Fostex FE208EΣ rear loaded horns


Bluebell Audio "Loftin White" 2A3 SET valve amp
TubeCAD Headphone amplifier
Fostex T90A supertweeters
Bluebell Audio "WE91" 300B SE valve amp monoblocks

Garrard 401

Goldring Lenco GL75
Unipivot tonearm - the Stick
Passive preamp
C3g driven 2A3 SE valve amp
Gordon Rankin's Bugle - a super 45 SET amp
Aikido Headphone Amp
C3g-Aikido Phono Amp
Experimenter's Chassis
Audio Meets
REM Discography
Favourite Music
Hi Fi Links
Other Links
Contact Me






Fostex T90A supertweeters

As much as I enjoyed my rear loaded horns with Fostex FE208EΣ drivers hearing other high quality speakers such as Steve Shiels' Lowther EX4 horns and James D's Quasars I became increasingly aware of the rolled off treble of the FE208EΣ.

The obvious solution was to add a supertweeter (or supertwater to the cognoscenti :-) ). The Fostex recommended enclosure suggested a T900A supertweeter could be added to extend the treble response out to over 20kHz. Looking at Wilmslow Audio’s website the T900A was priced at a not inconsiderable £289 each. This seemed a rather disproportionate cost compared to the FE208EΣ and cabinets (under £500 total). Wilmslow Audio also listed other Fostex supertweeters though, and the T90A was a much more reasonable, though still not cheap, £120 each. Looking at the specs there didn’t seem to be a huge difference other than the massive construction of the T900As, which Wilmslow Audio confirmed.

Fostex T90A supertweeter on supplied stand

For the simple filter to connect each T90A to the FE208EΣ I got a couple of 1uF Soniqs SAX polypropyline caps from World Designs to protect the supertweeter from the lower frequencies. Before settling on 1uF I tried values from 0.22uF up to 1uF, and although the difference wasn't great 1uF sounded the best. I suspect I would detect a clearer difference over a long period.

I spent some time thinking about the resistance of the caps and the effect this might have on an L-pad attenuator but Petercom of World Designs confirmed that whilst caps do have equivalent series resistance (ESR) the actual resistance of a cap can be considered as zero. This made it a lot easier to calculate the required reduction in sensitivity of the T90A from the quoted 106dB to the FE208EΣ’s 97dB by using a simple L-pad attenuator. There are several handy calculators out there on the web. I took the formulas and put them in a spreadsheet so I could play around with values and combinations of resistors.

The filter for the T90A supertweeter. The FE208EΣ is run "wideband" with no components in the signal path.

The web calculators gave theoretical values of R1=5.161ohms and R2=4.400ohms. Resistors were Maplins 3W coffin types as I could get these quite easily. Paralleling 10ohms with 2 @ 22ohms gives R1=5.238ohms, and paralleling 4.7ohms with 100ohms gives R2=4.489ohms. The components were fastened together using the tweaker’s favourite, clip leads. This allowed easy changing of resistors to see what effect different values had on the sound. Unsurprisingly the values providing 9dB of attenuation worked best.

The supertweeter filter components

The Sound

Describing the sound of a supertweeter isn't easy. With a 9dB attenuation the integration is seamless; other than the extended treble I can’t detect there is a supertweeter taking over i.e. no nasties in the filter. There isn't a great deal of scope to setting the supertweeters up. What can make a difference is moving them forward of the front of the baffle a touch. This has the effect of shifting the focus of the combined sound of the supertweeter and main driver down towards the listening position. Aiming the T90A down towards the listening position has a similar effect.

Strangely, I lived with the rolled-off treble of the horns for a couple of years without missing the higher frequencies at all, now the sound seems so well balanced I can’t imagine listening without the supertweeters. It’s not a cheap upgrade but it has transformed the speakers.

Future Improvements

One day I must get around to soldering everything together and putting the bits in nice little boxes. It’ll be interesting to see if I can hear any difference after soldering…

The resistors are wirewound types which are increasingly inductive at the higher frequencies associated with supertweeters. In theory non-wirewound resistor may sound better, so I may try paralleling some metal films. But as the T90As come in at a frequency quite a bit lower than most supertweeters it will be interesting to see if there is any difference.

Now I’ve settled on the degree of attenuation I may try using a pair of 2 gang 6 way input selectors to make a variable attenuator of small steps, perhaps 1dB, to see if a small change is beneficial, or indeed audible.

The T90A in residence



Bluebell Audio WE91 300B SE Valve Amp Monoblocks

TubeCAD Headphone Amplifier