Much thanks to Bob Riley for the wiring diagram and layout / drilling guide images!
Here is a picture of the completed jig.
PARTS LIST ITEM RS STOCK DESCRIPTION ---- -------- ----------- B1 270-1804 Black box, dimensions 2" x 6" x 1 1/8" SW1, SW4 275-626 DPDT micromini toggle switch SW2 275-625 SPDT micromini toggle switch SW3 275-325 SPDT center-off mini toggle switch R1 271-120 8 ohm non-inductive 20W resistor R2, R3 271-1126 10k 1/2W resistors R4 thru R7 271-1121 2.2k 1/2W resistors R8 thru R11 271-1103 22 ohm 1/2W resistors J1 thru J4 274-346 RCA jacks BP1 thru BP4 274-662 5-way binding posts (two pair needed) DIODES 276-041 Red LEDs
Notes:
Here is a picture that might help you correctly position the power resistor. There is actually room under it where I mounted some of the calibration reference resistor. Do a trial fit after you mount each resistor so you are sure that everything will go together at the end without bulging the box out.
Here is the schematic for the jig, along with a brief description of what each switch, binding post, and jack on the jig does:
Here is a suggested layout of the switches, jacks, and binding posts. If you don't use these guides I can't guarantee that R1 will fit in the box!
Here is a suggested wiring diagram. This is perhaps a more intuitive look at the jig circuit, and should help you decide how to position components and bus wires. Note that the protection LEDs aren't shown here.
Key: U = up
M = middle
D = down
X = don't care
MODE SW2 SW3 SW4
---- --- --- ---
CAL1 D D D
CAL2 U D D
LOOP X U D
IMP X M D
MIC X U U
The CAL1 and CAL2 modes place 5.5 ohm and 11 ohm resistors,
respectively, across the test terminals for calibration
purposes. The LOOP mode is used to calibrate the software in
order to compensate for the frequency response of the setup,
and input and output of the soundcard. IMP mode is used to
determine the impedance of a component connected to the test
terminals. MIC mode selects the line-level microphone
input.
MODE OHMS ---- ---- CAL1 5.5 -- record exact reading and write it on the jig! CAL2 11 -- record exact reading and write it on the jig! LOOP INFINITY IMP INFINITY MIC INFINITYNow place the ohm meter test leads on BP1 and BP3, the two positive or red binding posts. It should read:
MODE OHMS ---- ---- CAL1 8.0 -- record exact reading and write it on the jig! CAL2 8.0 -- should read the same as CAL1 LOOP 0.0 IMP 8.0 -- should read the same as CAL1 MIC 0.0Check the rest of the wiring until you are certain that the jig is wired correctly.
Open a second copy of Windows' Volume Control and click on menu item Options, Properties... then click on the radio button Recording. This should change its name to Recording Control. Again, place a check in all of the boxes so that all of the controls are displayed. Make sure that only Recording and Line-In have a check mark in their Select boxes. Now hide all controls except for Recording and Line-In and make sure that the balance slider is centered for these two controls. Center - and if possible disable - any Advanced controls. Place the Recording Control at the upper right of your screen. Finally, set the Line-In slider to full-up. My Recording slider is locked at mid-position for some reason. Perhaps this is normal.
Ok, now that you have been sufficiently softened up by that brow-beating, it's time to check the soundcard input and output levels. Let's create a signal, and then play it and record it. This will allow us to set and verify the input and output levels. From the SW menu, select Resource, New, Signal and name it "junk". This opens a signal which should be a sine wave by default. In SW you must create a signal or a driver and have it open in order to do any recording or to perform any impedance measurements. (This, I believe, is because SW employs a common - yet changing, depending on what type of window is active - menu area for all elements in the design tree. A better design might be to have a separate and specific sub-menu for each child window.). You should now see the name of the signal in the design tree - that thing that looks like a directory structure in the "Root" window. Now right-click in the chart area of the signal window and select Properties... from the pop-up dialog. Pick the Sine tab and set the Frequency to "500" Hz. The Phase should display "0" degrees. Click the OK button.
("Opening" a signal on the design tree is not as easy as double-clicking on it. You must first select it with the left mouse button, then click on it with the right mouse button, then click on the word Open in the pop-up dialog. Kind of complicated, but there it is. Note also that the menu items change whenever a signal or driver is open. If you can't find what you want in the menu, you probably don't have the right thing open. These must be the two queerest "features" of SW, but once you are aware of them you have the keys to the city. While we're at it, a signal window displays a time-domain signal, while a driver window displays a frequency response curve.)
Place the jig in LOOP mode by adjusting SW2, SW3, and SW4. Make sure the signal window is active (if not click on it) and select the menu item Sound, Record... and up should pop a dialog box. In the Output section, the No Output box should not be checked, the Volume entry area should have "100" in it, and the Channels entry area should have "Both" in it. In the Input area, the Calibration entry area should have "Left" in it, and the Data window should have "Right" in it. In the Time area, set the Play time to "0.01" Sec and the Record time to "0.2" Sec. The Type radio button should be "Frequency". Click the OK button and recording should start. If you have a speaker hooked up to the jig (across BP3 and BP4) you will hear a brief 500Hz "blip".
Now look at the design tree. It should have some new time domain measurements (denoted by a little page with the letter 't' on it). Open the one named "junk.in.l" and resize and move it so that it covers the lower left quarter of the SW work area. Right-click on it and chose Chart Properties... and a dialog should pop up. Click on the X Axis tab and in the Scale area place "10" in the max entry area. Next click on the Y Axis tab and put the numbers "32k" and "-32k" in the Maximum and Minimum entry areas, respectively. Click the OK button. You should see about 4 1/2 cycles of a sine wave in the window. Now open up the measurement named "junk.in.r" and place it in the lower right quarter of the work area. Set its chart properties the same as you just did for the l signal.
OK, now you are ready to explore the proper level settings for your soundcard. I would recommend that you leave the Line-in level on Windows' recording control to full up, and the sound record volume in SW set to 100. I would also set SW1 to the "divide by 10" position. Now explore where the maximum setting of the Volume Control slider in Windows' volume control produces no obvious flattening of the tops of the sine waves. To do this, inch the Volume Control slider up or down a bit and select the menu item Sound, Record Again, repeating as necessary. For my Sounblaster 16 this is almost 1/2 way up and produces a +/- 20k signal at the left and right inputs as reported by SW in the "junk.in.l(dataset)" and "junk.in.r(dataset)" windows. Any higher and the output clips no matter where the Line-In slider or sound record volume in SW is set. Kind of strange. Anyway, note the optimal location of the Volume Control slider and always put it there when you are using SW.
If it isn't already open, go to the design tree and open the "junk" signal that we were using before.
With the signal window open and active, flip the switches so that the jig is in the IMP mode and place a capacitor, inductor, or resistor (you should verify that all three do in fact work) across the test terminals. Select the menu item Measure, Passive Component and this should initiate the testing. A dialog box should pop up indicating the particulars of the component under test, including DCR for inductors and dynamic resistance for capacitors. This procedure occasionally farts, producing strange results. Note that if you have the L and R reversed on line-in cable to your soundcard, inductors will read as capacitors, and vice-versa, with a negative DC or dynamic resistance. Note also that performing this procedure with the jig in one of the CAL modes should produce the calibration resistor values associated with those modes (~5.5 ohms for CAL1 and ~11 ohms for CAL2).
I believe that only components with impedances on the order of the reference resistor (nominally 8 ohms for the jig) can be measured with the the stock jig. However, this should be fine for all of the values of resistors, inductors, and capacitors that you typically run across in a crossover. Just don't expect to be able to measure a 1pF cap with this setup!
Close any signals and create a driver by selecting Resource, New, Driver, and giving it the name "trash". This should open up the driver window with the name "trash (driver)", and it should be blank at this time. Leave this window open. Hook the speaker up to the test points BP3 and BP4 and put your jig in the IMP mode.
Next, check the preferences. In SW, click on Options, Preferences..., select the Measurement tab. Make sure the Sample Rate slider all the way to the right, and that the Volume is set to 100. The Sample Size slider should be set to 32,768 for good resolution in the bass region. The box Use preemphasis in the MLS Signal area should not be checked. This is not life-or-death, but if preemphasis is used the SNR of the very highest frequencies will suffer somewhat. Click the OK button.
Now select Measure, Impedance from the menu. You should hear a burst of noise, and then SW will take some time processing the signal. Check the design tree for the new chart named "trash.Impedance" and open it up. You should see what looks like a typical impedance curve for a speaker, with peaks consistent with the type of enclosure it is mounted in.
To tidy it up, right-click in the chart area and select Chart Properties..., then click on the Data Sets tab and click on the Phase box to deselect it. Next, click on the X Axis tab and set the Minimum entry area to "20". Then, click on the Y Axis tab and set the Minimum entry area to "0" and the Maximum entry area to something reasonable like "30" or "40". Set the Major entry area to "5", then click the OK button. Finally, smoothing the data might make it easier to use. With the "trash.Impedance (dataset)" window active, select Transform, Smooth... from the menu and click the 1/3 Octave radio button, then click the OK button. Now sit back and stare at the fruits of your labors. Was that easy, or what? Go ahead and tweak that port length!
Oh well, we have to go with the way the software is I guess (though this would be a great thing to persuade the author to do in a more conventional way).
We are now faced with two options: either ignore the Amplifier Reference Response calibration, or go ahead and use it. If we use the calibration, we can use the preemphasis option on the MLS signal if we wish. Preemphasis amounts to a lowpass filtering the MLS test signal. It makes the MLS sound less harsh, and thus less irritating to those around you when you are performing frequency response tests. It's purpose is to increase the SNR of such tests for the lower frequencies. If we do not perform the calibration, we cannot use preemphasis, since the calibration in effect applies the "deemphasis". Either way, we should be able to get decent frequency response measurements from SW. You should play around with both options. The moral here is that if you are looking at frequency response curves that unexpectedly have the high frequencies rolling off like mad, then you probably have the preemphasis box checked and aren't using the calibration.
If you want to delete this calibration, simply select Options, Calibrate... and click on the None button in the Amplifier Reference Response section. If you do this, then this calibration will not influence any frequency response measurements.
I recommend that the preferences be set the same way as they were for the Amplifier Reference Response Calibration. A very important thing that we must set now is the gating time. For these types of measurements we want to gate (or truncate) the recorded signal so that no room reflections enter into our measurements. In this way we can get a anechoic frequency response in a reverberant environment. The way to calculate the gating time is the following: measure the distance that any reflected sound wave would have to travel and subtract it from the direct path. Since sound travels at approximately 1 foot per milisecond, we set the gating time in milliseconds for this difference in feet. For example, if the the distance between the speaker and microphone is 1 meter, this approximately 3 feet. If the speaker and microphone are 1 meter off of the floor, this produces a travel distance for the first reflection of approximately 2 meters, or 6 feet. subtracting 3 feet from 6 gives us a total of 3 feet difference, so we gate the signal at 3 miliseconds.
Another more "seat-of-the-pants" way to set the gate time is to examine the impulse response for obvious reflections, and then set the gate so that these are not included in the FFT data. You can play with this on your own if you like, SW will easily record, calculate, and display the impulse response based on a test performed with the MLS signal. Do this by having the "trash" driver window open and active and selecting Measure, Pulse response from the menu and the test will be performed. You will hear a burst of noise. Make sure that the recorded signal neither clips nor is too low. On the design tree open "trash.Pulse" and right-click on the chart area. Click on Chart Properties... and then on the X Axis tab. Put the value "10" in the Maximum data entry area, then click on the OK button. I don't quite know what constitutes a clear reflection in this mush, but it is interesting to look at.
Note that the width of this gate will influence our low frequency resolution. SW takes this into account automatically by limiting the lower frequency displayed on any frequency response chart based on the time markers. You can calculate this lower limit by taking the inverse of the gate time. For example with our 3 ms gate time above we get a lower frequency of: 1 / 3E-3 = 333Hz. Now you know why John Whittaker spends so much time in the gym!
Let's set that gate time, shall we? Choose the menu item Options, Preferences..., and pick the Markers tab. Click the Visible box in the Time section so that it has a check mark in it. Also in the time section, make sure that there is "0" msec in the "1" marker entry box, and enter "3.000" msec in the "2" marker entry box. Then click the OK button. Any signal window that is open now should have red markers at 0 and 3 msec.
Now select Measure, Frequency Response, On Axis, and watch the recording level. You should hear a burst of noise and the recording level should neither clip nor be too low. You can adjust the recording level with the Line-In control in Windows' Recording Control, and adjust the level of the signal going to the speaker with SW's Options, Preferences..., Measurements tab, with the Volume entry box in the I/O area. Make sure that your microphone preamp is not overloading either.
After the test, open up the "trash.OnAxis" window which should be named "trash.OnAxis (dataset)" and take a gander at what was recorded. I usually use the Chart Properties... to hide the phase data set, as well as to lock the Y axis min and max to values that will display the data appropriately. I also usually set the Y axis major grid lines to 6 dB intervals. You can smooth the data if this window is active by selecting the menu item Transform, Smooth..., clicking on the 1/3 octave radio button, and then clicking OK. If you want to repeat the measurement, the "trash (driver)" window must be open and active. I recommend you perform this test many times repeatedly, and at different input levels, until you get a repeatable result that makes sense. Another thing to try is placing the microphone at 0.5 meters from the speaker and seeing what you get there.
After you do both the 1 meter and the close-miked responses, these two curves should be spliced together into a composite response. Audua has a way to do this, but I haven't quite figured it out yet.
> I would like to ask you the following:
> About MICROPHONES
> Im planning to use a normal mic with it - must it meet any specs? Or do
> I have to buy a an electret like the Panasonic WM60xx? If this is the case
> can it be the uncalibrated version? or do I need the cal curves and / or the
> mic preamp?
The jig is designed to plug a preamplified mic into. You could
probably just use a 3V to 9V supply (2 AA bats or a 9V bat)
along with a capacitor and resistor in order to power the
Panasonic mic cartridge. They are very cheap from Digikey (<$3
in single quantities). Engaging ASCII-CAD:
power switch
/
+-----o o-----+
| |
(+) 5k
3Vbatt |
(-) +------- 10uF -------(+)
| |
| (+)
| Panasonic cartridge To mic input on jig
| (-)
| |
+-------------+--------------------(gnd)
If you use the mic input on the soundcard, you might have to
monkey around more with the mixer controls before you make an
acoustic measurement. I'm not sure if the soundcard will
provide the "phantom" voltage necessary to power an electret
cartridge such as the Panasonic.
The reason the Panasonic cartridges are so nice is that they
are typically flat to within a couple of dB or so over the entire
20Hz-20kHz range and don't require much in the way of
calibration. For much work, calibration will probably be
unnecessary if you go with this cartridge.
> My soundcard is a Soundblaster SB16 that I know is compatible with the
> program but I don't have the specs (input impedance, etc) of the card but
> maybe you can help me with this also.
I use a value of 25k when calibrating the jig; this figure is
not that important as long as it is >10k or so. My soundcard
seems to roll off around 10kHz, not a huge deal when designing
xovers.
Sorry to hear that Audua is not addressing your requests. Hope
that doesn't mean that the whole Speaker Workshop thing will
die on the vine :-( The list seems to be the perfect place to do their
beta testing, don't you think?
-------------------------------------------------------------------
< - the following was a very helpful suggestion - edw >
I had a problem with Speaker Workshop 0.61:
I placed my mic in front of a driver and measured the
frequency response using an MLS signal.
The result astonished me because I know this driver
and have a reference chart for it. The amplitude
decreased dramatically with rising frequency.
After some searching within the program's setup
windows I found a 'Preemphasis' check box within
the 'Options'/'Preferences' pulldown menu.
After deactivating preemphasis the problem was gone.
It seems that Speaker Workshop doesn't compensate =
at the input for the filtering of the output signal.
(The preemphasis lacks the deemphasis).
I did *not* perform any calibration before starting
with my measurements. Just because my mic has a flat
response and the sound card has it as well; checked
that with other equipment.
Probably usually no one runs into problems with
that, because he or she performs calibration.
However, I think it is a flaw of the program to =
compensate for preemphasis by means of calibration
data. Someone may toggle this switch later on, after
having performed some measurements with the other position
of the switch.
Whether a response was measured using preemphasis or =
not should be stored in the measurement data file.
---------------------------------------------------------------------------
> I'm working on putting together your impedence jig for Speaker Workshop
> and have a few comments, questions, etc...
>
> Comment #1: 1/4" holes aren't big enough for the binding posts you
> recommend. Probably want to update that on the drawing.
The 1/4" holes shown on the drilling guide are the base hole
size. After drilling, the holes then must be elongated with a
round file in order to mount the binding posts. This keeps the
posts from "spinning" when tightening or loosening them.
> Comment #2: The correct part # for the 22ohm resistors is 271-1103, not
> 271-1130 as listed
Oops! Thanks for pointing that out. The parts list is now
corrected.
> Comment #3: I can't get everything to fir in that box without the side
> bulging out!
The 8 ohm power resistor has to lay down the middle of the box
with sw1, sw2, and sw3 on one side, and j1 and bp3 and bp4 on
the other. You may have to rotate it a bit about its main axis
also. It should fit if you used the drilling guide and the
same exact switches, as well as the black plastic top (not the
aluminum top).
--------------------------------------------------------------------
> Question: I'm trying to use SW on my Toshiba laptop, which does check
> out to have full duplex capability. I'm having problems getting the jig
> calibrated. I'm getting values like 1.5k and 500 ohms, etc for the
> resistors. The values change significantly as I change the Volume
> setting. The catch is that, for some reason, the volume control doesn't
> offer a recording option to view the level. I can only view the line-in
> level. Therefore, I can't match the levels correctly. (Incidently, I
> can get the recording level meter on my desktop, so I know what you're
> talking about.) Any thoughts? I verified that I'm using the same
> version of the volume control app.
If you can't see the recording level display in some way, using
the program will be somewhat more difficult. You might be able
to play with the line-in level until you get repeatable
results.
Try this: create a new signal *not* in the System folder and open
it. Make it a 1kHz sine wave by left clicking on the open
signal and using the properties item. Use the Sound menu item
to record the signal. After recording, several signals are
created around the original with extensions *.in.l and *.in.r
with * being the original signal name. Open these to see if
the signal is either too small, or so large that it is
clipping. The largest values are +/-32,000 since these are 16
bit numbers. Keep recording and ajusting until you get a good
large signal.
If you don't get any input signals, try hooking your line-in to
the line-out directly on your soundcard and performing the
recording process again. If you get signals with this direct
connection, then there is either a wiring error in your jig, or
one of your cables to/from it is bad.
---------------------------------------------------------------------
> Eric as a BassList lurker and frustrated Speaker Workshop user I have
> read your tutorial on the Speaker Workshop and tonight went out and
> bought all the goodies to build the test box.
> I have spent several unfruitful
> hours trying to get any kind of intelligible response using a reference
> resistor and an 8 ohm 5.5" midwoofer to no avail.
Same here, I couldn't get anything to work at first.
> The program truly
> looks like it is exactly what I need to help me improve my skills at DIY
> speaker building and add a touch of science to what I am doing.
It's a pretty good program, especially considering the price!
> The problem I seemed to have was a lack of any sound from the speaker at
> all. I am no rocket scientist, though do
> have a math and science background so can understand the basics of what
> we are trying to do.
>
> I wonder if the sound card I have is perhaps not powered or does not have
> the power to perform what is needed by the program. It is an Ensoniq
> Soundscape VIVO90 Plug and Play sound card, hooked up to a set of Altec
> Lansing (ACS45W/Portable 3W) powered multimedia speakers w/ sub-woofer.
I had an old original Ensoniq Sounscape, and it had *no*
internal amplifier (strictly line-out). I suspect your card is
the same.
> Coming out of the back of the sound card is a mini plug male to male
> cable that runs over to the sub-woofer. Will I need to take a similar
> cable and strip it to send a left channel to the jig and leave the right
> channel un hooked?
Yes, but you can probably buy a cable ready-made from a
computer store. Those 1/8" mini plug to RCA jack cables are
pretty standard. You would run this to an external amplifier,
and then run the speaker output from the amp (just one of the
channels) to the jig.
> Likewise the line in is only a single mini plug. the
> Wizard tells me the sound card is full duplex so should not be a problem
> from that standpoint.
This should be a stereo input (I'm almost certain that it is).
Use the same type of cable as above here to interface the jig
to the soundcard line-in.
> P.S. a number transposition occurred on your part number in the Parts
> List for the 22 ohm 1/2W resistors, the correct number is 271-1103 (the
> 271-1130 is a 47 ohm resistor).
> Eric thanks for taking the time to post the interesting info on your web
> pages, I may attempt to do the RS Digital SPL mods also, but first Audua.
Whoops! Sorry about that. I think someone already brought
this to my attention, but I'll check it out.
----------------------------------------------------------------------