Frequency response analysis. The concept of amplitude-frequency and phase-frequency characteristics of the system. Methods for experimental reading of frequency response and phase response Frequency response analysis
I bought Motorola Pulse Escape Bluetooth headphones. Overall I liked the sound, but one thing remained unclear. According to the instructions, they have an equalizer switch. Presumably, the headphones have several built-in settings that switch in a circle. Unfortunately, I could not determine by ear what settings there were and how many there were, so I decided to find out by measuring.
So, we want to measure the amplitude-frequency response (AFC) of headphones - this is a graph that shows which frequencies are reproduced louder and which ones are quieter. It turns out that such measurements can be made “on the knee”, without special equipment.
We will need a computer with Windows (I used a laptop), a microphone, and also a sound source - some kind of player with bluetooth (I took a smartphone). Well, the headphones themselves, of course.
(There are a lot of pictures under the cut).
Preparation
I found this microphone among my old gadgets. The microphone is cheap, for conversations, not intended for recording music, much less for measurements.Of course, such a microphone has its own frequency response (and, looking ahead, directional pattern), so it will greatly distort the measurement results, but it is suitable for the task at hand, because we are interested not so much in the absolute characteristics of the headphones, but in how they change when the equalizer is switched.
The laptop had only one combined audio jack. We connect our microphone there:
Windows asks what kind of device we connected. We answer that this is a microphone:
Windows is German, sorry. I promised to use improvised materials.
Thus, the only audio connector is occupied, which is why an additional sound source is needed. We download a special test audio signal to the smartphone - the so-called pink noise. Pink noise is a sound that contains the entire spectrum of frequencies, and equal power over the entire range. (Do not confuse it with white noise! White noise has a different power distribution, so it cannot be used for measurements, as this may damage the speakers).
Adjust the microphone sensitivity level. Right-click on the speaker icon in Windows and select adjust recording devices:
Find our microphone (I called it Jack Mic):
Select it as a recording device (bird in a green circle). We set its sensitivity level closer to the maximum:
Microphone Boost (if present) is removed! This is automatic sensitivity adjustment. It’s good for the voice, but during measurements it will only interfere.
We install the measuring program on the laptop. I love TrueRTA for the ability to see many charts on one screen at once. (RTA - frequency response in English). In the free demo version, the program measures the frequency response in octave steps (that is, adjacent measurement points differ in frequency by a factor of 2). This, of course, is very crude, but for our purposes it will do.
Using tape, secure the microphone near the edge of the table so that it can be covered with an earphone:
It is important to fix the microphone so that it does not move during the measurement process. We connect the headphones with a wire to the smartphone and place one earphone on top of the microphone, so as to tightly close it on top - something like how the earphone covers the human ear:
The second earphone hangs freely under the table, from which we will hear the test signal turned on. We make sure that the headphones are stable and cannot be moved during the measurement process. We can begin.
Measurements
We launch the TrueRTA program and see:The main part of the window is the field for graphs. To the left of it are the buttons for the signal generator; we don’t need it, because we have an external signal source, a smartphone. On the right are settings for graphs and measurements. At the top are some more settings and controls. Set the field color to white to better see the graphs (menu View → Background Color → White).
We set the measurement limit to 20 Hz and the number of measurements, say, 100. The program will automatically make the specified number of measurements in a row and average the result; this is necessary for a noise signal. Turn off the display of bar charts, let graphs be drawn instead (the button at the top with the image of bars is marked in the next screenshot).
Having made the settings, we make the first measurement - this will be the measurement of silence. We close the windows and doors, ask the children to be silent and press Go:
If everything is done correctly, a graph will begin to appear in the field. Let’s wait until it stabilizes (stops “dancing” back and forth) and click Stop:
We see that the “volume of silence” (background noise) does not exceed -40dBu, and we set (the dB Bottom control on the right side of the window) the lower display limit to -40dBu in order to remove background noise from the screen and see the graph of the signal we are interested in in a larger view.
Now we will measure the real test signal. Turn on the player on your smartphone, starting with low volume.
We start the measurement in TrueRTA with the Go button and gradually turn up the volume on the smartphone. A hissing noise begins to come from the free earphone, and a graph appears on the screen. Add volume until the graph reaches a height of approximately -10...0dBu:
After waiting for the graph to stabilize, we stop the measurement using the Stop button in the program. We also stop the player for now. So what do we see on the graph? Good bass (except for the deepest ones), some roll-off towards the mid-range frequencies and a sharp roll-off towards the high frequencies. Let me remind you that this is not the real frequency response of headphones; the microphone makes its contribution.
We will take this graph as a reference. The headphones received a signal via wire, in this mode they work as passive speakers without any equalizers, their buttons do not work. Let’s save the graph into memory number 1 (via the menu View → Save to Memory → Save to Memory 1 or by pressing Alt+1). You can save graphs in memory cells, and use the Mem1..Mem20 buttons at the top of the window to enable or disable the display of these graphs on the screen.
Now we disconnect the wire (both from the headphones and from the smartphone) and connect the headphones to the smartphone via bluetooth, being careful not to move them on the table.
We turn on the player again, start the measurement with the Go button and, by adjusting the volume on the smartphone, bring the new graph in level to the reference one. The reference chart is shown in green, and the new chart is shown in blue:
We stop the measurement (you don’t have to turn off the player if the hiss from a free earphone doesn’t irritate you) and are glad that via Bluetooth the headphones produce the same frequency response as via wire. We save the graph into memory number 2 (Alt+2) so that it does not leave the screen.
Now we switch the equalizer using the headphone buttons. The headphones report in a cheerful female voice “EQ changed.” We turn on the measurement and, after waiting for the graph to stabilize, we see:
Hm. In some places there are differences of 1 decibel, but this is somehow not serious. More like measurement errors. We put this graph into memory, switch the equalizer again and after the measurement we see another graph (if you look closely):
Well, you already understand. No matter how much I switched the equalizer on the headphones, it made no difference!
On this, in principle, we can finish the work and draw the following conclusion: These headphones do not have a working equalizer. (Now it’s clear why he couldn’t be heard).
However, the fact that we did not see any changes in the results is disappointing and even raises doubts about the correctness of the methodology. Maybe we measured something wrong?
Bonus dimensions
To make sure that we measured the frequency response, and not the weather on the Moon, let's turn the equalizer in another place. We have a player in our smartphone! Let's use its equalizer:Amplitude-frequency response of headphones (abbreviated frequency response, also “frequency response of the system”, in English - frequency response) is the dependence of the oscillation amplitude (volume) at the headphone output on the frequency of the reproduced harmonic signal. The amplitude-frequency response shows the tonal balance. From the amplitude-frequency response, a frequency response is obtained, which is also called the frequency range, indicated on the boxes or in the documentation for the headphones.
The frequency range is divided into low, medium and high frequencies; the picture above shows how the frequency grid and the names of the frequency ranges relate. Below are the meanings of each range. As you can see, frequencies are perceived in a logarithmic representation - through doubling the frequency. The frequency range in which the upper frequency is twice the lower frequency is called an octave. For example, octaves are frequency ranges: 20 ~ 40 Hz, 250 ~ 500 Hz, 3 ~ 6 kHz.
Common names of frequency ranges |
||
20 - 40 Hz | Low Bass | Sub bass |
40 - 80 Hz | Mid Bass | Midbass |
80 - 160 Hz | Upper Bass | Upper Bass |
160 - 320 Hz | Lower Midrange | Lower middle |
320 - 640 Hz | Middle Midrange | Center mid range |
640 Hz - 1.28 kHz | Upper Midrange | Upper middle |
1.28 - 2.56 kHz | Lower Treble | Bottom high |
2.56 - 5.12 kHz | Middele Treble | Mid high |
5.12 - 10.2 kHz | Upper Treble | Upper high |
10.2 - 20.4 kHz | Top Octave | Upper octave |
To evaluate the sound of instruments and different sounds, we offer the following diagram for your reference:
Green color - the main sound range (gray-green - non-dominant low frequencies), orange - aftertones, overtones, additional. harmonic series, (gray-orange - upper non-dominant range).
The vertical axis of the graph indicates the volume level, usually expressed in decibels (dB). A double change in sound pressure corresponds to 6 dB. Subjective perception of loudness depends on many factors (equal loudness curves, spectral composition, etc.), but in general cases we can roughly estimate that a double change in sound pressure will correspond to a double change in loudness.
The values can be relative or absolute in SPL (Sound pressure level). The SPL level can be used to determine the sensitivity of the headphones.
This example shows the frequency response of two headphones, A and B. Headphone A reproduces low and high frequencies quieter than earphone B, but at the same time reproduces mid frequencies louder.
This shows the deviation between headphones and shows more clearly that earphone A is up to 6 dB quieter at low frequencies, and up to 6 dB quieter at the highest frequencies (upper octave). But on average it’s louder by almost 6 dB. In other words, Earphone A plays the low and highest frequencies twice as quiet, and, conversely, the mid frequencies are almost twice as loud.
To assess sound evenness, we offer several general graphs.
Sony MDR-1AM2 Black
14 990 .-
Add to cart
Add to favorites
Compare
Sennheiser HD 800
Product available in the online store
64 900 .-
Add to cart
Add to favorites
Compare
General types of frequency response for open, closed and on-ear headphones. Peculiarities.
Here you can observe several characteristic types of frequency response. The green graph is a subjectively flat frequency response; at the highest frequencies you can see a decline; it is perceived smoothly due to the fact that we are accustomed to perceiving the smooth frequency response that is reproduced by acoustic systems located in front of the listener. In relation to the ear, it is at 60 degrees. If acoustics with a direct frequency response are placed on the side, at 0 degrees, then an excess of the highest frequencies will be perceived. Therefore, thanks to the smooth roll-off, a subjectively smooth sound is achieved. The yellow graph is usually audiophile headphones with accentuated low and high frequencies.
Such headphones are especially in demand among those who listen to recordings of live music, in which the lowest and highest frequencies are minimal. The blue graph is headphones with an emphasis on the upper mid frequencies; usually this graph is found on monitor headphones for musicians for whom it is important to hear their voice as clearly and intelligibly as possible. This can also be found in audiophile headphones for those who prefer listening to vocals. The red graph is a special dip that can serve as a solution against sibilance or other sound emphasis. which does not suit listeners when listening to certain genres. Having decided for what tasks you want to purchase headphones, you can select a number of models based on their characteristic frequency response characteristics.
In the high frequency region, unevenness can usually be observed. It is not worth calculating the exact frequencies and heights of peaks and dips, because they depend on how the headphones are worn. At our special stand there are much fewer variations to put on headphones than at simpler stands, and the stand is closest to reality. However, if your auricle is different, and you wear headphones slightly differently, then this unevenness will only be approximate. Also, depending on the volume level, this unevenness will be subjectively perceived a little differently, as follows from studies of equal volume curves.
The graph line may have some irregularity. Unevenness in the frequency response can appear either from resonances that decay for a long time, or from the interference of sound waves (which is typical for headphones with complex profiles of protective grilles). In the first case, this indicates worse sound, in the second case it does not affect the sound. For a complete picture, you need to look at diagrams of the cumulative spectrum (which is a three-dimensional sonogram) or the attenuation of resonances depending on periods at specific frequencies.
A number of dips are caused by wave interference. On graphs without smoothing, they represent dips in a narrow frequency range. Such dips are not significant and strongly depend on the fit of the headphones.
General types of frequency response for in-ear headphones (plugs). Peculiarities.
Here you can observe several characteristic types of frequency response. The green graph is a subjectively flat frequency response; at the highest frequencies you can see a decline; it is perceived as flat due to the fact that we are accustomed to perceiving the smooth frequency response that speaker systems reproduce when they are in front of the listener. In relation to the ear, it is at 60 degrees. If acoustics with a direct frequency response are placed on the side, at 0 degrees, then an excess of the highest frequencies will be perceived. Therefore, thanks to the smooth roll-off, a subjectively smooth sound is achieved.
The orange graph shows headphones with increased output at low frequencies; such headphones are preferred mainly for portable use when listening to music from a mobile phone or player. Many players and phones have a roll-off in the low frequencies (for example, to save batteries) and more bassy headphone models can correct this deficiency. The blue graph is headphones with an emphasis on the upper mids; this graph is usually found on monitor headphones for musicians for whom it is important to hear their voice as clearly and intelligibly as possible. This can also be found in audiophile headphones for those who prefer listening to vocals. Having decided for what tasks you want to purchase headphones, you can select a number of models based on their characteristic frequency response characteristics.
Unevenness above 10 kHz is highly dependent on the fit of the earphone in the ear canal, and a shift of half a millimeter completely changes the graph. For this reason, it is worth evaluating the graph as the average value in this area.
On the frequency response one can observe from one to several resonances, depending on the landing depth. The frequency of such a resonance is purely individual for each listener, so this resonance is excluded on the graph, but the typical resonance value is shown in dim color. Ideally, it is better to choose headphones that have little or no such resonances.
Dependence of frequency response on amplifier and headphone impedance
The type of frequency response depends on the impedance of the headphones and the impedance of the amplifier (output impedance). As a rule, the frequency response of headphones remains unchanged when the output impedance of the amplifier is close to zero, as well as when the impedance of the headphones has a minimal deviation close to resistive in nature. The higher the output impedance of the amplifier and the more the Rz curve fluctuates, the more the frequency response of the headphones changes.
When measuring the amplifier output in RMAA with an active load, where the load is headphones, you can see the frequency response with a hump in the low-frequency region. In this case, it shows how the frequency response of headphones changes against an amplifier with zero resistance. The error of such a graph depends on the input resistance of the sound card, and the higher it is, the lower the error.
In the example, we will consider the dependence of the frequency response on amplifiers with different output impedances. In our example, the headphones have an impedance of 20 ohms with a maximum value of 60 ohms at 60 Hz.
In the Rz graph, the resistance changes to 60 ohms at low frequencies. Along the horizontal axis of frequency, along the vertical axis - resistance in a logarithmic scale.
When connected to amplifiers with different output impedances, you can see how the frequency response changes. You can see that when you connect headphones to an amplifier with an output impedance of 300 Ohms, the frequency response at 60 Hz changes to 7 dB.
The frequency range indicated on headphone boxes does not show the amplitude-frequency response, but only shows the extreme frequencies after which a decline is expected. For amplifiers that usually have a flat frequency response, limits are indicated in dB, for example -1 dB, -3 dB or another number. For example, 20Hz - 20kHz - 3dB, will mean that already at 20Hz and 20kHz the signal amplitude is 3dB lower than at frequencies around 1kHz.
Frequency analysis. frequency response
15. Save the text from the output file in the report template, having previously removed empty lines from it. Highlight in the text the results of calculating the small-signal transfer function in the analysis mode for direct current, input and output resistance (Fig. 13).
** Profile: "SCHEMATIC1-post" [ C:\OrCAD_Data\test-
* pspicefiles\schematic1\post.sim ]
****JOB STATISTICS SUMMARY
Total job time (using Solver 1) = .02
Rice. 13. Output file fragment
The text interface of the PSpise A/D program, working with *.cir and *.out files, and modeling directives are described in more detail in .
Frequency analysis. frequency response
16. Transform the diagram in accordance with paragraph 3 of the laboratory assignment. Instead of the input source, put a VAC or IAC source (in accordance with the option), set the amplitude of the variable component arbitrarily, but not equal to zero. Other sources should be excluded from the diagram.
The current source has infinite internal resistance (open circuit), and the voltage source has zero (jumper).
Since the circuit is linear, and it is necessary to remove the frequency response and phase response, the amplitude of the input influence does not play a role (within the limits of values permissible in
PSpice, for voltages and currents - 10 10 volts or amperes).
VAC and IAC are harmonic signal sources for frequency analysis and can be used for DC analysis.
17. Create a new modeling profile. 3
18. Select analysis type AC Sweep – analysis of a circuit in the frequency domain. Set the initial analysis parameters as shown in Fig. 14.
Selecting a frequency step: Linear – linear, Logarithmic – logarithmic. For a linear step, the total number of points per scale (Total Points) is indicated, for a logarithmic step the number of points per decade or octa-
wu (Points/Decade (Octave)).Start Frequency – initial frequency of analysis, cannot be equal to 0.End Frequency – final frequency of analysis.
Laboratory work No. 1. Static, frequency and timing analysis of a passive RLC circuit
Rice. 14. Simulation settings window. Setting up AC Sweep Analysis
19. Run the simulation. 2
20. Open output file ( Output File)4 find and copy the section with analysis directives to the report template (Analysis directives).
Frequency domain analysis is specified by the .AC directive.
21. Construct frequency response graphs.
The frequency response is the dependence of the modulus of the complex coefficient
The frequency transfer ratio can be defined as the ratio of the amplitudes of the input and output signals.
21.a. Open the Add Traces window. In PSpice A/D, the command Trace>Add Trace..., the Insert key or a button on the toolbar (Fig. 15).
In OrCAD 16, you can also add a graph through the context menu, called by right-clicking on an empty plot area.
Rice. 15. Calling the Add Traces window
The functions of constructing graphs and post-processing of simulation results are performed directly by a graphical post-processor
Probe built into PSpice A/D.
Laboratory work No. 1. Static, frequency and timing analysis of a passive RLC circuit Customizing the appearance of the plotting area and graphs
21.b. In the Add Traces window, using the keyboard or mouse, enter into the Trace Expression line expressions for the frequency response of all outputs (Fig. 16), as the ratio of output, input voltages (even version) or currents (odd version).
The left side of the Add Traces window lists all the currents and potentials of the nodes in your circuit. On the right side is a list of mathematical functions and connectors that Probe can apply to individual graphs.
Rice. 16. Entering graph expressions in the Add Traces window
IN result of analysis AC Sweep nodal voltages are calculated
And branch currents, which are complex quantities. In mode AC Sweep Probe supports calculations with complex numbers. Entering expressions for complex values into the Trace Expression line of the Add Traces window without using any mathematical functions or Probe operators displays the result module. If an expression is entered for a real value, for example the phase of the complex transmission coefficient, then the result may be negative. If the expression is complex, for example, the complex voltage transfer coefficient V(N1)/V(N4) - defined as the ratio of the potentials of nodes N1 and N4, then its module is displayed, which is always non-negative.
To access the real and imaginary parts of the calculated quantities, the R and IMG functions are used, respectively.
IN The Probe program also uses the ABS (absolute value) function - absolute value and similar to it M (magnitude) - module, corresponding
valid expressions: V(N1)/V(N4), M(V(N1)/V(N4)), ABS(V(N1)/V(N4)) and SQRT(PWR(R(V(N1)/ V(N4)),2)+PWR(IMG(V(N1)/V(N4)),2)) – completely equivalent
valence. The SQRT function is the square root, and the PWR function is the exponentiation, in the example given, the square.
Laboratory work No. 1. Static, frequency and timing analysis of a passive RLC circuit Customizing the appearance of the plotting area and graphs
21st century Analyze the form of the obtained frequency response, open the simulation profile settings window (Simulation Settings) and change, if necessary, the limiting frequencies of the analysis, the type of frequency step, the number of points so that the graphs take on the most informative form.
You can call the Simulation Settings window and change the simulation directives directly from the PSpice A/D program by clicking the corresponding toolbar icon (Fig. 17) or using the command Simulation>Edit Profile….
21. In the Simulation Settings window, on the Probe Windows tab check the box Last plot in the Show group (Fig. 18 ) – displays graphs for the last entered expressions.
21.d. If the simulation directive has been changed, run the simulation again.
You can start the simulation directly from the PSpice A/D program by clicking the corresponding button on the toolbar (Fig. 17) or using the command
Simulation>Run.
Rice. 17. Calling the Simulation Settings window (Edit Profile command)
and starting the simulation (Run command) from the PSpice A/D program
Rice. 18. Simulation Settings window.
Probe Window tab – setting up the display of simulation results
Laboratory work No. 1. Static, frequency and timing analysis of a passive RLC circuit Customizing the appearance of the plotting area and graphs
After each simulation, information about the expressions entered in the Trace Expression line is reset; the Show Last plot option allows you not to enter expressions again.
Customizing the appearance of the plotting area and graphs
21.e. If necessary, change the display scale along the axes (linear or logarithmic) (Fig. 19).
Rice. 19. Changes the display scale along the axes.
Opening the Axis Settings window
21.g. Remove intermediate grid lines.
Open the window for setting up grid and axes parameters (Axis Settings). Command Plot>Axis Settings..., or double-click the left mouse button in the value area of one of the axes, or select the context menu item available by right-clicking on the grid line (Settings... item) (Fig. 19).
In the Axis Settings window on the X Grid and Y Grid tabs in the Minor Grids section check the box None (Fig. 20).
21.z. Configure the display of graphs.
Open the chart properties window (Trace Properties). Right-click the graph line or icon in the line with the graph legends, X axis (Fig. 21). In the context menu that appears, select Properties….
In the Trace Properties window, change the graph display parameters: increase the thickness of graph lines, change the color and type of lines.
Repeat the steps for all graphs.
The display settings for frame and grid lines can be configured in the same way.
Laboratory work No. 1. Static, frequency and timing analysis of a passive RLC circuit Frequency analysis. FCHH
The thickness of the lines affects the quality of printing and perception. Line colors should be selected that, when printed in black and white, provide acceptable clarity and contrast against a white background.
Rice. 20. Axis Settings window. Setting the display of intermediate grid lines
Rice. 21. Setting the appearance of graphs
21.i. Save frequency response graphs. Command Window>Copy to Clipboard (save to clipboard), in the window that opens, in the Foreground section, check the box change white to black (change white with black), click OK (Fig. 22). Paste the picture from the clipboard into the report template (Ctrl+V
or Shift+Ins).
The construction area, including axes, grid, graphs, axes labels, legend and text notes is copied to the buffer (Fig. 23). The size of the image in the buffer depends on the actual size of the construction area at the time of copying.