Measuring driver parameters vs the data sheet values

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm making the transition from building established designs to starting from scratch. I'm currently working on a low cost full range/neo tweeter design for a computer (learning at a low cost point) and I just measured the free air parameters of a Fountek FR88EX full range driver using DATS2. I calibrated the leads and the 1K ohm resistor measurement and was careful with the weighted VAS measurement. I am surprised at how different my measured numbers are vs. the data sheet. Is this typical or have I likely done something wrong?:

-------Meas----DS
Re----6.15----6.0
Fs-----133-----95
Qts---0.97---0.47
Qes---1.27---0.62
Qms--4.11---1.96
Vas---0.59----1.27


Thanks,

George
 
Last edited:
Just another Moderator
Joined 2003
Paid Member
Hi George, It is worth having a look on Fountek's site to see if they have any information on how they do their measurements. Not all manufacturers do it the same way. SB-Accoustics, and Seas both have information on their sites about their testing method. I found out recently that Seas uses 2V for their drive signal which is much higher than what something like DATS uses I believe.

break-in and even pre-measurement warm up procedures may affect the results.

Does the DAT's use MLS signals or swept sine for doing the measurement? I found that some (and only some) drivers give widely different results with mls compared to swept sine. Other drivers will give almost identical results with either.

Your fs above is the biggest discrepancy, break in may be affecting this, but fs was the thing that I found with some drivers was way off with MLS testing.

Tony.
 
Last edited:
Just another Moderator
Joined 2003
Paid Member
Hmm I was thinking that I picked that up from SB accoustics tech note http://www.sbacoustics.com/index.php/download_file/-/view/191/ and that it was to be done before each measurement.. however it seems maybe they are suggesting that it will actually break in the driver (I thought it would take longer than that).

They do state it should only make 10-15% differnce, which seems in your case to not be enough to account for the differences.

This is the link to Seas info on their T/S measurements DATASHEETS

As you can see Seas and SB are using different drive levels (which will definitely give different results). a woofer measured at 1V will almost certainly show different T/S params to one measured at 2V.

I guess the other thing I should ask is what your mounting method is for the driver?

Clamped securely (in some sort of frame) in free air away from any nearby objects is ideal. Don't sit the driver on it's magnet, especially if it has a vented pole piece.

Tony.
 
Life is a lottery... Specs are Puffery.... ;) Work with what you have..... measure over several days and variables=driven/ambient/up/down/flying around those magnificent men in... whoa...!!!
Then average and build... Don't let the math get in the way of the enjoyment... :)

Entropy rules... Devices 10 years on a shelf will not measure as manufacturers specs... work with it
YouTube
 
Last edited:
frugal-phile™
Joined 2001
Paid Member
I hav eposted this before but can’t find where:

T/S parameters, extremely useful though they are, are not perfect: they are a snapshot of a driver's behaviour under a given set of circumstances; it does not necessarily follow that these hold good under other conditions (they rarely do), so it is advisable to make allowances for these variations in the design stages.

Most people look at T/S and see a set of scalar numbers, but in reality they are a function of current (or voltage) -- they change with weather too. If one does not take care to account for these changes you end up with something that doesn't work all that well. For instance a typical tightly tuned bass reflex, with a driver with a non-flat T/S curve, will only be correctly tuned for one drive level. Not only does this mean that they will be in tune at only one volume level, given the dynamics in music, they will only be spot on at one level in the music and be progressively out of tune with changes in the dynamic levels.

You have no doubt noticed that Scott & my designs favour fairly high ratio vent aspect ratios (or volume fill damping). This adds an R to the vent and, analogous to a grid-stopper, and gives them a greater tolerance to non-flat T/S curves. Nothing models this, and requires an experienced eye to extropolate from what is modeled.

These are the numbers i measured:
Re: 6.17
Fs: 107.14
Qes: 0.90
Qms: 3.12
Qts: 0.70
Vas: 1.23
dB/2.83v: 85.1
Sd: 2762
BL: 3.00
Mms: 1.95
Cms: 1130

We used the factory numbers to design the Fountek Five and that worked out well.

dave
 
Life is a lottery... Specs are Puffery... <snip>

I couldn’t agree more but some of my numbers are so different from the data sheet that my box volume calculations are different by 2X. Because of this I need to make sure that I am measuring correctly before moving on.

Thanks,
George

Clamped securely (in some sort of frame) in free air away from any nearby objects is ideal. Don't sit the driver on it's magnet, especially if it has a vented pole piece.

Tony.

I did have the driver sitting on its magnet so I’ll try different mounting. I did notice that the weighted VAS measurement is very finicky so maybe I still have some errors in technique.

I have noticed a huge difference in listening impressions after a week of break in on a completed set of speakers so maybe breakin is part of the issue. Do you guys break in the raw driver in free air or mount it? Do you use a signal generator or music material?

Thanks,
George
 
Last edited:
I took an Aura Sound NS3-193-8A and tested it:

A) Data sheet
B) In a sealed cabinet
C) Mounted unsealed cabinet
D) Free air suspended by faceplate
E) Free air sitting on the magnet

Most of the measurements were closest when free air sitting on the magnet (E) except Qms:

--------A------B------C-------D------E
Re---7.60---7.79---7.71---7.72---7.72
Fs------80---137----121----125----123
Qts---0.67--1.07---1.02---0.94---0.87
Qes---0.73--1.28---1.22---1.12---1.08
Qms--8.00---6.61--6.15---5.79---4.49


It's on to music break in and another free air measurement.

George
 
I burned in the driver for 12 hours with music material. It changed the numbers, but not enough to explain the large differences from the data sheet (A). I added those values as a new column (F):


--------A------B------C-------D------E-------F
Re---7.60---7.79---7.71---7.72---7.72---7.78
Fs------80---137----121----125----123---112
Qts---0.67--1.07---1.02---0.94---0.87---0.84
Qes---0.73--1.28---1.22---1.12---1.08---1.03
Qms--8.00---6.61--6.15---5.79---4.49----4.5


What is a cost effective alternative to my DATS2 measuring rig for getting TS parameters from a raw driver that is known to be accurate?

Thanks,

George
 
frugal-phile™
Joined 2001
Paid Member
T/S are curves, not scalars.

Different measuring kit pulls the number from a different part of the curve, so measures can be significantly different. Also whether the sw estimates the parameters or actually measures them. Short of the pro stuff the factories use (ie $5k-ish), Smith+Larson Woofer Tester 2 is probably the best short of them.

I always start with the factory numbers,i have rarely had to redo a box. Also keep in mind that depending on the manufacturer parameters will vary from plus-minus 5% to plus-minus 20& or more.

I use S+L for driver matching and unknown drivers.

dave
 
I am beginning to think that measuring TS parameters at my equipment price point is a waste of time and that as you suggested, perhaps I should just go with the spec sheets. The crossover design is different because we measure and listen to the drivers in the box but it sure would suck if the initial box sizing was way off because the data sheet specs are BS.

It almost appears to be easier to have a bunch of crude boxes built at a few different volumes (or make one that is configurable with partitions or something to find the right tuning through experimentation and forget making a commitment to a full up furniture grade cabinet build based on just these wildly variable TS measurements.


George
 
Just another Moderator
Joined 2003
Paid Member
On the question of whether there are any cost effective alternatives to your DATS 2 for getting a different reading, then provided you have a sound card already with line out and line in then the answer is yes.

The most cost effective option is to make up a simple cable with one resistor and to use software such as REW or ARTA to measure the impedance and calculate the T/S params.

You need the impedance cable here Cables

Note that for REW and ARTA you ideally should use a 100 ohm resistor not 10 ohms as per the diagram.

Some soundcards struggle to drive a speaker directly. If that is the case then using an amplifier to drive the speaker may be needed, but make sure to keep the levels low. Use a 100Hz sine wave and an AC voltmeter to set your voltage to 1V or less. There is a risk that you can fry your soundcards input if you use and amp and don't have some sort of attenuation.

I myself use a Walin Jig II which I made. http://www.raymondaudio.nl/files/jig.pdf

edit: I posted some rubbish which I have deleted ;) Oh now I see, I'm not going crazy, your latest table has the wrong specs for manufacturer params ;)

Tony.
 
Last edited:
many years ago i used clio lite for measurements, but i found that justmls in lspcad gave me almost identical results so i stopped using it, dos-based as it was. Today i find that justmls gives me almost identical results to stepped sine in limp (arta), mls in i limp differs though.
justmls does not do any calculation so i do the calc by myself, i put the numbers in a excel spreadsheet.
stepped sine in limp works poorly on my computer, it often crashes. limp do t/s calc, but it reads the inductance value in a non standard way, i am not sure what is going on there.
clio seems still to be highly regarded and trustworthy so i would go for clio pocket if i was to buy a new system today, even though i am not sure if the chirp stimuli is the best stimuli for impedance measurment of some drivers
 
What's going on with vintage clio system is that its math procedure to calculate L1 and L10 inductance values within TSP evaluation is simply wrong. These figures are useless anyway. I have checked with clio pocket, and there it is set right. I still have got 2 machines to keep the clio board functioning powered by pentium slot1 300Mhz processor. The best investement I made diy wise, clio I mean.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.