Getting to know LS3/5As

[S-Man]

Here’s a ‘psychoacoustic’ smoothing which might make it easier to gauge the character traits. I suspect the big tonal difference is between 350Hz-1kHz. The crossover region around 3-4kHz is noticeably different too, and again I suspect there is more to be told there as that is the make/break of any multi-way speaker to my ears. This smoothing also highlights the LS3/5A’s bass bump at 110Hz or so.

Interesting comparison!
IMO:
Up to 200Hz can be changed dramatically by positioning/room.
The region between 300Hz and 900 Hz will be the most obvious difference. There is 3 to 4dB difference over >1 octave.
3KHz to 4KHz is < 1 octave so the 3dB difference will be less audible... unless the off-axis curves (power response) are very different.
The 1.1KHz peak of the LS3/5A might cause brass instruments to sound more brassy.

Altbough they have the same output at 1KHz, so nominally would have the same specified sensitivity, I would expect the JRs to sound slightly louder.

 

[Tony L]

3KHz to 4KHz is < 1 octave so the 3dB difference will be less audible... unless the off-axis curves (power response) are very different.

True, but that dip or rise may have something to do with phase doubling/cancelling etc as it is the area both drivers overlap, and to my ears that is often way, way more obvious than any tonal stuff elsewhere. These two speakers both have a really good crossover region to my ears, but many don’t, and whilst that area is only an octave it is right in the vocal & sibilance region so we are ultra sensitive to it. Many speakers are made or broken right there!

PS Yes, the 149s are subjectively slightly louder, plus maybe a bit more dynamic, though the LS3/5As are still surprisingly good on drums etc. They have some punch on the right stuff (that 1kHz lift will help there.

 

[Tony L]

Thinking some more I’m curious why the BBC stuck that 1kHz lift there? I’m sure they did so deliberately as the whole design process was just absurdly cost-no-object (over £1m development costs in today’s money). If they didn’t want it there they’d have notch-filtered it out!

PS - I assume the Falcons were measured with tygan grilles on?

Sorry, missed this previously, yes, both speakers grilles on. The LS3/5A grille is an important part of the BBC design, take it off and it isn’t an LS3/5A! Obviously I can’t guarantee my JR149 grilles are identical to the 1970s originals as none survive, but I can’t see them being much different. They were always acoustically transparent, I couldn’t hear them with my first pair back in the ‘70s when I had good young ears, I sure as hell couldn’t hear them now!

 

[Stunsworth]

Thinking some more I’m curious why the BBC stuck that 1kHz lift there? I’m sure they did so deliberately as the whole design process was just absurdly cost-no-object (over £1m development costs in today’s money). If they didn’t want it there they’d have notch-filtered it out!

To make vocals a little clearer?

 

[SwapShop]

To my mind the name is just a marketing exercise. A very nice little speaker for sure, but it isn’t an LS3/5A, and it has no direct connection at all to the BBC research department. Tonally similar, but I’d be amazed if you could use a Stirling on one channel and a real LS3/5A on the other and they’d be pair matched, and that is what the BBC spec actually requires! Some folk seem to like them better, so I’m not criticising the speaker itself, only the marketing.

Ok but they do have a direct connection. As I understand it Stirling used to make standard LS3/5a's until they ran out of KEF drivers.(Falcon now manufacture replica drivers). The V2's were approved by the BBC as tonally equivalent, and use adapted SEAS drives to replicate the BBC sound. The rest of the construction is to LS3/5a spec. No I wouldn't want to use them as matched pair with other LS3/5a's but as i understand it all LS3/5a's sound slightly different anyway.

The sale of all hi fi is ultimately a marketing exercise really.

 

[Tony L]

Ok but they do have a direct connection. As I understand it Stirling used to make standard LS3/5a's until they ran out of KEF drivers.(Falcon now manufacture replica drivers).

Again a marketing connection. The LS3/5A was a BBC in-house project. It was developed at the BBC Research Dept at Kingswood Warren by BBC engineers and after the design was formalised to an absurdly tight degree third parties could apply for licences to make it. Everything was nailed-down, the drivers weren’t just Kef B110s & T27s, the former were carefully selected, the latter slightly modified, every detail of the crossover and cabinet was precise to an absurd extent. Change any aspect and it isn’t an LS3/5A.

The early Stirlings were late-period 11 Ohm LS3/5As as they were made from redundant stock from one of the original suppliers. They should be just the same as a Rogers or Spendor of that period. The V2 is but a marketing exercise, it is not an LS3/5A as it doesn’t contain the bill of materials one needs to make an LA3/5A. Sadly the BBC, now they no longer use LS3/5As (they moved to active Dynaudios decades ago), seem happy to sell licenses. They no longer use the speaker for monitoring so no longer need to care about spec, the perfect pair matching between random units etc.

As stated earlier I’m sure the Stirlings (and Grahams) are very nice speakers indeed. I’m not knocking them beyond the marketing strategy. I’m just arguing that they aren’t LS3/5As, as, well, they aren’t! I’m prepared to bet if you stuck a Stirling V2 or V3 on the left and a Falcon on the right they wouldn’t behave as a matched pair any more than if I say stuck a JR149 on one side and an LS3/5A on the other. An LS3/5A needs to be an LS3/5A, the BBC ensured the spec was tight enough that if one broke in the field you only replaced it, not the pair. Any two, regardless of brand, was a matched pair! I bet the active Dynaudios they use now meet this requirement too.

 

[Tony L]

Ok graph fans, incoming!

LS3/5A RT60

JR149 RT 60

Not sure how to interpret these other than as decay waterfalls, but they is teh pretty colours.

 

[Tony L]

LS3/5A Impulse Step Response

JR149 Impulse Step Response

I picked these as they are obviously different between speaker whereas most measurements I looked at seemed very similar. I don’t really know how to interpret them.

 

[Tony L]

Distortion. I have no idea as to the validity of this (or any other) plot. I assume, as ever, it is assessing the room more than the loudspeaker as clearly none of this was measured in a controlled anechoic/scientific environment. It may however be implying the LS3/5A has a good few db lower distortion across the all important crossover region, which is very interesting indeed IMO. To my mind it is here that the LS3/5A magic lies. I also find it interesting that despite the LS3/5A having a bit of a tonal spike at 1kHz it is actually lower in distortion at that point than the 149.

PS If anyone wants to see any other presentation of these measurements just let me know and I’ll post it. I don’t want to have to measure again, but happy to present what I have however it is most interesting or revealing. Annoyingly I haven’t been able to make head or tail of phase yet, it just looks a total mess to my eyes, like a four year old scribbling with a crayon!

 

[Martyn Miles]

Ok but they do have a direct connection. As I understand it Stirling used to make standard LS3/5a's until they ran out of KEF drivers.(Falcon now manufacture replica drivers). The V2's were approved by the BBC as tonally equivalent, and use adapted SEAS drives to replicate the BBC sound. The rest of the construction is to LS3/5a spec. No, I wouldn't want to use them as matched pair with other LS3/5a's but as i understand it all LS3/5a's sound slightly different anyway.

The sale of all hi fi is ultimately a marketing exercise really.

I think you’ve got a good point, as all LS3/5as sound slightly different.
I’ve had experience of that.

Some might argue that the BBC ones ( mainly Rogers, with XLR fittings ) were especially selected to
be on spec. as is possible.

 

[ToTo Man]

@Tony L,

You must be running a snazzy new beta release of REW? - The version I'm running displays RT60 as a line graph!

Hopefully those with more knowledge of REW will chime in, but wrt the step response graph, it'll become more meaningful if you narrow the X-axis so that the left limit is around -3ms and the right limit is around +10ms. If it still looks a bit squashed then try -1ms left and +5ms right. This will reveal the difference in arrival time between the T27 and B110 at the listening seat.

@Tony L Sorry I meant X-axis not Y-axis, I've corrected my above comment accordingly!

 

[Tony L]

I downloaded a fresh copy of REW this morning! I’ll have a go at altering the step scale ranges later.

 

[Tony L]

...Is this what you meant:

LS3/5A

JR149

Again I’m not sure what I’m looking at here, though shortening the time domain has made the two plots look much more similar to my eyes.

 

[ToTo Man]

Technically it should be 0ms on the left but whenever I take measurements I always seem to get a bit of pre-ringing in the negative time domain (no idea why!!), that's why I tend to include at least a couple of ms before 0.

EDIT - I think I might have got my decimations mixed up! Play about with the limits until you get a response curve that looks something like this.

 

[Tw99]

I'm not sure if you have these in the same room as your Tannoys, but I'd be interested to see a comparison between those and the LS3/5as.

 

[Tony L]

EDIT - I think I might have got my decimations mixed up! Play about with the limits until you get a response curve that looks something like this.

That is beyond me I’m afraid, I can’t get that many zeroes to stick in the limits field. I really do not get on with the REW user interface at all and very regularly end up totally unable to get back from the mess I’ve caused trying to enter scale values etc short of deleting the global preference file and starting again. Just done it again now attempting this.

I'm not sure if you have these in the same room as your Tannoys, but I'd be interested to see a comparison between those and the LS3/5as.

I very briefly tried them plonked in front of the Tannoys (picture upthread) and they sounded really good once I’d adjusted the internal gain settings on my preamp for the massive difference in efficiency. They are astonishing for their size and sounded great with some good vinyl upstream. I’ve no plan of measuring them in that environment as I’d have to take the Tannoys out of the room for it to have any meaning, which is way, way too much effort, but they are a great little speaker for sure. Putting the Tannoys back you have a sense of endless headroom and truly effortless dynamic range and ease, but the LS3/5A didn’t sound small or squashed.

 

[ToTo Man]

That is beyond me I’m afraid, I can’t get that many zeroes to stick in the limits field. I really do not get on with the REW user interface at all and very regularly end up totally unable to get back from the mess I’ve caused trying to enter scale values etc short of deleting the global preference file and starting again. Just done it again now attempting this.

Just tried this myself and am experiencing a similar issue!

 

[Tony L]

If you PM me an email address I can send you the raw data file if you like.

 

[ToTo Man]

If you PM me an email address I can send you the raw data file if you like.

Sure, PM incoming.

 

[Tony L]

Cool, I’ll get that over to you. In the near here’s some more pretty graphness:

LS3/5A

JR149

I’ll let others interpret that.