My Dream Project: Planning a TL with Fostex Full Range Driver

[Hari Iyer]

This project has been on my mind for atleast 4 years now and I think that it will be taking shape in a couple of weeks /month(s).

I am considering the Fostex 6.5" FR driver FE166en for this project. I also did analysis of other FR available like Tangband 5" FR (tough decision to not buy this one), Dayton 7" (did not want the sharp / bright sound signature of a neo driver), Voxactive & lowthers (high cost). The Fostex sigma could have been one more option but would require more passive compensation due to bright sound signature above 5KHz and resonance issues. The FE166en was reasonably flat in this region and has a better cone control at this freq.

Also the FE166en has a lower xmax of only 1mm which will make the mids and the highs less distorted due to no rapid cone motion (low xmax).

Now another bigger aspect is the box loading. Lot of discussion on other forums about FE166en is suitable only for a BLH or a frugal horn. But unfortunately the dimensions of the speakers become too huge for my home and will have a poor WAF. Hence i have no choice but to go for a traditional TL (Alpha, MLTL or TQWT). The stuffing requirement for MLTL & TQWT are more of a trial and error to achieve a critical damping. Alpha TL is a bit overdamped and i am not sure how it will work if i remove some damping to make it critical. Still am considering various loading options.

I had ordered this drivers 3 weeks ago and the driver arrived last weekend. I quickly done a break-in and measured the TS parameters. I found the measured parameter agreeing to the published values with only 5% difference overall. It will be quite safe to use the published values imo for this driver.

I have done a rough-cut drawing for my proposed box and have not started any work till now on them. Hence am open for suggestions from FMs for the box loading and TL geometry. Will post the box drawing later on the day. I will start building the box only after 2 months (October) after i am back to Mumbai from Bangalore.

 

[jls001]

Interesting project, Hari :thumbsup:

Roughly what kind of box dimensions did you arrive at in your sim? And how low will it go?

 

[Hari Iyer]

Thanks Jousha.
Cross-section is around 82 sq.in (golden ratio). Line length is around 43 inches. After the spikes the final height is 45 inches. Tuning freq is 72Hz. Hence i am expecting half octave below 36Hz @ -6dB reference level IME. Damping density is around 1.75 pounds/cubic feet.

 

[Hari Iyer]

Those who have followed my speaker building posts in the past would know by now that i do a lot of homework before I actually start building the speakers.
First i filter and narrow down on the driver by having a look at its published TS-parameters, and after that lots of what-if simulations on the box dimensions and the equalization / crossover network and later look at various objective simulated measurements of FR, impedance, step response, impulse response, square wave response. After i am objectively satisfied with the results go and decide to purchase the driver.

I have noticed that the measured TS-parameters differ around 10%+ with the published results, but nothing much can be done about it and we will have to live with this issue. But to my surprise the FE166en measured quite close to the published values and FMs who wish to try with this driver can safely consider the published values.

Below are the simulations that i get with the Fostex full ranger after using my equalization network.

A 3 dB equalization for the baffle step and room correction is planned.

Flat impedance curve and a close to resistive load for the amplifier

The impulse response is very clean with the rise time and fall time amplitude is very symmetric

The square wave is very symmetric with the rise time and fall time in sync

step response is shows a near perfect right triangle

The above responses show the coherence and integration in both frequency and time domain and have potential to sound equally good.

Only after subjective listening we will be able to co-relate the objective findings and tweak for any room anomalies.

Am excited to start and complete this project asap but am constrained due to work related priorities.

 

[Hari Iyer]

I have been debating about the box geometry and loading in other forums and am soon am narrowing down on a ML-TQWT enclosure for the Fostex. The Alpha TL is a bit overdamped and the qualities of the full-range driver will be compromised in an overdamped enclosure.

The ML-TQWT is a cross between a horn & TL and it behaves like an oscillating horn. At times this enclosure will have a tendency to boom and can be dominating in a low-end on a particular tuning resonant mode. Careful location of the pipe, the pipe size and internal stuffing can do the trick of a well balanced system. Keeping my fingers crossed as of now.

 

[Sumanta]

Good going Hari.

 

[Hari Iyer]

I have been doing box simulations last and this week and still not able to decide the geometry and loading profile. I simulated with TQWT, ML-TQWT and MLTL. All gave me results that still require many tweaking. I think this is going to take longer than anticipated.

In the meanwhile i have also modeled my current TLFS with the software and results shows some tweaking will be required to tune this box too. So i may go ahead with modifying my current box and then apply my learning to the Fostex. This way any chance of possible errors can be reduced in the new box which usually are difficult to implement once built.

thanks for looking.

 

[venki7744]

Hi Hari,

Really amazed by the amount of planning that goes into designing your speakers . I just spend sometime designing then right away start with the build. Do you mind sharing the software you use to model the enclosures as most of them (read Free) only simulate sealed and ported. Few others that model the response that I know are Hornresp and Akabak, but I don't think these are capable of modeling BS and overall response.

Cheers,
Venki

 

[mpw]

good luck and have fun !

 

[Hari Iyer]

Hi Hari,

Really amazed by the amount of planning that goes into designing your speakers . I just spend sometime designing then right away start with the build. Do you mind sharing the software you use to model the enclosures as most of them (read Free) only simulate sealed and ported. Few others that model the response that I know are Hornresp and Akabak, but I don't think these are capable of modeling BS and overall response.

Cheers,
Venki

Hi Venki,
You can check Leonard Audio (LA) software. Its primarily a tapered quarter wave TL simulator. But you can also design sealed, Bass-reflex, Horn and TL (tapered, straight, exponential) with this software.

IMO this software is good but has a flimsy interface and there is lot to learn from stuffings and tuning provided you do that correctly.

Good luck.

 

[venki7744]

Hi Venki,
You can check Leonard Audio (LA) software. Its primarily a tapered quarter wave TL simulator. But you can also design sealed, Bass-reflex, Horn and TL (tapered, straight, exponential) with this software.

IMO this software is good but has a flimsy interface and there is lot to learn from stuffings and tuning provided you do that correctly.

Good luck.

Thanks Sire, I have already used LA for couple of projects in the past along with Hornresp. However I would really like to know the toolset you use to determine the expected response of a loudspeaker system. E.g. the shape of enclosure, size of port everything would affect the Baffle step response and in turn affect the overall far field response. Do you actually build a prototype of your design, measure and then fine tune it or do you check all these parameters in simulations, build the enclosure and finally fine tune it to meet your requirement. Sorry for asking too many questions .

Cheers,
Venki

 

[keith_correa]

The Leonard Audio TL software has an excellent interface IMO. There is no comparable software for a similar function that provides the ability to add/size/resize/draw sections/ports and their locations by copying/pasting/dragging/dropping. But the BIGGEST drawback of it is that it doesn't simulate the effect of stuffing accurately. If one really knows what they're doing one can second guess what one will need for stuffing.

The best software for doing accurate sims is Martin J. Kings Mathcad sheets. No GUI though. Not free either. But it is the most accurate there is. Initial sims using the LATL software [because of WYSIWYG to an extent] and then fine tuning it with MJK's sheets is the way to go. MJK's MathCad sheets also can simulate to an extent the effect the location of the speaker will have in a room. A lot of time/research has gone into it - superb stuff!

My 2 p worth.

 

[Hari Iyer]

MJK mathcad sheets are good to simulate the response IMO but the LA gives very good approximations too. I am not sure about Keith's personal experience about the stuffing simulation. I am yet to build and measure as per the software's suggestion. Hence it will be unfair on my part to comment on them.

What i will be doing after i am in Mumbai - I will model my current TLFS speakers using the LA software and measure them using REW and check for the tuning and stuffing. If the simulations and measurements agree within 5% i am okay as there will always be a margin of error in the TS parameters measurement too which we need to factor in the calculations.

@ Keith - Have you simulated using the LA software and measured using any objective measurement software like CLIO or REW? Were you getting too much different results on the impedance / phase curves. Did you also measure with and without stuffing to check the result. Keen to know. Please advise.

Thanks Sire, I have already used LA for couple of projects in the past along with Hornresp. However I would really like to know the toolset you use to determine the expected response of a loudspeaker system. E.g. the shape of enclosure, size of port everything would affect the Baffle step response and in turn affect the overall far field response. Do you actually build a prototype of your design, measure and then fine tune it or do you check all these parameters in simulations, build the enclosure and finally fine tune it to meet your requirement. Sorry for asking too many questions .

Cheers,
Venki

The LA itself is a tool to simulate a horn / TL / Bass reflex / Sealed enclosure. You need to play with various box geometry / tapers, position of the drivers, port size, stuffing to arrive at the tuned response. The Baffle step response is more depended on the room mode than the tuning of the box imo.

To get the correct tuned response without stuffing is important. Its important to see the terminus response for any noise. Also adding the right amount of stuffing is important to get a well balanced SPL and impedance curve.

Start with the white papers published by MJK and understand the theory behind the TL anatomy. Once you are familar with that only then you will be able to know what to expect from the LA.

Hope this helps.

 

[keith_correa]

@ Keith - Have you simulated using the LA software and measured using any objective measurement software like CLIO or REW? Were you getting too much different results on the impedance / phase curves. Did you also measure with and without stuffing to check the result. Keen to know. Please advise.

All I've done is to compare multiple MLTL simulations between LATL and MJK's sheets along with a TL guru [not from this forum] who has extensively designed MLTL's for years using MJK's sheets, builds his own speakers and who has also designed MLTL's for commercial speaker "shops". We concluded that the MJK and LATL sims without stuffing were 99% - 100% similar. Stuffing is where the LATL went berserk. Nothing matched.
What exactly are you hoping to achieve/tweak by measuring in room and then adjust stuffing? Adjust towards achieving what? The "TL guru" whom I extensively interacted with said that once simulated using the MJK sheets and stuffed using the proper density stuffing he never changes stuffing. All the rest is taken care of in the crossover. I never questioned that because maybe I didn't know how to ask the right questions being the novice that I am.

Measurements are currently beyond me - am keen to start doing this once I start to build the definitive speaker for my room. Soon. Maybe.

 

[venki7744]

I had done some comparison in the past between Hornresp simulations and LA response. Both are different, I had asked this question (about the difference in simulated response) in another forum but never really got a proper response. I have two speakers sitting in my room both designed with LA one a tapered sub and the other a straight TL. Unfortunately I am not finding enough time and patience to measure these using REW. I would love to but then true FR can be only obtained by properly splicing the near field and Far field response. There is a white paper on this published by Jeff Bagby, using the baffle step simulator and FRD blender from audio.claub.net. Yet to experiment with it.

PS: I first learned about TL through MJK site but his spreadsheet not being cheap prevented me from buying it .

Cheers,
Venki

 

[keith_correa]

Let me understand one thing upfront. What does no/less/more/adequate stuffing affect in lets say a MLTL?

 

[venki7744]

Let me understand one thing upfront. What does no/less/more/adequate stuffing affect in lets say a MLTL?

My basic knowledge tell me that stuff would affect the overall efficiency.reduce ripples in the FR for higher frequencies (or rather damp or block HF resonance in the chamber) and also extends the lower range a bit since stuffing makes the box appear larger.

 

[keith_correa]

I corresponded again with my friend, the TL "gooroo", about the need for experimentation with stuffing and he says that was the case in years past when the math was not "exactly" in place and so things were a bit of hit and miss. He firmly believes that with MJK's worksheets there is almost no need for multiple iterations of experimentation with stuffing. Yes, there is a price to pay for the worksheets but that in his opinion is small considering the time it saves and it's accuracy. He says that there are a lot of excel sheets, pdf's etc available for free that will allow one to build a TL and will allow one to come close to 90% but then for the remaining 10% one has to pass through multiple iterations of stuff/listen/measure etc so ultimately it depends upon what one wants/likes to do.

 

[Hari Iyer]

Let me understand one thing upfront. What does no/less/more/adequate stuffing affect in lets say a MLTL?

You need not measure the FR response to check the TL measured and simulated response. The FR response will not show much about the tuning and stuffing. This will be shown by the impedance curve.

You need to measure the impedance of the speaker with stuffing (two humps), low stuffing (one hump low and other hump high), adequate stuffing (one clean hump without any distortion on the curve ie. critical dumping), over stuffing will be indicated by the impedance @ resonance close to Free Air impedance.

In critically damped TL the impedance @ resonance (tuning freq of the line) is atleast 40% to 50% lower than the Free Air impedance.

my 2 p.

 

[keith_correa]

You need not measure the FR response to check the TL measured and simulated response. The FR response will not show much about the tuning and stuffing. This will be shown by the impedance curve.

You need to measure the impedance of the speaker with stuffing (two humps), low stuffing (one hump low and other hump high), adequate stuffing (one clean hump without any distortion on the curve ie. critical dumping), over stuffing will be indicated by the impedance @ resonance close to Free Air impedance.

Yes, I know but my laptop does not have line in hence cannot measure impedance. I think the jig needs Line-in + Line-out to measure, right? At least that was what I'd concluded when I looked it up a long time ago.
But my point is - the measures need not really be done - if one uses MJK's sheets - you're set. There are enough people who have verified this. Of course one can always go OCD in which case this never ends.

All this measuring, tuning etc only comes when you use the classic alignment tables or Shultz's paper for designing boxes for TL's - those are a bit of hit and miss.