'LGT' Construction Diary

[ShinOBIWAN]

Originally posted by 5th element
For some perspective the lowest note a soprano is normally called on to sing is a C4, also known as middle C and has a frequency of 260hz. For pop and dance this is somewhat 'high' and one could take a G3 as a rough base line. A G3 has a frequency of 195hz.

Top end soprano, highest note, is a C6. Rarely in certain pieces of music, such as Mozart's the magic flute, the soprano is asked to hit notes higher, in this case a F6 corresponding to a frequency of 1396hz. We can assume around an E5 to be roughly the max a pop or dance singer will hit in what is considered to be 'full voice', = 659hz. Occasionally they go higher, but normally these are not sustained notes.

Quite clearly if we ignore harmonics, the fundamental notes are produced by the midrange. The fundamental is, of course, what will supply the weight to the voice, the harmonics changing the flavour of the voice.

Thanks for the explanation. I didn't realise most of what you've just outlined there. Certainly I thought the problem to be higher up than 800hz and it was, although the problem was still mid driver related.

Problem is now solved.

Uli, who wrote Acourate, has generated a set of crossover/correction filters from my raw driver measurements and these sound MUCH better than my attempts.

The problem isn't something to be seen in the FR measurements and so I missed this. If you look at this data:



You can clearly see the black line describing the driver correction filter. Uli mentioned that its important to not correct the whole of the stop band(part where the driver is attenuated with the crossover filter) because 1) its not a significantly audible proportion of the sound so correcting information below -30dB can cause problems and 2) measurements as level decreases are less reliable so more error prone and you could be correcting a problem that is background noise related rather than that of the driver. Signal to ratio is important for measurements.
Its very clear to see that I corrected a driver resonance dip with a very steep inverse peak at 8-10Khz which was at -47dB. This caused some large unnatural sound from the midrange drivers and subsequently gave the effect I was objecting to.
I'm a little wiser and won't fall into that trap again but its interesting to see that despite near perfect FR the speaker had a significant problem not shown by it!

Uli did a fantastic job with the filters. The midrange is very creamy and lucid, completely gone is that thin nature. My only problem now is the recording quality since the design is quite ruthless.

Feed in quality though and you get results. I was listening to the Manger test CD and tracks 12 & 13 - Livingston Taylor jazz on the Chesky label. This was stunning, the sax is fantastic and certainly the most realistic approximation I've heard. Listening to the percussive sounds of track 14 & 15 - The O-Zone group were excellent fun. The cymbals and other sharp transients were created with superb speed and have harmonics and decay I've never heard before on my previous systems. The weight of the multiple large drums provided a rush of low frequency energy that travelled through the room with an amazing sense of dynamic weight but never overshadowing other elements of the sound, just natural. Combine both the sharp transients with the bass and the whole experience was something else really, like nothing I've heard before. You hear that rubbish about "being there" all the time but I genuinely felt that. I went back and played those tracks 3 times because it was such an experience. I like electronic music but these speakers have given me a new appreciation of the skill and beauty of these acoustic performances. Its quite addictive because each time you play its like some fascination of admiring someone talented playing an instrument.

I could finish up right now and leave the speakers as is and be happy but there are still things left to do on the check list so I can only imagine where things will go from here. An exciting prospect.

[Salas]

Aha. A hidden Gremlin was at work...

You sound thrilled. We are all happy for you!

What was your distance and axis for the measurements you gave to Uli? Single speaker I suppose.

[tinitus]

good for you ... really amazing what is possible with such tools, but it also shows that it should be handled very carefully

[Hezz]

Shin,

Glad you got that Gremlin out of the way. Don't know what was causing that big suckout around 600 Hz but that was definately right in the range you were describing. Sounds like you are really enjoying them now. Keep us posted on your continual tweaking process. It is very interesting and enlightening.

It seems like your PC crossover method would be really good to use even if you were going the standard passive route to help model the filter response in help in building the filter.

[wigginjs]

Shin:

You have a very unique exposure to lots of different kinds of loudspeakers. It sounds like you've listened to a lot of high end retail kit as well as having built two (so far ) world class loudspeakers. Have you heard the Linkwitz Orions? What was your impression.

[chrismercurio]

but my wife and stayed in the Sea Ranch home owned by Siegfried Linkwitz and outfitted with the Orion's. I stayed there to see if I wanted to build a pair, because it is a beautiful place to get away from it all, and is only a few hours drive.

Having owned a sold a great deal of high end "kit" and built a few home a car systems, I can say without a doubt that they are a beautiful sounding full range speaker. They are extremely low in coloration and perform a variety of music extremely well. I don't find them particularly atractive physically, and have listened to several speakers that are quite a bit more dynamic. So to sum up:

Very revealing
Low distortion
Full bandwidth (Almost)
not really dynamic
not attractive

My 2 cents,

C

[ShinOBIWAN]

Originally posted by m0tion
Shin:

You have a very unique exposure to lots of different kinds of loudspeakers. It sounds like you've listened to a lot of high end retail kit as well as having built two (so far ) world class loudspeakers. Have you heard the Linkwitz Orions? What was your impression.

I heard the Orions at the UK DIY meet but in less than favourable circumstances. The real problem was the room they were used in. I feel it would be a great disservice to pass judgement on them from that less than steller experience. I would love to hear them correctly setup in a good room and Vik, who built them, isn't a million miles away so maybe I'll get the chance someway or another.

Virtually all of the really, really hi-end speakers, ie. £15k+, I've heard have been at hifi shows which again are a bit hit and miss for assessing critically. Then again my own listening room is far from ideal as you've seen so, in a good room, I've no doubt this design has more to give even though I use room correction.

As far as what is better or how they compare, I hate to say it because it sounds like I'm blowing my own trumpet but after that issue was sorted things very sharply came into focus and just snapped together. The result is these are undoubtedly the best loudspeakers I've heard yet. That's not just in my living room but period. The combination of quality drive units, well suited and virtually inert cabinets along with the power of carefully crafted linear phase filtering has resulted in something greater than the sum of its parts.

I'm incredibly pleased with the outcome and compared to my last design, it isn't really a contest. These are another class or maybe two up IMO.

I often sell whatever I've built but have this feeling that I'm very much attached to these and that attachment isn't related to the time or effort spent but the sound and the feel good factor you get when listening to them. It really is quite amazing at times. I started this project with high expectations and the outcome has raised my own personal bar on what I thought possible.

Some might read that and think its only a loudspeaker FFS! Well that maybe but it connects me to music in a new way and I think that's worth shouting about.

[Salas]

Shin we are very happy for you, and we would find it very helpful for the conclusion of the thread, if some technical resume of the whole system chain along with some photos could be posted. Something compact so someone can know about LGT in one forum page. Short descriptions of this and that, working principles, you know....

P.S. What was the arbitrary measurement point that you based your final crossover approach? Do you have listening position single speaker and/or stereo plots to post?

[wigginjs]

I agree. A one or two page organized cut sheet describing all of the technical details of this design with measurements showing whats going on would be great. Maybe something like that would be better created after a while once the project is "complete".

One this I find sort of interesting is I have followed both of your build threads from beginning to end, but there are still somethings about your design I don't know. For instance, I know you are using Acourate to generate your FIR filters, but what convolution engine are you using? Do you still use console? Are you still using the RME card? With what optional boards? Do you use an external DAC? Wordclock? I clear concise document with all of the technical details of how you do what you do would be excellent. You know, all of the "ingredients" to your "secret sauce".

On second thought, documenting all of the technical details of this design is not a trivial or particularly fun task so I would completely understand if you said "heh, ok, maybe I'll get around to that".

[ShinOBIWAN]

Originally posted by salas
[B]Shin we are very happy for you, and we would find it very helpful for the conclusion of the thread, if some technical resume of the whole system chain along with some photos could be posted. Something compact so someone can know about LGT in one forum page. Short descriptions of this and that, working principles, you know....

Good idea Salas and once again thanks for all your support through this project.

P.S. What was the arbitrary measurement point that you based your final crossover approach? Do you have listening position single speaker and/or stereo plots to post?

I'll compile all the measurements I have, title them correctly and then post.

But in a nutshell.

The driver linearisation took place at a distance of 40cm on axis and was done for each individual driver. However I only have 8 channels of processing available due to my sound interface. To fully linearise the speaker I would need 10 channels ie. 1 for each driver.

So the tweeter, upper mid and lower mid all have individual tailored correction filters. The bass drivers have no linearisation at the moment because its far too difficult to get a good in room measurement but once I get around to taking them outside then I plan to measure at a distance of 1m on axis with the tweeter to provide a mid point between the two bass drivers. A further upgrade would be a sound interface upgrade to allow 10 channels and individually correct each bass driver instead of summing the pair.

From there the design has just been time aligned by playing and recording a logsweep at the listening position from 20hz - 22000hz through the loudspeaker with deliberate delays added to the crossover filters to allow identification of the individual pulses correlating to the drivers. From here those delibrate delays can be used as a baseline to calculate near perfect time arrival at the listening position from all drivers. Once done these delays are added to the outputs of the filters creating a time coherent speaker around the listening position.

You still have the room to deal with so one final correction remains and that is from the listening position recording a full range sweep and then generation a correction filter that is applied to each of the driver correction filters eventually creating the final crossover filters which incorporate driver linearisation, level matching, time alignment and room correction. Optionally you can include a target curve for the loudspeaker if you'd like to 'tune' to your room and tastes. Even with such FR shaping the great thing is the loudspeaker still maintains its excellent time-frequency relationship because of the linear phase filtering.

OK that wasn't in a nutshell at all Hope things are a little clearer.