Matching up Advents for Stacking

[Guest boseguy]

I have never tried stacking Advents, so I have no first-hand experience with why so many listeners seem to think there's some intangible "magic" to their sound. I have heard that stacking applied to the original Large Advent, not to later versions.

Speaker Dave's suggestion to reduce/eliminate tweeter lobing makes perfect sense from a technical standpoint.

However......I suspect that one aspect of that "intangible magic" had to do with the HF phase anomalies that stacking produced, and the resulting "spaciousness" that ensued.

Plus, of course, the reduced distortion of driving two speakers/ch to the same SPL as one speaker/ch.

Speaker Dave's approach would probably rob the Advent aficionado of the "mysteriousness" of the stacked sound. Now they'd just have a Dyna A-50 compared to an A-25.

boseguy

[speaker dave]

However......I suspect that one aspect of that "intangible magic" had to do with the HF phase anomalies that stacking produced, and the resulting "spaciousness" that ensued.

boseguy

Killjoy!

You may be right about the lobing causing a bit of spaciousness. I always wondered about systems like the Rectilinear III that had the tweeters way far apart. This would give a dense comb filtering whereas a pair of tweeters close together run the risk of one broad cancellation dip.

David

[Doug G.]

Interesting. I prefer my stacked Advents with both speakers upright so the tweeters are NOT close together.

I know, I know, it goes against ALL convention and I will probably be declared a pariah but, there you have it.

Doug

[Zilch]

I don't know that a whirling, swirling festival of phase interference translates to enhanced "spaciousness." It's not a generally recognized cue for that, I don't believe; perhaps Ken can clue us in. A rational argument might be made that it enhances "realism," but once the soloist or vocalist begins, it just translates to crappier imaging.

Lowpassing one member of the stacked pair does alter its impedance, in this case intentionally, but may be counterproductive with respect to sending full power to one tweeter, when the apparent intent of stacking, in some instances, at least, is to increase the available SPL before distortion limits the applicable "crank."

Yes, I can set this up and measure, but don't be surprised if the lowpass gets mighty low, as the woofers are lobing nicely in this combo as well. I think both Speaker Dave and Boseguy are correct; the alleged and elusive "magic" will be gone....

[speaker dave]

Lowpassing one member of the stacked pair does alter its impedance, in this case intentionally, but may be counterproductive with respect to sending full power to one tweeter, when the apparent intent of stacking, in some instances, at least, is to increase the available SPL before distortion limits the applicable "crank."

Yes, I can set this up and measure, but don't be surprised if the lowpass gets mighty low, as the woofers are lobing nicely in this combo as well.

I was only thinking of correcting for a little bit of offset in the treble range. If I can nick one of your curves (attached) maybe enough correction to cover the 3 and 6 degree curve. Both tweeters could work down towards the 1500Hz crossover and preserve some of the power handling.

The objective would be to reduce the critical nature of exact time allignment currently required to preserve treble. I think most users could get the time allignment good enough to have no problems in the woofer range. Your curves don't really show a woofer problem for small offsets.

One caveat on the capacitor bypass method. I'm not sure if there is some inherent phase lag for one system so the best response might be on a tilted axis. (I'm sure you'll figure it out )

David

[kkantor]

I don't know that a whirling, swirling festival of phase interference translates to enhanced "spaciousness." It's not a generally recognized cue for that, I don't believe; perhaps Ken can clue us in. A rational argument might be made that it enhances "realism," but once the soloist or vocalist begins, it just translates to crappier imaging.

Lowpassing one member of the stacked pair does alter its impedance, in this case intentionally, but may be counterproductive with respect to sending full power to one tweeter, when the apparent intent of stacking, in some instances, at least, is to increase the available SPL before distortion limits the applicable "crank."

Yes, I can set this up and measure, but don't be surprised if the lowpass gets mighty low, as the woofers are lobing nicely in this combo as well. I think both Speaker Dave and Boseguy are correct; the alleged and elusive "magic" will be gone....

:>)

Take the technically ill-defined nature of "spatiousness" or, really, the many types of things that listeners all respond to under that catchphrase of "spatiousness." (HRTF, lateral energy, ITD, IACC, strong reflections and image sources, etc, etc.). Add to this "a whirling, swirling festival of phase interference." Hmmm... not much of a stretch to image some cues swirling by the head every now and then....

-k

[kkantor]

David,

I've been pondering this casually in the background as I drag little schematic symbols around a CAD screen. (A very meditative process sometimes...) Here are some fairly indefensible thoughts:

1- I could design a decent 2-way speaker without measurements if, and only if, I knew and liked the drivers very well.

2- It would take about 10x longer than doing having both ears and analyzers available. I'd need a very good, dead listening room that I had worked in a lot. A big pile of clip leads and other xover proto tools, a bunch or R's, L'c and C's. A shop to make 5 or 6 test boxes.

3- There might be some small-ish residual refinements in the response that I just couldn't debug by ear alone, but they wouldn't be deal breakers.

4- A 3-way >might< be possible, but that would be a torturous and lengthy process. I would have to rely a great deal on my internal memory of measurements made in the past, and how they corellated to the sound. The results might be acceptable, but I know the design would be sub-optimal.

5- On the flip side, I think I would stand a pretty decent chance of doing an excellent 3-way by measurement alone. Fully optimized? Of course not. But close. I would be relying very heavily on memory and experience of >100 previous designs, and how people reacted to them. And would need a good battery of gear, and a few months.

Bottom Line: you are right. Both arms untied. Ears directing the meters, meters working to calibrate and satisfy the ears.

Maybe we should try it some time? Like start with a built-up 3-way with identical drivers, and try and do a crossover by ear only. Ther add measurment to the process.

-k

[Zilch]

Maybe we should try it some time? Like start with a built-up 3-way with identical drivers, and try and do a crossover by ear only. Ther add measurment to the process.

I dunno; are measurement guys allowed to have fun, too?

[speaker dave]

David,

5- On the flip side, I think I would stand a pretty decent chance of doing an excellent 3-way by measurement alone. Fully optimized? Of course not. But close. I would be relying very heavily on memory and experience of >100 previous designs, and how people reacted to them. And would need a good battery of gear, and a few months.

Interesting thoughts, Ken.

I think you are right about a "measurements only" approach relying heavily on past experience. Most likely when you do your 20th 6 1/2" two-way you have learned a measurement curve that "pleases all" and a duplication of that curve gives a pretty good probability of success.

As counter proof, I remember the first Uni-Q models from KEF. At KEF the measurement systems were evolved to the point where Laurie just didn't believe in listening. "Too many traps and pitfalls in informal listening tests, better to trust your measurements." Well, the first Uni-Q models, based on drivers with an inherently non typical dispersion characteristic, were designed to the same measurement balance as previous models but came out very mid range forward and dull. (UniQ placed a dome tweeter at the apex of a fairly deep 8" woofer cone. Tweeter directivity would be higher and the power response and near off axis curves would be strongly different.) We, the US distributor, complained mightily and a redesign was forced through. We ended up opening and modifying hundreds of crossovers of the early models.

Maybe we should try it some time? Like start with a built-up 3-way with identical drivers, and try and do a crossover by ear only. Then add measurment to the process.

I'd be game for a 2-way but I'm not sure I could ever get a 3 way finished!

David

[Carlspeak]

Hey David, I've got a pair of Uni-Q 65's with B160 SP1374 drivers dated 1996. Are they the old crap or improved style?

[speaker dave]

Hey David, I've got a pair of Uni-Q 65's with B160 SP1374 drivers dated 1996. Are they the old crap or improved style?

Yours are fine. In fact, none of the worst ones ever got to the market. We heard some advanced samples and didn't like them and modified them all before they got into the market. These were 8" units (B200) of the first generation of Uni-Q. Not clear on model numbers but maybe C40, C60 C75?

David

[Pete B]

Interesting thoughts, Ken.

I think you are right about a "measurements only" approach relying heavily on past experience. Most likely when you do your 20th 6 1/2" two-way you have learned a measurement curve that "pleases all" and a duplication of that curve gives a pretty good probability of success.

As counter proof, I remember the first Uni-Q models from KEF. At KEF the measurement systems were evolved to the point where Laurie just didn't believe in listening. "Too many traps and pitfalls in informal listening tests, better to trust your measurements." Well, the first Uni-Q models, based on drivers with an inherently non typical dispersion characteristic, were designed to the same measurement balance as previous models but came out very mid range forward and dull. (UniQ placed a dome tweeter at the apex of a fairly deep 8" woofer cone. Tweeter directivity would be higher and the power response and near off axis curves would be strongly different.) We, the US distributor, complained mightily and a redesign was forced through. We ended up opening and modifying hundreds of crossovers of the early models.

I'd be game for a 2-way but I'm not sure I could ever get a 3 way finished!

David

I'd be up for doing a 3-way by ear as long as I could use a reference that I know well

for listening/tuning.

I did a 4-way, first by ear, then with some limited measurements

back when I was a teenager. I actually heard some issues in the upper midrange that I

traced to the cupped faceplate on the 2" dome that I used - removed it and made a flat plate.

I found the problem by ear, then measured a deep notch with the driver out of the

system. I used a pair of 8" transmission line loaded woofers inspired after seeing the

low measured distortion of IMF TL systems - and hearing one, and was 4 way 1st order

inspired by the DQ-10. It was biamped with a passive line level XO before the amps so

that I did not need large components for the LF XO.

Don't think this would ever happen given the logistics issues.

How do you feel about having a reference for comparison purposes?

[speaker dave]

I'd be up for doing a 3-way by ear as long as I could use a reference that I know well

for listening/tuning.... How do you feel about having a reference for comparison purposes?

Everywhere I've worked we have always compared systems in progress to other speakers via listening tests. Of course the other systems were never absolute references but, if you assume speaker faults are random, comparisons to other known good systems should be useful. Of course you always seem to want something half way in between any two systems you have in the comparison! "Not quite as forward as A or as distant as B". At PSB, for some reaons of politics, we started doing our listening tests behind curtains, single blind. You would like to think of yourself as being unbiased but blind tests can really shake up your beliefs. We also found we were fooling ourselves by always liking a sytem in a particular room position.

I did a 4-way, first by ear, then with some limited measurements

back when I was a teenager. I actually heard some issues in the upper midrange that I

traced to the cupped faceplate on the 2" dome that I used - removed it and made a flat plate.

I found the problem by ear, then measured a deep notch with the driver out of the

system.

Doesn't that fall under my comment on using measurements to see the aberations, and listening tests to define which need to be fixed, even if done in the reverse order? I think you needed to see the measurement, the significant dip, and then based on frequency you could sleuth out that it was a reflection due to the faceplate. I'm guessing you even had an inkling this was the issue ahead of time, based on a visual inspection of the driver.

I still think that, for me, a 3 way would be way to many variables. Next time I work on a system I will try it purely by ear to see how far I get. I am curious how it would turn out. For example, the order of tasks would likely be to get a "good sounding" lowpass on the woofer, then a "good sounding" highpass on the tweeter, then choose the best phasing (polarity) for the combination, then adjust the pair for best blend and overall balance, while improving the combined listening axis. (Is it a good amplifier load? How do I listen to the impedance curve?)

And what does a "good sounding" low pass sound like? And how do you set the crossover frequency by ear? I know from listening to filtered noise that the corner frequency gives a strong pitch component to the sound, unless the corner Q is low. This would likely force you towards very soft corners. The system design task is really much more complex than, say, setting a graphic equalizer by ear. There are so many things you would see in response curves that are known pitfalls to avoid. For example, if filter shunt elements are too low in reactance, then you will get voltage gain above the input level. This is not good for the impedance curve. Its easy to see, but not something you could specifically identify by ear. In a listening test you may be unhappy with a result but that wouldn't point out the underlying cause.

There must be some blind speaker designers out there. How do they do it?

David

[Pete B]

Everywhere I've worked we have always compared systems in progress to other speakers via listening tests. Of course the other systems were never absolute references but, if you assume speaker faults are random, comparisons to other known good systems should be useful. Of course you always seem to want something half way in between any two systems you have in the comparison! "Not quite as forward as A or as distant as B". At PSB, for some reaons of politics, we started doing our listening tests behind curtains, single blind. You would like to think of yourself as being unbiased but blind tests can really shake up your beliefs. We also found we were fooling ourselves by always liking a sytem in a particular room position.

Doesn't that fall under my comment on using measurements to see the aberations, and listening tests to define which need to be fixed, even if done in the reverse order? I think you needed to see the measurement, the significant dip, and then based on frequency you could sleuth out that it was a reflection due to the faceplate. I'm guessing you even had an inkling this was the issue ahead of time, based on a visual inspection of the driver.

I still think that, for me, a 3 way would be way to many variables. Next time I work on a system I will try it purely by ear to see how far I get. I am curious how it would turn out. For example, the order of tasks would likely be to get a "good sounding" lowpass on the woofer, then a "good sounding" highpass on the tweeter, then choose the best phasing (polarity) for the combination, then adjust the pair for best blend and overall balance, while improving the combined listening axis. (Is it a good amplifier load? How do I listen to the impedance curve?)

And what does a "good sounding" low pass sound like? And how do you set the crossover frequency by ear? I know from listening to filtered noise that the corner frequency gives a strong pitch component to the sound, unless the corner Q is low. This would likely force you towards very soft corners. The system design task is really much more complex than, say, setting a graphic equalizer by ear. There are so many things you would see in response curves that are known pitfalls to avoid. For example, if filter shunt elements are too low in reactance, then you will get voltage gain above the input level. This is not good for the impedance curve. Its easy to see, but not something you could specifically identify by ear. In a listening test you may be unhappy with a result but that wouldn't point out the underlying cause.

There must be some blind speaker designers out there. How do they do it?

David

David, I pretty much agree with everything you wrote there. I've read

about the turntable at the NRC where you can move in and out different

speakers under blind conditions - not sure if that is what you used.

I've done a lot of simulation of speaker systems in CALSOD, and I have

a few what I'll loosely call prototype XO designs in my head. I frequency and

impedance scale them to get a rough guess and then often use the CALSOD

optimizer to get the actual target that I'm looking for. I also got to where

I could manually adjust things to get even closer without the optimizer.

I'm not sure how we got on this discussion, I simply think it is an interesting

challenge, but using the modern tools certainly makes it much easier as you

have said.

I'd expect that we'd be given at least the DC resistance of the drivers, perhaps

an impedance curve as if we selected them out of a catalog. But it would not

be absolutely required. I don't think it would hurt to allow the use of noise

sources, or we could say any test sources that you can get on CD - just a

suggestion.

Interesting challenge.

[Zilch]

Two-way, I'll provide the waveguides....

[Pete B]

Two-way, I'll provide the waveguides....

Selenium D220Ti-8 drivers just arrived but the horns are back ordered at PE.

Don't know when I'll have the time to try these.

[Zilch]

Selenium D220Ti-8 drivers just arrived but the horns are back ordered at PE.

Don't know when I'll have the time to try these.

Not much to do to try them but build the crossovers, flip the Advents upside-down, connect to the woofers (just remove one of the input screws for lead access,) stick the waveguides on top, and listen....

[EddieO56]

Interesting. I prefer my stacked Advents with both speakers upright so the tweeters are NOT close together.

I know, I know, it goes against ALL convention and I will probably be declared a pariah but, there you have it.

Doug

I'm a little late to this discussion but I had seen this a couple of years ago when I acquired a second pair of large Advents. My first pair are New Advent Utilities that I got new in 1977. The second pair were the 25th Year Anniversary model.

I was motivated to get the second pair after seeing this thread and reading the review of stacking large Advents. I tried stacking the tweeters close together for about a year. I liked the way it sounded but when I saw the comment about stacking both speakers upright I decided to give it a try. To my surprise I liked that set-up better. I've been running it that way for the last year and a half.

I have a Sony TA5650 that I use as a pre-amp. I have two Sony TAN55-ES amps for power.

I recently acquired a third set of large Advents. Nice walnut cabinets. Date stamp on the tweeters are April 11, 1975. Fried egg tweeters and Masonite ring woofers. They sound really nice.

I subbed in the newly acquired OLA's for the 25th Anniversary model. My set up went from sounded great to awesome.

I have a third Sony TAN55-ES amp that I'd like to add into the system so I can run all six speakers.

What is the best way to connect up all three amps and the third set of speakers? The two amps are currently connected in mono with one amp per side.

The 77's with the 25th Anniversary speakers:

The 77's with the 75's:

Downstairs set-up:

[Doug G.]

How many WPC are those amps in stereo?

OK, I was going to wait until you answered but I think I will tell you to use each amp in stereo on a pair of speakers and stack the three pairs vertically just like two pair. It will sound even more awesomer ()

I did that for awhile a few years ago until I decided to go full blown quadraphonics again.

And welcome to the forum.

Doug

[EddieO56]

How many WPC are those amps in stereo?

OK, I was going to wait until you answered but I think I will tell you to use each amp in stereo on a pair of speakers and stack the three pairs vertically just like two pair. It will sound even more awesomer ( )

I did that for awhile a few years ago until I decided to go full blown quadraphonics again.

And welcome to the forum.

Doug

Awesomer....nice. I guess I'm the antithesis of somebody like Zilch and his technical jargon. I just know what I like. I've been stuck on Advents since I first heard them back in 1976. I've never owned anything but Advents.

What really makes adding the OLA's into the mix for me was how I acquired them. I was driving through my neighborhood last Wednesday morning and that pair of speakers were sitting out on the curb. A well timed find. They were a little dusty, and they appeared to have been there overnight.

Got them home, removed the screens and the tweeter and woofer looked good. Foam surrounds appeared to be in good shape and they sound good. No problems. I figured at the very least I'd need to replace the foam surrounds on the woofers. More good luck. Cleaned them up, put on a couple of coats of wax and they look and sound "awesome" stacked with my NLA's. A little smoother than the 25th Anniversary models.

The amps are 110wpc.

So I just need to find a three way signal splitter for the amps to run them in stereo. Any suggestions?

[Doug G.]

Do the amps have tape output/input facilities? If so, you can just daisy chain them together for the input signals.

Feed the signals to the first amp, output from the tape out jacks to the tape in of the next amp and so on.

Doug

[EddieO56]

Do the amps have tape output/input facilities? If so, you can just daisy chain them together for the input signals.

Feed the signals to the first amp, output from the tape out jacks to the tape in of the next amp and so on.

Doug

The Sony TA-N55ES amps have only R/L inputs.

You can see what they look like here:

http://www.thevintag...y-TA-N55ES.html

I have the amps connected to a Sony TA 5650 integrated amp which I use as the pre-amp.

http://www.thevintag...ny-TA-5650.html

[Doug G.]

Ok, I see. Just basic amplifiers. When you use two, are you just using a Y connector to connect the preamp to the two power amps?

EDIT: Oops, I forgot you said you were using one amp for each channel so no Y connector necessary. You can use a Y connector to drive two amps from one preamp output and you may even get away with using two Y connectors to drive all three amps from one source but, depending on the output and input impedences, you may end up with loading effects which would reduce the output of the preamp and affect its frequency response.

Doug