Crazy Impedance Curve

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm modeling an Alpair 6.2P driver in a closed box. When I pull up the impedance curve I get the graph below. Does it cause a problem when the impedance goes straight through the roof??
 

Attachments

  • alpair_6P_Impedance_Curve.jpg
    alpair_6P_Impedance_Curve.jpg
    146.8 KB · Views: 201
Account Closed
Joined 2001
Most likely you have the wrong information for Le in your driver parameters.

I was looking at the data sheet for that driver and it shows Le of 6.94 microhenries. If I plot an impedance file of a driver with Le of 6.94 millihenries it shows me something nearly identical to yours. :)

Hope that helps.

Dave.
 
Davey,

Good catch, thank you sir! You are correct, the difference between micro and milli is significant. See corrected graph below (looks much better now!).

Next time I'll do a better job of keeping my units straight!! :eek:

Thank you again for pointing my error out to me, much appreciated my friend!

bp
 

Attachments

  • Corrected_Alpair6P_Impedance.jpg
    Corrected_Alpair6P_Impedance.jpg
    160.2 KB · Views: 182
OK, but that 6.94μH is at DC, correct? Anydriver will have a rising Le with rising frequency. A plot that levels out a the DC Le at high frequency is also very wrong.
Quite true. Modelling a drivers impedance curve based on Re and Le is an exercise in futility, as the Le is not constant with frequency.

The only way to do it accurately is actually measure the impedance curve of the driver, which is not difficult if you have one already.
 
Account Closed
Joined 2001
Yeah, I have my doubts about the listed value of 6.94uH on the driver data sheet. In fact, if you look at the impedance graph on the data sheet it does show the curve rising above 500Hz. You are correct that a proper measurement of an actual driver should be made to confirm.

Anyways, the gross error at higher frequencies in his simulation that bp was seeing was certainly just a pilot error on data input.
Which, I think, is all he was really asking about. :)

Cheers,

Dave.
 
Last edited:
Quite true. Modelling a drivers impedance curve based on Re and Le is an exercise in futility, as the Le is not constant with frequency.

The only way to do it accurately is actually measure the impedance curve of the driver, which is not difficult if you have one already.

Indeedy- I tend to look at the Le at a couple points on the rising portion of the curve, and select a value based upon the averaging of Re + Le that most closely matches the curve. Given that most of the increasing inductance in all but fullrange drivers (which is of course what we're discussing here) is above the passband anyway, this tends to make a good approximation for lightweight modeling purposes.
 
OK, but that 6.94μH is at DC, correct? Anydriver will have a rising Le with rising frequency. A plot that levels out a the DC Le at high frequency is also very wrong.

Bob

First a minor correction to what I said: I meant Re not DC Le.

Second, and not particularly relevant here, you need to use the Le at the cross-over frequency if you want to try textbook XO's.

Bob
 
Yep, it's extremely time consuming to do accurately :D

Also most multimeters other than high end ones have a limited frequency response and will be inaccurate above a few hundred Hz.

If you have a PC with a sound card and a couple of resistors that's all you need along with some free software to do it the easy way...
 
Last edited:
Simon, somebody did tell me this, I guess I'm missing something - it only takes a few minutes to set up. (which have done over 100 times)

The HP tube generator seems to work perfectly.... and fast

What could I be missing?

Found this to be a good way to measure coils also (almost same set up)
 
Last edited:
Account Closed
Joined 2001
You're not missing anything.

Nowadays they have fancy computer programs/gagdets that generate nice plots and can do the measurements quickly, but it's still, essentially, a simple measurement technique. You insert a series resistor with your device-under-test and measure the voltage across it while varying the frequency.

Cheers,

Dave.
 
I would have to disagree. There is one great big advantage. You understand what is really going on and then appreciate how fantastic free software and a PC really is!

Life is too short and beer is waiting. This is how we did it in the bad old days. Set freq. Verified level, measured. Move freq 1/3 octave, do again. Plot it and connect the dots with a french curve. ( I had, and as a matter of fact sold on e-bay yesterday, a wide band VTVM, so I could measure 10Hz to 50K.) When you get close to a peak, go slow. You also don't get phase measurements. Love my WT-II, but you can do the same with a sound card and several programs for free and it will be more accurate that I ever was with the help of the fine folks in Benton Harbor. I'll help set you up.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.