Contents - Index


Instructions for special measurements not described on other pages

Index
How to...

Splice two or more frequency response curves together in a graph
This can now be easily done using Added Curves.  Just save the responses to FRD files, then use the "Added Curves">"Add" menu to bring in both curves, selecting them to have the same color.  Configure their frequency ranges so that one ends where the other starts.  If you need to readjust either of them, just select the "Added Curves" menu and click on the curve name at the bottom of the offered submenu to bring up the properties form for the Added Curve.

Is there a fast way to disable sound output when I'm using the computer to generate test signals directly from Omnimic?
Yes, simply press on the [Esc] key on your keyboard.  Press [F2] to start it again.

Point the microphone
Omnimic is a small capsule, narrow tip measurement microphone and its response is essentially omnidirectional at most frequencies.  At very high frequencies, there is about 3dB reduction in response at 13kHz, as can be easily determined by making a set of comparative measurements with the mic pointing at and then perpendicular to a high frequency speaker.  Here is a result of such a measurement (showing the effect of perpendicular to directly aimed response):



Align two speakers in "time of arrival" at a listening position
Arrange your system so that the Test Track sound plays out of both speakers you are working with, but with any other speakers silent.  Choose the Frequency Response page and play the indicated track with the microphone set at the desired position.  Set the control that is near the top to "blended" and observe the Impulse Response display.  If there is a time of arrival difference between the two speakers, you should be able to see two impulse peaks.  The one to the left (near the 0 msec point) is from the speaker with the earlier delay time, the other is for the later one.  If you have an electronic delay unit, adjust the delay applied for the earlier speaker so the peaks become superimposed.  If you have more than two speakers you wish to align together, choose the latest (the one making the right-most peak) to be the reference and delay the others to align to its peak

Optimally position speakers within a room
Speaker and listener positioning within a room is probably the most significant factor determining how a system will sound.  This can usually be best adjusted using the Frequency Response display and either the Pseudo-Noise or the Sine Sweep signal.  Set the control near the top to "All" for a display in which all echoes, reflections, and room effects are included and apply 1/3rd octave smoothing.  Start with one speaker playing at a time.  The response will normally be irregular, but try moving the speaker position or listening position for the most uniform response, particularly at lower frequencies below 200Hz.  For stereo, you will need to find two locations which can give a good stereo image together while optimizing the low frequency response.  Very often such factors are not mutually attainable.  Low frequency positions may also be optimized to minimize standing waves by use of the Bass Decay analyzer.

A strong advantage of using separate subwoofers (rather than only full range speakers) is that they allow you to optimize the mostly non-directional subwoofers for smoothest bass response and the upper frequency speakers for best stereo imaging.  While bass tends to be nondirectional, the in-room response of the bass speakers is extremely sensitive to placement.  You will almost always be able to get the smoothest overall low frequency response by using two or three subwoofers optimally placed in different parts of the room, rather than a single subwoofer - even if the single subwoofer were of much higher quality.  The room position and diversity of placements will have significantly more effect on bass frequency response than could subwoofer response flatness.  

Adjust an Equalizer

Omnimic now has a very sophisticated feature for manually or automatically configuring an equalizer, and downloading settings to MiniDSP type equalizers through its Biquad Parameters. 

When adjusting equalizers:
  • adjustment at lower frequencies should use a measurement setup similar to the one for positioning speakers (but the placement should be determined first).  Strong boosts with the equalizer should be avoided at any frequency, but response peaks can be effectively removed to any degree necessary.
  • Equalization at higher frequencies  are usually best done only in moderation, and generally adjusted with echoes removed by setting the Frequency Response analyzer to "blended" and clicking the mouse within the Impulse Response display on a point just before the first strong reflection.  Avoid sharp narrow-band adjustments on the equalizer at higher frequencies because they can cause significantly worse sound at slightly different listening positions.

    Adjusting for more than one listening position
    For theaters or other situations with multiple listeners, compromise settings must be made to provide the best sound overall for all listeners.  Begin by optimally positioning the low frequency speakers (or main, if full range) to avoid major problems at various seats, particularly those furthest from the center.  Again, use of several well-placed low frequency speakers will be most effective for giving good results for all listeners.  

    Then, with the Frequency Response analyzer set to "All" and with 1/6th octave smoothing use the Averaging features.  Make a series of measurements at all (or at least a good representative number) seats, starting with "Clear Averages" then "New Average" for the first, and clicking "More Averages" (or pressing the spacebar) once with the OmniMic placed at each position.  This will result in an overall average room frequency response, which you should save to disk or print for future reference.  You can then adjust your equalizer to provide approximately the opposite of the resulting response; the new Equalizer Configuration form makes this easy.  That is, if there were an 8dB peak (relative to response at most frequencies) in the 4kHz range, then if a flat response is desired, then the equalizer should be set for -8dB at 4kHz.  Avoid applying large increases (more than about 6dB) with the equalizer and use only moderate and broad (not sharp in frequency) settngs at frequencies above about 500Hz.

    Arrange and test whether a frequency response falls within a specified range
    The "Evaluate within" function (in the "Math" section of the Frequency Response menu will show a small rectangle at the top left of the frequency response graph that indicates whether measured responses fall within ("good") or outside ("out") of the first two curves specified in the Curves menu.  The evaluation is performed only over the frequeny range currently being displayed.  To use this to determine whether new measurements fall within (for example) 3dB of a reference measurement, first make the reference measurement and save it to disk using the "File->Save Curve" menu.  Then, in the "Curves" menu, choose "Add" and then select the file you just saved and then select an offset of 3dB on the form that appears.  Use the "Curves->Add" menu again to choose the same curve, this time with an offset of -3dB.  The two curves should show on the graph.  Next, go to "Advanced Mode" (if you aren't already there) using the checkbox provided and in the "Math" menu, select "Evaluate within".  You can select whether to allow the program to try to fit the curve by shifting it up or down by up to some maximum dB value. Finally, set the frequency range of the graph (using the yellow arrow buttons) to display the frequencies over which you wish the evaluation to be performed.

    You can alternately define the two curve files to be used for limits by generating them using a text editor (simply load an FRD file into your text editor to see the simple format.  All frequencies need not be present, but they do need to be in increasing order).  Users familiar with spreadsheets such as Microsoft Excel can use these to edit FRD data for special custom limit curves, after resaving the resulting data with an FRD extension (or saving with a .txt or .prn extension and then changing the extension to .frd).

    Measure frequency responses of Tweeters
    OmniMic normally tests loudspeakers with a full range signal that briefly passes from below 2Hz to above 20kHz.  Some speaker drivers, particularly tweeters, compression drivers, ribbons, or electrostatics cannot be safely driven at low frequencies.  When testing devices such as these, you should always drive them through series capacitors (at least) to limit low frequency drive.  You can calculate the approximate value to use as C=1,000,000/(6.2*f*Z) where f is the desired cutoff frequency, Z is the driver impedance, and C is the value in uF.  Smaller values of capacitor are safer
    .
    Using such a capacitor causes a problem, though -- it also affects the measured response.  If you test with a cutoff capacitor that acts far below where you tend to use the speaker then those effects might be ignored, but in that case you would be less protected.

    Another option is to test with a pre-filtered (but still syncrhonous with OmniMIc) signal that has a known highpass response.  Several of these are provided on the V3 CD (or DVD) test disks, some with "bass removed" and some with "bass and midrange removed".  The "bass removed" tracks roll off below 300Hz, and the "bass and midrange removed" tracks roll off below 3kHz.  The effect of the rolloffs from using these tracks can be corrected by Normalizing the measured responses (either at measurement time, or after Adding an uncorrected file) with either the "Bass Removed.frd" or the "Bass and Midrange Removed.frd" files that can be found in your C:\Users\Public\OmniMic folder. 
    IMPORTANT NOTE:  You should still use a series capacitor (perhaps something large such as 100uF) to protect your sensitive driver from transients, hum, or other signals from the setup which might cause damage.

    Note, however, that when you use those test signals, the measurements will be much more susceptible to noise below the rolloff frequencies, so you should adjust the diplayed frequency range above the lower frequency range.

    Measure sensitivity of loudspeakers
    Loudspeaker sensitivity is normally expressed as the SPL level that is sensed at 1 meter from the loudspeaker when it is driven from a voltage of 2.83Vrms.  When you are using a CD as the signal source with OmniMic, the nominal level of the swept sine signals can be determined (or set) by use of the 50Hz reference tones that are at the end of the Version 3 OmniMic Test Signal CD or DVD.

    To use these, you may want to disconnect the speaker of the channel being tested (the output voltage of quality amplifier should not be affected by whether a speaker is being driven).  Then attach a Digital Voltmeter (also known as a "DVM") to read AC Voltage across the amplifier terminals.  Be sure to configure the voltmeter for AC Volts and that the probes connect to the meter's Voltage (and not its Current!) reading terminals.  

    Turn any equalizer or tone controls off.  You can then play the 50Hz reference tone tracks and measure or set the voltage level from the amplifier.  Next, reconnect the loudspeaker, set OmniMic at the 1 meter distance from the loudspeaker and play the desired sweep track of the test CD.  Use the OmniMic software in its Frequency Response measurement mode for Sine Sweep, and the plot should then be a reading of SPL sensitivity.

    Some loudspeakers reference to a 1 watt level rather than to a voltage level.  2.83Vrms is equivalent to 1 Watt into an 8 ohm (resistive) loudspeaker.   For a 4 ohm loudspeaker at 1 Watts, the voltage to use would be 2.0Vrms, for a 16 ohm speaker use 4.0Vrms.

    getting FRD files for crossover simulations

    Designing a crossover is much easier if you can do it using simulations rather than by trial and error and physical parts.  To design via simulation, you need to have both magnitude and phase responses for the drivers you will be using.  For passive crossovers, you will also need impedance data for the drivers -- for that, an electrical impedance measurement such as the Woofer Tester 3 is ideal.  See "Determining Z-Offset" for a technique that provides accurate phase responses for crossover designs using PCD and similar programs.

    You can also do the adjustments within the PCD program using files made with OmniMic.  For this you need three measurements of each pair of drivers -- one of each driver alone, both with phase response, and one of both playing together (without a crossover, which isn't designed yet).  You should use a blocking capacitor on the tweeter to avoid hitting it with very low frequency energy, and you should also keep the test level down to the lowest level that provides a clean measurement.  The phase response can be made using OmniMic in its "Advanced Mode" and with phase enabled, but without any "added tweeter" -- this gives a true phase response measurement, though the delay between driver and microphone isn't accounted for -- and that is what Jeff's technique addresses.  For the measurement with both drivers playing, try both with the tweeter inverted and with it non-inverted to try to find a configuration which gives a null where both drivers are playing.  Then using the measured response FRD files from each of the individual drivers, simulate in PCD the results of both playing (and with the tweeter polarity as used for the combined measurement).  In PCD, vary the simulated delay between the two drivers to make the results (in dB curve shape) match as closely as possible the shape of the dB curve measured when both drivers were playing.  This delay value then will correct the relative delay between the two drivers for complete crossover design in PCD.