Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Tom Danley]

Hi

It is unfortunate that an apples to apples comparison is not possible here, however it is worth looking a little further.
Another thing to keep in mind is how the distortion is displayed is different in the two systems (as a fraction and relative to the fundamental).
As one can read directly, the Labs measured around 3 / 10 of one percent down to about 35Hz.

With the 12pi being driven harder and measured differently, isn’t a direct comparison.
Several spot frequency comparisons the pi shows a difference between distortion and fundamental of about 40dB in the 80-100Hz area or about 1%.
At 30Hz, the Pi distortion has risen to about 10%, to more than 30% at 20Hz and distortion equals fundamental around 15Hz.

Tom Danley wrote on Sun, 04 November 2007 19:18
Have you built and measured a 12 pi that wasn’t in push pull to see how much difference it actually makes when “that” is the only thing different?

”I plan to do that in the next few months. I'm actually making a new horn prototype, and in the process of doing so I'll be able to take parts of it to do the exact test you're talking about. I can make the equivalent of a 12Pi without push-pull drive from it.

Care to place a wager with me on the results? I'll bet it confirms once again how the 12Pi push-pull drive is what is responsible for its ultra-low distortion at low frequencies and high power levels.”

Wayne, the driver is the primary source of distortion in bass horns, which is how BT-7’s can be less than few% at rated power and why the labs were the lowest distortion when measured so far.
I don’t see “ultra low” distortion levels in what you posted and besides, push pull ONLY reduces one distortion component and its not one that sounds bad.

Tom

[Wayne Parham]

Tom Danley wrote on Mon, 05 November 2007 07:30

With the 12pi being driven harder and measured differently, isn’t a direct comparison. Several spot frequency comparisons the pi shows a difference between distortion and fundamental of about 40dB in the 80-100Hz area or about 1%. At 30Hz, the Pi distortion has risen to about 10%, to more than 30% at 20Hz and distortion equals fundamental around 15Hz.

That's not true.  The chart I displayed was done at 28.3v (over 200 watts) and at that level, you can't see the distortion present because it is below the noise floor.  I guess you missed that I said the noise floor was about 70-75dB.  So unless the distortion rises above that level, you can't see it.

You can clearly see distortion when it rises above 75dB though, as it does in other basshorns without push-pull drive.

Tom Danley wrote on Mon, 05 November 2007 07:30

Wayne, the driver is the primary source of distortion in bass horns, which is how BT-7’s can be less than few% at rated power and why the labs were the lowest distortion when measured so far.

Driver non-linearity is the primary source of distortion in basshorns, and this is exactly what push-pull drive corrects.  The LABhorn has no mechanism to counter this kind of distortion at low frequencies.

The Michigan measurements of a group of four LABhorns run at a mere 25 watts show that the LABhorn generates more distortion than a 12Pi does when run at ten times that much power.  The LABhorn was run at a flea-power 25 watts, yet it still displayed measurable distortion particularly at low frequencies.  That's why I've said I don't think it's a good idea to bump up the 20Hz EQ on LABhorns.  Distortion goes through the roof at those frequencies.

12Pi distortion stays under 1% - in most cases, below the noise floor and unmeasurable - until power is increased to several hundred watts.  In Tulsa, we started our measurements at 100 watts and went up from there.  In the 12Pi measurements, You can't even see the distortion until power is high, because it's below the noise floor.

To illustrate, let's take the measurements on the other end of the scale, where distortion is at its worst.  You really have to hit the 12Pi's with several hundred watts before distortion starts to creep up above the noise floor.  So lets look at the 12Pi chart taken at 1600 watts:

Now you can see some distortion rising above the noise floor.  See down there at 10Hz?  It has started rising there.  You also see some at 30Hz and 50Hz, about 20dB below the fundamental, which is about 10%.  So by 1600 watts, we're seeing 10% distortion.  It stays under 10% down all the way to about 20Hz, where it finally rises.  Below 20Hz, the horn is actually very quiet, there's not much output either fundamental or distortion, but as you can see from the chart, at 1600 watts, you don't want to push 10Hz through them.  That's not what they're designed for anyway.

Now let's compare that to LABhorns, shall we?  We can't do the same 1600 watt sweep we did in Tulsa, because LABhorns won't survive it.  Not a single sweep.  If you set the amplifier voltage for 65v and run a sweep, you'll smoke the drivers before the sweep ends.  Without cooling plugs, the LAB12's just won't survive it.  I know, I've tested these things a lot, and the stock LAB12 just won't do it.

But we can run the sweep at 800 watts.  That would be fine for comparison.  I expect you must have some measurements of the LABhorn at high power levels somewhere.  Please dig them out and display them.  They aren't available online anywhere, but you must have some.  Please post them at your earliest convenience.

[Tom Danley]

Hi Wayne

Perhaps there is some confusion about what the two different types of distortion measurements show.
“Now you can see some distortion rising above the noise floor. See down there at 10Hz? It has started rising there. You also see some at 30Hz and 50Hz, about 20dB below the fundamental, which is about 10%. So by 1600 watts, we're seeing 10% distortion. It stays under 10% down all the way to about 20Hz, where it finally rises”

In a measure like your 1600W rating, where the distortion is shown relative to the fundamental and not as a percent, I get different numbers.
For example at 50Hz, its about 10%, at 30Hz, one has about 15 dB between them (about %17)  and at 20Hz its about 4dB or about 40%.
If your plot were done as “percent” instead of dB difference,  it would also tip up at the bottom end like the Lab subs.

“I expect you must have some measurements of the LABhorn at high power levels somewhere. Please dig them out and display them. They aren't available online anywhere, but you must have some. Please post them at your earliest convenience.”

Actually I have never built any or measured any myself.  I had enough “mileage” in the software designing horns to know what needed to be done and specify a proper driver without building one.  I only needed to measure the driver.
With both the Lab and your pi, it is the driver / horn alignment that makes it good.

With the push pull thing as well as what you believe to be the result of the cooling plugs, a proper or scientific comparison requires that one compare with only one thing different.
You can’t compare to a different horn and different driver and conclude the difference is “push pull”.  Your non-push pull pi, if otherwise identical should actually show what difference it really makes. Keep in mind, when measured side by side with other boxes, so far, the labs have had the lowest distortion, it is after all mostly the driver and loading ratio.
As for the plugs, same thing, you have to compare them in use with only that being the difference.
It is possible I suppose but unlikely I would think that you have discovered something that bypasses the nature / speed of the temperature rises and transfer mechanisms described here. Look how fast the coil can heat up relative to the metal bits.

http://klippel.net/pubs/Klippel%20papers/Nonlinear_Modeling_ of_Heat_Transfer_03.pdf


Tom

[Jeff Babcock]

Thanks both to Tom and Wayne for your "debate", any time these ensue they prove to be highly informative.  Thanks Tom for that Klippel link.

Best regards to you both!
Jeff

[Wayne Parham]

At the 2007 Prosound Shootout, our noise floor was 70-75dB.  This limited our distortion resolution to about 1%, depending on the SPL of the box tested.  If a cabinet produced 110dB, then the lowest distortion we could measure was 1%, at 115dB, 0.5% could be reliably detected.  But if only 90dB, then the lowest level distortion you could see was 10%.  So whatever SPL level you see on the response curve for any given box, that sets the resolution of distortion that can be measured at that particular frequency.  As a result, the resolution is greatest in the passband, less at the lower frequency limits.

All measurements were taken with the microphone 10 meters away.  We could have increased resolution of our distortion measurements by moving the microphone closer.  This would increase the signal-to-noise ratio, and would have allowed us to see distortion at lower power levels, like 25 watts.  But this was a Prosound Sub Shootout, not a single-ended triode shootout.  We weren't interested in what these speakers do at 25 watts.  All the subs we measured would do pretty well at such small signal levels.  There's no challenge in that.  

The information we gathered at the Prosound Shootout tells us what the subs would do at power levels they will be used at, at least a hundred watts, up to several hundred watts to the maximum rating of each sub.  Even the best subs generate double-digit distortion figures at low frequencies and high power levels.  Most subs generate triple-digit distortion figures, with distortion higher than the fundamental at the lowest frequencies.  So even with resolution dropping at the lowest frequencies because of the reduced SPL (signal-to-noise), we still could see distortion at high power levels at sub-30Hz frequencies on all boxes but one.  It isn't difficult to draw a conclusion from that.

Tom Danley wrote on Mon, 05 November 2007 13:58

For example at 50Hz, its about 10%, at 30Hz, one has about 15 dB between them (about %17)...

That's right.  At 1600 watts, the 12Pi measurements show about 10% distortion and maybe a smidge more at 30Hz, still less than 20%.

Tom Danley wrote on Mon, 05 November 2007 13:58

...and at 20Hz its about 4dB or about 40%.

No, the distortion at 20Hz is below the noise floor and so we can't reliably determine what it is.  The noise floor was 70-75dB, and output of both the fundamental and harmonics is falling below 30Hz, so distortion becomes indeterminate at that frequency because it is below the noise floor.  My guess is distortion at 20Hz is about 20%, maybe 30% at 1600 watts.  Even if it is 40%, just under the noise floor at that frequency, it would still be impressive.  Most hornsubs are well into triple-digit distortion when run that far below cutoff at full power.

Look at that squarely, Tom.  We're talking about a 12Pi basshorn sub generating 30% or maybe 40% distortion at 20Hz at 1600 watts.  What we've seen from the Michigan sub shootout is that a group of four LABhorns generates 30% distortion at 20Hz with only 25 watts input.  That means even by your own characterization of the data, the 12Pi basshorn subwoofer running full tilt at 1600 watts produces no more distortion than a LABhorn sent a mere 25 watts.

At 1600 watts, distortion of a LABhorn at 20Hz will be far higher than the fundamental.  Sorry, I misspoke.  At 1600 watts, output from a LABhorn will be zero after about the first half of the sweep because the voice coil will have burned open.  But at 800 watts where it can survive the sweep, the harmonics will be far louder than the fundamental at 20Hz.

What would be really nice is to see a LABhorn measured at 100 watts, 200 watts, 400 watts and 800 watts.  Knowing what it does at 25 watts tells us very little, particularly in terms of distortion, power compression, electro-mechanical parameter shifts and the ripples in response caused by parameter shifts at high temperatures / power levels.

Tom Danley wrote on Mon, 05 November 2007 13:58

Actually I have never built any or measured any myself.

[Al Limberg]

Sometimes I feel like I'm driving by a car wreck with these posts... I know better but I just have to slow down and look!
Not to throw a monkey wrench into things, but as probably the first person to actually complete and put Labs into use on a regular basis, I must say I have never seen anyone recommend running them below 28-32hz for pro use, even in groups of four or more.  I don't doubt for a second that the drivers would self destruct at 800 watts input, or even less with a 20hz sine wave sweep.   Then again, I don't expect a Porsche pancake six would make it a quarter million miles in a Kenworth hauling a fullload.
It just doesn't make alot of sense to me.  Talk of eq'ing the Labs to 20 hz was always in the context of home theatre and frankly, a 20hz sinewave at even 50 watts would have earned me castration from the ex back in the day!
One other point - as a participant at the Michigan shootout, to the best of my recollection all tests were done on single cabinets other than listening tests when we cranked up 4 Labs and the pair of David's Bassmaxx cabs and perhaps 4 of our host's modified MT quad 18 cabs along with a pair of TD1s. I could be wrong.  At the first shootout, our dear friend Too Tall had Murphy as his chief assistant, and even he may have been of more help than me!  On the other hand, even then a good deal was learned including some guidelines of how to run future shootouts.

Best to all,

Al

[Wayne Parham]

Jeff Babcock wrote on Mon, 05 November 2007 15:34

Thanks both to Tom and Wayne for your "debate", any time these ensue they prove to be highly informative.

I hope our back and forth discussions are productive for he and I and everyone else, and not boring to people.  The only thing we have to be careful of is the tendency to fall into a pattern of saying the same things over and over again.  When you're passionate about something and really sure you're right, it's easy to get into that mode.

Jeff Babcock wrote on Mon, 05 November 2007 15:34

Thanks Tom for that Klippel link.

It is a great link:

• Nonlinear Modeling of the Heat Transfer in Loudspeakers, by Wolfgang Klippel

The point is made that cooling vents work well for woofers with excursion to pump air around the voice coil.  But when excursion is reduced, like what happens in a basshorn, the effectiveness of forced air convection is reduced.  Basshorns increase efficiency, which reduces heat.  But they do not work at 100% efficiency, so the heat generated must be carried away in some way.

Heat is also radiated from the voice coil into the magnet, and there is direct heating of the pole piece and magnet through eddy currents.  That's why removal of the heat using a cooling plug is very beneficial, especially in systems like basshorns where excursion is reduced and so forced air convection is too.

The vent in a woofer isn't just there for cooling, but also to prevent pressure from building up behind the cone.  This can cause harmonic distortion because it can introduce asymmetrical cone motion.  So even if cooling vent air is stalled by a basshorn limiting excursion, it is important that the vent be there for other reasons.

Here's another good link:

• Loudspeaker Power Handling Vs. Efficiency, by Rod Elliot

Forced air convection cooling can be optimized with careful vent size, shape and placement.  One can tune the cooling vents like the intake and exhaust manifolds of a car, but these are resonant systems so what is good at one frequency will not necessarily be good at all frequencies.  The cooling vents are definitely a tunable parameter, with length, size, position and number all being important variables.  In some cases, plugging a vent will improve cooling.  The proper optimization isn't trivial, but rather fairly complex.

Removing heat from the pole piece doesn't require any tuning.  All you need is a good conductor having large surface area contact with the pole piece.  If you measure a speaker and find the magnet and pole piece get very hot, it's probably a good idea to add a cooling plug and a heat sink of some sort.  For loudspeaker motors, heat is not your friend.

[Tom Danley]

Hi Wayne

“At the 2007 Prosound Shootout, our noise floor was 70-75dB. This limited our distortion resolution to about 1%, depending on the SPL of the box tested. If a cabinet produced 110dB, then the lowest distortion we could measure was 1%, at 115dB, 0.5% could be reliably detected. But if only 90dB, then the lowest level distortion you could see was 10%. So whatever SPL level you see on the response curve for any given box, that sets the resolution of distortion that can be measured at that particular frequency. As a result, the resolution is greatest in the passband, less at the lower frequency limits.”

Its too bad you didn’t use a TEF machine, they have superb noise immunity and it doesn’t suffer from these severe limitations.

“The information we gathered at the Prosound Shootout tells us what the subs would do at power levels they will be used at, at least a hundred watts, up to several hundred watts to the maximum rating of each sub. Even the best subs generate double-digit distortion figures at low frequencies and high power levels. Most subs generate triple-digit distortion figures, with distortion higher than the fundamental at the lowest frequencies.”

Ah, duh.   Why do you think I have been making horn loaded subwoofer that have low distortion for a living 20 odd years.

“Look at that squarely, Tom. We're talking about a 12Pi basshorn sub generating 30% or maybe 40% distortion at 20Hz at 1600 watts. What we've seen from the Michigan sub shootout is that a group of four LABhorns generates 30% distortion at 20Hz with only 25 watts input. That means even by your own characterization of the data, the 12Pi basshorn subwoofer running full tilt at 1600 watts produces no more distortion than a LABhorn sent a mere 25 watts.”

Look at it squarely Wayne, until you actually measure labs side by side or make your modified pi, you have not demonstrated anything that says the push pull arrangement is responsible for what you measured.  
You can’t compare your box to other horns with other drivers and conclude the reason for low distortion is “push pull”.
You can’t disregard your measurements on one hand and then use it on the other.
You also keep bringing up eq’ing below cutoff, like Al said, this has always been in regard to home use. In live sound one normally puts high pass filter at the corner, not boost it.

Normally, a measurement like yours would have used a tracking filter to extract harmonics. I suspect it is showing the proper thing where at 10Hz, the distortion has risen 10-15dB above the fundamental.
Push pull only reduces the 2nd harmonic while drivers tend to produce mostly third, which (like harmonics other than the second) are not reduced by push pull.

I am curious what you actually see when you build your non-push pull 12pi to compare to.

“Tom Danley wrote on Mon, 05 November 2007 13:58
Actually I have never built any or measured any myself.”

Yes Wayne. I designed this stuff most of my life and don’t t need to build a simple bass horn to know how it would work, that’s my computers job.

Tom

[Wayne Parham]

Al Limberg wrote on Mon, 05 November 2007 16:28

as a participant at the Michigan shootout, to the best of my recollection all tests were done on single cabinets other than listening tests when we cranked up 4 Labs and the pair of David's Bassmaxx cabs and perhaps 4 of our host's modified MT quad 18 cabs along with a pair of TD1s. I could be wrong.

I pulled the information off this site:

• Michigan Subwoofer Shootout

Way down at the bottom, in the data section, there's a link called Al's four LAB subs (3.08MB).  Is that yours?  When I click on it, it downloads a compressed archive with the measurements inside.  Also inside is a document caled "Lab sub Al.doc" that says "All acoustic measurements done on a block of four cabinets."  That's where I got the information.

[Wayne Parham]

Wayne Parham wrote on Mon, 05 November 2007 11:06

Look at that squarely, Tom. We're talking about a 12Pi basshorn sub generating 30% or maybe 40% distortion at 20Hz at 1600 watts. What we've seen from the Michigan sub shootout is that a group of four LABhorns generates 30% distortion at 20Hz with only 25 watts input. That means even by your own characterization of the data, the 12Pi basshorn subwoofer running full tilt at 1600 watts produces no more distortion than a LABhorn sent a mere 25 watts.

Tom Danley wrote on Mon, 05 November 2007 17:41

Look at it squarely Wayne, until you actually measure labs side by side or make your modified pi, you have not demonstrated anything that says the push pull arrangement is responsible for what you measured.

OK, Tom, so why do you think a group of four LABhorns generates 30% distortion with only 25 watts input, yet a single 12Pi basshorn produces no more than that at 1600 watts?