The Subwoofer DIY Page v1.1 - Projects : An INF10 Bandpass Subwoofer
A project that I started several years ago. I found some time to do some more measurements today.
This bandpass sub is showing an extended upper frequency response that's not predicted by the LDC's tables (or my equations - which produce identical results to the LDC). In some cases the response is over 6dB higher. Measured Fb also proved to be significantly lower than predicted. Any ideas what actually might be causing this? I've made a couple of guesses, but I would like to know for sure what's causing it, and if the effect can be accurately predicted.
A project that I started several years ago. I found some time to do some more measurements today.
This bandpass sub is showing an extended upper frequency response that's not predicted by the LDC's tables (or my equations - which produce identical results to the LDC). In some cases the response is over 6dB higher. Measured Fb also proved to be significantly lower than predicted. Any ideas what actually might be causing this? I've made a couple of guesses, but I would like to know for sure what's causing it, and if the effect can be accurately predicted.
Banned
Joined 2002
I used to drool over all your projects
I also remember talking to you long time ago too, MAN time fly's.
Hi Brian,
I'll be interested in responses to this too!
I built a 6th-order bandpass sub that had a similar result:
Actual measured in blue
WinISD predicted in green
WinISD predicted after adjusting volumes to get frequency peaks to line up (in case my volume calcs were out)
Like you, the HF peak was higher in frequency and value than expected.
Mine had both chambers ported, - the LF chamber had the same issues as the HF chamber.
I'll be interested in responses to this too!
I built a 6th-order bandpass sub that had a similar result:
Actual measured in blue
WinISD predicted in green
WinISD predicted after adjusting volumes to get frequency peaks to line up (in case my volume calcs were out)
Like you, the HF peak was higher in frequency and value than expected.
Mine had both chambers ported, - the LF chamber had the same issues as the HF chamber.
I used to drool over all your projects
LOL - I didn't think that my projects were actually drool-worthy...
Hi Brian,
I'll be interested in responses to this too!
I built a 6th-order bandpass sub that had a similar result:
Actual measured in blue
WinISD predicted in green
WinISD predicted after adjusting volumes to get frequency peaks to line up (in case my volume calcs were out)
Like you, the HF peak was higher in frequency and value than expected.
Mine had both chambers ported, - the LF chamber had the same issues as the HF chamber.
Did you measure the T/S parameters of that driver to see if they corresponded with the published specs? Assuming that WinISD's modelling is fairly accurate, those peaks suggest that the driver's Qts might be higher than the published figure.
Just looked over your design...
The Qts of that driver is a bit higher than what I'd use for a 6th order BP. I tried modelling your alignment and got 5dB peaks at the ends of the passband.
Did you measure the impedance to see if you hit the target resonance frequencies? As you used flared vents, this might have shifted the resonance frequencies upwards a bit unless you took this into consideration in the port calcs.
Eeek, no, don't do that, LOL. Actual specs could sometimes be off published specs by quite a bit for various reasons (including some specmanship on behalf of the manufacturer, LOL). It's particularly important that if you going to tackle one of these higher-order alignments, you use actual t/s parameters. In fact, when I was a lot less lazy, I used to manually measure the parameters at a power level somewhat higher than "small signal" and use those results to generate an alignment. It's not that difficult to do, and would save you a lot of grief later on when an alignment doesn't measure as expected.
Banned
Joined 2002
It stopped raining for once, so I've been doing some measurements of a front loaded horn versus a tapped horn. Both enclosures use the same woofer, and the foot print is nearly identical.
I noticed a few things:
- At some frequencies, the TH has 12db of gain over the FLH. This is waaaaaay beyond what you'd expect. Tapped horns have some crazy gain.
- The FLH is smoother than the sims predict
- The sims are "in the ballpark", but the measured response is a lot different than what you'd expect.
Since front loaded horns and single reflex bandpass boxes are basically the same thing, and tapped horns are basically dual reflex bandpass boxes, I thought this might be of interest of you.
One thing that was particularly "fun" was adding pieces of plywood to the tapped horn to change the response. By rearranging segments and angles you can change the response quite dramatically, both for the better and for the worse.
Having a good measurement setup makes this hobby a lot more rewarding
Hmm... will be more thorough next time! - did take into account the flares though.
Impedence measurements shouldn't be too hard to do. Am waiting on building a high-pass filter for this sub, so will do some then.
btw: I like your worthy addition to the english language - "specmanship"
I suspect that may be because of losses. Perhaps Hornresp does not include losses in its model? I'm not sure. Losses will tend to flatten the peaks a bit.
In this particular mystery though, the upper cutoff point of the bandpass system is quite a bit higher than predicted. If anything, I expected it to be slightly lower, as my model does not include the effect of Le on the FR.
I've tried fudging around with box dimensions and Fb in the model, but can't emulate the response I'm seeing, which basically suggests that the model is for some reason or the other failing to sucessfully predict the upper cutoff point with good accuracy. Not that I'm bothered TOO much (the lower cutoff point is of considerably more importance, LOL), but I would really like to know what's invalidating the model.
It definitely isn't midrange leakage. The driver's cone is not visible through the vent, and response increase is occuring in and just above the passband, not out of band, where midrange leakage effects usually appear.
The 400Hz bit is the organ pipe resonance of the port - must be 13572/800=~17" long? The lower tuning frequency puzzles me - could the ports have a longer effective length in the environment you have them in? - OIC you have 14" ports - a pair of 3's is like a 4.24"er - so 14"+.73*4.24~=17... OTOH, Vented boxes with enclosure damping can have the actual FB a bit away from the minimum in impedance, IIRC.
The broadening of response may indicate a bit of acoustic mass loading, which raises Q and lowers Fs and sensitivity. Try adding a bit of mass to your sim to see what happens...
Last edited:
The lower Fb appears when there's no damping in the vented section of the enclosure, so I don't think that's applicable here.
Tried that. Can't seem to get a good emulation for the measured response at all. Either predicted Fb is wrong or the pass-band is wrong (shape & cutoff points).
BTW, I took a closer look at the pass-band this afternoon. It's slightly drooping with increasing frequency. If I blissfully ignore what's happening at the upper end of the passband, the sims suggest that I might losing about 2dB midband because of the lower tuning (see attached graph). That's a lot for a driver rated at 87dB/1W/1M. I think I'm going to try to move Fb up a bit closer to 58 Hz to get rid of the droop and gain back some of that lost efficiency. The vent calcs are totally off - sims suggest that the current 12.25" long vents would actually have to be 19.5" long to tune the vented section to the measured 50.9 Hz, but that's clearly not the case. That's a HUGE difference between sim'd vs. measured. So, it looks like I'll have to toss the calcs aside and use the "old-school" trim-measure-trim-measure-trim-measure approach to getting the pass-band closer to what I'd like to see
.Attachments
Is this further testing for the MCM? I think quite a few of us would be interested in the designs, particularly as compared to a 6th order bandpass (which I still haven't quite figured out how to design in a controlled way).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.