energy around 10-11khz, and this profile serves to balance that out a bit for that head.
But interestingly, while there does seem to be some excess energy on the 5128 in the upper treble as well, it was more the 9.5khz feature that stood out there. So here’s the profile based on the 5128 result.
I think both should be personalized to each individual’s preferences however possible, and I’d say neither are ‘wrong’ here… just specific to each head.
Other
Here’s the Susvara’s resonance frequency, which helps indicate driver damping:
Basically when there’s an air gap, the bass will boost since there is effectively no driver damping in this case. A narrower peak (high Q) entails less damping, while a wider peak or no boost with a gap indicates more critical damping.
Thankfully, the Susvara also has a particularly low resonance frequency meaning that if you break the seal (like by wearing glasses), you’ll actually get an increase in bass rather than a decrease.
Notes
- The Susvara has a more ‘modal’ response. You’ll see some of those wibbles throughout. What some may not realize is that these features usually aren’t perceptually relevant, since they’re quite fine-grained. And you can actually check this yourself by doing a manual sweep - you can’t hear most of these features even with test tones. Is it a good thing? certainly not. But it’s also not something to be too concerned about in practice.
- Subjectively exceptional. I’m glad to be reacquainted with this headphone, as a reminder for what sound can be. I was thinking a re-evaluation might change my mind about this one… it hasn’t.
- Large cups, but reasonably comfortable. I’ve always hated the wood inlays for these headphones, but apart from that they look and feel nice.
- They weigh 487g on my scale with the cable attached.