OBL-7, open baffle experiments

Copyright 2007 Troels Gravesen

Goodmans Axiom 150 mkII + Supravox 215RTF64 + JBL LE26-waveguide

Also view OB9, 2009 open baffle construction and OBL-11, the final OBL

The OB7 is my first attempt to make an open baffle speaker and having a range of suitable high-efficiency drivers, making a pair of suitable baffles seems straight forward. My role model is the vintage Wharfedale SFB (sand filled baffle), which can be studied in detail from here.
The baffles I made are a little bigger than the SFB as the drawback of open baffles is size. Ideally the baffle should infinite, but less can do as can be seen from commercial open baffle speakers and quite a number of diy project that can be found on the web. Having a narrow open baffle, e.g. Jamo, calls for more bass drivers and serious equalisation - and frankly - the efficiency of the JAMO open baffle system is not impressive. My target here is something in a little above 95 dB/2.8 volts and my expectation was limited bass extension, but this proved not to be as severe as feared. Having the bass driver close to the floor and the baffle flush to the floor adds quite some room gain and the quality of the bass is quite different from anything else we may be used to. Although deep bass may be low in level, it feels like the bass driver just goes on and on downwards. No 24 dB roll-off as is so well-known from vented enclosures.

The LE26 fitted with a waveguide makes a suitable option for high-efficiency systems, read LE26 article. Details on the Supravox 215RTF64 can be found at Supravox website. A classic, extremely sensitive high-Qt driver. Goodmans Axiom 150 mkII, details can be found here.
Given the TS data, this driver may not be particularly well suited for open baffles. Drivers should have a high Qt for OB application. However, they do darn well.

The crossover

We do need a few crossover components. I've tried running this from a single cap to the tweeter and it just doesn't work. The Supravox has a rising response towards higher frequences and makes the upper treble too aggressive. From a single cap (3.3 uF) the LE26+waveguide does well down to 1-1.5 kHz, where it really shouldn't mingle at all. And the Goodmans does well up to 2.5 kHz where it should leave it to the Supravox. So, basically the bass and mid now runs 1st order after adding impedance correcting RC circuits and the tweeter needs a 3rd order high-pass filter at around 6 kHz to mate the Supravox. The Supravox is really handling most of the treble area here. Let's reiterate frequency bands, so we know what we're talking about:

The simulation shown above is made from measurements done on drivers mounted in closed or vented cabs. Measuring drivers on an open baffle is major trouble as has been discussed in Acapella files and the Dipole Study. Read intro page for links.
First of all I wanted to try the most simple set-up possible and by this I mean the Supravox run almost fullrange with a point of crossover as high as possible like the French PHY-PH drivers. By adding an impedance correcting circuit (10 ohm + 10 uF across speaker terminals) and a series 0.47 mH coil, the Supra will run to 5-6 kHz and display a ~3rd order roll-off at around 7-8 kHz. The LE26 tweeter needs a similar 3rd order high-pass filter to match the Supra. No major problems here.

The big question in this set-up is the Goodmans bass driver. It's a capable driver up to 2-3 kHz but it doesn't sound good playing upper mid/lower treble. The 3.9 mH coil shown here makes a rather shallow roll-off of 6 dB/octave starting around 200-300 Hz, thus still provides quite a contribution to the whole midrange. Problem with this is that it's slightly out of phase with the Supravox. Adding a high-pass filter to the Supravox would mean inverting polarity of mid and tweeter and I didn't want to do this. The bass and mid are both connected with positive polarity and provides a bass membrane area close to a 15" driver.

The baffle

Baffle construction should be self-explanatory. Two layers of 22 mm MDF were used.
Rear panel is added egg crate foam to reduce reflections between panel and rear wall.
And the final speakers including drivers are heavy like h....I had to remove the Goodmans
drivers to move the panels from the workshop to the living room, not to break by back.

I told you! WAF is ultra-low.

So, how does these monster panels sound?

Well, I'm ready for surprises when it comes to speakers. I've been reading a lot about similar systems using the single-cap approach, and I've always been more than sceptical and the first single-cap run confirms my scepticism. At least for these drivers it doesn't work.
The sensitivity of this panel is phenomenal, reaching a healthy 100 dB/2.8 volts in all of the midband. Actually the JBL tweeter is working hard to make up to the two other drivers, but despite the apparent sloped response towards the top octave, treble seems to be well balanced from having no attenuation on the tweeter.
Reaching ear-shredding levels, I measure no more than 5-6 volts peak RMS on speaker terminals, accounting for some 5-6 watts. This means that even flea-powered amps can be in here and I haven't been able to drive the MAE1 amp to any noticeable distortion levels.

Obviously there's no boxy sound from this panel. The bass comes out clear and punchy and listening to acoustic bass is most enjoyable. It's also obvious that the bass is not as deep as had the Goodmans been placed in an e.g. Onken vented box. So, how about the mid running so high? Well, I hear treble crisp and clean, but dispersion - or lack of the same - in upper mid and lower treble is very noticeable. In that sense it resembles electrostatic speakers. Move your head a little and the soundstage changes a lot. There's a hot spot, no doubt about it. However, due to the dipole status, this is more tolerable than expected. A dipole tweeter might be interesting to try, but where do we find a suitable planar at this level of sensitivity?
Transparency is excellent but suffers from near wall placement, and it's obvious my current placement is not ideal for the panels. They need more space to the rear wall and would probably perform even better from a placement like this:

Should you by coincident have these or similar drivers at hand, try it out and prepare for some enjoyable moments. Be also prepared to experiment a lot on room placement and possible damping sheets on nearby walls. It may shake your views on conventional hifi speakers, what they can and what they certainly cannot. Be prepared that WAF is ultra low. And it takes will not to play these panels loud - because they can play very loud - so take care of your hearing.


Left: SPL from "single-cap" set-up. Doesn't look too bad, but the 5-10 kHz region is too high, making an aggressive sound on most recordings. The LE26 contribution all the way down to 1-1.5 kHz may be part of the problem.
Right: Impedance of system from "single-cap" set-up. Shouldn't be a problem for any amplifier. This is a 5 ohms speakers due to relatively low impedance of the Supravox.

Left: Response from Goodmans Axiom 150 mkII with and without crossover.
Right: Measurements made at 0.25 meter of all drivers with crossover - and heavy smoothing. Possibly the LE26 response can be further improved. Inverting tweeter polarity suggests decent performance.

Left: 64 ms reading at 2 meter distance with octave smoothing of response in workshop. Left and right speaker shown. Now, my workshop is definitely not ideal for these panels with an average huge lift in the response from 100-400 Hz. However, I do warn not to put too much into these measurements. Sounds better than it looks.
Right: Impedance of system with full crossover. Still an easy load for any amplifier.

Philips 9710 fullrange on OB7

I've had a request on setting up my 9710s on the OB7 and give my opinion on performance. Without too much trouble the 9710s would fit into the Supravox rebates.
I started out with my current classical favourite: Heinz Holliger plays J.S.Bach/3 Oboe Concertos/Philips/6514304. Vinyl obviously.
Playback level was moderate. After listening a few minutes on-axis I had to leave the room because my ears started hurting! If we look above, we see the frequency response of the 9710 (not from the OB7) and we have the whole 2.5-15 kHz region some 5-10 dB above average level. We have a huge peak at 3.5 kHz, right where the ear reaches it's highest sensitivity and where it hurts the most if things don't perform smoothly.

Some good things too. The bass isn't bad at all despite a modest membrane area trying to push the room air. It gives you a nice idea about what is going on in the lower registers. The whole of the midrange is good too. No problem here. Perspective is seriously compromised from treble level being way too high. Treble as such isn't bad at all, just much too much of it. I wasn't even thinking in terms of tweeters to supplement the 9710. Going back to my working desk - in the next room - and "listening-from-the-other-room" actually was pretty joyful. I can imagine the 9710 pointing upwards in the vintage Carlsson construction wasn't bad at all because you would never meet the 9710 "face to face".
Long time ago I had a mail from a guy in Indonesia using two of these 9710s - connected in series - on an open baffle. And he was writing positively about their sonic performance. Well, I have my doubts. By any standard it's seriously low-fi.

This experiment took me to the next one: The Supravox run fullrange. Supravox 215RTF64 directly connected to my 6AS7 single ended triodes. So back in went the Supras without any equalisation at all.

Supravox 215RTF64 fullrange on OB7

To the right the frequency response of the 215RTF64 is seen. A rather wide measuring window is used with modest smoothing.
This driver does well all the way up to 9 kHz with no nasty peak above 10 kHz. It simply dies from lack of ability to move mass above 9,000 cycles per second. Excellent! And very much like the French PHY-PH/H 21 LB15, which I'd like to hear. The guy asking for the 9710 set-up also told the PHY-PH drivers, despite showing similar frequency response as seen above, didn't need a tweeter at all. Well, I guess this Supra is a more affordable way of rendering a taste of the PHY-PHs. The Supras run fullrange are way better than the 9710. No doubt about it.

With a frequency response up to 9 kHz, most treble information is there. It sounds slightly "vintage" if you get my meaning. Not like old AM radio, far from, but you miss some airiness enhancing the spacial properties of the presentation. I can to some extent understand why "uncompromising" audiophiles settle here. "Uncompromising", because this approach is full of compromises - to my mind. I too like things to be as uncomplicated as possible. Simple solutions can hold a lot of beauty. But the price to pay for this single driver set-up is high.
Adding the equalising circuit seen below, provides a presentation being smooth and without edges. Simply adding a coil won't work, because we'll start loosing treble already at 6-7 kHz. With a series LCR circuit we can maintain full level up to 9 kHz.
The Supra on a 1.5 m^2 baffle simply lacks bass. Adding more baffles doesn't help. In this respect, the 9710 appeared slightly better. The Supra now sounds like what it looks like on paper - an extremely broad-banded midrange driver. And it's good - very good.

Inserting the Goodmans again via the 3.9 mH coil including impedance correcting RC circuit is a major step up. From resembling an indeed very good "transistor radio", it now has body and weight and gain momentum in the entire frequency band it covers. Listening to my favourite jazz LPs, I still miss the airiness of the cymbals and instruments with high frequency information. They become distant and somewhat closed-in due to lack of the upper octave. Very noticeable on piano too, by the way.
Well, unless we can find a fullrange driver including a whizzer cone that performs e.g. +/- 2 dB from 500 Hz to 15 kHz, the single driver concept calls for serious compromises. Living with a frequency band from 100-10,000 Hz is like going back to the Fifties. They made some terrific recordings in that late Fifties from which we today can fully extract all the goodies if we don't necessarily also adopt the standard household music appliances of that time. We can do much better - and with simple means as has been demonstrated.

The final experiment

After the fullrange experiments I went back to the 3-way system and as a final option changed the point of crossover between mid and tweeter to approx. 3.3 kHz. Impedance still remains good although dipping to 3 ohms at 8 kHz. My amps didn't mind.
What immediately was clear going from a) Goodmans + Supra fullrange to b) Goodmans + Supra up to 6.5 kHz and finally to c) Goodmans + Supra up to 3.5 kHz was the gradual increase in level of transparency. The sensation of detail and micro-dynamics gradually increases taking this journey and somehow I feel certain this is not due to reduction in rear radiated energy reflected back from the rear wall. It's more subtler than that. The way 8" fullrange drivers radiate energy in the treble area is my prime suspect parameter. No matter how an 8" fullrange driver is made, beaming will start around 2 kHz, where the wavelength is smaller than the cone diameter. For the Supra this will occur already below 2 kHz having a cone diameter of ~17.5 cm. Beaming smears detail and makes complex musical events sound congested and distorted when we reach a certain level of playback. I've read the same reservation in a PHY-PH review in a Swedish magazine. This from a 12" PHY-PH driver. Taking a 12" to 7 kHz calls for a few compromises. What usually knocks people off their feet from these high-efficiency drivers is the dynamic properties, the phenomenal transient attack - but again, it comes at a price.

I do not hear any disadvantage of using these drivers with conventional crossovers. For the purist, even running the Supra to 6-7 kHz may be acceptable. For most musical genres this provides significant improved performance compared to the non-equalised fullrange presentation. For the Supra, very few components are needed to render a balanced sound of instruments and vocals. There's no reason these drivers cannot be used in ways where we can start talking "high fidelity" in the true sense of the words. I mean, the musical signal is passed through countless wires and connectors, transistors, valves, resistors, capacitors, etc., before reaching the speaker. Not equalising frequency response from your loudspeakers - if need be - is like drinking an expensive vintage wine from a mug.