In this blog post I look at horn No.2007 and it's resulting performance. I had featured this new horn design in my previous blog post. I would like to use this horn in a few upcoming projects, one of which is the 1736 project. This is one of many attempts at finding a high frequency solution that will work with the new Purifi 8" woofer. One of those attempts can be found here.
For now, I will test with the Peerless OC25SC65-04 25mm dome tweeter originally featured here. I initially decided to test this tweeter after seeing good results at AudioXpress Test Bench of a similar tweeter, the Eminence SD28 soft dome. It performs well despite it's low cost.
The idea with horn No.2007 is to increase it's overall size compared to Horn No.1900 which will allow for pattern control down to 1kHz. This will allow perfect polar response matching with the 8" Purifi woofer in terms of LF and HF behavior at the crossover point. I've summarized my goals for this design below. All of these criteria need to be met, which is not small challenge.
- 1.5kHz crossover frequency using LR4 slope
- This means the response is -6dB down at 1.5kHz.
- Pattern control down to 1kHz
- This ensure no widening of coverage as the horn takes over from the woofer. Amir at Audio Science Review refers to these issues as "directivity errors"
- I had originally considered using a tweeter like the SEAS Prestige 27TBCD/GB-DXT however it only has pattern control down to 2.5kHz.
- Low harmonic and intermodulation distortion, particularly near the crossover region of 1.5kHz where diaphragm movement is greatest
- However! We can see artifacts modulate into the upper treble from stress related issues near the crossover region...so watch out!
- Flat frequency response (±)1dB across the bandwidth
- A flat linear response is always an indication of minimal time domain issues such as resonances and stored energy that would smear transients.
Below is the raw frequency response of the 2007 horn using a 30cm mic distance.
If you recall from a recent blog post I tested horn No.1900 which is shown below.
Horn No.2007 featured in this blog post is 60mm overall depth while Horn No.1900 (pictured above) is 50mm deep. This results in a lower Fc for the 2007 horn. The response graph below is a comparison between the two horns. (2007 is green, 1900 in red) As you can see from the result below, the region of bandwidth boosted by the horn has shifted to a lower frequency. Typically a horn can only provide gain across 2.5 octaves, which we clearly see in the result below, with the green trace shifted slightly lower into the frequency range.
Besides the lower cutoff frequency, the latest horn also smooths out the small dip at 10kHz.
When dealing with low crossover points near the tweeter's FS, it can sometimes be necessary to implement a conjugate filter. This type of filter notches out the impedance spike at the FS. The schematic below shows the conjugate filter that I settled on after extensive impedance sweep measurements of various component values. In other words, the conjugate filter is very specific to this horn/driver combination.
The conjugate filter stabilizes the impedance curve to allow predictability for the next stage of design, which is to implement the actual high pass filter elements. The conjugate filter does not effect the frequency response at this stage. However it's impact will be noticed once we start adding the high pass filter elements.
But before we get to the high pass filter design stage, I want to show the affect of the conjugate filter on the impedance sweep. The response below is the raw impedance sweep and phase response without any filtering elements.
The overlay is shown below with the conjugate filter in place. The raw impedance is shown in green with the blue trace showing the affect of the conjugate filter. The phase response is shown in the top half of the chart with red showing the affect of the conjugate. The phase response is flattened to some extent with the conjugate in place.
High Pass Filter Elements
With the conjugate filter in place, we can now look at adding the high pass components.
With the high pass elements in place we get the following frequency response.
Out of interest's sake, we can remove the conjugate filter to see the response without it. (shown in black below)
Without the conjugate filter it would be difficult to achieve the target response for the desired 1.5kHz LR4 crossover point.
Below is the horizontal off-axis colored polar map for horn No.2007.
No conjugate filter With conjugate filter
Below I test harmonic distortion using an 85dB SPL test signal at 1m. The mic was placed in the near field. We see -89dB for H2 and -96dB H3.
Increasing the test SPL to 95dB we see H2 at -89dB and H3 at -89dB. This is the noise floor for the Focusrite Scarllet Solo Mic Preamp/Audio Interface used in the test. Information on the distortion performance for the Focusrite can be found here. A search for a lower noise mic preamp is under way.
Looking at intermodulation distortion using a 12 band/octave multitone test signal at 85dB test SPL at 1m is shown below. We see distortion averaging -65dB across the spectrum, improving in some areas such the 3kHz region which is -71dB.
I listened to Stacey Kent's Dream In Concert which is a live recording with great drum brushes. The tweeter had no problem extracting the full detail from this recording. When the audience claps I could hear every clap distinctly and I could also hear the venue ambient cues. Everything was sounding excellent, on par with the some of the best tweeters available.
The Peerless OC25SC65-04 on Horn No.2007 achieves the sound quality objectives mentioned earlier. Directivity is wide and well behaved, with a very linear response. Distortion is also near the limit of my measurement system’s electronics.