The World of Computer Software/World_Of_Computer_Software-02-385-Vol-1of3.iso/t/t_s_wk1.zip/HOWTO.TXT

Text File | 1992-03-16 | 16KB | 343 lines

              THIEL-SMALL PARAMETER SPREADSHEET INSTRUCTIONS

HISTORY:

This procedure and spreadsheets have been evolving for more than
12 years.  The proceedure is based on Ashley & Swan's 1969 AES
paper "Experimental Determination of Low-Frequency Loudspeaker
Parameters".  Numerous refinements have been added over the years.
Most of the improvements came from Small's work, a few from unknown
sources, and a couple from me.

The arithmetic involved is straightforward but lengthy and tedious.
By automating the calculations, mathematical errors are eliminated.
Additionally, the spreadsheet organizes the gathering of input data,
makes all the unit conversions, and prints the results.

ACCURACY:

All input data must be accurately measured for the results to be
valid.  Frequencies, resistances, diameters and mass should be
measured to the precision recommended for the test equipment.  Once
the user becomes proficient with the technique, accurate results can
be produced in as little as 2 minutes per driver.

The beauty of this procedure is that it uses general purpose test
equipment rather than exotic, and expensive loudspeaker specific equip-
ment.

EQUIPMENT REQUIRED:

     Stable sine wave generator that goes from 2 or 3 Hz to 20000 Hz.
     (10 Hz to 200 Hz will work for 95% of all woofers)

     A frequency counter with 0.1 Hz resolution or better.
     (Don't even think about using the oscillator's dial)

     Digital ohm meter with .1 ohm resolution or better.

     Oscilloscope with balanced vertical input, good horiz sensitivity
     (AF "interstage" transformer can eliminate the bal input reqmt)

     2 AC milivolt meters -- must have identical low freq rolloff.

     An 18" steel rule calibrated in .01 inch increments.

     A gram scale.  0.1 gm resolution. (not a postal scale!)

     A power amplifier capable of > 20 volts rms output.  (50 watts)

     A precision power resistor - 10 ohms, 20 watts.

     Some permanently stick putty.  About 50 gms.

     Test boxes -- not needed if "delta mass" compliance method used.

     Means of suspending the driver with cone in vertical plane.

     An attenuator if sine wave generator does not have one.

     A computer with Lotus 1-2-3 or ".WK1" compatible spreadsheet.

CONFIGURATION:

One of the two AC milivolt meters must be "floating".  I.E. not
grounded.  Use a grounded plug adapter to "float" the power plug or
power it from an isolation transformer.  Be sure the meter's case does
not touch other equipment or metal parts of the work bench.  The float-
ing meter will be used to indicate |Z|.

The scope should have a balanced vertical input.  Otherwise, use a
high quality "interstage" transformer to isolate the scopes vertical
input.  This transformer should be a 1:1 isolation transformer with
impedance in the 5 K ohm range.

Place the scope, oscillator, amplifier, and meters close together so
that you can reach all the controls and see the scope and meters all
at the same time.  You will be adjusting level, and frequency while
simultaneously watching the scope and both meters.

Connect the frequency counter to the oscillator.  If the oscillator
has a square wave output, attach the counter to it.  Otherwise use the
sinewave output.

Hook the audio generator output to the amplifier input.  (Through the
attenuator if you are using one.)

Prepare two test leads.  Put alligator clips at one end of each and
strip and tin the other ends.  Length should be about 3 feet.  Attach
the tinned end of one lead to the (+) amplifier output.  Run a wire
from the (+) amplifier output to the input of the "|Z|" meter -- the
one that is floating.  Run a wire from the (+) amplifier terminal to
one of the scopes floating vertical inputs.

Run a wire from the "|Z|" meters "ground" to the hot input of the "I"
meter.  Connect the "I" meters ground to the amplifiers ground.  Con-
nect the precision 10 ohm power resistor across the "I" meters ter-
minals.

Hook the junction of the "|Z|" meter (-), "I" meter (+), and power
resistor to the scope's other vertical input.  Run a wire from this
vertical input over to the horizontal input.  Connect the other test
lead to this point as well.

Attach a woofer to the two test leads.  Set the volt meters to the 1
volt scale.  Turn oscillator and amplifier all the way down and power
up everything.  Set the scope for X-Y operation.

Here is a description of the hook-up using the SPICE node listing for-
mat.  Note, it will not run in spice.  Its only purpose is to unam-
biguously describe the hook-up to users who are familiar with SPICE.

Vgen      1 0 ; audio generator
Counter   1 0 ; counter w 3 Hz low freq cutoff, .1 Hz resoln
*
*      common input output
Atten     0     1     2        ; attenuator (if used)
*
*      common input  output
Pamp      0     2     3        ; Power amplifier 50 watt or more
*
Meter1    3 4  ; "Z" meter, must have "floating ground".
Meter2    4 0  ; "I" monitor meter.  Can have grounded chassis.
Rsense    4 0  ; 10 ohm 20 watt 1% current sense resistor
Spkr      3 4  ; driver under test
*     scope connections
*     Vin1  Vin2  VinGnd  Hin  HGnd
Scope  3     4      0      4    0 ;  Balanced V in, X-Y capable scope

When everything is working you should see a lissajous pattern on the
scope.  The trace will look like a diagonal line, a figure 8, or an
oval.  As you tune through driver resonance, you will see the oval col-
lapse into a diagonal line and then expand out again on the other side
of resonance.

Start your spreadsheet program and load 0MASS_TS.WK1 or 0_BOX_TS.WK1.
Be sure your computers clock is set.  Both spreadsheets include an
automatic date stamp that reads the internal clock.  As soon as the
spreadsheet loads, the macro will take over and present you with a
custom control bar.  Enter your initials, then use Ctrl Break to stop
the macro.  Use /wgpd to unprotect the worksheet.  Go to A74 and enter
the reset code for your printer.  Consult your printer manual, or try
||\027@ for dot matrix printers, or ||\027E for laser printers.  Hit
"Home" then move the cursor back to the right of "TYPE: ", and turn
protection back on with /wgpe, then save the file under its original
name.  Use /fr to re-load the spreadsheet and try out the print routine
to make sure your print code is working properly.  If you have difficulty
with the printer codes, simply put the cursor on A74 and use /wp to put a
page break into cell A74.  This works except that it leaves a blank
page in some laser printers.

In 0MASS_TS.WK1, Vas is calculated from the resonance shift due to
attachment of a known mass to the cone.  In 0_BOX_TS.WK1, Vas is calcu-
lated from the difference between free air resonance and resonance in
a box of known volume.  Each technique has its pluses and minuses.
The delta mass method is quicker, but is quite sensitive to errors in
the "DIA" measurement which must be estimated by the user.  The resonant
frequency with mass attached as well as the mass itself must be very
carefully determined.

0_BOX_TS.WK1 requires a test box of known volume and calculation of the
volume inside the cone.  This method is sensitive to air leaks.  Either
method is capable of excellent accuracy if all measurements are done
meticulously.

If you are using 0_BOX_TS.WK1, calculate the volume change that will
be caused by the drivers presence.  This volume will add to or sub-
tract from the test box volume depending on how the driver is mounted.
Use consistent units.  Cubic inches, liters, cubic meters etc.  If you
mount the driver facing out, remember to subtract the volume occupied
by the magnet as well as the cone.

Press <ENTER> to begin filling out the sheet.  You will be stepped
through the spreadsheet by the macro.

If you make a mistake entering a number simply backspace over the num-
ber and re-enter it.  If you hit the enter key before noticing
your error, you must stop execution of the macro by using control-
break.  You can start over, or simply make the correction
and then complete the input portion of the spreadsheet manually.
To print the results, manually activate the print macro using Alt P.

Once you have filled out the sheet, the command bar will appear.  High-
light PRINT and hit <ENTER> to print the filled out worksheet.
"PRINT" prints 2 pages.  The first page is the full spreadsheet with
all input and calculated results.  The second contains just the key
parameters printed in a small area.  This area can be snipped out and
attached to the driver for reference.

STEP BY STEP INSTRUCTIONS:

1.   Press <ENTER> to select "BEGIN".

2.   Fill in the type number and hit enter.  Likewise, fill in the
     sample number, and the manufacturer names.

3.   Measure the effective piston diameter which includes part of the
     surround.  If the surround is foam measure DIA from the top of
     the roll to the top of the roll.  If the surround is rubber
     measure from approximately 1/2 way up the side of the roll to the
     corresponding point on the opposite side.  Enter this dimension
     in decimal inches.

4.   Measure the dc resistance of the voice coil and enter it.

5.   Suspend the driver in free air with the cone in the vertical
     plane.  Note: a "hangmans noose" works great for this -- just put
     it around the driver's magnet.

6.   Connect the test equipment.  Increase the drive level while watch-
     ing the lissajous pattern on the scope.  Adjust the frequency of
     the oscillator to resonance.  The loop will close to a straight
     line at resonance.  Continue increasing level and readjusting to
     keep the loop closed.  At larger excursions, the loop will look
     more like a script "figure 8."

     Set the drive level for the highest "1" reading on the "I"
     monitor that still shows good linearity of the lissajous pattern.
     The "1" on the "I" monitor represents the current through the
     driver.  It will be 1 mA, 3.2 mA, 10 mA, 32 mA, 100 mA, 320 mA,
     or 1000 mA, etc.  Do the test at one of these levels.  100 mA or
     320 mA is suitable for most woofers.  1 volt across 10 ohms is
     100 mA, or on the next scale, "1" will actually be 320 mA.

7.   Attach enough mass to the cone so that the resonant frequency is
     decreased substantially.  Enter the mass in grams at "DELTA
     MASS".  Tune for a diagonal line, check the frequency,  and enter
     the resonant frequency at "RESONANCE W MASS".  Remove the mass.

8.   Weigh out the mass for the fsb measurement.  The required mass
     appears to the right of the cursor.  Attach this mass to the
     cone.  Keep the current constant (at "1"), and find the new
     resonant frequency where the loop closes.  Enter it at "Fsb".

9.   Remove the fsb mass and retune to resonance.  Double check the
     current on the "I" monitor.  Make sure it is 1.00.  Enter the fre-
     quency at "Fsa".

10.  Read the impedance magnitude from the "Z" meter.  Enter this at
     "Z MAG @ Fsa".  When taking this measurement, note if the scales
     for the "I" meter, and "Z" meter are the same.  If they are then
     the "Z" meter will be 30 ohms full scale.  If the "Z" meter is on
     the next higher scale, then full scale will be 100 ohms,  If it
     is 2 scales higher then full scale will be 300 ohms.

11.  The spreadsheet will calculate the magnitude of the -3 dB points
     on the impedance curve and wait for you to enter the lower of the
     two frequencies where the impedance magnitude is equal to this
     value.  Keep the current constant, tune through resonance and
     find the other frequency where |Z| = this number.  Then enter the
     upper frequency.

12.  The spreadsheet will now calculate the Thiel-Small and mechanical
     parameters.

13.  Highlight "PRINT" and hit <ENTER>.  The second page of the prin-
     tout has the key parameters printed in a small area.  Cut this
     out and attach it to the driver.

General advice:

Practice reading impedances from the "Z" meter.  Start with several
fixed resistors in the 20 to 100 ohm range.  Resistors simplify the
learning process because their resistance is not a function of fre-
quency.  Put a 22 ohm resistor in place of the woofer.  Set the cur-
rent for a "1" reading on the current monitor.  Look at the "Z" meter
and notice that the pointer is at 2.2 on the 3 scale.  (22 ohms).  Try
a higher resistance.  Perhaps 50 ohms.  Increase level until you have
"1" on the current monitor.  Click the "Z" meter range switch up to
the next range, and notice that the pointer indicates 5 on the 0-10
scale, or 50 ohms.  Try clicking both meters down 1 range.  Readjust
level for "1" on the current monitor.  Notice that the correct resis-
tance is still in the same spot on the "Z" meter scale.

Remember, your hand and the attenuator function as a "constant current
source".  At first you will find it difficult and frustrating coor-
dinating one hand on the attenuator and the other on the oscillator
while trying to find a frequency where current and impedance are what
you want.  But after a while, this all becomes second nature and much
more efficient than it is the first time.

It is a good idea to save backups of these spreadsheets.  Store them
in the directory with Lotus using some extension other than WK1, or
keep them on a write protected floppy disk.  That way the original
blank spreadsheet can be restored if disaster befalls your working
copy.

When using the "delta mass" method to find Vas, first measure fsa,
then stick on enough dmass to drop fm to about 1/2 fsa.  Measure fm
carefully and weigh dmass carefully.  DIA is critical with this
method.

With small drivers, the box method often gives better Vas results.
The larger the ratio of surround area to cone area, the more difficult
it is to correctly estimate DIA.  The drawbacks of the box method come
from uncertainty about the effective test box volume which either in-
cludes the volume of the cone, or if the driver is mounted the other
way, is reduced by the space occupied by the magnet, basket, and cone.
Air leaks also reduce the accuracy.

One way to know if Vas is in the ball park is to compare calculated
sensitivity to measured.  If the 1W-1M SPL is close to the measured
sensitivity then Vas is probably close too.  Note, the SPL calculation
is for radiation into 1/2 space.  It is a function of fsb cubed so
tiny errors in fsb can throw it way off.

If you want to use the box method, you should construct some air tight
boxes with openings to fit the drivers you expect to be testing.  5
liters, 10 liters, 15 liters, and 30 liters will permit measurements
on most woofers and open back midranges.

The macro is stored way off to the right and down from the working
area of the spreadsheet.  In fact it is at GF 5000.  This cell is
named "macro", so to examine the macro, hit F5, type in macro and hit
enter.

Lastly, do not make changes to the spreadsheet, its macro, or this
documentation.  Do not separate them.  If you discover any problems,
please let me know.

I reserve all rights to this work.  I am making it available to the
loudspeaker manufacturing and hobbyist community in the hope that it
will contribute to better sounding speakers for all.  You may dis-
tribute it so long as there is no more than a $5.00 charge to cover
the costs involved in duplication, and there are no changes made to
the spreadsheets, or the text files.  If you find this work useful,
send money.  I gladly accept donations.  I do not suggest any par-
ticular amount, nor do I offer anything beyond my most sincere thanks
in return for the donation.

Michael Thompson
Director of Engineering, KLH loudspeakers
11131 Dora St
Sun Valley, CA 91352
Tel: [phone]  Fax: [phone]

Version 1.1 March 16, 1992