Vintage Snell A3i review
Copyright 2011 Troels Gravesen
       

Some important links here: 
http://www.snellacoustics.com/History.asp
     http://www.stereophile.com/content/snell-type-loudspeaker-peter-snell-1946ndash1984

Lots of information on the Snell A3 as can be found on this Swedish website (in Swedish).

I had a mail from a Snell A3i owner being short of a new pair of 11F-M, H143, mids for his A3i speakers and some years ago I bought a (NOS) pair second hand - just because I liked having a pair, but eventually never used the drivers. So, NN got the mids and they will probably last some years before the foam surrounds are rotten like his old ones. I made a deal with the owner to come and hear and measure his speakers, as I never had the opportunity to hear these famous speakers of Peter Snell's hand. I guess these speakers were the ones he really wanted to build for himself, incorporation all his experiences in speaker building.

Apart from the conspicuous appearance, there are two things that make the Snell A3i unusual. One thing is having a so small middriver covering such a wide frequency range as this one does, i.e. ~275-2700 Hz. Next thing is the unusual even power response making it anything but a hot-spot speaker. If we have to look for a similar design today, the ATC speakers with their 3" dome is the way to go. The huge ATC dome covers very much the same frequency range and may be one of the reasons that some people take a particular interest in this brand, because it does sound special when a very small driver covers most of the midrange. It's close to a point source midrange.
To me this is the A3i's strength and weakness. Strength already mentioned, weakness because it puts enormous strain on the quality of the middriver in terms of power handling and distortion.

Peter Snell had read his Harry F. Olson papers of 1950 - and taken the consequence in making an almost hemispherical baffle for his midrange driver. Check the link to learn why.
This doesn't mean the middriver can be operated from a very simple crossover, on the contrary. The A3i's middriver crossover is an elaborate design with serious equalisation to the driver's intrinsic response from the use of several notch filters.
To make a smooth transition to the floor-mounted bass driver, high-slope crossovers are needed and from my measurements I think we're close to 4th order acoustically, somewhere between 250 and 300 Hz.

As can be read from the Stereophile links, some reservations were stated with regard to the A3's ability to render soundstage depth, but this was not my initial experience throwing my JVC Audiophile version of Jazz at the Pawnshop into the CD drawer. Maybe Snell dealers' recommendation of putting the A3s against the wall (- some WAF soothing bla bla..) was partly responsible for this. The A3s need to be off the front wall as do all other speakers unless particularly designed for this application.
Next the A3s will feed you tonnes of details, but this comes at a price. Generally the upper mid and treble region has too much energy for my taste and a "BBC tilt" may render a more balanced presentation of basic notes and upper tones. It just becomes too energetic on certain recordings although Siri's Killer Note passed better than anticipated.
What really made me drive 100 km to hear was the integration of the tiny 4" mid and bottom-mounted 12" bass driver. How could Peter manage to make a seamless integration of drivers' response with this relatively high point of crossover and such a distance between drivers. I think I can say he couldn't. There was a certain discontinuity in upper bass/lower mid with a lack of energy and slam on e.g. tom toms. It has to said the 12" RCF bass driver had been "re-rubbered", which may not have been for the benefit of the driver's performance. The RCF drivers originally had foam surrounds. However, bas-mid integration was better than expected has to be said, but not the best I have heard from 3-ways.
With regard to mid-tweeter integration I hold my peace. No problems here except for the choice of overall voicing, which is too bright to my taste.
The bass goes deep, but not exceptionally deep. We're down to simple physics here and the price to pay for deep bass from a 12" driver in a relatively small cabinet is efficiency. The RCF driver has a small magnet and overall system efficiency appears to be around 85-86 dB. From measurements shown below it appears we have decent bass down to around 30-35 Hz; not bad at all, but this speaker is not for flea-powered valve amps and bi-amping seems an obvious choice. Driving the mid and tweeter should be possible even from 10-20 wpc valve amps.

Is this a speaker for cloning? Yes, and no. Cloning the A3 as-is would be fairly easy and picking suitable drivers looks easy, even the middriver, as numerous really good 4" drivers are available. How about an Accuton  C-90-6-70 for mid? And a ScanSpeak 25W/8567SE for bass? It appears the original A3 had a 10" bass driver, by the way, and also used an el-cheapo Philips 5" paper midrange and why this at a later stage was replaced by the smaller 11F-M SEAS I don't know. Maybe because Philips ceased producing drivers in the Eighties - as far as memory goes. But why not the a Vifa or SEAS 13-something for mid? A 5" driver may have improved response in 250-500 Hz range, but maybe Peter Snell just loved the vivid presentation of the tiny 11F-M.

Thanks a lot to NN for letting me audition his speakers - and to his wife for baking a delicious cake for our afternoon coffee break.


Measurements


Left: Only mid and tweeter shown here, measured at meter distance at tweeter height normalised for 2.8V/1 meter.
Right: Nearfield response of drivers driven from crossover. Disregard the relative level between drivers as I was only interested in rendering possible points of crossover. The bass response was measured with the microphone as close as possible to front grille. The bass upper roll-off more seems like 6th order acoustically from this measurement.


Left: Mid and tweeter measured at 0, 10, 20, 30, 40 and 50 deg. horizontally.
Right: MT vertically measured at meter distance at M (red), MT and T height.