"ER-4J"
Aug 31, 2007 at 10:39 PM Thread Starter Post #1 of 3

j-curve

100+ Head-Fier
Joined
Apr 15, 2002
Posts
489
Likes
11
Here's a different way to drive any of the 4-series transducers by Etymotic.
The idea is to reduce the driving impedance without sacrificing the treble/bass balance.
This will improve the damping factor at the transducer's resonant frequency, to produce a more natural sound.

Unfortunately, there is a catch. Unless you have access to the rare (unobtainable?) machined pins Etymotic uses to connect their cables to the earphones, you'll be required to perform a "podectomy"
eek.gif
, i.e. removal of the series resistors and injection-moulded pod from the cable. However, this is nowhere near as scary as I anticipated. See below for some tips.

Nonetheless, you probably need a good reason to go hacking into the cable of your $200 earphones. My reasoning was as follows:-
i) Impression of the ER-4P: rolled-off in the top-end, but with a rich mid/treble which makes oboes and violins particularly sweet. That same effect tends towards treble colouration as the music becomes agitated.
ii) Adding series resistance to emulate the ER-4S, highs are better supported but the treble peak also becomes accentuated and annoying at times.
iii) Headroom's impedance graphs point to a driver resonance at 2.5kHz. This resonance cannot be controlled by the amplifier when a high series resistance is present. The ER-4P could be electronically equalised, or better still...
iv) Remove the pod resistor. Find a way to achieve the necessary upward treble tilt in a low impedance circuit.

attachment.php


Crucial Quantities
Input impedance: 36 Ohms, almost constant across the range, due to R6/C6 impedance compensation.
Sensitivity: Around 101 dB/V, i.e. a notch below the HD580, and no apologies! [See the hissbuster link for justification.]

Construction
For portability, I wanted to rig this into a cable, but a small box would have made things infinitely easier.

Sub-assemblies, front
attachment.php


and back. These had to fit into a length of 8mm heatshrink tubing.
attachment.php


Wired, front
attachment.php


and back.
attachment.php


Luckily I realised it was going to be a squeeze, and had the sticky heatshrink primed with chalk dust before wiring the cable.

attachment.php

Not quite as attractive or robust as hoped. [Do you like the DIY strain relievers in 6mm half-shrunk heatshrink? Classy, I thought.
redface.gif
]

PS: I just saw setmenu's Ety filter build. Very neat, wish I'd seen it before I put this thing together.
 
Aug 31, 2007 at 11:13 PM Post #2 of 3
First Listen
From past experience I know not to expect too much from modifications like this, and sure enough, there was momentary subjective disappointment as some of the all-too-familiar Ety "flavour" was effectively suppressed. This is especially the case when comparisons are made with short segments and A/B tests. By extending the duration, using a variety of material, and experimenting with algorithmic vs. circuit EQ, a better perspective can be reached. In summary, I believe the ER-4J is very close to what I'm looking for, certainly a step ahead of the ER-4X (Rpod=47 Ohm) which was taken as a reference point due to its favourable treble/bass balance.
In the process I have heard my Etys sound even better than the ER-4J, but in a configuration which is not at all portable.

Podectomy Tips
Wanting to trace every wire from start to finish, I painstakingly sliced away at the green pod until it was open - quite a chore. With the benefit of hindsight, I would advise simply cutting the main cable about 15mm from the pod, then marking one of the earphone cables red or blue where it exits the pod before cutting them close to the pod too, but with the nuance of cutting one of each pair a few millimetres longer than the other, in both cases. Then with a multimeter you can easily figure out which of the short or long wires running into the pod is the common earth, meaning the opposite (long or short) to the earphone is the earth connection. The main thing is not to reverse the polarity on one side only.
The rest is easy, with some heatshrink, but note, the cable has an unusual colour coding as follows:-
Shield = No Connection
White = Common or Earth
Red = Right
Black = Left
It's a quality cable, the outer insulation being very thin and flexible. The shield itself is there for durability and looks like it will withstand many thousands of flex cycles. The conductors are silver alloy, prone to dry joints perhaps but mine are holding together well so far with regular solder. [Just don't tell anyone.]
 
Sep 2, 2007 at 7:14 AM Post #3 of 3
attachment.php


I hope this project isn't perceived as some kind of evil EQ attempt. The equivalent circuit of an ER-4 transducer is not so different from that of a loudspeaker, basically an inductor with a little bit of resistance and a springy bit in the middle. The objective was to take some of the bounce out of the spring, but when this is done by reducing the external resistance, the rising frequency characteristic due to the driver's inductance is lost. Hence, the external inductor L4 is needed to lift the treble. I don't think of this as EQ, more as "restoring the balance".

attachment.php


The impedance of the earphone varies greatly over the audible range, from around 4 to 80 Ohms. With the ER-4P plugged into an 8-Ohm source, the driving impedance seen by the transducer is source+pod+cable=29 Ohms, which is greater than the transducer's impedance at the 2.5kHz peak, so the damping factor will be less than unity there, not to mention the rolled-off treble of the 4P.

By contrast, the driving impedance from the ER-4J, including the cable, is kept below 16 Ohms in the range to 4kHz. Why so high? Why go to the trouble of removing the pod, only to add back an 8.2 Ohm resistor in series with the output of the network? Initially, I aimed for a very low driving impedance so the first prototype had an output impedance below 8 Ohms up to 4kHz. Listening tests proved that you can have too much of a good thing. The resulting midrange suck-out left a very clinical sound. Violas and Cor Anglais sounded anaemic. Although the 2.5kHz peak was greatly reduced, I was also losing 2.5dB of output at 1.2kHz, and the Etys lost their character. Compromise was necessary, and it seemed reasonable to maintain a damping factor no worse than one in the vicinity of the peak, hence the blue trace on the graph above.

Relative Frequency Responses
attachment.php


What we're looking at here is the voltage across the terminals of the earphone as compared to the "ideal" voltage generated inside an 8-Ohm source, the analysis based on the equivalent circuit above. One way to quantify the treble tilt is to take the output at 12 kHz less the output at 120 Hz. Without this difference the ER-4 produces no highs to speak of, or to listen to either.
smily_headphones1.gif
For the ER-4P the treble tilt is 15dB, ER-4S requires 20dB, ER-4B pod produces 23dB. The ER-4J matches the ER-4X at 18dB.

The dark blue trace at the top compares ER-4J to 4X: -3dB @ 2.4 kHz but only 1.2 dB down at 1.2kHz so our mids are preserved.
Note: The two ER-4J traces are raised 11dB in this graph for easy comparison to the ER-4X. A raw plot has the pink line meeting the axis at -34dB, less sensitive than all of the official Etymotics. My portables can handle it, with volumes set near max.
 

Users who are viewing this thread

Back
Top