Lowther Fidelio in my living room

(Randolf Werner, 02/2021)

Lowther is a traditional and legendary loudspeaker manufacturer in Great Britain with a history back to 1934. It is especially famous for its full range drive units fitted in folded backloaded bass horn cabinets.  I came in touch with it almost 40 years ago and still use these loudspeakers. I like to share my experience with one of the more recent products the Lowther Fidelio MKII using an EX3 drive unit. Over the decades I used many different cabinets Acousta 109, Acousta 115, Classic 200, Classic 400 and drive units PM6C, PM2C and PM6A with minor problems and excellent results. Unfortunately, using the EX3 in the Fidelio in my new living room was initially rather disappointing. Anyway, I finally found a surprisingly simple (but for Lowther rather unusual) solution for the problems. My journey and the tools I used may help you to improve your own Lowther or not Lowther-based system at home as well.

Equipment

The equipment I used consists of the following additional components:

 

  • Yamaha R-N803D Receiver. It's a solid build Stereo Receiver that combines traditional amplifier technique with the digital world, offering digital inputs, streaming, digital sound processing and YPAO room correction. In contrast to a typical AV Receiver it is not a 7.1 channels surround home theater system, instead it is fully focused on excellent 2 channel Stereo.
  • Quad Z-1 excellent bookshelf speaker. I purchased them for comparison and because I initially run into serious problems with the Fidelio in my listening room.
  • Hifime Sabre 9018 USB-DAC combined with USB isolator to transfer digital compressed (MP3) music from JRiver Media Center on a Windows laptop to the Yamaha via optical toslink. JRiver offers excellent sound processing and the ability to apply a wide set of free adjustable filters (in particular parametric equalizer with almost arbitrary parameters, but also e.g. convolution). 
  • MiniDSP UMIK-1 measurement microphone used to perform measurements with REW measurement software.

Lowther

Lowther offers a wide range of full range 22.5 cm drive units. The diaphragm is very lightweight twin paper and similar if not the same for all types. All have very strong magnets, high magnetic flux and narrow air gap width. This results in very high sensivity requiring only small (but still reasonably good) amplifiers. There are 4 product lines C, A, DX and EX using 3 different magnet types Feroba (C), Alcomax (A) and Neodymium (DX and EX). In each line different magnet sizes/strength are available resulting in even higher flux density for a higher price. The EX is basically a DX but adds an additional acoustic chamber to the rear of the unit. All units use a phase plug to better reproduce higher frequencies. You can choose between a traditional elliptic shaped one and the newer “Phase Equalizer” design. Additional options are Silver vs. Aluminium voice coils and 8 vs. 15 Ohm coil impedance. Over the decades, the diaphragm shape has changed, today the edge of the inner cone is rolled back. But I assume that it is still possible to order the older shape if you want.

None of the above units can reproduce lower frequencies with the required SPL by its own. They must be fitted in a folded bass horn cabinet to bring lower frequencies to the required level. Lowther offers many different cabinet layouts wonderfully designed to host one or two drive units. All of them use the full range feature of the drive units without any crossover network. The size of the cabinets varies from middle to huge. Many of them require a corner or at least a wall behind the speaker. The models with a second drive unit on the rear, if correctly placed in a “compatible” room, can offer an even more realistic and deeper stereo image. However, it can become tricky to find a good position and it can be contra productive in some rooms. In addition, there are some very special and huge cabinets like the Audiovector, TP1 and Opus One. I cannot say very much about these since I never had the opportunity to listen to one of it. Lowther offers detailed plans of the cabinets together with the units allowing to build them by your own. If you decide to do so this saves a lot of money. You are totally free in the finish and finally it can be a lot of fun to build it by your own. Depending on the model moderate to high skills might be required to do so. You can purchase the parts directly from Lowther or selected local dealers like Audio Technik GmbH here in Germany. Don’t hesitate to contact Lowther or your local dealer, all of them are very friendly and helpful. Finally, Lowther and your local dealers offer a renovation option for old drive units for a very attractive price by reusing the magnet of your old unit. It is entirely possible to renovate a unit several times over many decades if required. Lowther also offers interesting exchange options with a significant discount by returning your existing drive units.

Loudspeaker comparison

Comparing loudspeakers can be difficult. There are little objective measurement criteria that finally match well with the personal human perception. Sensitivity and maximum SPL may be two of them, but only tell you how much power you need to reach a certain volume level and how loud the speakers are able to play at most. All Lowther speakers have excellent sensitivity and can (if you want) play very loud. You probably will never need more than 1-10 watts. This can reduce costs on the amplifier side and enables you to use tube amplifiers if you like. The Quad Z-1 is a lot less sensitive, the difference to the Fidelio is about 12dB, which means the Lowther plays about 4 times louder with the same amplification. Anyway, it is still good enough to work well with affordable amplifiers in most listening situations. Frequency response is often measured and many are looking for a flat response from 20-20000Hz. Anyway, it is very difficult to tell how the speaker really sounds by looking just at frequency responses for many reasons:

  • Frequency is just one component of a signal, other important ones are phase, timing, dynamic and distortion. The human ear analyzes all of these in different ways.
  • Human ears work very different from a microphone. Our ears are a lot more sensitive in the midrange of the frequencies. They tend to tolerate even deep narrow bandwidth drops but notice a peak of the same size much more.
  • You never just listen to a loudspeaker you always listen to a loudspeaker in your room. This is even more true for the bass section. The bass reproduction of an excellent speaker can easily be ruined by resonances in your room caused by the dimensions of the room, materials in the room, listening or loudspeaker position in the room. When you walk around in your room bass reproduction varies a lot. A speaker with a flat response at e.g. 45Hz may sound completely wrong placed in a room with a 10dB peak at the same frequency.
  • Frequency response curves are smoothed in order to see something meaningful. Constant 1/12 octave smoothing is often used (which btw. matches nicely with 12 keys per octave on a piano), but other constant or variable (psychoacoustic) smoothing is also widely applied. So watch out when you compare curves from different sources.
  • When measuring a loudspeaker not in an expensive anechoic chamber sometimes a (short) time window is set to only include the direct response of the loudspeaker without any room reflection (which arrive later at the microphone). This has 2 effects, you cannot collect meaningful data for low frequencies and the frequency response for lower/midrange frequencies gets smoothed (in addition to other smoothing). So, watch out when looking at frequency responses without low frequencies. Anyway, this technique is very (cost) effective and still accurate for measuring high frequency response of tweeters.
  • The goal for an optimal frequency response curve is not necessary a totally flat one. Often a slight decay to higher frequencies is preferred.
  • Different measurement techniques like log swept sine chirp  vs. Maximum Length Sequence (MLS) can be used.

Typically, the frequency response is measured on axis at tweeter Hight, additionally often slightly horizontal of axis (~10-30 degrees) responses are measured to give a hint about the expected sweet spot and optimal tilt angle for positioning the loudspeakers. Anyway Standard Method of Measurement for In-Home Loudspeakers ANSI/CEA-2034 defines a more sophisticated measurement standard for loudspeakers which requires to measure 70 different frequency responses at 35 different horizontal and 35 vertical angles each 10 degrees apart from the other in an anechoic chamber. It than defines how to calculate “Listening Window”, “Early Reflections”, “Vertical Reflections”, “Horizontal Reflections” and “Sound Power” frequency responses out of these raw measurements as well as “Sound Power Directivity Index” and “Early Reflections Directivity Index”.  The overall goal is to be able to predict a likely to expect frequency response in an average listening room. The assumption is that a loudspeaker with little and smooth directivity will produce better results in an average room. Depending on the particular room and loudspeaker type this may be more or less actually the case or not. Anyway, which conclusion to draw from this so called Spinorama measurements is not generally accepted, but some reviewer and a few manufacturers provide that data (see https://www.spinorama.org/)

 

The Klippel near field scanner takes the idea of CEA-2034 even further. It performs a fully automated near field measurement on a high resolution (>500 measurements) 3-D sphere around the loudspeaker without the need of an anechoic chamber. From that data far field, CEA-2034 compliant and many other data (e.g. horizontal and vertical directivity plots) can be calculated matching very well with more expensive anechoic chamber measurements. However still the near files scanner costs about 100000 Euro. Some reviewers (https://www.erinsaudiocorner.com/loudspeakers/, https://www.audiosciencereview.com/forum/index.php?pages/Reviews/ ) are using it.

 

Some tips for a good comparison:

  • Before comparing try to position both speaker pairs as good as possible.
  • Use an A/B setup that allows you to directly switch between the 2 speakers without interruption.
  • Adjust the volume level when switching so that both speaker pairs will play equally loud, otherwise your ears will prefer the louder one most of the time.
  • Use both rapid switching between the 2 speakers as well as listing to one pair for a longer time before switching. Your ears will adapt to the sound over time, sometimes something sounds initially good but after a while this will change and you will notice this when switching back.
  • Use a wide range of music for testing. You will probably prefer your favorite music. But sometimes it can be good to use others as well, in particular if they are recorded well and contain challenging instruments like violins, female voices or sometimes even an electric guitar. For some music types tonal balance is most important for others it is dynamic.
  • Listen to a single speaker pair for a longer time, even after listening for hours to well recorded music the sound should neither become boring nor annoying.

 

The Problem

I have been using Lowthers for decades and has always been very satisfied with the result. In particular, I used an Acousta 115 with PM6a drive units for almost 40 years. For the first drive unit renovation many years ago I sticked to the same spec and the result was as expected very satisfying. The second time (in 2010) I decided to upgrade to the latest EX3, 8 Ohm, Aluminium voice coils since Lowther offered a replacement option. I was able to do a short A/B mono comparison of the EX3 with the PM6a in the Acousta 115 and noticed a very big difference, especially in midrange. I very much preferred the old PM6a. The EX3 sounded hard and aggressive to me in the midrange, especially most female voices sounded wrong to me. The EX3 and PM6a were different in many details:

  • Runed in for many years vs. brand new
  • Alcomax vs. Neodymium magnet type
  • 1,8 Tesla vs. 2.1 Tesla magnetic flux
  • Traditional diaphragm shape vs. new one with rolled back edge of the inner cone
  • Open rear side vs. acoustic chamber
  • Traditional phase plug vs. new Phase Equalizer plug

Before sending back the old PM6a to Lowther, I was able to rule out that the phase plug shape causes the difference, simply by switching it between the two drive unit types. Later on, I removed the acoustic chamber from the rear of the EX3:

I think this slightly improved the situation, but the fundamental difference was still there. Even after a long run in phase I was not really satisfied. Meanwhile I had moved to a new apartment with a much bigger living room. This hadn’t really improved the sound but enabled me to place bigger corner horn speakers. I choose to go for the Fidelio MKII and built it on my own from the cabinet plans Lowther kindly had provided to me: 

The bass section was very different, more, deeper and better bass on the Fidelio. However, the midrange problem remained, maybe a little less obvious since the extra bass could better compete with the midrange. Unfortunately, I noticed a new problem, there were big peaks in the bass section at certain frequencies, probably caused by room resonances triggered by the deeper bass of the Fidelio. So, in total the sound was even more out of balance than before.

Analysis

I decided to analyze the problem in more detail. Therefore, I used REW freeware software, the MiniDSP UMIK1 microphone and my Hifime DAC to measure the frequency response at listening position. One obvious finding was a huge about 10dB peak at about 45Hz and some more moderate peaks at higher bass frequencies. To my surprise at that point the Yamaha YPAO calibration was not really able to fix the situation, it only slightly improved it. Another obvious finding were peaks in the midrange at ~2KHz, YPAO calibration did not change that at all.

It was time to find out if it’s the room or the loudspeaker causing so much trouble.  I decided to purchase a Quad Z-1 bookshelf size speaker pair. I expected such a small speaker should never run into the problem of two much bass. I was wrong, too much bass and exactly the same high peak at 45Hz. Changing the speaker position (away from corners/walls) had little effect on that. To my surprise the YPAO correction worked far better this time, resulting in an excellent and balanced frequency response. The measurement shows the frequency response of the Quad Z-1 without YPAO (blue), with YPAO (green), YPAO + 40Hz cutoff (purple), YPAO + 60Hz cutoff (red):

Together with a good speaker position (the ribbon tweeter of the Z-1 has an excellent horizontal radiation but needs to be positioned at the height of your ears) the Z-1 sounds great, probably better than most would expect from a speaker of that size. My conclusion is the room has serious bass resonance problems which YPAO can fix pretty well. The midrange problem is due to the Lowther or perhaps combination of Lowther and my room.

Room simulation

Let’s have a closer look at my room. It is an L shaped living/dining room combination of 35m2 and 2,44m height. It has an open staircase to the second floor, separated by a small wall and an open passage to the kitchen. Mostly there is hard material in it with many huge floor-to-ceiling windows, leather sofas, wooden tables, parquet, only little thin curtains, plastered walls mostly without wallpaper.

In total there is little damping in the room. REW offers simplified room simulation for standard rectangular shaped rooms. Let’s try it by just using the living room part section (4,16m x 3,48m):

Wow a huge peak at 42Hz, a deep drop at 70Hz and a peak at 100Hz, that matches surprising well with the Quad Z-1 measurement without YPAO! What would happen if I would remove the small wall behind the sofa. Let’s approximate that by extending the room length from 4,16m to 6,22m and keeping speaker and listener position as is:

What a big difference the peak a 42Hz would change into a deep drop. I haven’t tried this out because it would be an expensive (new staircase required) and destructive experiment. Another option would be using bass traps to improve the room acoustics, but this requires extra care to work well for such low frequencies. I may try this in the future, e.g. by placing a Helmholtz resonator or broadband bass trap in the corner between the 2 sofas. This may improve frequency response as well as poor/long reverberation time at 45Hz. The position between the 2 sofas is the only available option to place a bass trap since the Fidelio is a corner bass horn.

So, when you hear that some speaker has “no bass” or sounds “boomy” this most likely means this particular speaker in a specific room at a specific position at a specific listening position sounds that way. When looking at manufacture or test frequency responses of speakers do not complain too much about smaller drops and not very deep bass, it can easily turn out that this becomes an advantage in your listening room. The Acousta 115 is an example for this, due to its rather limited bass range and strength it had little problems with bass peaks in my room.

 

Fixing the Fidelio im my room

REW allows you to design filters in order to fix problems in the frequency response. It can do so automatically, which didn’t work well for me with the Fidelio. I tried to apply multiple narrow bandwidth filters manually which didn’t work well either. Finally, I decided to use a single notch filter with broad bandwidth to only fix the most obvious midrange problem at 1900Hz. I ended up with parameter 1900Hz, -5db, Q=0,4. This looked good from a frequency response perspective in REW but much more important sounded great.

Using REW you can design filters based on your measurement and simulate the result. To actually hear the result, you need some player allowing you to apply the filter parameters to your music. JRiver is an excellent example of such a player. It offers an excellent parametric equalizer allowing you to define almost arbitrary filters, so you can simply enter 1900Hz, -5db, Q=0,4 and start listening when playing music via JRiver. With one click you can enable/disable individual filters or the entire equalizer to compare the effect on the fly. You can also adjust the filter parameters on the fly and play at least with the damping factor perhaps in the range of -1 to -6db. This entirely fixed my midrange problem, everything sounded perfectly right without almost any negative effect. Of course from a theoretical point of view the filter worsens the phase response. The disadvantage is you now need JRiver for play back all the time. You can configure JRiver that all sounds of your laptop go through the JRiver sound engine in order to apply the filtering to other music sources as well. You can extend it even further by installing an Airplay receiver like Shairport4w on your laptop. This allows you to stream music from your iOS device via Airplay to the laptop through the JRiver filter to your amplifier.  Although this works great, comes (almost) for free and is very flexible it has some disadvantages:

  • There are probably other sources you like to play through your filter, e.g. a CD player, TV or turn table connected to your amplifier.
  • In addition, you have to boot/start the laptop whenever you want to use the filter.
  • You cannot use the digital streaming capabilities of your amplifier.
  • You cannot apply YPAO correction of the amplifier on top of your filter (e.g. to fix bass problems).

Fortunately, it is possible to build an analog filter with almost the exact same properties as our digital notch filter. In fact this type of notch filter is widely used with other full range loudspeaker but very rarely with Lowther. There are many online tools to design this simple filter type, a good one is e.g. provided by  HIFI AUDIO DESIGN (http://www.mh-audio.nl/Calculators/parallelnotchfilter.html). You need to enter peak frequency (1900Hz), damping (5,0dB) and loudspeaker impedance. A single Lowther unit is either 8 or 15 Ohm. However, impedance depends on the frequency. I measured the Fidelio (red) and Quad (green) impedance using REW (http://roomeqwizard.com/help/help_en-GB/html/impedancemeasurement.html):

It just requires a soundcard with headphone output, line-in input and a 100 Ohm resistor to perform an accurate measurement. The huge impedance peaks of the Fidelio at low frequencies are due to the horn cabinet. A good approximation of the 8 Ohm EX3 in my Fidelio at that frequency range around 2000Hz is about 10 Ohm. Once you entered the parameters you need to extend the bandwidth of the filter to match well with Q=0.4. So I finally end up with 7,8 Ohm, 6,22uF, 1,13mH. Matching this against available values I started with 8,2 Ohm, 6,8uF, 1,2mH. I used Audaphon MKP capacitor, Mundorf air coil with 0,56 Ohm and a 20 Watt ceramic resistor. The price to build a pair of these parallel notch filters is about 60,- Euro and looks like this:

I measured the effect of both the digital (blue) and analog (green) filter, the red curve in the original frequency response:

The effect on the frequency response is almost identical to the digital filter, slightly less damping on high frequencies due to the increasing impedance of the speaker towards higher frequencies. You can also easily do an A/B comparison of the filter effect by short cutting the circuit using a standard cable as “remote control” from your listening position. 

After listening for many days to this setup I noticed that there is a problem with this filter layout. The peak around 8 kHz is not damped by it and becomes a lot more noticeable because of the up to 5db damped midrange. A more balanced filter layout with just about 3db damping of the entire midrange up to about 10 kHz gives better results and avoids harsh sizzling “s” in vocals. This can be achieved by reducing the resistor and capacitor. My final layout parameters are 4,7 Ohm, 1,8 uF and 1,2 mH. With this parameter the filter becomes similar to a -3db high shelf filter but avoids the damping on very high frequencies. In fact just setting the simple treble control of the amplifier to -3db could already improve the situation to some degree, but the fine-tuned notch filter gives better results. The frequency response (all with YPAO correction) without filter (green), -3db notch filter (red) and -3db treble control of the Yamaha Receiver (purple, ~ 3db high shelve at 1 kHz) looks like this:

Now we can also apply the YPAO correction on top of the filter. This time the correction worked very well almost completely fixing the bass problems. You can further improve it by adjusting some YPAO settings after the measurement:

  • Enable the subwoofer option.
  • Switch the speaker type from full range to bass limited.
  • Choose 40 Hz cutoff frequency;  this will bring the still slight peak at 45 Hz a little bit more down, you may also want to try 60 Hz.
  • Disable YPAO Volume in case do not want to use the digital loudness function.

I am not sure why YPAO did not worked well with the unmodified Fidelio. Perhaps it had problems to identify the target SPL correctly due to the too high level at the midrange frequencies and therefore applied to less damping of the bass resonance frequencies.

 

The final frequency response measurement (1/12 octave smoothing) of Fidelio (red) and Z-1 (green) now looks very promising, typical for the speaker types and well balanced. The graph shows the averaged response of left and right loudspeaker. The upper 2 graphs show the response with YPAO and for the Fidelio with additional parallel notch filter, both with and without 40 Hz cutoff. The lower graphs show the original response without any filtering.

Using the 40 Hz cutoff is a compromise, it’s a little too less damping at 45Hz and too much damping below 40 Hz. Furthermore neither YPAO nor the 40 Hz cutoff can improve the long reverberation time of the room at this frequency range of about 18dB in 300ms / 7,6dB in 160 ms which we can see in waterfall and decay diagrams:

A better approach might be to use a bass trap instead, which can reduce SPL and reverberation time. A narrow bandwidth tuned bass trap like a Helmholtz resonator placed in the corner between the 2 sofas may do the job. From the frequency perspective -5dB, 45Hz, Q=3 would be required which may match well with a 216L Helmholtz resonator with 2,2cm x ø17,2cm tunnel (=> Q=2,7, K=0,21), but also broadband absorbers using Rockwood or Basotect may work. A first experiment with just filling the corner (~70x70x70 cm) with duvets already shows some improvement in the measurement.

 

Helmholtz Resonator

I finally decided to build a Helmholtz resonator to eliminate the 45Hz room resonance problem and get rid of the 40/60 Hz cutoff workaround. I used 22mm MDF to build a box of 69,6 x 69,6 x 44,6 cm inner dimension (= 216 litre) and placed it in the edge between the 2 sofas. I started drilling 68mm holes in the cover panel and measured the effect on the frequency response. The measured resonance frequencies matched very well with the ones calculated from the formulas. Using 6 68mm holes I was able to match the 45Hz room resonance perfectly. The graph (no smoothing) shows the result without (Helmholtz resonator box without any cover panel, green line) and 3-6 68 mm holes without any damping in the box:

As we can see the 44,7 Hz peak in the green line (without Helmholtz resonator) matches perfectly with the drop in the red line (Helmholtz resonator with 6 68mm holes). 

The resonator is very sensitive to damping on the holes. By using ~3cm polyester wool damping material on the 6 holes I was able to smooth the curve perfectly. 

I also build a cover panel with a single 17cm hole for testing, but this resulted in a too low resonance frequency of ~40Hz. For such big diameters and short tunnel length (22mm) the Helmholtz formulas do not match as good as they did for smaller diameters. If the box is accurately built it is possible to do the initial measurement without having to glue the cover panel, a few weights to weigh down the lid were sufficient for testing. This makes it easier to test different diameters and damping.

 

The final frequency response measurement of the Fidelio (1/12 octave smoothing, no 40/60 Hz cut off) with and without Helmholtz resonator was:

 

REW can also generate a spectrogram graph. The spectrogram is like a waterfall viewed from above, with the level indicated by colour. It makes the 45 Hz room resonance very obvious (left Fidelio 1/12 octave smoothing without Helmholtz resonator):

 

And shows how effective the Helmholtz resonator fixed the problem (left Fidelio 1/12 octave smoothing with Helmholtz resonator):

 

To summarize the procedure to optimize the Fidelio in my living room:

  1. Apply a parallel notch filter (4,7 Ohm, 1,8 uF, 1,2 mH) to the loudspeaker.
  2. Carefully position the speakers. In particular, find a proper tilt angle. On axis Lowthers typically sound too bright and aggressive. For me the best compromise was at about 10 degree out of axis.
  3. Apply the YPAO measurement. Take extra care to position the Yamaha microphone exactly at listing position at the same distance to both speakers. YPAO will show you the measured distance after the measurement. It should be almost the same for both speakers (+/- 5cm). Also, the level for both speakers should be the same (+/- 0.5dB). You can manually correct small differences, in case of big difference check the setup or try to apply another YPAO measurement. Verify the result by enabling and disabling YPAO, the stereo image and balance should not change.
  4. Add a 216 litre Helmholtz resonator with 6 68 mm holes, 22mm tunnel length and 3 cm polyester wool damping on the holes to the room corner.

Comparison

Finally, everything was prepared to compare the Fidelio and Z-1:

The Yamaha Receiver stores independent YPAO settings for speaker pair A and B. For each configuration you can manually adjust the volume of left and right speaker within +/- 10 dB range. An equal volume level (additionally verified via SPL measurement using REW) could be achieved by setting the Fidelio to -10dB and the Z-1 to +1dB. Now you can seamlessly switch between the speakers from you listening position using the Yamaha MusicCast App on your iPhone while listening to the music.

I will not bother you too much to explain with many words how it sounds. First of all, both are great speakers and sound very well. Tonal balance is very good, almost any reasonable good recording will just sound right. One obvious difference is the sweet spot. The Fidelio has a narrow sweet spot, you need to sit in the middle of the speakers, the sound will significantly change when you move to the left or right, however it will never become really bad. The ribbon tweeter of the Z-1 has an excellent horizontal radiation, resulting in a wide sweet spot which changes little when you move from left or right. However, on the vertical axis this is different (typical for a ribbon tweeter). If you stand up the Z-1 can easily sound too dull. The Z-1 can have problems to fill the room between the speakers, they sound better a little bit closer together. In the picture above they e.g. sound worse when positioned in front or on top of the Fidelio. After Listening to both for many hours I prefer the Fidelio in almost any discipline. I love the solid bass, open and very dynamic sound with excellent stereo image. Of course the Fidelio can also play significantly louder, but this of little interest for me. There is one little compromise. Compared to really excellent tweeters (like the ribbon tweeter of the Z-1) they sometimes lack a little bit of details at very high frequencies and sometimes can sound a little bit too aggressive. On the other hand, if full-sized big to huge speakers do not fit in your listening room the Quad Z-1 is a good loudspeaker to consider for small to medium sized rooms and moderate listening volumes. They are well designed and manufactured with attention for details. In addition, they look great with the curved shape cabinet and black piano lacquer. Only the acoustic quality of the cabinet is a little bit questionable to me. The cabinet is noticeable vibrating.    

The Fidelio cabinet is clearly superior to the Acousta. In case you can place them in a corner or at least in front of a wall this is probably the best reasonable sized cabinet for a single drive unit Lowther. Sometimes people complain about too little bass of Lowthers, I hardly can image that this could ever be the case with the Fidelio (unless it is due to the room acoustics). In case you look for something a little bit smaller and you can place the speaker in corners, the Bicor 200 is probably a good alternative still superior to the Acousta.  

It is hard to recommend a specific drive unit. As you have read, I had my problem with the EX3. But it is hard to tell which of the many parameters really caused my midrange problems. I would probably recommend smaller magnet types, especially the PM6a eventually with Silver voice coil, PM6c, DX2 and DX3 may be an alternative. I don’t see any benefit of the EX line in the Fidelio nor the Acousta. A detailed A/B comparison of the drive units would be interesting, but this is difficult to perform using corner bass horns. A detailed mono A/B comparison in the Acousta would be easier to achieve but is still quite expensive. Lowther offers detailed technical specifications and frequency responses for all drive unit types on their website. For better comparison I digitized the published frequency responses of 8 Ohm PM6c, PM6A, DX2, DX3, EX2 and EX3 with aluminium and silver voice coil and arranged them in a single graph:

I don’t know the exact measurement conditions Lowther used, but I guess they used the same conditions for all drive units, most likely the drive unit alone in an open baffle on axis with some smoothing applied to the curves. So these conditions are very different from a unit fitted in a corner bass horn at some tilt angle in a living room. Anyhow it may help us a little bit to understand the differences you could expect. All units have a typical similar “Lowther shape” frequency response, which is a very low SPL on frequencies below 100Hz (which is why a bass horn cabinet is required),  a flat response up to 1 kHz, a rising frequency response between 2 and 10 kHz and some roll of beyond 10 kHz. Especially above 1 kHz the differences are significantly, and we can choose between a wide range of different frequency responses. Let’s have a closer look at some differences. The effect of a stronger magnet can be seen by comparing a DX2 and DX3 with aluminium voice coil:

The DX3 has an extended high frequency range and higher SPL beyond 2 kHz. The influence of the silver voice coil on the frequency response can be seen by comparing the DX3 with both different voice coil types:

The silver voice coil gives a smoother but a little bit more limited frequency response. Lets finally compare the DX3 aluminium voice coil (which is probably identical to my EX3 without rear chamber) to the PM6a with aluminium and silver voice coil:

Perhaps this at least partially explains the big differences I noticed when switching to the EX3 (I don’t know if my old PM6a had an aluminium or silver voice coil).

My conclusion so far would be that bigger magnets and aluminium voice coils may results in a more detailed but eventually less balanced sound, which you may need to mitigate using a different tilt angle or even filtering like I did.

 

In Germany AER and Voxativ are 2 manufacturers building full-range drive units and cabinets probably inspired a lot by the Lowther design. AER is located in Stuttgart and Voxativ is located Berlin. At a first glance both look similar to Lowther drive units, in fact the AER BD drive units look very similar to Lowther drive units about 30 year ago (except that AER does not use a phase plug). However, both are independent unique designs like Voxativ stated on their website: “We build our own full-range drivers based on our own proprietary design. It's neither Lowther, nor Fostex.”  Most of these drive units are rather expensive. Currently Voxativ offers drive units from $900 up to $29500 and the price for the AER BD drive units is currently from 2320 Euro to 30000 Euro. In Poland Cube Audio offers drive units from 790 Euro to 2990 Euro. Unfortunately, I never had the opportunity to listen to any of these so far. The price range for Lowther drive units currently is about 450 Euro up to 1560 Euro. Fostex is another manufacturer of full-range drive units located in Japan with a long history back to 1973. In China Tang Band is offering a wide range of small full-range drive units since 1996, Lii Audio since 2016 and Markaudio since 2010. Commonsense Audio in USA is offering Lowther and Fostex drive units as well as their own Audio Nirvana drive units. DAYTON AUDIO in the USA offers several affordable full-range drive units, especially the Point Source Series. In France Supravox offers full-range drive units from 189 Euro to 1519 Euro. SEAS in Norway also has some full-range drive units in it's product line. Last but not least there is Esoteric Audio Devices (EAD) in Sweden offering 9-22 cm full-range drive units at a reasonable price level (100 – 230 Euro).

 

The Yamaha YPAO now worked rather well for my bass problems. Unfortunately, the R-N803D Yamaha offers little possibilities to manually adjust it. I am pretty much sure the DSP could easily execute a simple 1900Hz, -5db, Q=0,4 notch filter or add some more damping at 45Hz to avoid a 40Hz/60Hz cutoff. I would love to see some kind of expert mode to manually tweak more parameters. Other vendors may eventually offer more degrees of freedom here.

 

Amplifier hint

Most amplifiers will work well with Lowther loudspeakers. Like other high efficiency loudspeakers they require only a small amount of power. Also, the impedance curve is uncritical and easy to drive by almost any amplifier. Anyway, as with any high efficiency loudspeakers there is one issue to consider. The residual noise floor of the amplifier should be low. This is especially important when listening at low volume or at close distance to the loudspeakers. Otherwise, a permanent hiss can become audible when listening at low volume. Yamaha publishes the residual noise of most amplifiers. The value also depends on the operation mode like “Pure Direct”, “CD Direct” or “Main Direct”, the shorter the signal pass is the lower the residual noise will be. Here are some values of Yamaha amplifiers I used:

  • A-960: 220uV (mode?)
  • A-590: 35uV (CD Direct), 90uV (Pure Direct)
  • R-N803D: 70uV (mode?)
  • R-N402D: 70uV

The A-590 and R-N803D have an almost inaudible residual noise with the Lowthers. The very old A-960 from 1980 and the rather new budget R-N402D from 2021 have a clearly audible residual noise with the Lowthers. The difference of the R-N803D and R-N402D was surprising, since both have the same rating in the specs. But after looking into the R-N402D technical details I really doubt this value. It uses a BD3491FS as volume control, this device is rated with 5uV residual noise, which than will be amplified by the power amplifier section. The R-N803D uses a BD34703KS2 as volume control which is rated with just 1uV residual noise. This I guess explains the clear audible difference with respect to residual noise between both amplifiers when connected to the Lowther Fidelio. Some amplifiers like the R-N402D use an additional mute switch, which gets activated when the volume is turned down completely or a digital 0 source signal is amplified. This can fake a good residual noise level, which gets a lot worse once you only slightly turn up the volume or the music starts.

 

Building the cabinet by your own

Building the Fidelio cabinet by your own is not too difficult. Your local hardware store or carpentry can provide the 18 and 22 mm MDF boards and crop them for you (at least for the outer rectangular shape). For the inner parts you will need to crop it by your own due to the many nonrectangular angles. You can do so manually using a file or you can use a Mitre saw like the Metabo KGS 216 M that I used. It can cut angles from 0-47 degrees up to 305 mm length. This is enough for the inner width of 244mm of the Fidelio and most of the angles. For a few angles you still need to file. For the drive unit and cable connector holes you may want to use a router like the Einhell TE-RO 1255 E. For the finish there are many options, I used Easiwood veneer and lots of Pattex classic contact glue. The Lowther cabinet plans are very accurate and detailed. They use millings almost everywhere to further improve the strength of the construction. This is difficult to achieve using budget equipment; however, it is absolutely possible to butt glue the panels without compromising the construction significantly. Just ensure the internal dimensions of the cabinet do not change. The front panel is designed for 25mm MDF, since this wasn’t available, I modified it to 22mm MDF. The plans contain many mathematically correct angles like 56,65°. Try to work as accurate as possible but 56° or 57° degree works as well.

 

Lowther cabinets

Lowther has designed many different cabinets in its long history, from simple to very complex and from medium size to huge. The dimensions of the cabinets below are all given in width x hight x depth. Minor derivations due to wood thickness (18 vs.19 mm), grills, stands and minor plan modifications over the decades are possible. Cabinet plans can be purchased from Lowther and are offered for free when buying new drive units.

 

The Acousta 90 (342x632x238 mm) is a very simple bassreflex design for a single drive unit. The Accolade 1, 2 and 4 also have rather simple cabinets but with some internal labyrinth utilizing 2 or 4 drive units of 8 and 5 inch size on the front baffle. Although I never heard any of these, I would not recommend any of these cabinets since there exist better reasonable sized horn cabinet designs for the same drive units.

 

The Acousta 115/116 (419x813x381 or 468x830x373 mm) is a typical backloaded horn design for a single drive unit. There are 2 versions of the Acousta 115 cabinet, one with a perpendicular baffle and one with slightly upwards angled baffle and wider cabinet. The Acousta 109 (356x712x360 mm) is a smaller version of the Acousta 115. The Acousta 124 (472x834x444 mm) is also similar to the 115 but fits 2 drive units on the baffle. The Acousta 115 is probably the easiest to handle Lowther cabinet, it will work well in most rooms. However, the bass reproduction is rather reserved and not very deep. In case positioning in room corners or at least close to a back wall is possible the Bicor 200 and in particular the Fidelio can offer superior bass reproduction.

 

The Classic 200 is a small corner horn for a single drive unit with the horn mouth and an additional port at the backside. The Bicor 200 (250x718x320 mm) is the successor of the Classic 200 with only minor modifications. The Fidelio (280x1000x440 mm) is a bigger cabinet but follows the same design principals. Audio Technik GmbH in Germany offers a Bicor 250 in between the size of the Bicor 200 and the Fidelio. Due to the corner horn design they need to be placed close to a room corner or at least close to a back wall, of course with some distance.

 

The Classic 400 (250x808x368mm) is a small corner horn for 2 drive units, one unit at the front baffle as usual. At the rear baffle the second unit pointing diagonally upwards and the horn mouth are located. Both units are driving the backloaded horn. The Delphic 500 (362x808x488 mm) is a bigger cabinet following the same principal. The Bicor 2000 (254x814x368 mm) adds an additional port at the backside. The Academy (280x1000x440 mm) is the bigger version of the Bicor 2000. All cabinets utilize reflections from the rear and ceiling. The inclination angle of the rear driver differs significantly between the models, in the Classic 400 and Bicor 2000 it is 20 / 25 degrees (so pointing more to the wall behind the speaker), in the Delphic 500 it is 50 degrees (similar in the Academy) (pointing more to the ceiling). Positioning of these speakers can be very difficult (in some rooms even impossible) in order to find a good balance between direct and reflected sound. Due to the corner horn and reflecting design they need to be placed close to a room corner or at least close to a back wall, typically with some more distance (for the classic 400 it can be about 55cm distance from the back wall).

 

The TP1 London (800x1020x600 mm) is a huge double loaded horn utilizing a single drive unit for corner placement. The mouth of the backloaded bass horn it directed to the front. In addition, the same drive unit is driving a forward radiating mid-range horn.

 

The Audiovector Symphonic (650x860x460 mm) is a huge horn speaker utilizing 2 drive units. One unit is located at the front baffle driving the backloaded horn with the horn mouth at the front. The second drive unit is located at the top driving the backloaded horn and the upwards pointing midrange horn. The Audiovector utilizes reflections from the ceiling or a special reflector. It is a very complex to build cabinet (https://www.youtube.com/watch?v=9KSQcK5bw_A)

 

In principal all cabinets can be used with any of the 8 inch Lowther drive units, however not all cabinets provide sufficient space for drive units with bigger (deeper) magnets.

 

In 2018 Martin Thornton took over Lowther. Since than the Almira (TQWT with mouth at the bottom utilizing a DX-3 and a supertweeter, 300x1200x330 mm), Edilia (TQWT with mouth at the bottom utilizing PM7A, a supertweeter and a DX-2 bass driver, 300x1200x330 mm), Acousta 117 (modified Acousta 115), 205F (probably a TQWT with DX55 driver, 250x1000x350 mm) had been introduced.

 

HiFi Voodoo

I never invested significantly in chinch or loudspeaker cables and used budget equipment of reasonable quality.  Loudspeaker cable of 2,5 or 4 mm2 was always sufficient for me. On the other hand, I never really analyzed it a lot. The Quad Z-1 offers Bi-Wiring terminals. Let me quote the potential benefit of this from the Quad manual: “Where one cable feeds both bass and treble units, this heavy bass current can modulate the high frequencies. Using separate cables for treble and bass units reduces intermodulation effects and improves headroom and clarity.” Personally, I doubt that there is a significant effect. Even (expensive) Bi-Amplifying is mentioned in the Quad manual. For a single full-range drive unit speaker like the Lowther this of course makes no sense at all. I strongly recommend before investing significant budget in this area to consider investing in the loudspeakers instead. Paying 500,- Euro for a chinch cable doesn’t “sound” like a good investment to me and I guess my Yamaha is not internally wired with cables of this price.

 

I used a Hifime USB DAC to connect my laptop to the amplifier. Today most build-in sound cards of PCs and laptop are of fairly good quality, but sometimes they aren’t. Some models do extra internal processing like resampling everything to 48Khz (which is something you probably like to avoid for a 44,1Khz source). The effects are typically small. Depending on the model direct line-in connection from PC to analog amplifier sometimes cause extra problems like hum. Therefore, I initially purchased the DAC (and USB isolator power supply) while connecting via line-in to my old Yamaha AX-590 amplifier. With the Yamaha R-N803D I just use the optical digital toslink output of the DAC. Since the DAC accepts many different sampling rates and 16 as well as 24bit resolution I can be sure the signal simply gets forwarded to the Yamaha. There are partially very expensive DACs available. I directly compared 3 DACs by my own. The build-in one of my vintage Yamaha CDX-590 CD player from 1999, the ESS Sabre ES9018k2m from USB DAC and the ESS Sabre ES9006AS from my Yamaha R-N803D using my favorite headphones (a vintage Jecklin Float Model II and a Grado GS-1000). I cannot really distinguish this 3 DACs. So again, I would recommend to not investing to much in this area and saving your money for the loudspeakers instead.

 

Today I mostly listen to music from compressed and sometimes uncompressed digital sources. MP3 compression with reasonable bit rate simply sounds fine to me (if not messed up by some unnecessary processing or poor remastering). Of course, I used Vinyl for a long time, and I fully understand people still using it. Not necessary for its objective sound quality but for the far more emotional entire process. Putting your favorite LP on an excellent great looking turn table, cleaning disk and pick up and finally relax and listen concentrated for 20 minutes to a decent part of music with a cup of tea is very different to browsing through millions of songs of your favorite streaming provider on an iPhone APP, streaming it to your Stereo device and jumping from one song to another since you cannot decide which of the millions of songs to listen to. Anyway, I just gave my vintage Dual CS 741-Q with Goldring and Stanton pickups to the son of a neighbor simply because I haven’t used it for years and he was really interested in it. Before doing so I did some final comparison, it still sounds competitive but for me the digital alternative sounds even better (like it did 40 years ago when the CD was born). This is probably even more true for most modern recordings. For me it is hard to argue why a digitally mastered recording should sound better when D/A converting it, pressing it on Vinyl (with limited dynamic range and channel separation) and finally playing it at home on a turn table. This is probably even more true in my case since the Yamaha R-N803D would A/D convert the turntable signal again apply its YPAO correction and D/A convert it another time.    

 

2022 Update

I am now listing to my setup for more than 1 year and I am still very happy with it. The Quad Z-1 is still there but hardly ever used for anything else than demonstrating the speakers to my friends. Most of them are first surprised how similar the speakers often sound and after some more listing how much better the Fidelio performs in most disciplines. Anyway, there are news from Lowther. Meanwhile Lowther offers a so called Phase & Frequency Response Compensation Board (PFRCB) in 2 quality levels for 300 and 600 GBP. It is a little passive network with low complexity like my simple parallel notch filter, but it uses different values and probably very high-quality parts. The PFCRB seems to be prepared for adding a supertweeter. Lowther also provides an update on the Fidelio cabinet resulting in a Mark III version. They recommend equipping it with DX2 or DX3 drive units and PFRCB. They also provide some REW measurements of this setup with and without PFCRB on the website. So far I didn’t had the opportunity to evaluate the PFRCB by myself. The new Lowther Almira combines a slightly modified DX3 with an additional supertweeter and PFRCB in a voigt pipe cabinet (combination of transmission line, ported enclosure and horn). World’s Second Best Speakers! is a very nice video which explains voigt pipe principals very clearly.

 

Digital room correction (YPAO vs. Dirac Live)

Although the Yamaha YPAO was useful to correct some of the bass problems except for the 45 Hz peak which had to be addressed by the Helmholtz resonator it didn’t changed anything in the mid or treble section. In particular the remaining peaks at 2 and 8 KHz. So I was wondering if other DSP room correction could address that. In Sweden   Dirac is offering Dirac Live a highend room correcton software. Several manufactures started integrating it especially in there highend AV Receivers. The Pioneer VSX-LX305 Elite, Onkyo TX-NR7100 and Onkyo TX-RZ50 are 9.2 AV Receivers with Dirac Live integration for an attractive price and the miniDSP SHD Power offers a fully integrated 2 channel solution. All are utilizing Class D type power amplifiers. The NAD T 758 V3i is another option, for additional 99$ it can be upgraded to full frequency Dirac Live support. However less expensive integration e.g. with miniDSP DDRC product line and miniDSP FLex as well as a software only solution DiracLiveProcessor for Windows and Mac OS (standalone and AU/VST/VST3/AAX plugins) are also available. A 14 day trial license can be purchased for free. So, all it requires to start testing it is a PC and a microphone like my MiniDSP UMIK1. I used the Windows software solution with the trial license for testing it in comparison to the Yamaha YPAO of the R-N803. I applied it to the Fidelio with my parallel notch filter and the Quad Z-1 with my Helmholtz resonator in place.

 

The Yamaha YPAO of the R-N803 offers almost no options, you do a single point fully automatic measurement and have no manual control at all. You cannot even review the measurement result nor the calculated filters. Only in some AV Receivers Yamaha offers a multipoint measurement YPAO version.

 

Dirac Live is more advanced. You start with 9 point measurement

 

and when you are finished you can adjust target curve and operation range.

 

Lowther Fidelio measurement:

 

Quad Z-1 measurement:

Dirac comes with a reasonable default target curve designed for high SPLs with a slight decay towards higher frequencies.  It also suggests a reasonable frequency operation range which should be optimized by Dirac Live. Finally, Dirac Live generates the filters and you can apply it to your Dirac Live device, in my case this is the DiracLiveProcessor for Windows. 

 

 After doing so I measured the response without DSP (red), with YPAO (green) and with Dirac Live (blue) of the Fidelio

 

and the Quad Z-1 (without DSP (red), with YPAO (purple)), with Dirac Live (green)):

 

We can see that both YPAO as well as Dirac Live address the bass section, but only Dirac Live addresses the upper midrange a treble issue of the Fidelio and the lower midrange issue of the Quad Z-1. The overall result of Dirac Live is very linear with intended slight treble decay.

 

At a first glance operating Dirac Live may sound complicated, however it is not. Infact the 2 measurements above are my very first Dirac Live usage and worked very well out of the box.

 

I just started using Dirac Live and listen to the result only for a few hours so far. However, I am impressed, there is a very noticeable effect. The bass section does not differ very much from the YPAO result (both offer less booming and room modes and a little bit deeper bass), but the midrange is a lot different. Especially the combination with the Fidelio sounds very promising, no more discoloration nor sometimes aggressive treble.  The Quad result might be a little bit too smooth resulting in a little bit lifeless overall result. I guess that might be due to the 200-400 Hz section, perhaps some target curve adjustment could help.

 

I guess the overall Dirca Live result may depend on the capabilities of the Dirca Live device. It could be that some hardware devices offer less or more DSP capabilities than the software DiracLiveProcessor I used. E.g. a classical miniDSP 2x4 HD offers 10 biquad IIR filters per channel and in total 4096 taps (e.g. 2x2048) for FIR filtering. The corresponding Dirca Live enabled miniDSP DDRC-24 utilizes the same ADSP21489 400MHz Sharc DSP floating point engine.

 

The Dirac Live measurement results may also give you an easy to interpret overall idea of the speaker tonal characteristics by looking at the derivations from the target curve. In my room the Fidelio is overall very neutral but with some significant peaks as 2 and 8 kHz. This translates to an analytic sound but also to some midrange shout and aggressive treble. The Quad Z-1 has a broader range peak at 200 to 1000 Hz and a more significant bass peak around 60 Hz. This translates to warmer less analytic sound but also a somewhat muddy and more boomy sound. This matches very well with may overall impression when switching between the Fidelio and Quad Z-1 without using any DSP.

 

I highly recommend reading Mathias Johansson On room correction and equalization of sound systems paper. It’s an advanced paper (fortunately without too much math in it). But even if you do not understand all the technical details it gives you some important insights why time is so important for the human perception and only smoothing the frequency curve is often not sufficient. For me it e.g. explains one effect in my room: Typically fullrange speakers have an almost perfect very precise Stereo imaging. That is for sure true for Lowther. However in my room it is not. In some frequency ranges it is hard to say from where even a mono signal actual comes from. Normally it should be perfectly located in the middle between the 2 speakers. After reading the paper I have an idea what the root cause actually is. I guess it comes from reflections and uncorrelated sound from left and right speaker due to the asymmetric shape of my room. Let me quote Mathias Johansson:

 

“Let us now consider a traditional stereo loudspeaker set-up in a listening room. If the two sources have different transfer functions then they will change the correlation properties of the Left and Right signals. The signals at the ears are no longer correlated in the same manner as they were on the recording. If the difference is big, the L and R signals may have been completely decorrelated. In that case, even when L=R at the sources we hear two distinct sources instead of the expected phantom in the middle. The sound stage becomes diffuse. If we now do a proper impulse response correction of the individual transfer functions from the left speaker and the right speaker then we can reconstruct the original correlation properties from the recording and the sound stage will become distinct and coherent again. Unfortunately, the reverse situation is sometimes true in alleged room correction systems employing high-order filters. Instead of improving the correlation of the L and R signals, they unwittingly decorrelate the sound field.“

 

Infact Dirac Live improved the Stereo image of the Fidelio in my room a lot, with Dirac Live enabled a singer in the middle now clearly is always located in the middle regardless of the frequency range. 

 

Please refer to "Dirac Live DSP with Lowther Fidelio loudspeaker in my living room" for my latest detailed results using a miniDSP DDRC-24.

 

References