Text preview for : AKG K551 modifications.pdf part of AKG K551 AKG K551 modifications

Back to : AKG K551 modifications.pd | Home

AKG K551 modifications
The AKG K551 is a headphone intended for use with portable equipment and for Apple products specifically.
In the very thin cord it also has a microphone and (volume) control buttons when connected to certain Apple
products (have not used nor tested it).
It is a closed headphone with a pretty good isolation.

The drivers are 50mm in diameter which is quite large
meaning that when you move it around on your head the
sound doesn't alter too much.

The ear-pads are quite big, much bigger than most
headphones and are very comfortable and soft. Downside is
the ears get warm quite soon.
Getting a proper seal is PARAMOUNT to getting a good
sound. People with bony faces or small heads might not be
able to get a good seal. My head is average sized and
anything but bony, so little seal problems for me. Without a
good seal the K551 sounds 'cuppy' and has no body or bass to it.

Clamping force is very low so when moving your head it is quite easy to momentarily break seal which is
annoying (bass disappears on one or both sides). Working, or walking outside, (for which it seems to be
intended) isn't very do-able for me. It works well when you remain pretty stationary.
The cable is short and very thin but totally not microphonic. Yes, it has a microphone in it, but that's not what
is meant with microphonic. It means when you touch the cable that sound isn't mechanically transferred to
the cups and heard. For me this is an important asset (not the microphone/remote as I don't own Apple

Looks are appealing to me and the build/feel seems very durable.

Adjusting the cups isn't very easy to do and even have to forcefully 'bend' it in position to get a good seal. No
squeaking sounds though, or earcups 'falling' in unintended positions when trying to put it on.

Inside the cups (that can be swivelled 90 degrees so they can lie flat on a surface / transported) are VERY
big letters 'L' and 'R' indicating both channels. This is much better than small print on the inside or outside of
the cups or headband.

The headband padding is soft and comfortable.

The plastic covers on the inside of the headband 'slide' open pretty quick when attempting to adjust the
height. Very annoying and a little screw or other mounting method could have prevented that.


The K551 is a low impedance (35 nominal) headphone and doesn't react much to sources with different
output resistances between 0 and 100.
The output resistance of most portable equipment varies between (close to) 0 and 10 in general.
Nominal power rating is 200mW (0.2W) which is a pretty standard rating for portable headphones.
The efficiency is pretty good. With 1V applied it produces 114dB. 1V 30mW, this means at the specified
200mW an SPL is reached of 122dB... that's painfully LOUD !
As the 200mW is nominal it is quite safe to have peak signals in music that are much higher. Don't know if
your hearing likes that though.
To produce 90dB only 0,12mW is needed (= 65mV) which means on most portable equipment it will play
loud enough to even above decent levels.


First impression was nice. Bass is good and extended. Mids were decent but a bit 'off' it which is most
apparent on female voices and instruments. Highs bordering on sibilance and if there is any distortion in the
recording this becomes accentuated. Fake detail comes to mind but not in a very annoying way.
It is a better than average sounding headphone but IMO not 'reference grade' as it is advertised and the
price-tag suggests. Not a headphone I could listen all evening to, as I can with a few other headphones.

How does it measure ?

Left channel
Right channel

As can be seen it measures quite well from <20Hz to 1kHz with a slight 'bump' in the subsonic area. There is
some channel imbalance up to 2dB but isn't that noticeable.
Above 1.5kHz it drops about 4dB to 3kHz. From that point trouble starts.
The dip around 5kHz is a resonance yet manifests itself as a dip. From 7kHz to 16kHz the driver probably
does not behave quite uniform there. Also in the higher frequencies there is quite some channel imbalance.
Dips aren't very audible but peaks are. In the 8kHz region a peak usually provides the impression of being
'highly detailed' yet the details are fake and certainly after a while may give listening fatigue and 'sibilance'.
The 'Sssss' sound that is usually associated with sibilance isn't present but fake details and accentuated
recording flaws as well. The slightly raised lows and highs will be favourable on 'first listen' or when
auditioning it but might annoy once you have gotten used to its 'interpretation'.

Waterfall plot (CSD) Left channel Waterfall plot (CSD) Left channel

The waterfall plots (500Hz to 20kHz) shows the resonance at 5kHz and around 7kHz. Some channel
imbalance is also visible in these plots in the area below 2kHz. It is the peak in the frequency range what is
the most objectionable part. The ringing isn't that long lived but the fact it is there tells something about the
damping of the membrane though, which isn't on par with some other headphones in the same price range.

It looks like the peak, ringing and overall frequency response need to be addressed. This means ...
..................MODDING TIME................

Getting to know some components

The pads are something I would like to keep as they also determine the sound. I prefer velour type pads but
as the K551 is closed and isolates pretty well (also because of the pleather pads) I will not change those.
The driver is 50mm in diameter and is placed in a small closed cup
behind the driver with a rubber 'port' on the back of the driver. The
picture on the right shows the bottom which attaches to the rear of
the cup. There is a small hole that connects to the chamber which in
turn connects to 2 small holes in the back of the driver. One hole is
for the 'dust cap' and has a small piece of foam in it and the other one
is near the voicecoil. A construction I haven't seen before. There is also some damping
material on the magnet with a cut-out in it for the voicecoil. The hole in the rubber port does not line up with
either of the 2 holes in the magnet. There is a small 'chamber' behind the magnet where all holes converge.

What happens when you remove the rubber 'port' ?, which seems to end outside the plastic cup against the
metal name-plate on the sides of the cups (which seems an odd thing to do).

Rubber port removed, stock

Bad idea to remove the rubber port.
Removing it makes the K551 muffled and
'closed' sounding.
It seems to affect the area between 60Hz and
1.5kHz the most.

On the business end of the driver the fragile membrane itself is protected by a
plate with holes in it. Surrounding the driver there is a ring of slightly porous
synthetic fabric. In most headphones this is some kind of paper but not so in
this case. There are markings on it suggesting there are 10 holes underneath
that cloth. In reality there are only 3 holes (as big as the circles on the fabric
What happens if we seal these 3 holes/vents ?

Stock, vents closed

Bad idea to seal those vents/ports. It makes
the K551 sound a fair bit 'warmer' which is not
what I am looking for, on the contrary.
It seems to affect the area between 150Hz
and 1.5kHz the most and the effect is pretty
similar to removing the rubber 'port' inside.

It seems to be a bad idea to tinker with these two 'components' .

The 8kHz peak

Having modified a few headphones with an electronic peak filter, which ONLY targets the offending area, and
has worked well in the past, a simple notch filter was calculated and built.
(left channel only) stock , filtered.
The peak in the highs is lowered while the
frequency extension (above 15kHz) is not
affected. The dip around 5kHz is deepened
as well because the side bands of the peak
filter are also active. The frequency
response is a bit 'ragged' above 4kHz and
the resonance at 5kHz is still there. The
one around 7kHz is slightly improved.

Having used some 2 mm thick (medical) felt in front of T50RP drivers that also have a peak in the 8kHz area
has shown that this material has damping properties and also shaves off a few dB above 4kHz.
Below the effect of the white felt.
(left channel only) stock , white felt.
Above 4kHz there is a 5dB drop in level
but more importantly the frequency plot is
showing less 'ragged' behaviour indicating
the felt is damping the resonances better
than an electronic filter (as well as applied
EQ) can bring. Exit electronic filter options.
The K551 already sounds less 'edgy' on
some music but at the same time is
sounding a bit 'darker' as well with
sparkling highs to top it off. The highs
seem a bit 'unattached' to the rest of the sound and it still sounds 'off' compared to other top tier headphones.

Having modified a few Superlux HD681-EVO headphones, this has filled my 'acoustic
filter' stock. This stock consists of various Creatology felt pieces, various types of
foam and felt, and that piece of round (50mm diameter) 3mm dense thick felt from the
HD681-EVO. This was fitted in front of the driver to tame the highs. In the HD681-
EVO this did get in the way more than it did any good and replaced the felt disc with
an electronic filter that works much better there.
( left channel only) stock , HD681-EVO felt
Above 5kHz there is a 7dB drop but also above
15kHz. The felt has about the same amount of
peak reduction as the filter does.
The small excerpt from the plot below on the
right shows the differences between an
electronic filter and the HD681-EVO felt.
The extension above 15kHz is better in the
filtered version but more 'ragged' looking and
affecting the dip too much. The felt shows superior behaviour but drops the top end by
about 5dB. HD681-EVO felt seems to perform best.

The headphone now sounds a bit 'dull' and lacks sparkle, but this is caused by the
relatively 'raised' area between 20Hz and 1.5kHz.
This is impossible to cure correctly in a 'mechanical way'. Enter the electronics again....

20Hz 1.5kHz the electronic correction

If the region between 20Hz and 500Hz were dropped by about 6dB and the filter action slowly stops around
2kHz that would be a near ideal correction. This means a high-pass filter is to be constructed with about 6dB
reduction and it has to stop filtering at 2kHz. The driver impedance (including the headphone cord) is around
35. 6dB reduction equals a 'factor' 0.5 in voltage when acting as a voltage divider. This means the series
resistor must be around the same value as the impedance of the driver itself. A practical resistor value of
33 will give the desired amount of reduction. The +3dB point in the frequency plot (when looking from the
2kHz point downwards) is around 400Hz in case of a simple 6dB/octave filter. 1/(2fR) says we will need a
capacitor of 11F in parallel to that resistor. A practical value would be 10F. That capacitor has to be
nonpolar (or bipolar). If the filter is to be small and fit inside the headphone only electrolytic capacitors are
going to be feasible. A Bipolar 10F/25V (or higher voltage rating) must be used in that case. When size
isn't an issue in case the filter is made external (in an extension cord/box) one can use non-polar 10F
filmtype capacitors. Some people will shiver at the thought of adding a series resistance of 33 and thereby
reducing the 'damping factor' to around '1'. Surely that can not be helping.. judge for yourself.
The schematic diagram of a filter in an extension cord (both left and right channel shown).

When the filter is to be built into the headphone cups the
capacitors can be electrolytic bipolar ones. These caps
are not easy to find as most electrolytic capacitors are
There is a trick to use 2 polar electrolytic caps in anti-
series to form a bipolar capacitor of half the value of the
used capacitors, this is shown below.

It does not really matter if one connects the + of both
capacitors together or both sides but because the 2 pins that are going to be connected to the driver will be
bent and when soldered might melt the plastic foil around the cap and possibly make contact with the exterior
(housing) of the capacitor it is recommend to tie both + connections together as the 1 pin is electrically
connected to the housing. This way when there is an accidental short between the pin and the housing of
the capacitor this doesn't matter as they are connected anyway. If the + were to make connection to the
housing it would be 'shorted' and you would (unwillingly) be ending up with a 22F polar capacitor instead of
an 11F bipolar capacitor.
As can be seen from the calculations an 11F capacitor is what is needed so using two 'standard' 22F polar
capacitors makes sense and they can be quite small as well.

On the right a picture of what this would look like in a practical form. Note
the -- is indicated by a grey stripe nearest to the -- leg.
Both polar and bipolar caps have a long and shorter leg where the short pin
indicates what pin is connected to the housing. In case of polar capacitors
this means the short leg is the --.
The capacitors used in this example have a 100V rating (had these lying
around) but can also be 25V, 35V, 50V or 63V that doesn't really matter in this case.
The resistor is a small sized metal-film 0.4W resistor. The type is called SFR16.

Some people have reservations of using (any capacitor) in the audiopath and electrolytic ones are the
biggest no-no one can make. I don't share that point of view but those that do are free to use film type
capacitors. Film types also exist in small SMD sizes which could be used.
There is one little catch when using a series filter as this one, which is the output resistance of the used
source (amplifier, DAP e.t.c.) must be between 0 and 10, which usually is the case.

Final results
Below the frequency plot of the 'improved' K551 with the described filter (page 5) and Superlux HD681-EVO
felt disc used.


This looks more like it. The transformation is sound is remarkable. Below the before and after differences in
one plot (left channel only)


The frequency plot is looking quite flat . People that like big bass certainly must look elsewhere.
The sound is very similar to that of the venerable K501, except this one doesn't have lack of bass extension.
It sounds very airy and detailed without a single hint you are listening to a closed headphone.
Resonances are much smaller now and ringing shorter lived. The waterfall plots below look MUCH improved.

above right channel stock above right channel modified

above left channel stock above left channel modified
Open up the earpieces

The following pages show how to 'copy' what was done. In this case polar capacitors are used as well as the
felt disc from the Superlux HD681-EVO.
Those that want to use the exact same felt disc I would recommend to simply buy an HD681-EVO and use
it's felt disc and modify the EVO as described in the article ' Superlux HD681-EVO ' found here:
http://diyaudioheaven.wordpress.com/schematics/headphone/ .
This way you end up with 2 great headphones. The costs of the Superlux won't be an issue as tit is very
cheap, certainly when compared to the price of the K551 and you won't be disappointed in a modified
HD681-EVO as well. A soldering iron and some experience using it is highly recommended.

Removing the pads is quite easy and simply involves pulling of the pads
GENTLY. There is a plastic flange on the inside that provides some grip.
The mounting flange itself a soft pleather (fake leather) flap.

Remove the 5 screws (in
the red circles) shown on
the right. Be careful not to
drop them in the holes of
the driver, as the powerful
magnet will attract them
and you might end up with
a damaged membrane.

I recommend to open the right cup first as this one is easiest to
modify. The cups are made of plastic and you will find a rubber
component in there. It may either 'stick' to the driver or to the
cup or maybe even simply fall out when opened.
To reduce reflections from the back and sides of the cup I used 2mm thick self-adhesive felt. This material
damps sound pretty good, but other materials that do a similar thing can of course be used. Frequency plots
and waterfall plots showed absolutely no difference, but as reflections are smaller in amplitude they won't
show in these types of plots anyway. Less reflections is always better IMO and thus recommend this.

The area of the metal rods and to the left and right of it must not
be covered with felt (shown in the red circles) as 2 posts on the
baffle will make contact with that area.

Also note the position of the hole in the rubber
'port' (green circle) should be located away from
the headband as shown on the left.

The green encircled part of the rubber 'port'
must align with the small hole in the
cup (shown below on the right).
Strangely enough that hole is
covered by the metal nameplate so
it appears to be sealed there.
The rubber 'port' only fits correctly
when it is inserted in the proper
orientation though.

Mounting the electronic filters

To mount the filter on the PCB of the right driver you must pre-tin the two soldering pads on the right (red and
orange circle). De-solder the green wire that is connected to the pad in the small green circle. The green line
shows where the wire originally wss connected. The de-soldering has to be done VERY quickly. The reason
for that is the wires from the voicecoil, that are EXTREMELY thin and fragile, are also soldered on those
pads. These thin wires are 'protected' with some white paste (in the magenta circle) and should not be
touched. The solder underneath the white paste should not melt too much ! Once the green wire has
been removed it must be soldered onto the pad in the red circle (as shown below).

The pre-assembled filter components (shown on the left) must be
soldered onto the pad indicated with the green circle (see picture
below) and to the pad indicated by the red circle.
Solder the wires as far away as possible from the white paste.
Below a picture of how the filter is connected. Make sure the right
capacitor does not stick out past the PCB on the right (on the left is
no problem). Secure the capacitors with a dab of (hotmeld) glue or
other material that secures these parts. Keep them away from the
magnet as the rubber cap must fit around the magnet and there
must be free space around the magnet to allow for this.

below: capacitors fixed in position so they
can't move around.
(malleable eraser is used in this case)

For the left earpiece the filter is to be mounted in a different way, as the pads that were available on the right
driver are occupied with the wires that cross over to the right earpiece via the headband.
De-solder the blue wire which is soldered onto the most right pad (blue circle) when unmodified.
One side of the pre-assembled filter legs must be soldered onto the pad in the blue circle.
Make sure the top of the capacitor does not stick out past the PCB near the blue circle !
Solder the blue wire that came off the pad on the far right directly on the other side of the filter assembly.
This must NOT make contact with any of the other pads and must be 'floating'.

Because the capacitor on the left is literally 'floating' it must be secured to the PCB to avoid that components
can make contact with the PCB / wiring and to ensure the components it do not rattle around in the cup.

Capacitors can be secured with (hotmeld)
glue or other materials that can stop it
rattling around. In this case malleable eraser
is used so I could experiment and easily
remove the filter when needed but still have
the parts secured in their position.
Make sure to allow for enough free space
around the magnet so the rubber 'port' can
be installed correctly.
Perhaps check this by fitting the rubber 'port'
onto the magnet before permanently
securing the parts.

When you feel the cable needs to be replaced (for your own reasons) now is the time to do so.
Reasons for re-cabling could be to get rid of the mic/remote or if you want a longer cable on there.
The cable is single entry and the return wire is a common one. Given the thickness of the wires it stands to
reason the resistance lies in the neighbourhood of 1-2 per wire. This is rather high and could affect the
stereo image. I left the original cable in place as it is supple and low in microphonics and works fine.


To re-assemble the baffle it must be placed in the correct
orientation /position to the cups.
Twist the cups so they lie flat on a surface and mount the rubber
'port' in the correct orientation (see bottom page 7) and as shown
on the right.

Align the holes in the baffle with the posts in the cup and observe
how the 2 posts on the baffle have to fit over the metal rod in the
bottom of the cup.
Also make sure the wires are 'free' inside the cups and aren't
caught in the rubber 'port' or the edge of the cup.

Carefully reposition the baffle and feel how it 'slides' in its original

Screw back the 5 screws and make sure NOT to over-tighten
these screws as you will tear out the thread in the plastic posts. !

Do the same for the left cup and MIND the WIRES when re-assembling as there are more wires that can get
caught behind something.

When the baffles are back in their original positions it is time to put the felt discs onto the driver.
They can be secured with a few small drops of glue on the edges of the felt or simply put them on there when
the pads are re-mounted. Make sure NO glue seeps into the holes. Because the drivers of the HD681-EVO
are also 50mm the felt discs fit perfectly over the driver. The felt discs will stay in place even when they aren't
glued as they will be held in place by the dust screen that is integrated in the pads.

Driver on the left (right channel) has the felt disc lying on it, the cup on the right is still 'stock'.

Mounting the pads

Mounting the pads is easy to do. Use the small cut-out which is located on the right side in the baffle.
Keep the cups horizontal with the felt pad already lying (or glued) on it.
Simply place the 'flap' of the pad over one end of the cut-out (magenta circle) and rotate the pad till it is gone
360 degrees around and it is done.
Check if the felt disc is exactly in the middle of the
earpads. If not, gently work it in the centre position. It
will stay in its place once it is in the correct position.

Now it is time to enjoy this completely transformed
headphone. Nobody will believe this is a closed

Tweaking the filter circuit.

When you feel the headphone has too little bass/warmth it is simply a matter of changing the value of the
If an external filter is made the resistors could be replaced by small trimmer pots of 50 or 100 or one
could experiment with different resistor values to suit your taste.
Below is a small table with resistor values and corresponding attenuation values.
33 = -6dB
27 = -5dB
22 = -4dB
15 = -3dB
10 = -2dB
4.7 = -1dB

A downside of the used filter is the output resistance of the used source (amplifier, DAP e.t.c.) must be
between 0 and 10, which usually is the case for portable equipment anyway so won't be a problem in

When using it on higher output resistance sources the 'low reduction effect' becomes less and thus lows will
increase in level. Amplifiers with selectable output resistances will thus provide a form of 'tone control'.

Damping factor: Some people are of the opinion that the 'damping
factor' of an amplifier has a profound effect on the sound. The damping
factor story may be considered an old fashioned marketing ploy. What
is perceived as poor damping increased/fat/uncontrolled bass is simply
a matter of voltage division because of a rise in the drivers impedance.
Fortunately the K551 has a very constant impedance and thus the rise
of the source impedance (by the resistor in the filter) does not increase
a specific area in the lows. Instead when one listens to the modified
K551 one can only speak of tight and controlled bass which some may
even find on the 'thin' side of things. Enjoy ! ... I know I do.

Solderdude: http://diyaudioheaven.wordpress.com/ http://diyah.boards.net/ Dec 2013