How can I determine zobels for loudspeakers? Use a pulse generator like Quasimodo?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Member
Joined 2006
Paid Member
How can I determine zobels for loudspeakers? Use a pulse generator like Quasimodo?

After building the Quasimodo rig as per Marl Johnson's pdf http://www.diyaudio.com/forums/powe...-using-quasimodo-test-jig-20.html#post4220292

I found how easy it is to determine snubbers for TX so to avoid ringing triggered by diode shut off.

I tested the rig with some woofers and it seems to work the same way.

AndrewT posted: "one can load the speaker driver so that the rising impedance seen by the source due to inductance of the driver voice coil, is effectively counteracted by the added R+C, to give a near constant impedance of the combined speaker driver||R+C.
This allows for a simpler selection of crossover filter components to give a predictable roll off at the desired crossover frequency."

As I am now building a new speaker crossover I would really like to have your expert input on how to determine the best zobel to linearize loudspeakers response.

Ricardo
 
easy, peasy>
conjugate match of Re+ XLe
Xl= Xc at F where uncorrected impedance ~ 2*Re

now tell me why you need a zobel? it linearizes the (small signal) impedance not the response!
but Le is not a linear parameter it varies with cone position actually,

The above is what Dynadio uses for its so called 1st order filters
 
Last edited:
Member
Joined 2010
Paid Member
Possibly this is not the right forum for this topic.To determine the components needed to correct an impedance rise with frequency of a moving coil speaker,requires finding the speakers inductance and resistive components at the frequency of interest.This is normally at the crossover frequency with which the correction network will be applied.A Boucherot cell is the correct term for this application.
 
Member
Joined 2010
Paid Member
It is a well established practice in electrical engineering for power factor correction applications.Blushing aside,I have used this method since the early 70's well before the erroneous formula appeared in audio publications.For a vindication that this (Boucherot cell) is the correct method,Smith & Larson have utilised it in their Woofer Tester equipment and this is the easiest and fastest way to establish exact values of the components .Otherwise if you do not have access to the Woofer Tester then you are obliged to use a more tedious measurement and calculation approach.
 
.....As I am now building a new speaker crossover I would really like to have your expert input on how to determine the best zobel to linearize loudspeakers response.

Ricardo
Hi find this free XO program from bwaslo to be exelent finding right component values http://www.diyaudio.com/forums/multi-way/259865-xsim-free-crossover-designer.html. Put a little time in learning the program it's great for this, quick guide is pick a amp and perfect driver, now measure your real world drivers impedance in the final box design and make a dot zma extension file that is linked to the before perfect driver, add a RC network and toggle R and C up and down till curve suits you in the impedance window. Attach a example picture from my A10.2 driver where both inductance and resonance is compensated. Because changing something on the box design (ports, dampening material, any sizes) changes the measured impedance, the impedance measurement and final network components values has to be last point in the building process.
To measure impedance i use the DATS device, but free REW can do same and make a dot zma file.

Great!

Can you tell some more about your experiments?

Any ideas about the influence on impuls-reproduction of the speaker?
Setup from the attached picture did measure cleaner impulse/step response but both RLC and dual RC networks added this full range driver. On this driver it's a night and day listening result and raise sound quality. Think it's good idea measuring before after effects using REW to confirm impulse response and group delay is getting better and not worse by the network, acoustic domain verse electric domain can behave different and probably worse for full rangers that not behave perfect piston and minimum phase LF to HF. RC network in measurements i can't see it change in HF frequency response, the RLC network do some small change in LF area which depend design and speaker placement in room can be benefit or not, normal i use EQ so doesn't matter. On request when time allow can do some before after measurements in REW, or try linearize yourself and measure + listen effects before and after.
 

Attachments

  • ImpPlot.PNG
    ImpPlot.PNG
    27.5 KB · Views: 183
Last edited:
you guys make it too hard.
get a graph of the drivers impedance (its in the data sheet , no need to measure anything folks ) look where Le is causing a rise . copy the frequency Fz where the impedance is risen to 2*Re. Use Fz , Le from the T/S list and solve for Cz. knowing Xcz = Xle. anybody remember the formula?
The zoble network is Cz + Rz where Rz = Re.
wow ppl can do stuff without computer software amazing.
 
Last edited:
Member
Joined 2006
Paid Member
you guys make it too hard.
get a graph of the drivers impedance (its in the data sheet , no need to measure anything folks ) look where Le is causing a rise . copy the frequency Fz where the impedance is risen to 2*Re. Use Fz , Le from the T/S list and solve for Cz. knowing Xcz = Xle. anybody remember the formula?
The zoble network is Cz + Rz where Rz = Re.
wow ppl can do stuff without computer software amazing.

Would you care to elaborate on the formula ?

What is Re ?

Xcz ? Xle ?

Please excuse my lack of knowledge :)
 
conjugate means set Xl = Xc where X = imaginary parts of complex impedance ( setting them equal means they cancel out and impedance is real at that frequency )

Xl= 2*PI*F*L
Xc = 1/(2*PI *F*C)
use algebra to solve for C at F = Fz

Le & Re from drivers T/S data sheet & determine Fz from looking at impedance graph on data sheet.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.