I have recently purchased a pair of SB17NBAC35-8 midwoofers from Intertechnik, and have measured their TS parameters using the Dayton DATS V2 tester. Before the measurement, I gave them 1 hour of break-in at 10Hz and +/-4mm.
The results are atrociously bad. Here are a few of the parameters:
Spec:
fs=31.5Hz
Qts=0.35
Vas=45l
Mms=12.5g
Cms=2.3mm/N
Sample 1:
fs=35.7Hz
Qts=0.53
Vas=21.2l
Mms=18.4g
Cms=1.09mm/N
Sample 2:
fs=34.3Hz
Qts=0.56
Vas=21.9l
Mms=19.2g
Cms=1.1mm/N
As you can see, both drivers are very similar, so it is not just a rogue sample. In each case, the moving mass is about 50% too high and the compliance is half what it should be, leading to a huge Qts.
This means my intended ported cabinets are now useless. How can drivers be so bad ? This really does suggest staggering incompetence. Imagine if you bought a car which was supposed to do 100mph, but it actually only did 50mph. You would be outraged, and indeed the manufacturer would be prosecuted for deception and fraud. But it seems if you are a speaker driver maker, you can get away with it.
I have been in touch with SB, but no response yet.
The results are atrociously bad. Here are a few of the parameters:
Spec:
fs=31.5Hz
Qts=0.35
Vas=45l
Mms=12.5g
Cms=2.3mm/N
Sample 1:
fs=35.7Hz
Qts=0.53
Vas=21.2l
Mms=18.4g
Cms=1.09mm/N
Sample 2:
fs=34.3Hz
Qts=0.56
Vas=21.9l
Mms=19.2g
Cms=1.1mm/N
As you can see, both drivers are very similar, so it is not just a rogue sample. In each case, the moving mass is about 50% too high and the compliance is half what it should be, leading to a huge Qts.
This means my intended ported cabinets are now useless. How can drivers be so bad ? This really does suggest staggering incompetence. Imagine if you bought a car which was supposed to do 100mph, but it actually only did 50mph. You would be outraged, and indeed the manufacturer would be prosecuted for deception and fraud. But it seems if you are a speaker driver maker, you can get away with it.
I have been in touch with SB, but no response yet.
I'm guessing the first thing SB will say is have you done your measurements as per their document http://www.sbacoustics.com/index.php/download_file/-/view/191/
I'd also try measuring them with something else, like REW or speaker workshop to see what you get. I've found some drivers will measure wildly different depending on which method I use, whereas other drivers will be consistent across methods.
Tony.
I'd also try measuring them with something else, like REW or speaker workshop to see what you get. I've found some drivers will measure wildly different depending on which method I use, whereas other drivers will be consistent across methods.
Tony.
It is important that you measure the impedance with a voltage on the voice coil that is not too low. For some suspensions the compliance becomes lower when the excursion is low, especially for new drivers. As a consequence the fs and Qts do increase. I chose 2.83 Vrms for 8 Ohm drivers to measure the TSP. The suspension excursion should be linear as a function of the applied voltage but it isn't always like that.
I didn't notice any significant change in results before or after break-in, and that was after an hour, whereas SB say that 10 minutes should be sufficient.
The DATS V2 test system is thoroughly proven to be accurate, and there is no way I am buying another tester just to verify that the SB drivers are crap.
Incidentally, I notice that recent measurements of the SB17CRC drivers on the Audioexcite.com site show exactly the same issues. That is, moving mass 50% too high, compliance half what it should be. And that is using two different test systems too.
The DATS V2 test system is thoroughly proven to be accurate, and there is no way I am buying another tester just to verify that the SB drivers are crap.
Incidentally, I notice that recent measurements of the SB17CRC drivers on the Audioexcite.com site show exactly the same issues. That is, moving mass 50% too high, compliance half what it should be. And that is using two different test systems too.
Last edited:
I am not surprised. I designed many loudspeakers and a 20-30% difference in driver parameters is very commonplace.
If you need lower Qts and you can use a 4-Ohm driver in your design instead of the 8-Ohm one, you may try the 4-Ohm version of this speaker. See this homepage with measurements of the SB17NAC35-4: www.audioexcite.com >> SB Acoustics SB17NAC35-4, average Qts was 0.42 or 0.45 depending on the measurement system used. It's still higher than specified, but not as high as your 8 Ohm version.
In fact it's more difficult to achieve low Qts with 8 Ohm drivers, because Qes is proportional to voice coil resistance Re. Decreasing Mms and increasing Cms is the way to go to reduce Qes and Qms at the same time or increasing Rms to reduce Qms only. Since Mms is usually fixed, the only parameters to influence is Cms and Rms.
I think it is possible that the workers mistakenly used the 4-Ohm surround and spider (which should be stiffer) for the 8-Ohm version of the speaker. This would explain the big difference in parameters.
If you need lower Qts and you can use a 4-Ohm driver in your design instead of the 8-Ohm one, you may try the 4-Ohm version of this speaker. See this homepage with measurements of the SB17NAC35-4: www.audioexcite.com >> SB Acoustics SB17NAC35-4, average Qts was 0.42 or 0.45 depending on the measurement system used. It's still higher than specified, but not as high as your 8 Ohm version.
In fact it's more difficult to achieve low Qts with 8 Ohm drivers, because Qes is proportional to voice coil resistance Re. Decreasing Mms and increasing Cms is the way to go to reduce Qes and Qms at the same time or increasing Rms to reduce Qms only. Since Mms is usually fixed, the only parameters to influence is Cms and Rms.
I think it is possible that the workers mistakenly used the 4-Ohm surround and spider (which should be stiffer) for the 8-Ohm version of the speaker. This would explain the big difference in parameters.
Good input.
Would that also explain the very high moving mass, or have they used cone material which is thicker than it should be ?
So, it is more recommended to measure the large signal TSP and use these for the system design. It is more relevant for the system behavior at 1W input.
Erik, indeed ... that is a problem. In most cases the difference between small and large signal measurements is not that high. But there are drivers with that problem. I also had such driver in the past. Since then I always measure TSP large signal with voltage on the voice coil and current sensing. Looking to the SB measurements of CS, I thought it could be such "problem" driver and then you best measure also with large signal. If large and small signal results are almost the same, you are sure some driver parameters are not within specification in case of a measured TSP deviation.
Unfortunately I don't. I have an iMac for general computing and an old Sony laptop for use with the DATS. Only USB interfaces are available.
I have noted that driver suspensions are often a bit stiffer than they are supposed to be, but the moving mass is generally pretty close to spec. This leads to both fs and Qts increasing, so the overall response doesn't change much, and it normally only needs a slight tweak on the port tuning.
But with these SB drivers, the compliance is half what it should be and the moving mass is 50% high as well. Net result is that the fs goes up slightly, but the Qts is huge. I put my measured parameters into the MJK MLTL worksheet for my cabinets, and I couldn't really get the port to tune up properly at all. They would work in a sealed box, but then the low frequency cone excursion (and hence power handling) would be severely compromised.
Presumably some clueless numpty at the SB factory decided to use aluminium sheet for the cones which was 50% thicker than usual, without any regard for the consequences.
If cone mass is increased Fs would normally be lower. I'm presuming these drivers were OEM for Revel; the fact that Intertechnik have them cheaper than the NAC version suggests that stock may be limited.
I'm just guessing something happened when samsung bought harman group this year, which lead to SB releasing the black and ceramic cone variants for sale.
I'm just guessing something happened when samsung bought harman group this year, which lead to SB releasing the black and ceramic cone variants for sale.
Yes that's right, but in this case Cms is halved too, so overall the fs goes up slightly.
According to Ulrik Schmidt, who designed these drivers, the black version will become the standard product but the original uncoated versions will be available to special order. The black version also has a vented motor which eliminates the wobble around 600Hz - This mod will eventually be applied to all their woofers.
I used the added mass method.
Initially, I added 10g extra mass (80% of the expected Mms), but DATS reported a shift in fs of <25% and required the mass to be increased. I steadily increased the mass to 19g (cf. actual Mms of 18g), which then worked. The mass was in the form of blutack, firmly stuck to the dustcap.
I did the tests several times, and the results were repeatable and very similar for the two drivers.
Initially, I added 10g extra mass (80% of the expected Mms), but DATS reported a shift in fs of <25% and required the mass to be increased. I steadily increased the mass to 19g (cf. actual Mms of 18g), which then worked. The mass was in the form of blutack, firmly stuck to the dustcap.
I did the tests several times, and the results were repeatable and very similar for the two drivers.
Apparently there was an air volume resonance just behind the dust cap, so they drilled a series of holes around the circumference of the voice coil former to sort it out.
It's a pity their design thoroughness doesn't seem to carry through to their manufacturing, which is very sloppy. There really is no excuse. You only have to look at Peerless drivers, which are made in China yet achieve excellent agreement between spec and measured.
It's a pity their design thoroughness doesn't seem to carry through to their manufacturing, which is very sloppy. There really is no excuse. You only have to look at Peerless drivers, which are made in China yet achieve excellent agreement between spec and measured.