AVS Forum banner
1 - 20 of 27 Posts

·
Registered
Joined
·
2,246 Posts
Discussion Starter · #1 ·
What is up with this.

I know published and measures TS parameters will vary to some extent. Manufacturing, temperature, all kinds of things. Some just over optimism. No big deal. But Le? EVERY driver I measured recently is way off the spec. So, I first distrusted my measurement method, WT2. Well, it measures known coils just fine. So I used my B&K meter. Sure, it measures at 60 Hz not 1000 but not far off from WT2. So I built a bridge with precision resistors and my HP generator. Measured at 1K. All three methods were within 1%. Not only that, when I build filters and zobels, the result does match when tested. So, 4 Daytons and the Zaph all off by more than 2X. I can see VAS being not too close, but Le?
 

·
Registered
Joined
·
2,136 Posts

Quote:
Originally Posted by tvrgeek /forum/post/17008595


What is up with this.

I know published and measures TS parameters will vary to some extent. Manufacturing, temperature, all kinds of things. Some just over optimism. No big deal. But Le? EVERY driver I measured recently is way off the spec. So, I first distrusted my measurement method, WT2. Well, it measures known coils just fine. So I used my B&K meter. Sure, it measures at 60 Hz not 1000 but not far off from WT2. So I built a bridge with precision resistors and my HP generator. Measured at 1K. All three methods were within 1%. Not only that, when I build filters and zobels, the result does match when tested. So, 4 Daytons and the Zaph all off by more than 2X. I can see VAS being not too close, but Le?

Le and Xmax are parameters that are tempting to over or undersate.


Kinda like when guys post their height on a dating site instead of 5'10" they will write 6'0"

 

·
Registered
Joined
·
2,028 Posts
How you measure it is everything. Some just measure at 1khz which I don't believe is that useful. Some methods use curve fitting which can very accurate but usually requires some optimizing. And throw a full copper sleeve on the pole piece and it can be very basically impossible to a get single Le value that represents the whole frequency range as it will actually change. Same with shorting rings, but sleeves seem to have a much bigger impact in my experience.


For relative comparison just look at the impedance curves as whole to judge.
 

·
Registered
Joined
·
2,028 Posts
BTW how can you measure the Le at 60hz and hope to get an accurate result? The Le figure by nature is for the curve well above the driver resonant frequency. I would guess the meter is not actually sending/measuring a 60hz signal.
 

·
Registered
Joined
·
2,246 Posts
Discussion Starter · #7 ·

Quote:
Originally Posted by augerpro /forum/post/17009761


BTW how can you measure the Le at 60hz and hope to get an accurate result? The Le figure by nature is for the curve well above the driver resonant frequency. I would guess the meter is not actually sending/measuring a 60hz signal.

Quite right, not a valid measurement. Meter does use 60 Hz. It was just a sanity check. Handheld LCR meters are only valid for testing/tuning coils, not motor assemblies.


Drivers are specified at 1 Khz. WT2 records the measurement at 1Khz, and I used by bench setup at 1Khz getting similar results.


Interesting comment about the shorting ring. It just happens every driver I have measured with these tools have shorting rings of some sort. But the spec is for the complete motor, so it should be part of the measurement.


It still comes back to the overly large difference between published driver spec and measured. When I add a zobel based on my measurement, it is very close to providing the measured flat impedance, so I am having less doubt about my measurement. Le is not a big selling point like VAS, Fs and Qts so I don't know why it would be so far off.
 

·
Registered
Joined
·
2,136 Posts

Quote:
Originally Posted by augerpro /forum/post/17009752


How you measure it is everything. Some just measure at 1khz which I don't believe is that useful. Some methods use curve fitting which can very accurate but usually requires some optimizing. And throw a full copper sleeve on the pole piece and it can be very basically impossible to a get single Le value that represents the whole frequency range as it will actually change. Same with shorting rings, but sleeves seem to have a much bigger impact in my experience.


For relative comparison just look at the impedance curves as whole to judge.

can you elaborate on that a little ? i want to know how a copper sleeve would selectively affect inductance at different frequencies.


thanks.
 

·
Registered
Joined
·
2,028 Posts
Dunno Vas but if you take a look at some of the drivers on either my site or Zaph's you'll see the ones with full sleeves not only have a flat inductance, but occasionally seem to go up and down just a little bit through the frequency range. You tell me and we'll both know
 

·
Registered
Joined
·
3,801 Posts
We need Kyle to weigh in here. Le as a single number is oversimplifying it. As I recall, the more accurate model goes something like:

Code:
Code:
---L1--|--L2--|---
       |--R2--|
but I could be wrong.
 

·
Registered
Joined
·
2,028 Posts
Because you are measuring at 1khz and the others (at least Zaph) are using curve fitting. We'll need a little more data to show that others are off and not you.
 

·
Registered
Joined
·
2,246 Posts
Discussion Starter · #13 ·
Actually, every driver I am referring to is specifically mentioned at measured at 1Khz. Impedance will vary with Freq. ( you know 2-pi-f-l) but I do not see why the stated 1Khz readings are far away from measured.


Looking at the ZA14W08 sheet right now, yes he has a graph showing the changes in Z free and clamped, but the specification still lists "Le(1K): 0,37 (mH)" and I measure in the .167 to .180 range between drivers.
 

·
Registered
Joined
·
2,028 Posts
So you are measuring Zaph's new driver? I don't know why it says 1khz in his spec sheet. Maybe that was the spec he got from the manufacturer and not his own measured spec? At least not with Soundeasy as it uses curve fitting. I don't know what's going on there.


The Le(x) plot is unrelated to what we are talking about here. It's more to show how linear the motor is at xmax.


The proof is in the pudding and that is what gives the ideal response after the Zobel. Model a Zobel using both manufacturer's and your own results, build it, and remeasure the impedance. Plot the result and post it here and we'll see what we can come up with.
 

·
Registered
Joined
·
2,246 Posts
Discussion Starter · #15 ·
Yes. I am using them for the mid in my tweeter test. For testing, they will use an active crossover so the Le does not really matter.

I will certainly do the zobel test but it has to wait a tad. I ran short of caps. When I did on the last Daytons and Peerlesses, my measurement was closer. After the tweeter test, I plan on recycling them into a MTM or MMTMM with the winning tweeter to replace my Dayton/Vifa's in my guest room. For that I will need to do a passive crossover.


Curious on the Zobel. Some designs recommend using Re, some Re + 25%. I wonder if that is compensation for Le variance? I build and tune and it never comes out as either.
 

·
Registered
Joined
·
1,125 Posts
Le really does not tell the entire story, you have inductance which is your basic low pass filter - but you can also have a non-linear inductance issue which is more complex to measure and or report. If the inductor is causing distortion, you might actually hear it kind of like a cheep electrical inductor you buy at the store. You can at least get rid of the non-linear issues by trying to lienarize the electromagentic feild with respect to displacement (linear inductance coupling) and two, you need to use a linear magnetic material unlike low carbin steel, ideally something that is not electrically conductive such as powdered iron material commonly used in inductor cores - ATC implements this in thier drivers, and we have expiermened with such motors for subwoofres, but it tends to be more placebo for low frequencies - shorting rings seem solve the primary problem of lowering L all together. Inductance is an issue that relates to the coupling and proximity of the coil and the matter between and around it. There is another issue that solely relates to the size of the electromagnetic feild of the coil and the fixed DC field of the motor and that is the flux modulation - lower current or higher fixed B fixes this problem or again, shorting rings can help! Fun experiment: hook up your midrange driver to high inductance subwoofer thats not magantized in series... see how it sounds compared to the same driver with a equivalent pure resistive load in series.
 

·
Registered
Joined
·
2,246 Posts
Discussion Starter · #17 ·
OK. Notes to save. Fun to have. But for T/S small signal parameters I should be well away from the non-linear effects, and should be away from thermal effects. It still does not explain why if I measure Le at 1Khz ( either the old hard way or with WT-2) I get a value way off (3X on several Daytons, 2X on the Zaph, about 175% on a Peerless '500) that is published, and if I build a Zobel and plot the result with WT-2 I get a far closer cancellation than if I used the published spec. I would be skeptical of the computer, but when my old HP AC voltmeter and generator with .1% resistors in hand comes up the same There is something basic I don't get. Why can't we reproduce the specs as published? QA check if you will. There must be something about the test conditions I do not know.


I thought it could be the very low drive I am using or that I am not clamping the vc, but if that were so, then the Zobel would not be correct. Could we be back to "what is small" for T/S parameters? Should the drive be smaller or larger? As I measure and prototype, it is no big deal for me other than I don't get it. Woe be the beginner who has to rely only on the published specs.
 

·
Registered
Joined
·
2,246 Posts
Discussion Starter · #18 ·
Maybe a hint in test conditions. With WT-2, I can drop from "100% drive" which is about 25mV at 1Khz down to under 2 mv. As the drive drops, so does the measured Le. In this case from .169 to .164 Not a big change. Could it be that Le is commonly measured at 2.66V or something? I have to set up the bench again to try that. I would have a hard time calling that "small signal" he driver would be jumping off the table as that is a lot of drive.
 

·
Registered
Joined
·
3,801 Posts
This may seem like a dumb question but why do you care what the Le is? All that really matters is an impedance sweep, ohms vs. frequency. All the good crossover simulation programs use real impedance sweeps, not T/S parameters. And at the low frequencies that box modeling programs use, Le isn't really a factor for tuning the box.
 
1 - 20 of 27 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top