A 3-way built around the Fostex FE208ES

[Mario Pankov]

This is a multiway thread but I am intentionaly posting it in the fullrange forum, why will be seen below.
After many experiments with crossovers and a system based around the commonly spread woofer/midrange/tweeter combination, I am looking into a tweaked version of this allignment.
A friend of mine would ask me to build his system and following my request to experiment with his speaker, agreed that I build it in the way I want and see if my observations are in the right direction ( well, he`ll have ot buy the drivers so it was generous...).
I want to build a system around a fullrange driver. I believe fullrange is the only speaker able to be used as a very wide range mid-driver without noticeable shortfalls. 1st order with a sloped baffle is what I plan to do. After a review of quite a lot of drivers ( and without having heard even a single one ) I think I will go with the Fostex FE208ES. I like this Sigma driver - it has lots of Sd, works quite well ( based on comments from Mr. Pass ) and has good dispersion up to 10Khz. It is said to have some of the best midrange from the commercially available speakers. After some simulations with It I see that in a 3.5l sealed box its rolloff is -10db at 100Hz. Driver is rated at 97db, quite sensitive, but looking at the Fostex graph I think this is reached at freq. above 5Khz and the rising response looks quite consistent and at about +4-5db. A bigger coil or a shelving network should clear this out. Natural roll off starts at about 11-12Khz, which makes it ideal for my application.
The Scanspeak 25W/8565 ( I have heard it at a friend of mine ) is possibly the ideal companion and can do sealed/variovent. Crossed over at 80Hz, at least based on simulations, it would give an excellent acoustical 1st order Butterworth at 100hz to the Fullrange unit. Rolloff of the Fostex in the sealed box is without a crossover and has a 12db slope. This gets a bit acoustically asymetrical but is fine.
A tweeter ( a ribbon ) would be mated to the Sigma, again with a 1st order somewhere at 9-11Khz, based on listening tests. Acoustical on the fullrange should exceed 6db/oct which is good.

Question is regarding the Sigma driver. My only concern with it is the 1.2mm excursion. I am not sure how it would cope with the 200-600Hz range, up to 200hz the woofer would add up body to the sound, but above 400Hz it should be 12db down.
Also, how would you comment the use of this driver in a sealed enclosure down to 100Hz? I would appreciate input from people that have played with the 208 Sigma.

[DaveCan]

Dick Olsher did a 3 way with that driver some years back that's in high regard by many, here's a few links: Black Dahlia Acoustics - Loudspeaker Kit Plans

Tip_78

[dabbler]

Download WinISD Pro. It has the FE208Es in the driver database and you can plug in different values for power, simulate crossovers then check against the cone excursion graph to see when xmax is exceeded.

[901Fixer]

I agree this is a great idea. I'm working on a two way FAST using the Dayton PS220-8 and Dayton 12" ST305-8 woofer. The PS220-8 is a bit unruly without some attenuation but it should be minimal.

If you plan ahead and build an enclosure optimized for the full range with the woofers some how "added on" you can listen to just the full range without the woofer support when need be. Have your cake and eat it too!

[Greg B]

Having built a number of quasi 3 ways like that, and I have a few comments:

1. You are neglecting the baffle step. An additional 6db of rolloff to mid curve below this point. F3 = 115/width (in meters) is a good rule of thumb, typically around 300-500Hz. '

2. There is a huge gain in efficiency by crossing the helper woofer to the full range at the baffle step.

3. A passive electrical first order highpass at 100hz will not work at all for the mid/FR, because the impedance curve is nowhere near flat at that point. Putting the mid in an aperiodic or (better) a stuffed pipe will flatten the Z enough to make it work, if the crossover is higher.

4. You will need about 4 of those Scanspeaks to keep up with the 'mid'. A 10"/25cm woofer pairs better with a 4 or 5" FR/mid. You might consider the FE168 instead, to provide some extra headroom, but it will still need to be padded down.

5. Finally, it's pretty tough to make a ribbon sound good as a supertweeter. IMO and IME you'll get much better integration with a horn supertweeter such as the T90A. Bang/buck is also more favorable.

[901Fixer]

Great info!

1. My cabinets are ~22" wide so baffle step will be ~200Hz, correct?

2.ST305-8 is good with this.

3.PLLXO or FMOD for the PS220-8 in a MLTL. Plate amp with built in active XO for the woofer, in bass reflex.

4.95dB and 92dB efficiency.

5.I have more tweeter than I need with the PS220-8, it will have to be attenuated.

[Greg B]

1. My cabinets are ~22" wide so baffle step will be ~200Hz, correct?

Yeah, thereabouts.

PLLXO should work fine.

5.I have more tweeter than I need with the PS220-8, it will have to be attenuated.

Being that it is only around 90db at ~400Hz, and 103db at 3500Hz, I'd say your issue is climbing response rather than sensitivity. 2-3mH inductor in series would flatten it, maybe bypass with a small cap if the upper octave gets attenuated. Or use EQ. That peak at 3500 looks like it'd be audible.

Might be a fun driver to put in a ~200hz front horn, which would also flatten the response by bringing up the bottom end. Load the hole thing into corners. That's what I would do.

[Mario Pankov]

Good pints Greg, I have taken into account the BF but needed to get the drivers and do the measurements. After playing with WinISD it appears that a 1st order at around 150Hz might be a good starting point to protect the driver from over-excursion. However, the impedance peak at Fs and in a sealed will be too close so how effective this cap will be is questionable.
The mid will be padded down and possibly shelved with a bigger coil to compensate for the rising response. Overall, looking at the factory measurements, I see it to be at around 93-94db sensitive. Will have to reduce this significantly.
I understand your note about the Super Tw. but do you not think a 1st order crossover might not be enough for it? I have never played with supertweeters but Ihave noticed that they improve both the high and mid frequency ranges which makes me think the factory suggested and used crossovers are possibly a 6db/oct?

[Greg B]

how effective this cap will be is questionable.

Not in my mind. It will basically track the impedance curve. But don't take my word for it, try it out. If you flatten the impedance with an aperiodic or TL enclosure, or an RLC trap, it can work.

1st order will work OK on a ribbon if it is efficient and robust, and if you cross very high with say a 1uF cap. I just don't think they sound very good as supertweeters. The pattern is hard to integrate. I love my ribbons when crossed low as conventional tweeters. YMMV of course. I preferred a $10 CTS piezo to an AC G3 in a direct comparison.

[Mario Pankov]

RLC trap at this frequency would consist of quite impressive values Anything different from a sealed box is not an option, problem is size. The ss woofer needs at least 100l, the final speaker will look far too big I think. Most ribbons have higher efficiency that the targeted one, a 1st order will be somewhere at 10Khz which makes it 12b down at around 2.5k, most of them can do a 3rd order at 2.5-3k and get up to 60-70W ( at least what manufacturers suggest ), this is twice over what I will use. Can you comment on this impression?
Do you think the Sigma will have enough fundament in the high bass region? It`s got Sd but at 1.2mm xmax worries me a bit.