I’ve gone through 2-3 WinISD tutorials and there’s something I can’t wrap my head around. This is my first real experience with free software since I haven’t designed anything in years. Anyway, how can you tell when a design is bad or unrealistic? The only thing I can think of is that the f3 shouldn’t be any less than about 90% of Fs (e.g., 90% of 55 is about 50 Hz). What I see from WinISD seems hilariously optimistic.
I’ve also loaded the driver parameters into UniBox, which agreed with WinISD.
I’m working on 2 projects. One is a soundbar using 3 of the Dayton 4” coaxials and the other uses a pair of Aura NS6-255-8A 6" that I completely forgot about having. Inputting the Auras is more out of curiosity than anything since I’m building a pair of Maurbacs. Of course, these are fairly small drivers.
In each case, the ideal f3 is much, much lower than fs and Vb is big. Here are the ideal specs according to WinISD and UniBox:
Dayton CX120-8 4” coax Fs=83, f3=55 & Vb=5.7 l*
Aura NS6-255-8A 6" Fs=55, f3=35 & Vb=50.2 l
* According to the PE web page, BassBox 6 says these numbers should be 80Hz at 4.2 liters. Putting that volume in WinISD gives 70Hz, not 80.
Some questions:
(Fun fact: in a previous life I extracted BSIM3 MOSFET models for SPICE, so I’m familiar with the concept of model equations)
I’ve also loaded the driver parameters into UniBox, which agreed with WinISD.
I’m working on 2 projects. One is a soundbar using 3 of the Dayton 4” coaxials and the other uses a pair of Aura NS6-255-8A 6" that I completely forgot about having. Inputting the Auras is more out of curiosity than anything since I’m building a pair of Maurbacs. Of course, these are fairly small drivers.
In each case, the ideal f3 is much, much lower than fs and Vb is big. Here are the ideal specs according to WinISD and UniBox:
Dayton CX120-8 4” coax Fs=83, f3=55 & Vb=5.7 l*
Aura NS6-255-8A 6" Fs=55, f3=35 & Vb=50.2 l
* According to the PE web page, BassBox 6 says these numbers should be 80Hz at 4.2 liters. Putting that volume in WinISD gives 70Hz, not 80.
Some questions:
- Are the WinISD results possible or realistic? Like I said, it’s been a while since I’ve designed anything.
- Could I have just put in the wrong units for the parameters, or there’s a mistake in the spec sheets? I noticed that the calculated Vas doesn’t match the spec sheet.
- I thought that ideal sizes would have a perfectly flat output curve, then roll off about fs. Are these supposed to have a slight bump before rolloff?
- Can anyone recommend another software tool to try? I’ve tried Jeff Bagby’s spreadsheet and it’s a little over my head, but what I’ve seen is more realistic.
(Fun fact: in a previous life I extracted BSIM3 MOSFET models for SPICE, so I’m familiar with the concept of model equations)
Last edited:
Traditionally all rear volume simulators use the same simple formula to find the response of a driver/box combination. This gives consistent results. More recently, WinISD and some others have been simulating the result of stuffing which has occasionally caused some uncertainty. Some will automatically assume you want a Q factor of 0.7, and this is where you might want to decide for yourself what enclosure to use.
Well I have both WinIsd and Bass Box 6. I find Bass Box to be very good as far as simulation Vs. actual FB. I have measured FB when all was done and can reliably get within 1 Hz of target. I think I had a worst case of 2 hz off. I use normal fill and flared ports both sides.
As long as you are reasonably accurate volume wise it seems to work very well.
Rob
As long as you are reasonably accurate volume wise it seems to work very well.
Rob
Didn't sim yours, but as these basic programs go, the current version of WinISD is pretty accurate if all the options are turned on. No experience with the others.
Driver specs often aren't accurate, especially cheaper drivers and Vas is usually off at least a little regardless of price.
Once Qts' is > ~0.403, the cab net volume [Vb] is > Vas and Fb is < Fs, i.e. an under-damped alignment; so yes, there will be a bump in the response, increasing in amplitude with increasing Qts'.
Qts' = Qts + any added series resistance [Rs]: HiFi Loudspeaker Design
As long as you're doing basic speaker alignments, WinISD, Unibox is fine and if you want to do TL, horns, various Band-Pass alignments, then Hornresp, the most popular around here.
GM
....and more, always has been, but requires a large horn, otherwise best to limit it to a 1/2 octave max [0.707x Fs] and even then there normally won't be hardly any power handling capability, so generally only done when there's good corner loading.
Nowadays we normally use some form of TL for alignments with a > 0.403 Qts' to get extended ~flat responses to a < Fs tuning, hence the popularity of Hornresp
.
GM
No, I think that if you wish to make blanket statements without any semblence of context, caveats or anything else whatsoever, you cannot expect it to be recieved as anything other than the twaddle it is. For the vast majority of low frequency enclosures designed Le is of little consequence, and if you are claiming otherwise, you will have to overcome the formidable obstacle of literally millions of loudspeaker LF enclosures that work as the designers intended, which do not factor VC inductance into the lumped element LF modelling.
As far as LF enclosure design is concerned, Le primarily becomes of note when the electrical [and as relevant acoustical] corner created is sufficiently near the box operating BW to produce non-linearities in the frequency / amplitude response. Subwoofers apart these are atypical conditions, so a blanket claim that 'any box simulation that does not take driver voice coil inductance into account is pretty much useless' is complete nonsense.
As far as LF enclosure design is concerned, Le primarily becomes of note when the electrical [and as relevant acoustical] corner created is sufficiently near the box operating BW to produce non-linearities in the frequency / amplitude response. Subwoofers apart these are atypical conditions, so a blanket claim that 'any box simulation that does not take driver voice coil inductance into account is pretty much useless' is complete nonsense.
I wondered about that too. What if you add a reciprocal circuit to counteract the inductance (sorry, but I can't remember what it's called)?
you can model le losses in basta! and hornresp, do some simulations and then say this is not an important parameter!
no i think you need to adress this in the driver magnet system by using copper rings/sleeves
So what you are saying is that people should address VC inductive rise by ripping their drive units to pieces and somehow provide them with full symmetric drive (redesigning the rest of the motor, and potentially the suspension and cone also to account for the changes in behaviour in the process). Which even if it were realistic, is of scant relevance to the majority of back loaded or bass enclosures, which, as you know perfectly well, are only operative or useful over an extremely limited BW before they reach their upper corner (or GD becomes excessive), thus Le has little relevance to the acoustical behaviour of most loudspeaker cabinets in their operating BW. VC inductance (rather than a static figure of Le, which is typically only given for 1KHz or an alternative single frequency) is a major factor in other areas of loudspeaker design, but not for the acoustic behaviour of the majority of loudspeaker bass enclosures.
They're usually called Zobel networks -whether that's strictly correct is debatable, but it's a reasonable enough handle. Series RC circuit shunted across the driver with the object of flattening the VC impedance rise, although a series RC shunt is sometimes used for EQ / trimming the HF response also, although you have to go easy if you don't want the HF impedance load to drop too low.
If you're interested in TLs, you should read Martin King's pages Quarter Wavelength Loudspeaker Design if you want to know some of the physics. Start with the pdf 'Anatomy of a Transmission Line' under the 'Transmission Line Theory' tab, and move on from there.
no i do not say that people should break their drivers, just to take driver inductance into the modelling
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.