Thanks WTPRO!
Thanks WTPRO!
Indeed, many thanks, WTPRO!
[So many sweeps; so little time....]
I am a bit new to this message board style and almost missed the 'page 2'. I found you guys with a google and thought I might toss in a few pointers (and something to add to the Help file).
Also, If you keep an eye on C&S Audio's web site www.woofertester.com, a 1.02 'beta' will probably be posted in a few weeks.
Comprising 4 ea. 25-Ohm, 10W wirewounds in parallel with 6 mH, 0.83 Ohm laminate-core inductor, all in series with the LE14H-1 woofer. Hidden behind the insulation, it caused me three days of agonizing doubt regarding woofer and box measurements. Note polarity reversal to make AES Standard H-1 conform to JBL (-) polarity.
[Or something....]
FWIW:Originally Posted by Zilch
When measuring the impedance of a woofer with Clio I've noticed quite a difference between sitting the woofer on a surface... any surface, and hanging it in free space. I have had it on a foam covered surface and a carpeted cement floor... they both change the results compared to true free air tests.
How long does the woofer tester take to run a sweep? I have found that if I use a lower resolution I can hold the woofer under test for the few seconds required and the results aren't too much different than from a higher resolution measurement.
Widget
About 2 minutes with 80-point resolution starting at 10 Hz.
15 - 20 seconds using low resolution.
I'm elevating them off the floor on heavy stainless plates now, so the vent is not covered....
I think I'd also use a damping pad or blu tac or something...
Hmmm... 80 points sounds kinda low to me, I'd stay with that resolution. As long as there is no surface to reflect the sound waves back toward the cone you should be OK... obviously also not blocking any vents.
If you want, for kicks, you might bring a tested woofer over and we could run a Clio sweep to do a comparison.
Widget
That would be great. Do the WT2 test at the same time in the same environment. It's highly portable and can be up and running in a minute or two.
There are several WT2 settings that will allow the tests to run faster.
1) As the pivot point is searched, the search eventually falls out when the 'search ratio min' value is reached. The default 0.1% is pretty darn small. The idea was to make an impression.
2) If you close all child windows and look under options, the WAV buffer size settings and some filtering options can be set. This can effect speed and stability. YMMV (your milage may vary).
Cautions:
Some older PC's, typically with very slow graphics, can barf if the frame size is set where the frame rate exceeds the graphics card update rate. Increase the WAV buffer size to solve.
I cant recall if the values are checked all that well (making a note to myself), so keep to powers of two 8192, 16384, 32768... And if something does go wrong, simply delete the WT2.CFG file and restart to get the defaults.
3) Set SweepLo/SweepHi to bracket the required sweep range for T/S calculations. Keep in mind that in the future Lem, Rem (enhanced T/S modeling) will be supported, so you may want to get used to 20K as the high end. The points of interest for that calculation are the zero phase point above Fs and SweepHi.
4) Step ratio: The default is 1.300 and produces pretty clean looking results. Experimentation here may produce faster times.
Hope this helps
Best regards
Keith
PS: Another new feature that will be going into version 1.02 is a point and click frequency setting with the left mouse button. After the meter settles down, that data point can be added to the current data set with the right moust button. Real nice for cleaning up choppy graphs!
Also, since I see someone measuring crossovers... has anyone noticed that the Z of the supplied test leads can vary as much as j*0.250 ohms (inductive) at 20Khz? This is caused by the area enclosed by the wire loop, which is what goes into the equation for calculating inductance. If you go check various cables and things, you may be surprised. Example plots of this effect using the supplied test leads are given in the help file (the only two plots in fact). These plots are created by compensating the WT2 with a 'zero length' test cable. Basically you plug the CalR and a shorted banana jack directly into the box when asked.
If high frequency accuracy is required, the best solution is to make a test cable out of 3-10 feet of zip cord (any kind of crap cable should work). This keeps the wire spacing constant allowing the WT2 compensation algorithm to do its thing in a consistent way. As long as you compensate the cable, these effects will be nulled. If you compensate properly and do an arbitrary sweep on the calibration resistor you should get a flat line for both Z and phase from 1-20K. You should be looking for +/-10 millohms and +/-0.1 degrees flatness or better.
Hello Keith
I don't have a testor yet but plan too. Thanks for the great information you have supplied.
Rob
Comprising 4 ea. 25-Ohm, 10W wirewounds in parallel with 6 mH, 0.83 Ohm laminate-core inductor, all in series with the LE14H-1 woofer. Hidden behind the insulation, it caused me three days of agonizing doubt regarding woofer and box measurements. Note polarity reversal to make AES Standard H-1 conform to JBL (-) polarity.
[Or something....]You certainly don't need all that junk in your passive sub. That's what active filters or AVR's are for.
While I am a speed freak and will check the settings you recommend I also remember the days of using 50 pound HP vacuum tube test gear and graph paper before we got a Urei plotter. It's hilarious to be able to pick up this little doohicky and the CDROM and take them anywhere I feel like taking them.
Thanks for the tips Keith!
From the looks of that crossover sweep, I'd say we're getting good results.
Perhaps someone can compare it to sim predicted performance....
Here's the voltage drive with an 8 ohm dummy load.
Hello Giskard
The WT2 banana jack outputs are connected to a current source that is *not* ground referenced, which is why you dont want to connect it to something that is grounded. For example the ground clip from most oscilloscopes will be (eventually) connected to OSHA ground... AC wall ground... and maybe PC ground (laptops might be floating).
The mini-jack output is a voltage output with a 470 ohm series resistor for protection (simple, but it works). This *is* ground referenced to the PC ground. My only suggestion here is to check the PC grounds with respect to whatever else you are doing. PC's are notoriously bad for such things.
If I am guessing your intentions properly, you want to know the speaker attenuation for a given frequency. The WT2 cant measure this directly, but it can derive the electrical characteristics of all the components. From this, you would be able to use the circuit simulation program.
However, for best results you will need to wait for the software I am working on that includes Xem and Rem. You have no doubt noticed that a speakers impedance increases above Fs, but have you considered how simplistic including (or not) Le is? Xem/Rem are frequency dependent 'correction' factors that can go into a T/S model that more or less correctly resolve Z and Phase at all frequencies.
What you may not realize is that your speaker 'load' can also be represented as an RLC tank circuit in series with Revc. All you would need is something to generate those values. Hmm, this would be easy for a speaker in free air, but vented boxes might be a bit more challenging.
You will also probably quickly point out that the WT2 does not directly measure capacitance (there is a lame reason). You need to put an R in parallel with capacitor to keep the current source from swinging to the rail. Also, the math I was dinking around with was not giving me what I considered full precision at all frequencies.
Dont let this stop you! A 1K resistor will easily keep the current source from saturating and will have minimal effect on the capacitors reactance. A quick reverse calculation of capacitive reactance and voila... you have capacitance. IE, Xc=1/(2*pi*F*C), C=1/(2*pi*X*F).
BTW, the problem was that I wanted at least 3 digits of accuracy accross a fairly wide frequency range before I considered the compensation algorithm 'nailed'.
Or... quite possibly the capacitor I was checking was crap.
Cheers,
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
Reply With Quote