In this blog post I compare the Fostex FE108EZ against the Fostex FE108NS. The 'NS' series of full range drivers area a recent offering by Fostex that follows in the success of the limited edition 'SOL' series of drivers. NS stands for "New Sol" and adopts some of the technologies developed in the SOL series.
I will compare the objective test data between the drivers and offer my subjective listening impression on the differences as well.
I begin by comparing the test data which can be downloaded as a PDF by clicking HERE.
The Fostex FE108EZ is one of my favourite full range drivers for near field listening. This entails listening at levels of around 75dB with a listening distance of only 2m. In this application the driver provides a very low distortion listening experience along with good dynamic range.
When the NS series was introduced I was immediately interested to hear the sound character compared to the long standing FE108EZ. The two drivers share the same cast aluminum basket however the magnet is slightly smaller on the NS. The EZ has an eight hole bolt pattern while the NS has a four bolt pattern.
Looking at the objective test data shows that these drivers are not as far apart as Fostex's own published data would have you believe. Fostex recently started publishing more accurate test data. I commend Fostex for this. I hope other manufactures follow suit.
To highlight this, below is my measured response on the FE108NS.
Below is Fostex's own published response.
The two response graphs are very close and the differences are likely only related to the baffle mounting differences. I mounted my driver to a 20cm x 20cm baffle which represents a typical mounting. Fostex likely mounted their driver in an infinite baffle with no enclosure. My testing is with the driver mounted in a 3 liter sealed enclosure.
To contrast this against the published data for the Fostex FE108EZ you can see things are somewhat embellished.
Below is my own response measurement on the Fostex FE108EZ.
My response shares some vague similarities with the published data. For example there is dip at 1kHz shared between the two. However the dip is a full -10dB on my graph where the published is only -5dB. This is likely a result of the smoothing applied to the published data. Also there is a +4dB shelf starting at 3kHz. The -10dB dip at 6kHz shows only as small -2dB dip on the published graph.
This highlights the struggle in using published data (at least in the past with Fostex) in deciding what driver is better than another. Many in the DIY community are left helpless, as they cannot rely on other's opinions due to all the variables involved.
Subjective Listening
I listened to the two drivers by swapping the driver out of the enclosure which took about three minutes. I listened to Dana Krall's live performance 'Temptation' for my evaluation. The NS series driver had more presence when compared to the EZ version with Dana's voice. The tambourine also had a more realistic signature with the NS driver. The EZ driver had a more rolled off sound signature. I had to struggle to hear the detail in the tambourine. The NS series sounded like it had a little more dynamic range, with a little more sharpness to the leading edge of transients. Overall I prefer the sound signature of the NS version.
I was interested to see the step response measurement and how that compared between the two drivers. The test data confirmed what I was hearing. Below is the step response on the Fostex 'NS' where you can see the initial step rise is perfectly vertical. The settle time is also nearly vertical. This represents the driver's ability to reproduce transients with more accuracy (look up crest factor with music).
Comparing this against the EZ version things are not quite as good.
Another interesting aspect is the off-axis polar response between the two drivers.
Below is the off-axis response on the 'NS'.
The driver exhibits excess energy in the 2kHz region which is likely due to my baffle dimensions of 20cm x 20cm. Above 2kHz we see a gradual narrowing. At 10kHz we have a 60 degree listening window which is acceptable for a 10cm fullrange driver.
Below is the off-axis polar map for the Fostex FE108EZ.
We can see with the EZ version things are not quite as straight forward. There is the same excess of energy at 2kHz however we are getting a very narrow band of energy at 6.2kHz. By 10kHz the driver has narrowed to only a 40 degree listening window. This goes some ways as to explaining why the tambourine did not have nearly as natural character as the NS version. The reflected energy in the room had a significantly different sonic signature than the on-axis, rendering instruments less life like. Mounting either of these drivers in a better baffle will improve these aspects.
Below are some pictures of the new 'NS' driver unboxing as well as my test setup for reference.
I should also note that I purchased a pair of the Fostex FE108NS drivers with my own personal funds. The Fostex FE108EZ was on loan from my brother who uses them in his 'B' system at home.
Currently I am working on developing a DIY plan that utilize the FE108NS. I've developed the first prototype and my initial impressions are quite good. I've decided to include a woofer cabinet to help support the bass and mid-bass. This also aids in achieving my target frequency response. I prefer a nice hump in the bass and mid-bass region and so dual 5.5" Scanspeak woofers will support this.
Below is a rendering on where I am at with the design.
So far I am experimenting with different L-Pad circuits on the Fostex FE108NS. I've been experimenting with adjustments on the overall bass and mid-bass level by padding the FE108NS down. I still achieve 90dB@1 sensitivity in the bass. I have been experimenting with speaker position in the room along with trying different room sizes. This greatly effects the bass response level but generally I'm finding that a full -10dB pad on the Fostex provides an excellent 'full' sound where bass and mid-bass sounds come through with great richness. I am only evaluating this at lower listening levels (<75dB) to account for the loudness curves that we naturally experience. Below is the most recent ISO 226-2003 loudness curves for 60dB.
My subjective listening is aligned with the ISO 226-2003 data which suggests +7dB at 63Hz and +5 at 125Hz when comparing the differences between the 80dB and 60dB curve in the above chart. This assumes that the mixing and mastering process was done at 80dB.
Below is where I am at with my own in-room response which I find to be my subjective preference. Please note that his is at a 60dB AVG listening level. I suggest you try to replicate this with your own experiment.
The response below represents where I am at currently with my preferred response. This is with the measurement microphone at 1 meter distance from the speaker. My listening distance is 2m. My room size is 11 square meters. The speakers are positioned 30cm from the rear wall and 45cm from the side wall.
Applying 1/1 octave smoothing to see the general trend is presented below.
As you can see the dual Scanspeak 15W/4434G00 woofers area providing substantial aid in the bass region (35Hz-80Hz) as well as the mid-bass region (80Hz-400Hz). This approach differs from what what is known as FAST (Fullrange Assisted Subwoofer Technology) where a subwoofer is used to support the bass frequencies. Often overlooked is the mid-bass region that needs support as well. By adding two 5.5" woofers I am able to drastically reduce distortion in this area as well as provide a more natural falling response. Essentially what I am doing is creating the ideal loudness contour for this speakers intended application, which is near field listening. Many designs fall short in this regard resulting in a shouty, thin, forward, and aggressive sound character. My approach may seem strange to those that are not familiar with near field listening, which is a subset of the audiophile world. I've come to really enjoy this type of experience which is commonly associated with late night listening. I encourage you to try it yourself.