The measurements of the SB Audience ROSSO 12MW300 woofer:
https://hificompass.com/en/speakers/measurements/sb-audience/sb-audience-rosso-12mw300
https://hificompass.com/en/speakers/measurements/sb-audience/sb-audience-rosso-12mw300
Perhaps not the best place to ask but anyhow: what is a good way to measure the TS Parameters of a woofer in a way more or less equal to the normal usage case?
On Hificompass these was input from Seas showing that the res frequency varied significant with the applied voltage.
But how to do? With a low value sense resistor?
On Hificompass these was input from Seas showing that the res frequency varied significant with the applied voltage.
But how to do? With a low value sense resistor?
That is just the Cms not being constant.
Or in other words, drivers with a big change in the Cms curve, also will have a big shift in Fs.
Since; Fs = 1 / (2*pi * sqrt (Cms * Mms))
About what's the right voltage for normal usage.
That's unfortunately a chicken egg story with quite some manufacturers.
Many don't use the AES default of 0.1V, but seem to cherry pick. Sometimes at voltages of well over 2-3V or above.
Also Seas seems to derive their Re from measurements. Which leads to a value that's too high.
There is an online paper somewhere that describes this very well. Maybe you can use your plodding skills for that again?
I can try to have a look myself when I am back home, somewhere in my own pile of stuff lol
Anyway, this is one of the reasons why a good constant Cms (Kms) is so important. I have a feeling that it's maybe even more audible than distortion.
Btw , it's quite easy to do cheap poormans Klippel measurements at home.
The measurements of the FaitalPro 15FH500 woofer:
https://hificompass.com/en/speakers/measurements/faitalpro/faitalpro-15fh500
https://hificompass.com/en/speakers/measurements/faitalpro/faitalpro-15fh500
In my opinion, SEAS test conditions are closer to real application and more useful than others. The very small signal measurements are important too, but they are more of academic interest
I don't fully agree with this, but the reason why is very specific.
Yes, at first glance they use a value for like Re that resembles "maybe" the practical situation a TINY bit better.
The problem with this, this is NOT how Re is defined.
The definition of Re is very simple, just the resistance at DC. Nothing more, nothing less.
(well I guess at a given temperature, humidity etc would be nice).
The Re Seas is using can be seen more like Re(AC) at a given AC voltage and frequency.
Not only does this mean it doesn't follow definition anymore, it also means that we can't compare anything to it anymore.
Meaning that Seas "Re" will be totally different from other brands.
But most of all, it is literally still a value that doesn't say much in a practical sense, because for that we need Klippel and distortion measurements! (which they refuse to share btw!!)
The estimated value of Re at even a small AC voltage (or noise), can sometimes drastically differ from the actual DC value.
This is even more true for driver with a low Fs and a low Qt. This results in a impedance peak with a low Q.
In practice this means that it will takes forever for this impedance peak to actually flat-line with the actual Re.
Even at like 2-5Hz, this curve is still not flat-lined in some cases.
So in some cases I have seen that Seas even just uses a wrong value with quite some error in it.
One of the reasons why Re @ DC is so important, is that Rms and BL are determined from there.
It's also a value do determine thermal/power compression.
So I am fine that Seas is using "their" method of determining Re (even if it's wrong sometimes).
But please let them call it Re(ac) instead of Re(dc), because at this point it is confusing as hell.
And to be academic about it, it is even just plain wrong because how Re is defined.
Regarding the distinction between Rdc and Re, there is a good AES document:
https://www.aes.org/e-lib/browse.cfm?elib=14048
https://www.aes.org/e-lib/browse.cfm?elib=14048
From a practical point of view, I am gonna correct myself a little bit.
More about how important it is from a box/cabinet design perspective.
I have attached two graphs of a certain woofer.
The first one is a closed box, where I changed just the Cms to 75% (=10% distortion) and 50% (=20% distortion).
I also changed the BL to 82% (=10 % distortion).
The second graph is the same idea with a BR system.
All other variables being kept constant. (volume, tuning of the port etc etc etc)
As you can see, the Cms doesn't have a huge impact, even with a change in Fs from 28Hz to 39.5Hz (dF = 11.5Hz) at 20% distortion (50% of value Cms).
But only a little change in BL (equivalent to 10% distortion) has a huge impact.
So from a cabinet design perspective, you would rather like to have a nice constant BL over Cms.
Yellow = original
Green = Cms 75% (= 10% distortion)
Orange = Cms 50% (= 20% distortion)
Blue = BL 82% (= 10% distortion)
Your simulation assumes that Rms is constant, but it is not. Rms increases as the frequency decreases, so Qms and Qts would not remain constant compared to the case if Rms were constant. BTW, you are absolutely correct that the box tuning is quite invariant to Cms.
Nevertheless, in view of the non-linearity of the parameters, I believe that the most correct way to measure them under the conditions closest to those of actual operation
Nevertheless, in view of the non-linearity of the parameters, I believe that the most correct way to measure them under the conditions closest to those of actual operation
Right, that was one of the papers I was talking about before.
That companies take a different variable that maybe suits better (also open for debate, but different discussion), is fine.
At least in this paper they are clear about the difference and way there is a difference.
This paper was written by the people of Tymphany (Peerless), but guess what, they still use Re in their datasheets.
Not clear if this is just Rdc, or their new Re value.
In the old days, this was most definitely Rdc and I have a strong suspicion that they also don't apply this "new" Re to all their drivers.
Seas just calls it "voice coil resistance".
Point is, that we just don't know anymore what companies use plus it's extremely important to always specify the Rdc value as a starting point.
Therefor comparing is also completely impossible.
Last edited:
I totally agree, but this is also what Klippel assumes!!
I have also an hard time beleving that Rms will stay constant.
That being said, it won't be as significant I think. (that's also what Klippel claims, although there is no data to back that up)
It will mostly have an effect on Qms. I have even experimented with this by opening up the motor structure more and more.
(by drilling and closing more venting holes for example)
As I stated, this is from a box/cabinet design point of view!!
So it doesn't say ANYTHING about non-lineairities and even less so what is better or worse.
But I think a 2,5dB peak (= different total Q factor) can be pretty audible in some cases.
Ideally, in the end both need to be good enough obviously.
I guess my message is more that getting adequate data is just almost impossible.Very highly nonlinear systems, such as speakers, are very difficult to model over a very large amplitude range. The best way is to use the most linear speakers!
Kind of bizarre anno 2023, especially since ALL major speaker manufactures have a Klippel system anyway.
Also, not every project or budget lends itself for the most linear speakers (with the additional obvious price tag).
But you wanna at least know what you're up too!!
Luckily we still have HifiCompass to help us out here!
💙💛
@HiFiCompass
I do have a question, but I don't know how easy it is for you to implement.
But would it be possible to show the impedance graph at different voltages?
Or at least at a standard minimum and a maximum voltage?
That way we get at least a good sense of how much the BL and Cms are shifting.
This only works with a constant voltage source (amplifier) or when a very small reference resistor is being used.
I do have a question, but I don't know how easy it is for you to implement.
But would it be possible to show the impedance graph at different voltages?
Or at least at a standard minimum and a maximum voltage?
That way we get at least a good sense of how much the BL and Cms are shifting.
This only works with a constant voltage source (amplifier) or when a very small reference resistor is being used.
I don't have that option because I do impedance measurements with the WT3 tester. I see no point in spending time on more in-depth research, because that should be a priority for speaker driver manufacturers, not for end users which we al are.
But I once made short Fs measurements at different voltage levels for some speakers, and if you want, I can post them here.
But I once made short Fs measurements at different voltage levels for some speakers, and if you want, I can post them here.