The secrets of the technical characteristics of headphones
Opening our presence on GeekTimes, we decided to start with something technical. And what's better than having a conversation with a well-known expert in the field of technical measurements, asking various tricky questions?
We asked many questions to the founder of the Reference Audio Analyzer project in Russia, Roman Kuznetsov, known under the pseudonym romanrex, the author of many articles that have long been engaged in measuring audio equipment.
')
Some of the questions we were previously asked by clients during consultations and we accordingly included them in the interview.
Alexey: What are the most important characteristics of headphones?
Roman: Of those that are usually indicated on the boxes, only the resistance of the headphones (impedance). This is the only feature to which you can navigate.
Alexey: What about the frequency range?
Roman: There is no single “right” stand for headphone measurements. Today, there are many different stands for manufacturers of headphones for different tasks. This also applies to different methods of measuring characteristics from the stands.
Different stands give different frequency response (amplitude-frequency characteristics). According to one method, the range limit should be determined:
Neither the stand, nor the technique, nor the standard on the box is usually indicated and the numbers in the form of 20 Hz-20 kHz can mean anything. Quite often, a wide range can be seen either on very expensive high-quality models, or on cheap low-quality headphones for 100 rubles.
The difference in frequency response from the position of the microphone on the stand
Unfortunately, even within the range of the same manufacturer, the headphones are measured on different stands using different methods and the results between the headphones are incomparable. Typically, the numbers fall on the packaging of what are needed marketer. All specifications are optional and no one controls them. Some manufacturers do not give any characteristics (Beats Audio, Bose), while others print them only because competitors do.
Alexey: Then the characteristics are completely useless?
Novel: The characteristics of each other can be compared only within the same measurement conditions from one stand. Today on the Internet there are several large bases of measurements of headphones, where you can compare the characteristics of some headphones with others. Knowing which stand was used and how the measurements were taken, we can estimate the difference between the headphones.
Alexey: And which stands are the most correct?
Roman: There is not a single 100% correct booth. There are both universal, and specialized under high accuracy measurements of certain characteristics. For example, there was a situation when it was necessary to check the difference between the right and left channels in headphones with large soft ear cushions (Audez'e LCD-2). Ear cushions at the headphones always affect the fit and worsen the result, although subjectively all these differences level the ear.
Special stand for measuring the difference in the frequency response between the right and left channels of the Audeze headphones
Audeze cannot remove ear cushions - they are glued. In order for the right and left channel of the headphones to be in the same conditions for measurements, I had to make a separate remote camera. The frequency response line at this stand is far from subjectively perceived by man, but it shows exactly the difference between the channels. In the specific case, the difference in the frequency response was negligible.
Special stand for measuring the difference in the frequency response between the right and left channels of headphones Denon D7000
But not only Audeze should be measured without ear cushions, this rule applies to almost all headphones, the ear cushions are removed and measurements are made through a special camera with a microphone.
For measurements of the frequency response close to the perceived subjectively, you need a stand as close as possible to a human head. However, in the academic world there is no agreement on where the microphone should be located, many copies are broken, but there is no answer. Theoretically, the microphone located at the eardrum is obliged to give a more correct result, with a resonance in the region of 2-5 kHz due to the ear canal. The stand with the location of the microphone at the entrance to the auditory canal does not give resonance and is theoretically less correct. Listeners indicate that the received frequency response from the stand with the microphone at the ear canal is closer to what they hear, but there is no 100% hit. This is not surprising, because each auricle shape is unique.
Now in the Reference Audio Analyzer project, new generation stands are ready, in which high accuracy of acoustic damping is achieved. To set up new stands, headphones were measured on a person’s living head, with fixed microphones. The results of the new stand fully coincide with the “live” results.
It was possible to compare the results of test headphones from the standardized artificial ear from Brüel & Kjær. In a standardized stand, internal cameras are used to create acoustic damping equivalent to a living ear. The result was not so close to the living ear, than the new stand RAA HDM-X.
Anatomical accuracy of soft material ears
At the stand with auricles, the frequency response parameter between the right and left channels inevitably suffers.
Stand new generation RAA HDM-X
Alexey: Well, and what characteristics are enough to determine the quality?
Roman: Unfortunately, we can measure only part of the parameters. There are various methods for determining quality, for example, there is a good FSQ test, which was developed by the Acoustic Center of the Radio Broadcasting and Electroacoustics Department of the Moscow Technical University of Communications and Informatics (MTUCI). Subjectively, there is an assessment of about 14 basic parameters, where the frequency response of the frequency response directly or indirectly covers 3-4 parameters. A number of other parameters can be measured, but there is no intelligible psychoacoustic model. Some parameters can be measured, but with a high risk of damage to the headphones (for example, overload capacity), and some parameters require too time-consuming tests that are time consuming and difficult to process. For this reason, the characteristics serve as an addition or give a weighted choice for specific conditions of use of headphones.
Another important priority is the distribution of characteristics. Once it was thought that a flat frequency response is a guarantee of quality. Later it became clear that this is an erroneous view, because in the pursuit of a flat frequency response, other parameters were sacrificed.
All characteristics are auxiliary data, the evaluation of which must be approached carefully.
Alexey: But the more parameters, the more can be said about how headphones sound?
Roman: Of course, although confusion and substitution of concepts often arise. After all, we can measure not only what is directly connected with subjective perception, but also that which has no direct relation to sound. For example, phasing and the presence / absence of high-frequency resonance are determined by the form of the impulse response. But a number of militant audiophiles try to see “speed” and “attack” on an impulse, which is applicable only to amplifiers with a flat frequency response and not applicable to headphones with an uneven frequency response. Despite the name "speed", this does not replace the definition of an attack from psychoacoustics. This is like a characteristic for a car where the engine is located, in front or behind. And according to the audiophile approach, the best “fast” cars are a racing car of Formula 1 on a par with “Zaporozhets”, and in the worst “Zhiguli”, “Mercedes” and “Bentley”.
Alexey: If few parameters can be measured, is it worth it to do it at all?
Roman: That part of the parameters that can be measured can significantly help in the final choice of product. According to the frequency response, we can choose between “more” bass or “light” headphones, where for comparative analysis it is enough to listen to one of the models. According to the sensitivity of the headphones, it is possible to predict whether the amplifier of the player has enough power for specific headphones.
Aleksey: Ie Do not expect a simple quality rating system?
Roman: The only true and suitable rating system for all, without exception, cannot exist. But there is a fresh time ranking, where, based on the technical data of measurements, they can predict a number of recommendations for operating conditions. There will be something like a "rating with asterisks" for using headphones for different purposes. The model is based on statistical data of more than 500 models of headphones and 100 amplifiers and, accordingly, has a fairly high accuracy.
Alexey: And what is needed to start measuring headphones, what equipment?
Novel: Standardized equipment for headphone measurements is quite expensive and if you don’t need to blow your cheeks, you can start with inexpensive measuring microphones and flat-bed construction. Most of the problems at an early stage will not be connected with hardware and software, but with experience where it is important to distinguish whether the test passed correctly or not. From the software it is possible to recommend RMAA and ARTA.
Once the experience has been gained, it can be determined whether to go forward in the development of stands or not. In production, equipment is strictly divided into “partings” and “indicators.” The devices fall under a certain standard (we have in GOST) and the results are legally binding. Indicators can be several classes of higher accuracy in comparison with devices, but not legally binding. Bringing equipment into GOST is an expensive pleasure, and with small runs one can observe space prices. In order to consolidate the result when modding headphones, the race for standardized equipment looks unnecessarily expensive.
Bechringer ECM8000 microphone and Panasonic WM62 primer
From microphones, you can start with the popular Behriger ECM8000 and virtually any sound card that supports phantom power and ASIO. Later or immediately, you can buy microphone capsules, for example, Panasoinc WM62, each of which costs $ 1-2 and take measurements from them, calibrating or checking their performance through ECM8000.
The design of the flat bench is very simple, and on the Internet it will often be possible to find the results of similar headphones and, accordingly, estimate how well the measurements turned out to be correct. But certain discrepancies will still be because It's trite, everyone uses different amplifiers, and matching the full output impedance of the amplifier and the impedance of the headphones change the frequency response. In RAA, for example, this effect is compensated and the result corresponds to the measurement from the “ideal” amplifier with zero resistance. No software today is able to do such compensation.
Alexey: Are there any textbooks or other materials on measurements?
Roman: Unfortunately, no, as well as no brief instructions. Sometimes questions come in the style of “I have an E-MU1616m, downloaded RMAA, how can I measure quality ...”. When asked what the difficulty is: in RMAA or in setting up the E-MU1616m, there are tremendous answers: “Yes, I haven’t even started RMAA - I only heard that it can measure something, tell me what and how ...”. Here you can only say one thing, if there is no real desire in self-development, then forget about measurements right away. Many have a good enough equipment, suitable for measurements, but units measure, since Everything here rests on experience. It’s like taking a scalpel in your hands and naively believing that you’ve turned into a surgeon and is completely ready for a real operation. Mastering requires gradual practice. It is necessary to do step by step. Separate answers to various questions have been online for a long time.
Alexey: Have there been any breakthroughs in the field of headphone measurements lately?
Novel: No radically new tests have emerged over the past decade, but in the past, almost no one measured headphones beyond production. Today, several media outlets are doing this, plus there are separate bases for measuring headphones.
A vivid example of how the frequency response of the M-Audio IE-40 (UltimateEars TripleFi 10) with an amplifier like a Colorfly C4 Pro and with most of the amplifiers from sound cards with a constant resistance of about 20 ohms.
I consider as a separate breakthrough the popularization of the relationship between the impedances of the headphones and the amplifier, where a change in the frequency response appears. In the world of acoustic systems, this is not investigated in any way, since By default, amplifiers always have a high dumping factor. Headphone amplifiers have a completely different situation. Currently, no software can show the interconnection, and RAA is the first software and hardware complex in which it is.
Alexey: But RAA is not available for free use?
Roman: True, the software does not have a “boxed” version, and is currently closely connected with current stands. It is planned to open On-Line service with an interactive charting model. Development as a commercial project at the moment does not seem appropriate.
Alexey: Is it possible to give something to measure and how much does it cost?
Roman: Yes, I can always be contacted and if my work schedule is not overloaded, I can take measurements. For ordinary audio amateurs it is completely free.
Alexey: Were there any cases when measurements were useful outside of professional activity?
Roman: Yes, of course. Most recently, when trying to buy cheaply from the hands of Sennheiser ie 800, measurements on the spot showed that the seller had a fake Sennheiser ie800 . Measurements saved money from waste.
Alexey: What can you wish to the readers?
Roman: I wish not to dwell on any characteristics, but rather to listen with your own ears. And look not for reference headphones, but those that meet the tasks. If the task is to listen to music, then look for those headphones that give pleasure from listening, and not the right, but unpleasant sound. Fortunately, it’s not a problem to listen to headphones before buying, at least in the stores of Moscow and St. Petersburg in Soundpal . And what is the best way to listen, probably the topic is a separate interview.
Alexey: Thanks for the answers, I think that over time we will have new ones.
Roman: I will be glad to answer.
Do you think that some topics were left behind the scenes? Send us questions or write in the comments.
We asked many questions to the founder of the Reference Audio Analyzer project in Russia, Roman Kuznetsov, known under the pseudonym romanrex, the author of many articles that have long been engaged in measuring audio equipment.
')
Some of the questions we were previously asked by clients during consultations and we accordingly included them in the interview.
Alexey: What are the most important characteristics of headphones?
Roman: Of those that are usually indicated on the boxes, only the resistance of the headphones (impedance). This is the only feature to which you can navigate.
Alexey: What about the frequency range?
Roman: There is no single “right” stand for headphone measurements. Today, there are many different stands for manufacturers of headphones for different tasks. This also applies to different methods of measuring characteristics from the stands.
Different stands give different frequency response (amplitude-frequency characteristics). According to one method, the range limit should be determined:
- According to the first - on the decline of 1 dB,
- In the second - 60 dB,
- On the third - that the failure would be no more than 30 dB per octave.
Neither the stand, nor the technique, nor the standard on the box is usually indicated and the numbers in the form of 20 Hz-20 kHz can mean anything. Quite often, a wide range can be seen either on very expensive high-quality models, or on cheap low-quality headphones for 100 rubles.
The difference in frequency response from the position of the microphone on the stand
Unfortunately, even within the range of the same manufacturer, the headphones are measured on different stands using different methods and the results between the headphones are incomparable. Typically, the numbers fall on the packaging of what are needed marketer. All specifications are optional and no one controls them. Some manufacturers do not give any characteristics (Beats Audio, Bose), while others print them only because competitors do.
Alexey: Then the characteristics are completely useless?
Novel: The characteristics of each other can be compared only within the same measurement conditions from one stand. Today on the Internet there are several large bases of measurements of headphones, where you can compare the characteristics of some headphones with others. Knowing which stand was used and how the measurements were taken, we can estimate the difference between the headphones.
Alexey: And which stands are the most correct?
Roman: There is not a single 100% correct booth. There are both universal, and specialized under high accuracy measurements of certain characteristics. For example, there was a situation when it was necessary to check the difference between the right and left channels in headphones with large soft ear cushions (Audez'e LCD-2). Ear cushions at the headphones always affect the fit and worsen the result, although subjectively all these differences level the ear.
Special stand for measuring the difference in the frequency response between the right and left channels of the Audeze headphones
Audeze cannot remove ear cushions - they are glued. In order for the right and left channel of the headphones to be in the same conditions for measurements, I had to make a separate remote camera. The frequency response line at this stand is far from subjectively perceived by man, but it shows exactly the difference between the channels. In the specific case, the difference in the frequency response was negligible.
Special stand for measuring the difference in the frequency response between the right and left channels of headphones Denon D7000
But not only Audeze should be measured without ear cushions, this rule applies to almost all headphones, the ear cushions are removed and measurements are made through a special camera with a microphone.
For measurements of the frequency response close to the perceived subjectively, you need a stand as close as possible to a human head. However, in the academic world there is no agreement on where the microphone should be located, many copies are broken, but there is no answer. Theoretically, the microphone located at the eardrum is obliged to give a more correct result, with a resonance in the region of 2-5 kHz due to the ear canal. The stand with the location of the microphone at the entrance to the auditory canal does not give resonance and is theoretically less correct. Listeners indicate that the received frequency response from the stand with the microphone at the ear canal is closer to what they hear, but there is no 100% hit. This is not surprising, because each auricle shape is unique.
Now in the Reference Audio Analyzer project, new generation stands are ready, in which high accuracy of acoustic damping is achieved. To set up new stands, headphones were measured on a person’s living head, with fixed microphones. The results of the new stand fully coincide with the “live” results.
It was possible to compare the results of test headphones from the standardized artificial ear from Brüel & Kjær. In a standardized stand, internal cameras are used to create acoustic damping equivalent to a living ear. The result was not so close to the living ear, than the new stand RAA HDM-X.
Anatomical accuracy of soft material ears
At the stand with auricles, the frequency response parameter between the right and left channels inevitably suffers.
Stand new generation RAA HDM-X
Alexey: Well, and what characteristics are enough to determine the quality?
Roman: Unfortunately, we can measure only part of the parameters. There are various methods for determining quality, for example, there is a good FSQ test, which was developed by the Acoustic Center of the Radio Broadcasting and Electroacoustics Department of the Moscow Technical University of Communications and Informatics (MTUCI). Subjectively, there is an assessment of about 14 basic parameters, where the frequency response of the frequency response directly or indirectly covers 3-4 parameters. A number of other parameters can be measured, but there is no intelligible psychoacoustic model. Some parameters can be measured, but with a high risk of damage to the headphones (for example, overload capacity), and some parameters require too time-consuming tests that are time consuming and difficult to process. For this reason, the characteristics serve as an addition or give a weighted choice for specific conditions of use of headphones.
Another important priority is the distribution of characteristics. Once it was thought that a flat frequency response is a guarantee of quality. Later it became clear that this is an erroneous view, because in the pursuit of a flat frequency response, other parameters were sacrificed.
All characteristics are auxiliary data, the evaluation of which must be approached carefully.
Alexey: But the more parameters, the more can be said about how headphones sound?
Roman: Of course, although confusion and substitution of concepts often arise. After all, we can measure not only what is directly connected with subjective perception, but also that which has no direct relation to sound. For example, phasing and the presence / absence of high-frequency resonance are determined by the form of the impulse response. But a number of militant audiophiles try to see “speed” and “attack” on an impulse, which is applicable only to amplifiers with a flat frequency response and not applicable to headphones with an uneven frequency response. Despite the name "speed", this does not replace the definition of an attack from psychoacoustics. This is like a characteristic for a car where the engine is located, in front or behind. And according to the audiophile approach, the best “fast” cars are a racing car of Formula 1 on a par with “Zaporozhets”, and in the worst “Zhiguli”, “Mercedes” and “Bentley”.
Alexey: If few parameters can be measured, is it worth it to do it at all?
Roman: That part of the parameters that can be measured can significantly help in the final choice of product. According to the frequency response, we can choose between “more” bass or “light” headphones, where for comparative analysis it is enough to listen to one of the models. According to the sensitivity of the headphones, it is possible to predict whether the amplifier of the player has enough power for specific headphones.
Aleksey: Ie Do not expect a simple quality rating system?
Roman: The only true and suitable rating system for all, without exception, cannot exist. But there is a fresh time ranking, where, based on the technical data of measurements, they can predict a number of recommendations for operating conditions. There will be something like a "rating with asterisks" for using headphones for different purposes. The model is based on statistical data of more than 500 models of headphones and 100 amplifiers and, accordingly, has a fairly high accuracy.
Alexey: And what is needed to start measuring headphones, what equipment?
Novel: Standardized equipment for headphone measurements is quite expensive and if you don’t need to blow your cheeks, you can start with inexpensive measuring microphones and flat-bed construction. Most of the problems at an early stage will not be connected with hardware and software, but with experience where it is important to distinguish whether the test passed correctly or not. From the software it is possible to recommend RMAA and ARTA.
Once the experience has been gained, it can be determined whether to go forward in the development of stands or not. In production, equipment is strictly divided into “partings” and “indicators.” The devices fall under a certain standard (we have in GOST) and the results are legally binding. Indicators can be several classes of higher accuracy in comparison with devices, but not legally binding. Bringing equipment into GOST is an expensive pleasure, and with small runs one can observe space prices. In order to consolidate the result when modding headphones, the race for standardized equipment looks unnecessarily expensive.
Bechringer ECM8000 microphone and Panasonic WM62 primer
From microphones, you can start with the popular Behriger ECM8000 and virtually any sound card that supports phantom power and ASIO. Later or immediately, you can buy microphone capsules, for example, Panasoinc WM62, each of which costs $ 1-2 and take measurements from them, calibrating or checking their performance through ECM8000.
The design of the flat bench is very simple, and on the Internet it will often be possible to find the results of similar headphones and, accordingly, estimate how well the measurements turned out to be correct. But certain discrepancies will still be because It's trite, everyone uses different amplifiers, and matching the full output impedance of the amplifier and the impedance of the headphones change the frequency response. In RAA, for example, this effect is compensated and the result corresponds to the measurement from the “ideal” amplifier with zero resistance. No software today is able to do such compensation.
Alexey: Are there any textbooks or other materials on measurements?
Roman: Unfortunately, no, as well as no brief instructions. Sometimes questions come in the style of “I have an E-MU1616m, downloaded RMAA, how can I measure quality ...”. When asked what the difficulty is: in RMAA or in setting up the E-MU1616m, there are tremendous answers: “Yes, I haven’t even started RMAA - I only heard that it can measure something, tell me what and how ...”. Here you can only say one thing, if there is no real desire in self-development, then forget about measurements right away. Many have a good enough equipment, suitable for measurements, but units measure, since Everything here rests on experience. It’s like taking a scalpel in your hands and naively believing that you’ve turned into a surgeon and is completely ready for a real operation. Mastering requires gradual practice. It is necessary to do step by step. Separate answers to various questions have been online for a long time.
Alexey: Have there been any breakthroughs in the field of headphone measurements lately?
Novel: No radically new tests have emerged over the past decade, but in the past, almost no one measured headphones beyond production. Today, several media outlets are doing this, plus there are separate bases for measuring headphones.
A vivid example of how the frequency response of the M-Audio IE-40 (UltimateEars TripleFi 10) with an amplifier like a Colorfly C4 Pro and with most of the amplifiers from sound cards with a constant resistance of about 20 ohms.
I consider as a separate breakthrough the popularization of the relationship between the impedances of the headphones and the amplifier, where a change in the frequency response appears. In the world of acoustic systems, this is not investigated in any way, since By default, amplifiers always have a high dumping factor. Headphone amplifiers have a completely different situation. Currently, no software can show the interconnection, and RAA is the first software and hardware complex in which it is.
Alexey: But RAA is not available for free use?
Roman: True, the software does not have a “boxed” version, and is currently closely connected with current stands. It is planned to open On-Line service with an interactive charting model. Development as a commercial project at the moment does not seem appropriate.
Alexey: Is it possible to give something to measure and how much does it cost?
Roman: Yes, I can always be contacted and if my work schedule is not overloaded, I can take measurements. For ordinary audio amateurs it is completely free.
Alexey: Were there any cases when measurements were useful outside of professional activity?
Roman: Yes, of course. Most recently, when trying to buy cheaply from the hands of Sennheiser ie 800, measurements on the spot showed that the seller had a fake Sennheiser ie800 . Measurements saved money from waste.
Alexey: What can you wish to the readers?
Roman: I wish not to dwell on any characteristics, but rather to listen with your own ears. And look not for reference headphones, but those that meet the tasks. If the task is to listen to music, then look for those headphones that give pleasure from listening, and not the right, but unpleasant sound. Fortunately, it’s not a problem to listen to headphones before buying, at least in the stores of Moscow and St. Petersburg in Soundpal . And what is the best way to listen, probably the topic is a separate interview.
Alexey: Thanks for the answers, I think that over time we will have new ones.
Roman: I will be glad to answer.
Do you think that some topics were left behind the scenes? Send us questions or write in the comments.
Source: https://habr.com/ru/post/364447/
All Articles