Starling-2020 (Seas ER18RNX with Morel MDT29)


I first worked on the Seas ER18RNX in 2010. Over the years, I did a total of 4 versions. The Starling-2020 is my 5th and by far, my best.

In this latest version, I made a few changes. First and foremost, I enlarged the vented cabinet from 15 liters to 18 liters. This is to accommodate an over-damped alignment. The other major decision is in the treble. I replaced the Seas 27TDFC with my favorite tweeter, the Morel MDT29. With this revamp, the Starling-2020 finally came together. 

Seas Prestige ER18RNX

Fig 1 – Seas ER18RNX RAW Response  •  Baffle Width=8-1/2″

I have a love/hate relationship with the Seas ER18RNX. From the outset, she’s not a friendly woofer to work with (Fig 1). To be blunt, she’s awfully difficult. Normally, I would rather avoid working with such a driver but she has certain alluring qualities that continue to draw me in time and again. It is her spontaneity and naturalness in producing music that I find most beguiling. It’s like a turbulent love affair. 

Fig 2 – Seas ER18RNX RAW and LPF

The success in using the Seas ER18RNX lies in the control. If you fail to tame her, you’re finished. She’ll eat you up. She’ very unforgiving in that respect.

The Blue plot in Fig 2 is the ER18RNX with a new network. It took quite a bit of effort and components to get her to roll-off in this manner.

Fig 3 – Seas ER18RNX LPF and Morel MDT29 HPF

Fig 3 shows the acoustic roll-offs of the Seas ER18RNX and the Morel MDT29. My targeted crossover frequency is 2.5kHz and they do cross precisely at that frequency. But it’s more than just the crossover frequency. Notice how the two slopes are almost symmetrical. 

Fig 4 – Startling-2020 Crossover Passband

The Black plot in Fig 4 is exactly what I’m after, a beautiful summation in the passband. This has eluded me for a very long time. 

Fig 5 – Starling-2020 Frequency Response

Fig 5 is the final frequency response. Due to limitations in my room, measurements below 500Hz are in nearfield. See how flat the response is from 2kHz to 4kHz. In fact, it is incredibly flat from 2kHz to about 15kHz except for a small dip at 5.5kHz.

Fig 6 – Starling-2020 RAW and Nearfield

To check on the bass, I added the in-room response below 500Hz (Black plot in Fig 6). What I’m looking for is the level of the bass in relation to the midrange. It is only about 2dB less. My BSC network is about just right because my measurement is in Full Space (4pi). The bass will be louder in a smaller room or when the Starling-2020 is placed nearer to the back wall. What I want to avoid is the bass over-powering the midrange.

Fig 7 – Step Response

The step response of the Starling-2020 shows the ER18RNX is quite fast. The attack is smooth with no breaks in the leading edge. The apex is at about 250 microsec. This is impressive for a 6-1/2″ midwoofer. 

Fig 8 – Waterfall

The Waterfall plot in Fig 8 is quite unexpected. I don’t believe I’ve ever seen such a clean waterfall from a 6-1/2″ speaker before. It is the region below 3kHz that caught my attention. It seems to have very little decay.

Fig 9 – Toneburst Energy Storage

The Toneburst plot in Fig 9 shows some stored energy from 1kHz to 1.5kHz. I don’t think they will smear the midrange because they only last for 4 cycles. 

Fig 10 – Spectrogram

When viewed in a Spectrogram, we can see the stored energy at 1.2kHz. It is not harmful because it is fully dissipated by 4 msec. From the three plots (Fig 8,9,10), we can infer that the Starling-2020 is exceptionally clean.

Fig 11 – Impedance

The Starling-2020 is an easy load for power amplifiers. Most of the time, it is above 7Ω. It only dips to 4Ω at 3kHz and that’s the lowest. Box tuning with the two ports is at 40Hz.  

the Final Chapter

The Starling-2020 is one of the most musical speakers I have. It all has to do with the Seas ER18RNX. The dynamism and realism are out of this world. This kind of reproduction is at high-end level. The only other woofers I have that share similar attributes are the Scanspeak 15WU and the Scanspeak PL18. These Nordic drivers are in a class of their own. Once you hear them, you won’t want to listen to others anymore.

To those that intend to build the Starling, I strongly recommend the Starling-2020. This is the only version that I am truly satisfied with. You won’t be disappointed.

next > Time-aligning the Starling-2020

Unless otherwise stated, all measurements were made with the mic at 36 ins, tweeter axis. Impulse Window=5ms. No smoothing applied.