5/23/2009

Validation

Honestly, I cannot be really sure whether I'm doing the right thing (I’m sure many of you share the same feeling).

To be sure, I did simple measurements as below;

Testing method
1) Generates WAV files containing sinusoidal signals (30/40/50/70/100Hz, with 0/-6 dB).
2) Playbacks the test files with my usual amplifier volume position (approx. 1/5 angular position, full power=60W(spec.data)).
3) Records the sounds via a microphone just in front of a speaker.
4) Compares the recoded signals with the original signals.

Result s
1) 100Hz
0dB, std. volume
2) 70Hz
0dB, std. volume

3) 50Hz
0dB, std. volume

-6dB, std. volume
4) 40Hz
0dB, std. volume
-6dB, std. volume
-10dB, std. volume
5) 30Hz
0dB, std. volume
-6dB, std. volume

0dB corresponds to the absolute maximum signal magnitude in 16bit Wav data, thus any boosted signals cannot exceed this level. As described in previous articles, peak signal levels blow 50Hz in most of music sources do not exceed –20 dB , and it is exceedingly rare and quite short moment to reach to –10 dB. These levels correspond respectively to –10dB and 0dB with my standard boost setting (base level = -13.5dB, approx. +10dB at low end ). So basically blow 50Hz signal can be less than –10dB in most cases with my typical boosted listening conditions. If the signal reaches to 0dB, then I can know that via the AVC activations, which happen quite infrequently.

I would say, those conditions can be acceptable for any music sources as far as the volume level is maintained below this standard position, with which I can even play Stravinsky “Rite of Spring” without feeling problems (the signal level actually reaches to 0dB). I have the problem with this tune because the entire signal level of this CD is relatively low in order to record the extraordinary strong tympani within the 16bit dynamic range. In such case I have to turn up the volume beyond the standard position to listen it comfortably, then suffer the tragedy.

You can see a big distortion in 40Hz/0dB with slightly turned up volume position. "Rite of Spring" becomes the disaster.
40Hz, 0dB, std. + 1/20 volume - no mechanical attacking noise
Even with this volume position, I don't feel any critical distortions with classical music including LvB symphonies because of originally low signal level in low end, while sometimes feel a certain distortions with contra-basses in jazz tunes.

In other words, “maximum low boosting” means “reduction of mid-high range from the maximum powered condition of the driver". Therefore basically I would recommend to use Alpair6 or CHR-70 for this kind of extreme boosting applications, with which you could enjoy much more boosting potential (see below table).

driver / X-max(1way) / max. Power
Alpair5 / 2.5mm / 12W
Aplair6 / 5.0mm / 30W
CHR-70 / 4.5mm / 40W
(data from Mark Audio home page)

5/13/2009

If you don’t like to use a computer

or if nice software with English language are not available (Frieve Audio: only Japanese),
as an alternative way, you can do exactly the same process (automatic compensation + manual equalizing) by integrating an external digital equalizer to your existing audio system;
BEHRINGER DEQ2496 Ultracurve Pro

http://www.behringer.com/EN/Products/DEQ2496.aspx (English site)
no computers necessary
Digital In/Out and Analog In/Out available
approx. 40,000.- YEN in Japan
a microphone (option) necessary to perform automatic compensation

Although this equalizer is originally designed for music creators, many Japanese audio freaks are enjoying it quite comfortably.
http://audiolike.exblog.jp/ (Japanese site)

Possible Configurations;
(1) CD Player (Digital OUT) – (Digital IN) EQUALIZER (Digital OUT) – DAC – Amplifier
(2) CD Player (Digital OUT) – (Digital IN) EQUALIZER (Analog OUT) – Amplifier
(3) Analog Player or CD Player (Analog OUT) – (Analog IN) EQUALIZER (Analog OUT) – Amplifier

Sound Quality
(1) > (2) > (3)

(1) is strongly recommended.

5/11/2009

Final Equalizer Settings and Conclusions

I’ve settled the equalizer settings and finalized the system as below.

1. Analysis of Source Signal Level
Below figures are showing original signal levels (PEAK VALUE) in three frequency bands (-50Hz, 50-100Hz, 100-1kHz) measured from several single tracks. music sources(from left);
- "Beethoven Cello sonata No3 I (Classical)"
- "Beethoven Symphony No.8 I (Classical)"
- "Paul Chambers Yesterdays (Jazz)"
- "Chick Corea Return to Forever (Jazz Elec.)"
- "Beethoven Symphony No.5 IV (Classical)"
- "Jaco Pastrius Crisis (Jazz Elec.)"
- "Miles Davis So What (Jazz)"
- "Weather Report Volcano for Hire (Jazz Elec.)"
- "Stravinsky The Rite of Spring(Classical)"

Low end signal levels below 50Hz are generally quite low except in "Rite of Spring".

Below figure is another result measured from totally 45 CDs (including all tracks) (-50Hz only). music sources(from left);
- "Pink Floyd The dark side of the moon (Rock)"
- "Beethoven Symphony No.1-9 (Classical)"
- "Miles Davis 21CDs (Jazz, incl. Elec.)"
- "Weather Report 12 CDs (Jazz, Elec.)"
- "Stravinsky The Rite of Spring (Classical)"
- "Madonna Erotica 1 CD (Pops)"

These are only Maximum Peak values measured in a number of tracks. Generally, it is exceedingly rare and quite short moment to reach –10 dB in this frequency band (below 50Hz).

2. Final Equalizer Setting
Based on the results above I finally made two equalizer settings, i.e. Full Boost Version (30Hz flat) for most of music sources, and Limited Boost Version for e.g. “Rite of Spring”.

Base level of each setting is determined so as to suppress signal overflows (–13.5 dB for Full, -12 dB for Limited). This is necessary to avoid sudden volume down due to AVC (Automatic Volume Control) implemented in the software Frieve Audio, which automatically reduces the total gain (i.e. expands the dynamic range) at the moment of digital overflow.

Generally, signal overflow is inevitable in this kind of extreme digital boosting. Thus higher bit number (more than 24bit) and excellent S/N (120dB in my case) is required for DAC (digital-analog converter) to avoid significant deterioration of the sound quality.

Here is the compensated response curves;
Blue: Limited Boost Version
Red: Full Boost Version
Black: FOSTEX G2000, 4way, 20cm woofer x2
(catalog data, as reference)

You might be anxious about phase delay in low-end frequency range.
Below is the measured delay in nearly maximum excursion driving.
The result indicates clear delaying tendency in lower frequency (absolute vertical scale is not disclosed in the software), which is inevitable in any laud speakers. The DSP also compensates this delay.

3. Conclusions
Now, I’m going to finish this audio trial on my desktop. It is quite comfortable to listen the real full range sound (30Hz to 30kHz) generated from just a single cone with a quite small diameter (approx. 5cm) in a simple sealed box without any reverse ports. This was a kind of dream.

If you can accept this kind of new technologies (i.e. computer audio system with digital equalizing), limited sound volume (near-field listening style) and trivial sacrifices in the sound qualities (i.e. deterioration in S/N due to expanded dynamic range and distortions due to larger diaphragm excursion, honestly I don't feel any significant problems) , then you can realize the dream quite easily.

I selected Alpair5 originally because of the subwoofer integration. If you would not be so adhering to the super detail of Alpair5 in mid-high range, I would recommend Alpair6(fs=74Hz) or CHR-70(fs=70Hz) for desktop applications to reduce the boosting factors. I expect 6dB reduction with those drivers in my calculation.
If your listening distance exceeds 1 m substantially, then Alpair10 full range or Alpair10 woofer + Alpair5 could be suitable. In this case, I expect 10dB or more reduction in the boosting factors.

Let's try !

4/28/2009

Report of Alpair5 Aplication

Here is the total report of the Alpair5 application on my desktop system.
Please also check the system introduction to see the overview of my LEAN AUDIO concept.
http://lean-audio.blogspot.com/2009/03/hello-im-currently-enjoying-on-computer.html

1. Sealed box
I built a pair of sealed boxes for Alpair5 drivers, which have exactly the same dimensions with those for F80AMG drivers.

main dimensions: W=115, H=170, D=240 mm, t=15 mm (lauan plywood), Vin=2.5 L, sound absorbent material (fiber) =32g (adjusted based on hearing )


T/S parameters; F80AMG vs Alpair5 (gold);
http://hpcgi3.nifty.com/bachagi/calnt/calsptop.pl

Calculation result; F80AMG vs Alpair5 (gold); 2.5L sealed box Calculation results;
F80AMG: Fs=125Hz, F3=103Hz
Alpair5: Fs=152Hz, F3=129Hz

Measured responses at 20cm distance on the axial center;
black=F80AMG, red=Alpair5 (gray)

The software(Frieve Audio M-Class) plots the curves automatically adjusting the levels. The actual absolute level of Alpair5 is approx. 3dB higher at the same amplifier volume position.
Roll-off frequencies seems nearly identical despite considerable differences in the T/S parameters (fs=89 Hz vs 125Hz). It is remarkable that Alpair5 indicates nearly the same low-end response with F80AMG whose bass performance is highly reputed in Japan.

Results of the automatic response compensation (20cm, axial center, 30Hz…18kHz flat);
black=F80AMG, red=Alpair5 (gray)

Equalizing factors (20cm, axial center);
black=F80AMG, red=Alpair5 (gray)

This time, I tried extreme equalization up to nearly +30dB ! Surprisingly both drivers were capable of such tremendous low-boosting as shown above.

2. System Integration
This is the current system configuration;

Current system without subwoofer

I bought a new amplifier ONKYO A-905 FX (60W/ch at 4 ohm) to drive Alpair5 (4 ohm) with a plenty of power margin for extreme low-end boosting. The former KENWOOD KA-S10 is not so powerful (only 12W/ch, 8 ohm) and doesn’t support 4 ohm speakers.

Measured responses at the standard position (Alpair5, 80cm distance, 25deg inside offset);
black=LHS, red=RHS

The response becomes uneven due to the room effect, reflections from the desktop and axial offset.

Results of the automatic compensation (25Hz to 18kHz flat,);
black=LHS, red=RHS
NO subwoofer !

Equalizing factors;

F80AMG also worked in the same way as Alpair5.

Do you believe this kind of extreme boosting can be realistic?
I had never believed so. However, surprisingly it works quite well. I also tried a combination with the desktop subwoofer in the same way as described in previous articles in this Blog, and compared both configurations with the same response curve (flat to 25Hz). As the result I decided not to use the subwoofer because Alpair5 without subwoofer produces more natural, clearer and tighter bass. Alpair5 seems maintain its super detail character even in low-end area.

I can play most of music sources including classical symphonies with full boost setting (25Hz flat) without suffering noticeable distortions (i.e. excess driver excursions). I tried a very steep Low-pass filter with 50Hz cut-off on several music sources to know the actual signal level in low-end. Generally the signal level below 50 Hz is not so strong in most tunes that considerable boosting can be possible if you are listening the music in relatively low volume level (ex. near field listening).

As exceptional cases, some modern orchestra tunes, like as “Rite of Spring”(Stravinsky) and “The Planets”(Holst), include extremely strong signals (nearly full dynamic range) in very low-end, which come from bass drums(timpani).
Stravinsky: Le Sacre Du Printemps
Chailly / Cleveland Orchestra
Check its super bass !

For this kind of tunes, I have to reduce the boosting factor in order to avoid excess excursions (distortions). Subwoofer could be beneficial in this case.

Reduced boosting factor;

3. Summary
Alpair5s were implemented in my desktop audio system to playback lossless WAV files stored on an audio PC equipped with powerful digital signal processing software Frieve Audio M-class that offers automatic response compensation.

On the system, Alpair5 and F80AMG (highly reputed 8cm driver in Japan) were compared under the same conditions each other, i.e. under the equalized output response (25Hz to 20kHz flat) without a subwoofer, using the same 2.5L sealed boxes.

Both drivers were capable of extreme low-end digital boosting (equalizing) down to 25Hz without noticeable distortions under the limited conditions, i.e. low sound volume due to near-field listening, thus plenty of margin for maximum excursion of the drivers as well as for maximum output power of the amplifier. Further more 24bit DAC enables to maintain the lowest level signals (the least significant bit) of the original 16bit data even under such the extreme signal processing (Frieve Audio itself performs the internal process with 64bit resolution). For this kind of applications, sealed boxes should be suitable because of no phase reverse thus no steep roll-off.

Only for few music sources with exceptionally high magnitude signal below 50Hz (modern orchestra with very strong bass drums, e.g. “Rite of Spring / Stravinsky), the boosting factors below 50Hz had to be reduced in order to avoid excess excursions. Most of other music sources within my music library including classical symphonies, jazz (acoustic and electric), rock and pops do not indicate noticeable distortions even with the full boost setting.

4. Impressions of Alpair5
Comparing to F80AMG, Alpair5 delivers much higher clarity and details not only in mid-high range as already well known but also surprisingly in very low range below 100Hz with digital boosting. I’m quite sure that I can identify Alpair5 perfectly even under a blind test. The reputations of Alpair5 on its mid-high range is quite famous, so I don't like to repeat it here but would only say “that’s really true” and just add “also in low-end range if you boosted it digitally”. I don’t need F80AMG no longer for any music categories.

The boosted bass from Alpair5 is so tighter and faster that I can no longer be satisified with my own cheap subwoofer. The boosted bass from solely Alpair5 sounds much better in most cases except for above special music.
I didn't have such uncomfortable feelings with F80AMG + subwoofer combination may be because the sound qualities from both units (sub and F80) are not so much different each other.

see the next article for the conclusions;
http://lean-audio.blogspot.com/2009/05/final-equalizer-setting-and-conclusion.html

3/22/2009

Introduction of LEANAUDIO System

I'm currently enjoying a computer audio system consisting of powerful DSP software, DIY small full-range speakers and a compact desktop subwoofer. It’s a quite lean & compact desktop system for near-field listening with a super wide & flat frequency response.
I’m quite satisfied with the system and would like to introduce my audio trial on this Blog.

- Trigger
I started this audio trial approx. 6 months ago. Before that I had not been interested in audio issues at all and satisfied with DENON system compo (approx. 40,000YEN) until I tried a small MP3 player with earphones. It was a kind of cultural shock. It is super convenient because for example I can play LvB symphonies from No.1 to No.9 or Miles Davis from “Birth of the Cool” to “Bitches Brew” in a yearly order without changing CDs. Furthermore, the sound from high quality earphones directly vibrating my eardrums is very clear in whole range including super low-end. Wow, I've immediately recognized that super low must be necessary to really enjoy symphonies. - Target
Then I started this trial to realize clear & wide-range & direct sound like as earphones as well as that operative conveniences on my desktop, where I’m spending plenty of time for daily translation works.

I bought a small audio computer from ONKYO, downloaded a WAV playback software with powerful DSP functionalities that enable automatic response & phase compensation, built DIY small speakers with 3” full-range drivers and finally integrated a small desktop subwoofer to the system. After playing around and try-and-errors for 6 months, I’ve got a quite sufficient result with the system.

1. LEANAUDIO System Configuration
Here is the current system configuration;
---------------------------------------------------
- Audio Computer
ONKYO HDC-1L; Intel Atom processor 230, 160GB HD
- WAV file Playback Software
Frieve Audio M-Class; Shareware software with powerful DSP, automatic response compensation possible
- Main Speakers
8cm full-range drivers; DIY AUDIO SA/F80AMG, 2.5L sealed enclosures
- Powered Subwoofer
Victor SP-DW-1; 14cm woofer, 30W
- Amplifier
KENWOOD KA-S10; 12W/ch
- Desktop dimension
W 1045 x D 750 mm
---------------------------------------------------------------
Figure: System overview on my desktop
7.5kg weights on the subwoofer

2. Concept-1: Near-field Listening
As mentioned above, my target is to realize very clear sound field on my desktop, which is comparable to the high quality earphones. Thus I don't like to have any reflections from surrounding room walls. I just want to listen to direct sound waves form the speakers as far as possible.

With full size audio system, in many cases, you suffer with standing wave effects caused by sound wave reflections from every wall ,ceiling and floor in your room. Generally this phenomenon becomes serious in low frequency range (....100Hz) in which the wavelength reaches comparable to the domestic room size.

Here is a simple calculation result. You can see locations of speakers (rectangles red&black) and a listener (circle) in a square room (3.4m x 3.4m, approx) in the right diagram. If the speakers generate perfectly flat response down to 20Hz, then the listener at the location will get the actual response shown in the left diagram. The result is indicating that the listener under this specific arrangement will loose 70Hz response quite seriously due to the standing wave influence. For this calculation, reflection factor of each surface is set to 0.7, which corresponds to relatively dead condition. I found this value via a rough comparison with measured data in my room (many books and a bed stuffed in a small room).

Below is the measured response in my room.
Above figure indicates the measured responses in front of a full-range driver on its axial center with two different distances from the front baffle. With the distance of 135cm, you can see a big dip at approx 75Hz and a boosting effect in the range below 50Hz. The response with distance of 40cm looks nearly the driver’s original response.

Here is another calculation result. This time I moved the speakers towards the listener to approx. 30cm distance. The conclusion is clear.
So, this is the advantage of Near-field listening. With the shorter distance to the speakers, you can increase a ratio of Direct / Reflection the more.

Based on the consideration above, I decided to locate the speakers on my desktop nearer to me as far as necessary working space could be maintained.

3. Concept-2: Full Utilization of Digital Signal Processing
Many of traditional audio freaks in Japan seem still rather reluctant for Digital Signal Processing(DSP).

Nowadays, actually 100% of music source is available only in digital format. If thinking in logical way, it must be quite reasonable to perform all necessary filtering and equalization within digital phase because digital filtering offers highest flexibility and accuracy without phase errors.

However, still in the world, many of such processes are done after DAC (digital to analog conversion). For example, most of LCR circuits (e.g. channel divider, some kind of speaker response compensation, tone control of amplifier) and low boost mechanisms (e.g. bass reflex box, resonance boxes), with which phase problems are inevitable, can be partly replaced by DSP.
I’m much interested in comparison of two types of speakers, i.e. one with a simple sealed box with DSP low boost, and another with a bass-reflex box without DSP. Generally, bass-reflex has some problems in low-end range, i.e. larger phase delay, phase reverse thus steep response drop and steep increase of diaphragm stroke. Even with a sealed box phase delay in low range is not avoidable, however DSP can compensate it too.

Now, I’ll introduce powerful WAV playback software Frieve Audio M-Class (Shareware, 3,200YEN !, only Japanese version).
Figure: Overview of Frieve Audio

Its main functionality is playing back of lossless WAV file extracted from music CDs on Windows PC . However, Frieve Audio is outstanding because of its automatic response compensation ability. If you have a microphone for PC, you can make frequency response at your listening position completely flat. As mentioned above, the response can be relatively flatter with the near-field listening style. However, even within 1m distance (actually 85cm in my case), you will still see clear influences of sound reflections.

Using the automatic compensation you can reduce these effects very easily. The system generates a test signal and analyzes the response signal from the microphone at the same time, then calculates necessary equalization factors and also phase compensation factors at each frequency point (max 65535 taps, adjustable).

The internal process is done with 64bit resolution. The resulted factors can be stored permanently and recalled. All calculations and compensations are performed independently for L and R speakers.

Here you can see the effect of automatic compensation at my standard listening position.

Figure: Compensation OFF and ON

First time if you tried the flat response, you might feel unreasonable. However, once you switched off the compensation, you would immediately recognize that you should never switch it off again.
Most of people who tried this compensation are expressing similar experiences in their own Internet threads. Many of them including me use the flat condition as a kind of reference and modify it slightly as preferred. It’s a really useful tool.

4. Concept-3: Satisfactory Low-end Response with Desktop Subwoofer
From experiences with high quality earphones, I have learned the importance of low-end response for really enjoying music. So this is another target, i.e. “super direct low-end sound like as earphones”.

Down to which Hz should the response be flat?
A Japanese audio consultant Mr. Teppei Kado, I rely on his logical and reasonable opinions, indicates his own criteria for “satisfactory speaker” in his own Internet site. According to his opinion, 25Hz(-5dB) should be a guideline. OK, I’ll take it. I decided to integrate a small subwoofer to my desktop system.

I wanted a small powered subwoofer that could be put on my desktop but not on the floor. For such applications, the device should have all sound radiating elements (i.e. woofer and duct) on its front face. Fortunately I found an exactly right one with fairly low price.


Victor SP-DW1
14cm woofer
30W digital amp
approx. 15,000YEN

Why I put it on the desktop?
The reason is to catch the sound wave as directly as possible. If you put a subwoofer on the desktop, you can effectively utilize the first reflections (so called wave guide effect) from a desktop and a computer display surfaces just in front of you. Once I tried on the floor setting, but I’ll never do it again. If you put it on the floor, you will have to boost the subwoofer volume more to get the same level of low-end response because of more dispersion, thus you will suffer more reflections from your surrounding walls.
Subwoofer Desktop Configuration

How to optimize the subwoofer integration?
At first, adjust volume and cut-off frequency of the subwoofer based on measurements. The result is shown in a figure blow. The cut-off is the lowest (50Hz), the volume is nearly minimum (less than 1/4). It requires only minimum power thanks to above-mentioned desktop configuration. With this minimum setting, the subwoofer's effect can be limited up to approx. 100Hz.
Then, let’s make it flat by using Frieve Audio software.

Frieve Audio also compensates the phase delay that often be controversial on subwoofer issues.

The low-end of this bass-reflex subwoofer sometimes sounds unstable, how to say, boomy or untidy. To avoid this I had to apply very steep 30Hz digital low-cut filter on Frieve Audio manual equalizer.
Currently I have closed the duct by tightly stuffing felt mats, so now it is more ore less a sealed box. With this modified configuration, I can boost +10 dB from 30Hz to 20Hz without suffering any boomy problems.
Finally I’ve got;

Red: duct open(low-end cut-off), Black: duct closed(low-end boosted)

The difference might look small, however thorough-bass of symphonies sounds definitely different. It’s good. Now I have achieved the criterion of Mr. Kado, i.e. 25Hz/-5dB.

The direct and tight super low-end sound from a desktop subwoofer is quite cool.
Try it!

5. Concept-4: 8cm Single Cone Full-range Drivers for the Main Speakers
Now, the fundamental of “Music Reproduction” is firmly established by “Near field Listening Style”, “Desktop Subwoofer” and “Response Compensation”.
On this basement, the main speakers finally determine the tone character, taste, mood and so on of the total system, which much depends on the listener’s preferences.

As the main speakers, I have selected 8cm(3”) full-range drivers because many of them covers from 100 to 20kHz easily. For me, response below 100Hz is not necessary or even disturbing because of the subwoofer integration. 10cm might be also OK, but maybe less high-end quality. 2-way? I don’t like it for my near-field style. I cannot accept, just mentally, a fact that 3kHz sound wave that has 11.3cm wavelength comes from two different drivers distant each other approx. 10cm. I don’t like to listen it at 85cm distance (again in mentally).

I'm using DIY AUDIO SA/F80AMG that was the highest reputed 8cm driver at that time in Japan. I tried another 3 types of drivers, one from FOSTEX and two from TangBand, then finally selected F80AMG. It’s not so powerful, it’s not so brilliant, it’s not so amusing, however it delivers totally best sound for me. How to say, it’s gentle, accurate, moderate, well structured and stout. I would fully agree the reputations.

DIY AUDIO SA/F80AMG 8cm full-range driver

I did say above, “F80AMG was highest reputed at that time”.
Because I can no longer believe that's really true.
Just recently the reputation of MarkAudio Alpair is coming up significantly among Japanese DIY builders. I would say "like as the Golden rising sun".
For me the smallest Alpair 5 must be the target, which seems designed as a super wide-range tweeter for combining with a subwoofer or a large driver. It must be the best one in the world for my LEANAUDIO system, I believe.
MarkAudio Alpair 5 Gold


6. Summary Finally I’m quite satisfied with the system performance.

The target of this trial was to realize high quality earphones-like sound on my desktop. For that, first of all, influence of sound reflections from walls (i.e. standing-wave effect) was reduced effectively by means of “Near-field Listening Style” and “Response Compensation via DSP”.

On this base, a subwoofer was integrated additionally to fulfill the second target "earphones-like clear & tight super low-end". To achieve this, “Desktop Subwoofer Configuration” and “Optimum Subwoofer Integration via DSP” worked quite efficiently.

I believe above tow requirements, i.e. reduction of standing-wave effects and fulfillment of satisfactory low-end (also high-end of course, but easier) must be the very basic of music reproduction on a 2ch stereo audio system, which should be discussed and fulfilled before any trivial sound fidelity issues.

On this basic structure, 8cm single cone full-range drivers were implemented as the main speakers, which cover 100Hz … 20kHz and determine the total sound character and taste of the system. Although current F80AMG drivers are quite satisfactory for me, I’m plotting to try Alpair 5 in order to further improve detailed clarity in mid-high range.

to be continued....