Study on evaluation method of the pue tone fo small fan Takao YAMAGUCHI 1 ; Gaku MINORIKAWA 2 ; Masayuki KIHARA 3 1, 2 Hosei Univesity, Japan 3 Shap Copoation, Japan ABSTRACT In the field of audio, visual and infomation technology equipment, small fan noise which includes many pue tones becomes annoying components. Pue tones ae caused by the electomagnetic foce of moto, the flow intefeence between blades and spokes, the acoustic modes of the stuctue and so on. These tones become not only the main contibution to the oveall sound pessue level, but also unpleasant component. Some metics fo the pue tone have been pesented, howeve the one which specified fo small fan noise has not been developed. In this study, evaluation of multiple pue tones geneated by small fan was attempted and examined by using sensoy test. Keywods: Small fan, Fan noise, Tonal components, Tone to noise atio, Pominence atio, Sound Quality I-INCE Classification of Subjects Numbe(s): 12.3.4 63.7 1. INTRODUCTION Recently, pefomance impovement and downsizing of infomation technology o audio and visual equipment have been acceleated, the installation density of electic devices has also been inceasing and as a esult, the cooling fans ae equied so as to exhaust the big amount of heat. The usage of fan makes the heat design much easie, on the othe hand, the fan noise is getting big poblem. Especially, the tonal components of the noise ae annoying even though the level is low. In the standads such as the ISO 7779, ECMA 74 and ANSI S1.13., the judgment methods of the cetain pominent discete tone in infomation technology devices ae pesented as the Tone to Noise Ratio (TNR) and the Pominence Ratio (PR). On the othe hand, a psychoacoustic appoach to detec t the tonal feeling has been developed and the metics such as the Tonality has been used fo the evaluation of the tonal feeling. In the pesent study, the sound quality evaluation fo the small fan was done by using the metics based on the TNR and PR. 2. Noise chaacteistics of small fan 2.1 Fan noise geneation mechanism The Aeodynamic noise geneated fom fan can be classified into the otating noise and the tubulent flow noise. The fome is poduced by the peiodic pessue fluctuations due to the otation of the impelle and has the discete components at the blade passing fequency which is the poduct of the numbe of blades Z and otating speed n and its hamonics. It is so peaky that the contibution to the ovealls noise level is lage and is annoying. The otating noise is caused by two types of unsteady flow, one is so called the potential intefeence noise caused by the pessue distibution aound the impelle and anothe is so called wake intefeence noise induced by the sepaated flow fom the tailing edge of the impelle. In case of the small axial fan, when the obstacles such as statos and spokes ae located just behind the impelle, the unsteady flow intefee the obstacles and lage noise will be geneated. On the othe hand, the tubulent noise is caused by the andom tubulent flow in the fan. The noise souce consists of the pessue fluctuations induced by the tubulent bounday laye on the suface of 1 takao.yamaguchi.8u@stu.hosei.ac.jp 2 minoi@hosei.ac.jp 3 kihaa.masayuki@shap.co.jp Inte-noise 2014 Page 1 of 5
Page 2 of 5 Inte-noise 2014 the impelle, the fluctuation of lift caused by the votex shedding fom the tailing edge of the impelle and so on. It has boadband fequency components, the enegy is smalle than the otating noise and the contibution to the oveall sound pessue level is seconday. 2.2 Tonal components of centifugal fan The otating noise fom centifugal fan is caused by the intefeence between the peiodic flow fom the impelle blade and the tongue of the fan casing. The fequency of tonal components f ae shown as, Whee, i : Numbe of fequency components Z : Numbe of blades n : Rotating speed [ps] f i Z n (1) 2.3 Tonal components of axial fan The otating noise fom axial fan is caused by the intefeence between the peiodic flow fom the impelle and the stato o the spoke which suppots the moto. In geneal, the numbe of the blades and the spokes should be the pime numbe so as not to occu the intefeence of the blades and the spokes at the same time, so the fequency components ae moe complicated matte than that of centifugal fan. The fequency of tonal components f ae shown as the following fomula, Whee, j, k : Intege index( -2, -1, 0, 1, 2 ) Z, Z s : Numbe of blades and spokes n : Rotating speed [ps] f j Z n (2) j Z k Zs On both cases, the shape of the wake is not exact sinusoidal, so the hamonics components ae appeaed at the same time in FFT. In addition, these components ae influenced by the fequency esponse function of the fan system. The typical noise spectum of the small axial fan is shown in Figue 1. Figue 1 Noise spectum of small axial fan Page 2 of 5 Inte-noise 2014
Inte-noise 2014 Page 3 of 5 3. Evaluation methods of tonal components 3.1 Tonality Tonality povided by Tehadt o Aues is known as the metics that indicate the tonal feeling of the sound. It is calculated as the poduct of the vaious weighted functions to the tonal components such as the influence of masking by othe tones, the atio of tonal components and noise, the influence of heaing theshold and is applicable to the evaluation of subjective tonality of total sound. 3.2 Tone to Noise Ratio (TNR) and Pominence Ratio (PR) In the standad, the DIN 45681 Acoustics - Detemination of tonal components of noise and detemination of a tone adjustment fo the assessment of noise immissions defines how to detemine the tonal component to the noise. But this method shows only the excess of the level and does not include the psychoacoustic consideation. In geneal, the audible theshold o the pominence of a tonal component is decided by the elationships between the tonal component level and the suounding band noise level which is masking the tonal component. The fequency bandwidth is so called the citical band that is centeed at the fequency of the tone. In the standads such sas the ISO 7779, ECMA 74 and ANSI S1.13., the judgment methods of the cetain pominent discete tone in infomation technology devices ae pesented as the Tone to Noise Ratio (TNR) and the Pominence Ratio (PR). The TNR is defined as the decibel value of the atio of the powe of tonal component and othe noise component in the citical band. In the ECMA-74, when the TNR exceeds by 6dB, the tonal component is egaded as the pominent discete tone. In case that multiple peaks exist in the same citical bandwidth o the noise level next to the citical bandwidth is consideable, the TNR vales tends to show bigge o smalle. The PR is defined as the decibel value of the atio of the citical bandwidth powe including the tonal component and the citical bandwidth powe on both sides. In the ECMA-74, when the PR exceeds by 7dB, the tonal component is egaded as the pominent discete tone. Figue 2 is the schematic view of calculating the TNR and PR. Tone to Noise Ratio Figue 2 Schematic view of TNR and PR calculation Pominence Ratio 3.3 Total TNR(T-TNR) and Total PR(T-PR) In fan noise evaluation, it is impotant to quantify the tonal components fo the poduct quality contol, the identification of noise souce and the noise eduction. The Tonality is a little bit complicated and isn t good at the evaluation of the individual tonal component. On the othe hand, the TNR and PR ae the indicatos fo the pominent discete tone and don t cae fo the multiple tonal components. Hee, the new evaluation paametes so called TTNR and TPR ae pesented, in which the excess levels fo multiple tonal components ae calculated by the method of the TNR and PR and summed up the TNRs and PRs. Inte-noise 2014 Page 3 of 5
Page 4 of 5 Inte-noise 2014 n T TNR 10 log 10 10 i1 n T PR 10 log 10 10 i1 Whee, T TNR : Total tone to noise atio [db] TNR i : Tone to noise atio fo i th peak component [db] T PR : Total pominence atio [db] PR : Pominence atio fo i th peak component [db] i TNR i 10 PR i 10 (3) 4. Sensoy test 4.1 Noise souce As the noise souce, a small axial fan (The Oientalmoto MD825B-12) was chosen. The fan was diven by 3740pm without flow estiction and the noise was measued in font of the fan by the noise level mete (Rion NL-31) set at 0.5m apat fom the fan. The noise signal was ecoded by the PC as wav file. The ecoded noise was cut by 3sec duation and egaded as the Sound 1 which was oiginal one. Then, the Sound 2 to Sound 5 wee made by amplifying the peak components unde 4000Hz by 2 to 5times as the Sound 1. Finally, the sound pessue of the sounds was adjusted so as the oveall sound level wee the same. Figue 3 Spectum of test sound 4.2 Juy test Fo the juy test, the PC (Paied Compaison) method was chosen and 20 college students with the nomal heaing attended the test as juies. The adjective of Annoying Not annoying" was used fo the evaluation. In each test set, two sounds wee chosen at andom out of five sounds and the juy judged the impession of the fist sound to the following sound by five steps. In ode to eject the influence of the pesentation ode, all combination pattens wee poposed. By using the esults, the Page 4 of 5 Inte-noise 2014
Inte-noise 2014 Page 5 of 5 coelation between the subjective annoyance and the sound quality evaluation paamete as pesented in this pape was calculated. 5. Results and discussion Figue 4 shows the elationships between the pesented metics and the subjective annoyanc e. TTNR and TPR showed high coelation coefficients ove 0.99. Compaing the TTNR and TPR, TPR has highe sensitivity. In this case, since the fequency stuctue of the sample sounds was same and only the level of the tonal components was vaied, the level of the tones might decide the annoyance even the oveall level wee standadized. As the next step, it is equied the test with sample sounds which fequency stuctue of tones o each level of tones ae vaied. In addition, by standadizing the loudness level, the effect of tonal component will appea moe significantly.. Figue 4 Relationships between TTNR o TPR and subjective annoyance 6. SUMMARY Fo the evaluation of the sound quality of small fan, new metics based on the TNR and PR wee pesented and investigated the elationship to the subjective annoyance by the juy test. The pesented metics showed high coelation with the subjective annoyance. REFERENCES 1. Shoji Suzuki, Masahau Nishimua, Shinya Kijimoto and Gaku Minoikawa, Mechanical Acoustical Engineeing, Coona publishing, 1993 2. Reseach Committee of Sensoy Evaluation, Union of Japanese Scientists and Enginees, Sensoy Evaluation Handbook, JUSE Pess 1973 3. Seiichio Namba and Sonoko Kuwano, Psychological measuing method fo sound evaluation, Coona publishing 1998 4. ECMA-74 Measuement of Aibone Noise emitted by Infomation Technology and Telecommunications Equipment, ECMA Intenational 2010 5. DIN 45681 Acoustics - Detemination of tonal components of noise and detemination of a tone adjustment fo the assessment of noise immissions, DIN 2005 6. Aaon Hastings and Paticia Davies, Pudue Univesity, An examination of Aues s model of tonality, Poceedings of Intenoise2002 2002 7. Enst Tehadt, Gehad Stoll and Manfed Seewann, Algoithm fo extaction of pitch salience fom complex tonal signals, J. Acoust. Soc. Am 1982 Inte-noise 2014 Page 5 of 5