THAM15 - a compact 15" tapped horn

[Brian Steele]

Originally Posted by tb46

Hi,

Interesting stuff Brian, I think we're all waiting for your high SPL results.

As an aside, I compared the THAM15 to the SS15:

Regards,

Not surprising - isn't the SS15 a larger box? If we can successfully deal with the notch in the THAM15's response though, the result would be something with a wider passband and lower cutoff frequency.

Question: how do these compare to a simple vented box using the same driver?

[soho54]

Originally Posted by tb46

As an aside, I compared the THAM15 to the SS15:

You may want to build your own THAM15 model from the plans.

It should look something like this with 2.83v:


Xco1's model on the first page is pretty close. L34 is a little long, do to his measuring a right corner at the end. You want to measure without the line through the corner, and then place the area line to account for the volume in AkAbak.

[Djim]

Originally Posted by Brian Steele

Question: how do these compare to a simple vented box using the same driver?

[soho54]

THAM15 vs 60l vented sub
Both with 2.83v in
Both a 6ohm minima load

[tb46]

Hi guys,

I did build my own ACAD model, and arrived my Hornresp Input data from there. I even used soho54's corner method. I also started with an overlay on Xco1's nice drawing. For general comparison, and future development I used the 4-section model. Let me know if you see any mistakes.

I agree on the internal volumes: according to Hornresp using Par for both models: THAM15=149.7 Litres / SS15=209.757 Litres. I've designed quite a few folds for the 15TBX100 that don't have the big notch. I think it is mainly a matter of increasing L45 and S4/S5, and go with a slightly larger volume (about 190 ltr, or so).

Regards,

[soho54]

It is jam up except for the S3 point. You just picked the wrong spot. If you take your model, and unfold it you will see the S3 point should be at the last bend. Right before it is a large section with zero flare around the 180deg turn, and then a constriction followed by a large increase in volume over a short distance. You have too much volume in the 3rd quarter of the horn, before the tap to model the real horn.

The total length, and horn section area numbers are spot on though, and very close to what I got from my own model.

I don't have the THAM SketchUp here, but this should show what I mean.
These two pics illustrate the base issue. It is reversed here, but I think you will get the idea. I don't have time to draw up new gifs right now.

The second one is more correct. If you kept moving to the left you would start to get extra volume on the right side of the selection point.



This picture will show how I go about it. Using the Spud here.


You can eyeball it in the Xoc1 straight horn illustration. You just have to remember his third section is to long.


The dip in the finished THAM15 is a fold translation error. HR can't quite sim it perfectly because of the non flared section. You need one more segment to completely capture it. As such the second peak(3rd Harmonic) is a little overstated, and the first peak(1st Harmonic) is a little under stated. AkAbak handles things much better here.

EDIT: also, you don't want to use the supplied BL or Mms as Mmd. Let HR calculate those for you. It makes a big difference sometimes.

[soho54]

It should look more like the purple line here.


Also in the previous graphs I posted, they were at 2.43v not 2.83v. I typo'ed the voltage when I changed them from my default 1w/1m numbers. It should also be 5ohm minima, 6ohm would be the rated spec.

I should stop posting from abroad.

[Brian Steele]

Originally Posted by soho54

THAM15 vs 60l vented sub
Both with 2.83v in
Both a 6ohm minima load

Nice. The sim of the vented output might be a bit optimistic though, considering that HornResp does not include the effect of losses. Still, it shows what looks like about 2.5dB of gain across the passband.

[tb46]

Hi soho54,

Thanks for the analysis, I'll have to chew on that one a bit :-), but that's why I'm here.

Let me just think out loud: the way I picked the S3 point, is that it is a relatively clean horn section from S2 to my S3 with only minor (90 degree) bends at a relatively squashed cross-section. I can visualize what is acoustically happening in that section. The following two bends (from my S3 to S4) are roughly 180 degrees each, and I find it hard to see how they can be modelled short of a full FEA analysis, but I admit, I have to use AkAbak more. The way I understand what Tom Danley has said about the corners and bends: it is important to keep the volume about the same as the horn would have, if it were the same length (as the centerline?) for the given hron tapers (surrounding?: before and after the bends); (this is my interpretation, from memory, as I cannot find the respective TD comment).

I like your interpretation of the SPUD, but I still feel you should be able to place the S3 point anywhere between S2 and S4, and arrive at approximately the same response. I just don't like to pick a point in the middle of a bend for any of the definition points, unless it is unavoidable. Luckily the wavelength below 100Hz are so long, that they just flow over a lot of errors. One of the differences may the use of Par v. Exp.

As to the supply voltage and conditions, I just use 2.83V and 2pi for comparison. Modern amplifiers being what they are.

Thanks again, input, I need input..... :-)

Regards,

[Brian Steele]

Originally Posted by tb46

Interesting stuff Brian, I think we're all waiting for your high SPL results.

Gotta wait a little longer - the neighbours seem to be back .

After some more fiddling around, I settled for the simplest "fix" possible - a piece of board across the 3rd fold that cut the cross-section at that point by 2/3rds. I've included an image of the change below, along with the measured frequency response before (blue) and after (red) the change. In green is the response measured at the point - I noticed that it measured fairly flat before the addition of the extra panel, and the notch appeared after the addition. All responses are 1/6th octave smoothed. Most definitely the extra panel changes the response at that location. Perhaps taking measurements at that location could help to fine-tune the modification.

There's an audible difference before and after adding the "dogfood duct". Whether or not that difference remains audible once the TH is used as part of a full-range system remains to be seen.

Concerning SPL and compression measurements, as I mentioned the neighbours are back, so hooking it up to the car amp to give it a good 100W or so was out of the question. However, I did do some testing in-house and it seems to remain quite linear up to the output capabilities of my (admittedly small) temporary amp I'm using at the moment. Basically I used HolmImpulse to measure the FR and stepped the output level up by 10dB per step, and set HolmImpulse to auto-adjust to 0dB. My thinking was, if the resulting graphs overlay each other perfectly, then compression is basically non-existent, and they did up to the limits of my amp.

If the neighbours are out tomorrow, I'll repeat the test using my car amp, which should add about 6dB of output capability.