Hello,
Sorry for admins if i put in wrong section, you can move freely to better one (if correct section even exists on diyA). Anyway, i enjoy this forum a lot so i decided to share it here.
Anyway, as you know HD800 has terrible peak at 6kHz. Between HD800 and HD800S sennheiser has added an interesting thingamajig in the center helmholz hole of the driver assembly which has been previously emulated (i wont say copied, because for example the superdupont mod which this follows, differs in all but the principle)
Here is how a bare driver in an earcup looks, we will be retuning this resonator in a high Q manner so as not to negatively impact other frequencies, only to absorb resonance at 6kHz.
PROCEDURE
Before we begin,
Do not use my measurements as a metric for what they sound
The rig was a flat plate and does not capture any pinna or ear canal interactions. It does not have headphone compensation curve applied.
The only goal of the measurement rig was to be self consistent and repeatable, to give us a relative change (delta).
To test this, measurements were performed by taking away and putting back on the headphone, this would always capture a slightly different transfer curve
You can see for example that the bass area differs due to a different seal achieved on each run, and the very highs somewhat as well due to the different cup interactions with the mic when it is positioned slightly differently
Yet despite this the whole midrange and the 6kHz peak are in excellent agreement
Then an RMS average of these runs was taken to make it easier to interpret later
BEFORE measurements
AFTER
AVERAGES OF BOTH (Before = red, after = green)
Sweeps were performed with identical settings, and trace was not realigned, so seems like overall efficiency is slightly higher as one trace hovers above the other.
Peak seems to have shifted up by about 150Hz. There seems to have been created another peak at 11k. I cannot really hear it because its narrow, or maybe its a measurement thing, but who knows.
Midrange isnt attenuated and if anything got a little boost at 1.1k (1.5dB).
After re-aligning the peak difference (83.31-78.81 = 4.5 dB). This is DEFINITELY noticable, as it also affects things from basically 4 - 10k. Sounds a lot less harsh and agressive, a lot more listenable.
It is overall a positive change, but if i had to say anything, i would have liked the resonator to be just a little more high-Q, because i didnt necessarily want to decrease 7-10k part, but its all good in the hood.
CSD BEFORE
AFTER
All settings identical, nothing changed, zoomed, rescaled, whatever.
Also, ignore below 300Hz and above 10kHz because as we noted above, these were variable between individual measurements, and everything in between had excellent agreement.
The image is the same between all individual measurements, even if i wanted to i could not make the bottom look anything like the top.
A very significant reduction in stored energy and ringing. Note that this was already at a quite low level and wasn' t that obvious to begin with. Perhaps this is also a part of why i noted it to be less "harsh"?
Interestingly, it seems to work all the way down to 1khz, cleaning up it up quite a bit.
Sorry for admins if i put in wrong section, you can move freely to better one (if correct section even exists on diyA). Anyway, i enjoy this forum a lot so i decided to share it here.
Anyway, as you know HD800 has terrible peak at 6kHz. Between HD800 and HD800S sennheiser has added an interesting thingamajig in the center helmholz hole of the driver assembly which has been previously emulated (i wont say copied, because for example the superdupont mod which this follows, differs in all but the principle)
Here is how a bare driver in an earcup looks, we will be retuning this resonator in a high Q manner so as not to negatively impact other frequencies, only to absorb resonance at 6kHz.
PROCEDURE
Resonator is basically a similar take on SDR already. I will not BS and say they must be some gold doped magical materials, total cost of materials was probably 2€, just stuff that seemed vaguely appropriate i found in a random hobby store (only one in my city).
We have some felt material, creatology foam like material, and furniture pads (you put underneath to protect floorboards). Ignore the coaster mat, that is for another day.
I wont go into too much detail about construction but outer diameter of all is 14mm, we need hole punch of that size in everything.
Then we do 5mm punch in felt and foam, and 10mm punch of furniture pads.
Then find a random business card and make a 14mm donut with no hole from it as well.
The stack is felt-foam-pad-pad-pad-businesscard (depending on the thickness of your pads it might be one more or less). Total height needs to be 7-8mm.
Just cram this bad boy in there, so that bottom is pressing against bottom, and top is flush with top.
We have some felt material, creatology foam like material, and furniture pads (you put underneath to protect floorboards). Ignore the coaster mat, that is for another day.
I wont go into too much detail about construction but outer diameter of all is 14mm, we need hole punch of that size in everything.
Then we do 5mm punch in felt and foam, and 10mm punch of furniture pads.
Then find a random business card and make a 14mm donut with no hole from it as well.
The stack is felt-foam-pad-pad-pad-businesscard (depending on the thickness of your pads it might be one more or less). Total height needs to be 7-8mm.
Just cram this bad boy in there, so that bottom is pressing against bottom, and top is flush with top.
Before we begin,
Do not use my measurements as a metric for what they sound
The rig was a flat plate and does not capture any pinna or ear canal interactions. It does not have headphone compensation curve applied.
The only goal of the measurement rig was to be self consistent and repeatable, to give us a relative change (delta).
To test this, measurements were performed by taking away and putting back on the headphone, this would always capture a slightly different transfer curve
You can see for example that the bass area differs due to a different seal achieved on each run, and the very highs somewhat as well due to the different cup interactions with the mic when it is positioned slightly differently
Yet despite this the whole midrange and the 6kHz peak are in excellent agreement
Then an RMS average of these runs was taken to make it easier to interpret later
BEFORE measurements
AFTER
AVERAGES OF BOTH (Before = red, after = green)
Sweeps were performed with identical settings, and trace was not realigned, so seems like overall efficiency is slightly higher as one trace hovers above the other.
Peak seems to have shifted up by about 150Hz. There seems to have been created another peak at 11k. I cannot really hear it because its narrow, or maybe its a measurement thing, but who knows.
Midrange isnt attenuated and if anything got a little boost at 1.1k (1.5dB).
After re-aligning the peak difference (83.31-78.81 = 4.5 dB). This is DEFINITELY noticable, as it also affects things from basically 4 - 10k. Sounds a lot less harsh and agressive, a lot more listenable.
It is overall a positive change, but if i had to say anything, i would have liked the resonator to be just a little more high-Q, because i didnt necessarily want to decrease 7-10k part, but its all good in the hood.
CSD BEFORE
AFTER
All settings identical, nothing changed, zoomed, rescaled, whatever.
Also, ignore below 300Hz and above 10kHz because as we noted above, these were variable between individual measurements, and everything in between had excellent agreement.
The image is the same between all individual measurements, even if i wanted to i could not make the bottom look anything like the top.
A very significant reduction in stored energy and ringing. Note that this was already at a quite low level and wasn' t that obvious to begin with. Perhaps this is also a part of why i noted it to be less "harsh"?
Interestingly, it seems to work all the way down to 1khz, cleaning up it up quite a bit.
Attachments
Last edited: