[Jascheck-DD]

Hello Guys,

after playing around whith the REW programm for some time and doing some private research (i work as a professional acoustics engenieer in germany, but not in the Audio buiseness) while making all kinds of usefull and unuseful measurements i came to a problem really perplexes me.

I found some old coaxial drivers (caraudio) and decided to determin their Thiele Small Parameters. Maybe to build some monitors out of them .. sometimes. After some struggeling whith the cable setup (microphone inpput on Noteboocks are not to bee used at all) i got a pretty well running set up.
I use a Tascam US-1641 (now US-1800) interface.
The Signal comes out of the headphones output and goes in two Line inputs (-10) (not so good at high frequencies but useful). As sense resistor i use one 100 Ohm.

I made some measurements whith a 48kHz and 96kHz settings. And se results seem pretty reasonable.

http://img109.imageshack.us/img109/1909/coaximpedance.jpg

I use some kind of modelling clay (TerrostatX,whyich is often used for sealing leakages at acoustical purposes) to put an even load of 5µt (microtons=gramms :bigsmile.

My problem is, that when I enter the Data accordingly into the TSP-Tool it does no calculation at all. :hissyfit:

I dont know what happened. As I was testing the whole thing in advance I made some unusefull measuremnts whith varaiing preampgain fom the output. But it calculated some parameters anyway. It even warned me that my DC-resistance i entered would not match the measurements and i shoult enter somehat whith 1 ohm less, the error would be corrected ... and so on.

But since i perfected my setup ... nothing...!! I hit the calculate button, he window blinks once and thats it. No results.:huh:

I hope someone has done similar measurements too and will try to help me.

I have some theories for possible errors which some I already tried out.

1. high frequenzy slope (due to increasing inductivity of the coil) does not match lumped elements model (since the driver is a coax it has an internal crossover network) --> action: making measurement only up to 2kHz (should suffice for basic parameter) ---> result: no effect

2. problems whith the programm --> action: reinstalling, installing on different pc ---> result: no effect

3. low gain on referece channel in while testing the levels before a measurement i get the warning the reference channel (behind the 100 ohms) is below -39.9 dBFS, but since the results are good (in resonance it is about -10 - -15 dbFS) the measurements go whithout any warnings ---> action: could be reduced by changing the SensResistor to a lower Value ... that would be my next step but therefor i need to resolder my network :doh:

I am open to any suggestions.

Thx in advance

Jaschek-DD

 

[JohnM]

You need 5 posts to include links, you can use the post padding thread to get the required post count, then allow an hour or so.

If you attach the mdat file containing the measurements I'll investigate what is going on.

 

[Jascheck-DD]

Hi John,

thanks for the advice.
I attached the file whith the measurement using 48kHz.

http://www.hometheatershack.com/forums/attachment.php?attachmentid=35597&stc=1&d=1332970318

I repeated the measurement. the second as you can see is only to 2kHz.
The parameters of the speakers are:
Rdc ~ 3,5 Ohm
A - 129,6 cm²

load was 5 g = 5 µt

I tried also differnt variations of parameters but got still no result. :huh:

Thanks for taking a look at the data

 

[JohnM]

Only just noticed that the impedance figures appear to be offset by approx 100 ohms - that suggest the left and right connections are the wrong way around, you can either swap them over or change the input channel selection in the REW soundcard preferences.

 

[Jascheck-DD]

Many Thanks John,

I switched the main and the reference recording channel and now it works perfectly.

Dont know why I havent thougt of it myself. Maybe I got all confused rearranging the cable tree so often. :innocent:

Once again many thanks.

Please let me ask you just one mor thing:

I did some measurements and repeated them whith variations in gain (~10dB), different driverpositions (horizontal/vertical) and whith different preloading (resulting in different coiltemperature and bearing stiffness).

My measurements show all some slight variations. Now I tried to average the Different measurements and get to get some averaged TSP.

But unfortunately it wont calculate from averaged curves. Is there any way to do it? Or can i only average the exported files by hand.

The same problem I encountered while measuring some RT60s. I wanted to average the Reveberation of some different points in the room, but couldnt find out how to do it whith REW.

I hope I havent overlooked the obvious once again.

Greeting Jascheck

 

[JohnM]

You could average pairs of measurements using the Trace Arithmetic feature - click the cog icon on the graph that shows all the measurements and you will see the trace arithmetic controls, (A+B)/2 will produce an average that can be used with the TS parameter calculation tool. For more than 2 measurements you need to proceed in pairs, e.g. average A and B, then C and D, then average the two averages (consequently need to have 2 or 4 or 8 or 16 etc measurements to average and it will be somewhat tedious for more than a small number).

The same approach can be used with room measurements, but you must ensure that the measurements you average are time aligned or you can get quite unexpected results, since the arithmetic takes account of both magnitude and phase. One way to produce traces that should be aligned is to use the control to estimate the IR delay and then click the button to remove that delay from the measurement.