How to test T/S easy and cheap

[SPLEclipse]

After buying the cheap ebay subs a couple months ago, I've been trying to figure out a way to easily and cheaply measure TSPs on my own. After trying a few different programs and poring over literally thousands of pages of info, I've found a way to do it that's really inexpensive, easy to set up, and accurate. This guide is meant to be for someone who has no background in electronics whatsoever. First I'll go over the hardware. You'll need:

1) One 100ohm resistor. A 1/2 watt rating will do fine, I picked up a 1w because it was only a few cents more.
2) Some speaker wire. A foot will do.
3) A stereo (this part is important, it will have 3 bands on the connector instead of 2) 1/8" male-to-male cable. In lieu of that, you will need adapters to "make" it have two male 1/8" ends. As you can see in the pictures, that's what I've done. I got the sacrificial cable from Dollar General, you should be able to find one for under $5.
4) A way to connect everything - be it solder, wire nuts, tape, whatever. Keep in mid that it is essential they are solid connections. We're dealing with very minute measurements here, and some very tiny wires.
5) A computer with a descent sound card. If your computer is less than 5 years old, it will probably work just fine. The only things we are dealing with here are the "line in" and "line out" jacks. If you aren't sure if yours will work, try it. You won't mess anything up.
6) A test load resistor. This will be used to verify that you have things straight before measuring an actual speaker. I used a 10 ohm 1/2 watt resistor. You want to get something with a low resistance because it's more comparable to the ohm loads of a real speaker.

Optional, but very useful: Wire strippers, Alligator Clips (cheap, I recommend them), a multimeter, beer



First step is to cut up your stereo headphone cable. You will probably see three wires underneath the outer jacket: White usually indicates the left channel, Red indicates the right channel, and the other indicates the ground or shield. Some cables may have two ground wires, one for each channel. Mine did not, it was a single yellow wire. Carefully strip back the insulation on these inner wires. You'll need about 3/4" of exposed wire. Go ahead and strip the ends of your speaker wire as well.



Before you go on, chose one of the ends to be "in" (to go to the "line in" on your computer) and the other to be "out" (to go the the "line out"). Mark them clearly with tape. I can not stress enough the importance of keeping things labeled and organized, it will save you a lot of headache down the road. Now it's time to connect everything. Here's a simple diagram, courtesy of gougigaga on diyaudio.com:


The negative speaker wire (left side of speaker) wire connects to the ground. The positive speaker wire connects as shown to one side of the resistor as well as the right "in" side. The square thing on the right is the 100ohm resistor. It doesn't make a difference which way the resistor is oriented.

I printed this out and marked which wire was whit/red/yell to help me out. I recommend you do the same Note that on the "in" wire you will not connect the ground (in my case yellow wire) to anything. On the "out" wire you will not connect the right wire. NOTE: I was working too fast and reversed my channels in this diagram, so white is right and red is left. This doesn't make a difference as you can switch channels in the software. More on this later



Then connect everything. My soldering iron took a shit on me, so i used wire nuts. If you are connecting two smaller wires, you might want to put a small run of speaker wire in just to cinch everything tightly together. If you look at the left-most connector in the picture below, you'll see I did this with some spare blue speaker cable. This will not be used for anything, it's purely to create a strong mechanical connection.



Now take a brake, you deserve it. Up next is the software side of things.....

[SPLEclipse]

Software...

The program I use is called "ARTA" and can be found here:

ARTA Home

We are specifically using a program within this set of tools called "LIMP". Download and install, connect your "in" and "out" wires and open up LIMP. Do not connect anything to the ends of your speaker wire, just leave them bare. It should go without saying that you do not want to touch the ends of the speaker wire. If you have alligator clips attach them to the ends of the speaker wire now.

When you open it up, click "continue in Demo Mode". You'll need to set up a couple things. First, go to the Setup menu, and select "Audio Devices". Make sure that you "assign" the correct input and output. If you only have one sound card or onboard sound this should be easier. You might have to close out an media player to avoid an error. Select the default windows drivers and change the menu on the right to "float". Click "Ok".



Now Go back under Setup, but this time pick "Measurement". If you reversed you r/l wires like I did, you will want to change your reference channel to "left", otherwise leave it as "right". You can't break anything if this is wrong, you'll just get very strange readings. Leave the reference resistor value at 100ohm. I changed my frequency min/max to the setting you see because I'm testing a subwoofer. If you are testing a tweet/mid/fullrange you can adjust accordingly. You can leave it alone if you would like. Leave everything else as-is. Click "Ok".



Now you need to calibrate the system to account for any minute differences between channels. To do this, double check that everything is plugged in correctly and that the bare speaker wires are not touching. Click "CAL" at the top of the graph.



If your level goes red/yellow, you need to do two things: first turn the volume down on your computer. Second, adjust the gain down in the drop down menu in the calibration window labeled "Output Volume". I found that lowering the volume on my computer worked. There should be a very small difference between the channels, under 2db. Over that amount will trigger a pop-up telling you something is wrong. If that happens check your hardware jig, something is wired wrong or loose. If there is absolutely no difference (0.00 db) something is wrong as well.

Now that you have it calibrated it's almost time to test. This is where your 10ohm resistor comes into play. Hook it up to your speaker wire (polarity doesn't matter.



Now you have a choice to make. There are two methods of testing: Pink Noise and "Stepped Sine", both available from the "Gen" drop-down menu on the left of the screen, above the graph. Without getting too technical, you probably want to choose step-sine. It's more reliable if there will be any noise or air-movement in your environment. If you don't live in an anechoic environment (tip: you don't) you'll want this. Now click the red arrow. As it begins to measure, you should see two straight lines forming horizontally on the graph. One is phase (we don't care about that) and one is impedance. It should be the same value (within an ohm or so) across the graph. If it's really high or choppy, you might have reversed your "in" and "out" plugs on the computer. You might also have a problem with your jig or loose connections. If the line is straight but the reading is high, go back into your setup and adjust your resistor value from 100 to something lower. You might have to compensate for additional resistance in your sound card itself. Because the t/s specs will be a measure of comparison and not of absolute value, all we care about is having a base that's stable.



You can stop a few seconds into the test, you don't need it to run for more than 10 seconds before you can tell if everything is OK or FUBAR.

That's it for setup. Odds are you probably haven't made it this far on your first try, so sort everything out before you go further. Now I need a break, so I'll be back later with the actual testing procedure.

[SPLEclipse]

Now we test!

When you're calibrating or testing resistor, the environment around your test area doesn't matter. When dealing with actual speakers it's a big deal. It must be silent. That means to turn off any sources of air movement (fans, heat, air conditioning) and noise (talking, whispering, keyboard strokes, even noise from cars outside). There's a bit of debate on how to position you speaker when testing, so I'll share what I think: it doesn't really matter for a sub. Technically speaking, you don't want the speaker to be on it's back as gravity will be acting upon the soft parts and could skew the numbers down. The trade off is that it's more stable and will lead to less "jittery" readings (see my graph below). If you can build a frame to hold the speaker firm and on its' side, that would be the best and most accurate way to do things. However, regardless of method, if you're testing a subwoofer, I've found the difference to be less than 5%. That's a very small margin, so don't take it too seriously. Once you get the hang of it testing is a breeze, so try it a few different ways to see what happens.

If you haven't already, go ahead and calibrate and test your rig. Do this before every speaker you test to make sure your readings will be accurate. Even if it's only been a few hours since you last test DO IT ANYWAY.

Now connect your speaker wire (or in my case alligator clips) from your jig to the sub.



Make sure your test frequency boundaries are set according to what you are testing (here it's 20hz lower and 200hz upper) and press the red arrow above the graph. If you've gone through all the above steps and it all checks out, the graph should look like this. It will take a minute to complete testing.



This was with the sub on it's back. If you have the sub on it's side it might look something like this:



That's OK, but not ideal. Again, The difference between these two methods only resulted in about a 5% shift in final TSPs, so no biggie.

Now click on the "Overlay" tab and select "Set as Overlay". This will keep the first test set and let you make another run without deleting these results.

Because we are using the "mass added" method, we need to *guess what* ADD MASS! We'll do this by taping a know weight to the cone. If you have a postal scale, or something else sensitive enough, you can use anything. I don't have access to this, so I used quarters. Each quarter weighs 5.67 grams.

The United States Mint About Us

I used twelve quarters, in four stacks of three taped to the cone. Make sure you spread the weight evenly.





Position the sub in the same manner as you did for the first test. Now press the red arrow again. Wait a minute for the test to complete. If for some reason you didn't set an upper frequency bound, you can always stop the test once it's out of usable range. Notice that the line levels off after 150hz or so, so anything above that isn't useful to us. If you're testing a mid/tweet/fullrange that will change. Now you have a graph like this, with two peaks:



Click on the "Analyze" tab and select "added mass method". On the right you'll see three fields that need to be filled in. For the "Voice Coil Resistance" put in the resistance you are testing. If you have the ability to measure this with a multimeter then by all means use that data. If not just approximate it based on what you know. So if it's a single 4ohm sub put in "4". If it's a dual 4ohm sub wired in series put in "8", and so on.

In the second field you'll need to put in the diameter of the cone plus 1/3 of the surround on either side of the cone. For a 12" sub this is not "12". Guessing here will skew your results, so get a tape measure. Remember that it asks for this number in centimeters, not inches. One inch = 2.54 centimeters.

The last field is for the amount of mass you added to the cone in the second run. I used twelve quarters, so 68 grams. I figured on about 1 gram for the tape for a final number of 69g. Click "Calculate TSP" on the bottom and voila!!! You are done!!

A quick note about adding mass: In order to get the most accurate numbers, you'll need to add enough to skew the peak of the second run downward by at least 20%. If the first run showed a peak at 30hz, You'll need to get the second peak to: 30 x .2 = 6hz lower than the first, or 24hz.

Good luck!

[surreal]

Woo! Pretty in depth. I posted something just like this awhile back but It was more of a figuring it out thing then.

Hopefully people see the value in this!

[SPLEclipse]

Originally Posted by surreal

Woo! Pretty in depth. I posted something just like this awhile back but It was more of a figuring it out thing then.

Hopefully people see the value in this!

Thanks man! I really appreciate it, especially coming from you. Every resource I've found online can give you the same information, but it's usually waaay too advanced for most people. I just figured I'd share my experience in an easier-to-digest way.

[SINTORMAN]

Bookmarked this page. Will try this out for sure. You should use this with a sub that has been tested with like a woofer tester and check it's accuracy.

[TaylorFade]

Holy crap. That was the most confusing thing I have ever read.

Bad ass for the DIY'er. I'd rather just drop a 'C' note on a DAT though. Lol.

Sticky!!

[brandes.cm]

:thumbup: I like this, confusing was making the harness. Then I seen the drawing, that's all I needed.

[SPLEclipse]

Originally Posted by SINTORMAN

Bookmarked this page. Will try this out for sure. You should use this with a sub that has been tested with like a woofer tester and check it's accuracy.

I tested one of my Bravox mids because it's probably the only speaker I have with accurate TSPs on the web, lol. The results were the same as the manufacturer with only slight variations. You can also internally cross-check parameters from the same test to verify that they are correct. Honestly the software and hardware are doing the same thing regardless of what setup you use, it's the testing environment and method that make the biggest difference.

Originally Posted by TaylorFade

Holy crap. That was the most confusing thing I have ever read.

Bad ass for the DIY'er. I'd rather just drop a 'C' note on a DAT though. Lol.

Sticky!!

Thanks! I'd rather spend $5 and spend a bit of time to DIY it, lol.

Originally Posted by brandes.cm

:thumbup: I like this, confusing was making the harness. Then I seen the drawing, that's all I needed.

I went through a lot of schematics and failed jigs before I figured it out. I actually found that diagram after I was done building the jig, but it's the easiest one to understand conceptually. Here's a quick way to test the jig before you hook it up (you probably know this, but it will help others who try it):

Measure the resistance on the two outer rings of the "in" lead. It should be slightly more than the 100ohm resistor. My measurement wasn't actually this high in real life, it was extremely hard to hold this pose and take a picture.

[hawaiianmaxx]

Nice write up!!!