Best audio perfomance from 8Ω or 16Ω loaded compression drivers?

[1audiohack]

For the last month or so I have been hell bent on quantifying the differences I hear from the different diaphragm materials and driver designs of the two inch throat variety. I think if I can hear it, I should be able to measure it, please note that I said should.

The current test pile consists of 2441's, 2445's, 2446's and 2450's with D16R2441's, D8R2445's, D16R2445's, D8R2450SL's and Truextent Be4008"s 8Ω.

I have extensively measured all of the possible combinations with the exception of the Be's in the 2441's and some patterns are emerging from the data that agree with my ears.

What I don't have is 8 and 16Ω versions of all of the diaphragms. It has been pointed out to me that most all of the really good / high end / top shelf drivers are 16Ω. If that is indeed the case then I must wonder why?

I have always bought the 8Ω versions (for home use) as I like to keep the amps as small as possible (with adequate head room of course) eliminating excess gain to keep the noise floor as low as possible, half the impedance, half the required drive voltage.

Following this logic, or lack of it, my question is this, is there a real sonic benefit to the 16Ω variety? Theoretical sonic advantages? Unless I buy a bunch more diaphragms, I am unable to answer this question for myself. Who would really know?

Thank you,
Barry.

[Robh3606]

It has been pointed out to me that most all of the really good / high end / top shelf drivers are 16Ω.

What would those be?? What's the imp. value for the 435Be, 045Be and the 476Be? If the impedance is the issue then throat diameter is a moot point. The main difference is the wire gage used and the number of turns to keep things simple. Why would more turns of lighter gage wire guarantee you sonic superiority all other parameters being the same??

Rob:)

[Mr. Widget]

Until recently almost all JBL compression drivers were rated as 16 ohms. Isn't the 476Be a 16 ohm driver? I know that all TADs are 16 ohms... my guess would be that the prevalence of 16 ohms is of a historical nature as opposed to one of performance, but that is just a wild assed guess.


Widget

[Eaulive]

On a purely mathematical point of view, higher resistance for a given reactance decreases the Q factor, so there maybe a softer impedance curve... with a given crossover it could lead to a flatter response.

Anyways, what do I know... :blah:

[pos]

there was a thread on this topic on diyaudio recently:
http://www.diyaudio.com/forums/multi-way/163536-1-compression-driver-8-ohm-vs-16ohm.html


I have always bought the 8Ω versions (for home use) as I like to keep the amps as small as possible (with adequate head room of course) eliminating excess gain to keep the noise floor as low as possible, half the impedance, half the required drive voltage.

You can look at it the other way around:
- with 16 ohms you get 3dB more S/N from a given amp
- with 16 ohms you get twice the damping factor from a given amp (not really important for a tweeter I guess...)
- with 16 ohms the protection cap value is divided by two (-> lower price or better quality, especially if charge coupled)

of course you loose 3dB of headroom from your amp, but its power supply will be less stressed, and I cannot see any decent commercial solid state amp that would not be powerful enough to power a compression driver to ear splitting levels at home.


I have extensively measured all of the possible combinations with the exception of the Be's in the 2441's and some patterns are emerging from the data that agree with my ears.Please tell us more! :bouncy:

[yggdrasil]

16Ohm impedance represents an easier load for the amplifier than 8Ohm, giving the possibility of lower distortion, and/or the use of a simpler amplifier construction.

Given a very simple amplifer(e.g. SET), you will get better results by matching the amplifier bias to the represented driver / horn impedance.

[Eaulive]

- with 16 ohms you get 3dB more S/N from a given amp
----------------------------------------------------
of course you loose 3dB of headroom from your amp,

So at the end you're back to square 1 :dont-know:

[4313B]

Isn't the 476Be a 16 ohm driver?Yes.

[pos]

So at the end you're back to square 1 :dont-know:
You have 3dB less noise when connecting the driver without signal (input shorted), which is good for home use

[Eaulive]

You have 3dB less noise when connecting the driver without signal (input shorted), which is good for home use

Yes, but ony because you get half the power from a given voltage, so the amplifier will arrive faster to clipping for the same SPL.
All in all the dynamic range is the same, you could arrive to the same result by padding the driver ;)

[1audiohack]

there was a thread on this topic on diyaudio recently:
http://www.diyaudio.com/forums/multi-way/163536-1-compression-driver-8-ohm-vs-16ohm.html


Please tell us more! :bouncy:

I tried to glean what I could from that thread, thank you for that link. One thing of interest was the coments about Neo, I see something that puzzles me in the 2450 driver that I think is magnet related and I will follow up on it.

If and when I believe I have something conclusive to share I will be sure to share it. The last thing I want to do is possibly mislead anyone with incorrect information. At this point I think I need to purchase a pair of 16 Ohm Truextent diaphragms and keep on digging. I know it's a big puzzle, but I have now in place (at least for me) a couple more pieces.

Thank you for sharing guys,
Barry.

[pos]

Yes, but ony because you get half the power from a given voltage, so the amplifier will arrive faster to clipping for the same SPL.
All in all the dynamic range is the same, you could arrive to the same result by padding the driver ;)
The residual noise (in volt) will generate less acoustical noise, which is very important when driving the driver directly from an amp. If you use a lpad you also loose power (real power this time, in heat) and you lower the apparent damping factor for the driver (but how cares for a compression driver...).

[Eaulive]

The residual noise (in volt) will generate less acoustical noise, which is very important when driving the driver directly from an amp. If you use a lpad you also loose power (real power this time, in heat) and you lower the apparent damping factor for the driver (but how cares for a compression driver...).

Resistors are a common component in crossovers, of course they eat power, but all in all, I don't think it was the reason behind having compression drivers at 16 ohms instead of 8.

I'm not even sure there's a real reason, :D

[JeffW]

I tried to glean what I could from that thread, thank you for that link. One thing of interest was the coments about Neo, I see something that puzzles me in the 2450 driver that I think is magnet related and I will follow up on it.



I didn't see anything about neo magnets in that mess, but I kind of glossed over a bunch of it :blah:. You have 2446 and 2450 drivers, aren't they about the same except for the magnet? I hope it's not neo=bad, all my drivers are neo :eek:

[1audiohack]

I don't want to wade through all of that thread again, what I thought was implied was that (most?) Neo drivers skimp on the magnet because Neo is ??? Not all of that threads discussion was centered around compression drivers. But I see a ghost in the 2450 that is likely unimportant, I just want to know what it is.

[JeffW]

Ah, found it, post #106


There is one thing which has always been apparent and is again shown by some studies with these and some other compression drivers- neodymium magnet structures always skimp on the magnet. There are several reasons for this but suffice to say neodymium is never worth the money...so far. These neo transducers seem to almost always lead to much less than ideal results than their ceramic magnet cousins and could lead to false conclusions about driver usefulness and behavior. Use the ceramic for more accurate understandings until the neo magnet type are of the same or better performance than there presently superior cousins.

Not sure about the guy's credentials, and all he ever linked to was Parts Express cut sheets. It'd be interesting to hear what JBL's transducer engineers have to say on his opinion.

FWIW, JBL lists the 2446 and 2450 as both being 19,000 gauss. I guess you'd need to measure it to see if they skimped on the 2450.

[Eaulive]

I won't read the whole thread again but the neodymium magnets have more power per pound than ceramic and even alinco, having them smaller is not to "skimp on the magnet" this notion is wrong.

When they switched from alnico to ceramic, this was a good opportunity to skimp, some did maybe, JBL didn't that's why we ended up with woofers twice the weight.

At the end the important thing is the flux inside the gap.

[1audiohack]

At the end the important thing is the flux inside the gap.

A possible consideration is flux stability or the magnetic hysteresis of the motor. One could argue it's field coil, AlNiCo, ferrite, then Neo, and sheer mass may be a part of that. Not my speciality, just curiosity.

[Eaulive]

One could argue it's field coil, AlNiCo, ferrite, then Neo,

You mean from best to worse?

[tomt]

Ragnar Lian, co-founder of Scan-Speak,
said that speakers with alnico magnets had 10x less distortion than ferrite,
and neodymium had about 20x less.
If a copper ring is added to a ferrite magnet,
it can approach neodymium specs but only above a certain frequency,
below which the ring becomes ineffective.

http://audioroundtable.com/Speaker/messages/223.html

Greg Timbers on magnets -

Ferrite magnets do not demagnitize with time or drive. They are affected by temperature but that is reversible. They will return to normal when they return to room temperature. Ferrite is basically a lousy magnet material for speakers but it is cheap and readily available. JBL has done a ton of things within the magnetic circuit to make the material behave in a more stable manner. At 100 degrees F, a Ferrite motor will be down about 1.5 dB in level which means the midband of the woofer will be lower by that much and there will be increased output around the system resonance. The TS parameters will be completely different - as though the BL was reduced by about 18%.

Alnico magnets, by their nature are easy to demagnitize with drive. They will not change with time and their dependence on temperature is really small - maybe 1% at 100 deg.F. Alnico stability and resistance to back EMF is really good. This is why they make the best sounding magnetic structures. Unfortunatelly, given a big enough pulse of magnetic energy, they will demagnitize by up to 3 dB. The sensitivity to demagging is dependent on the specifics of the magnetic circuit and the length of the coil providing the field. Underhung woofers (LE15 and such) midranges, tweeters and compression drivers do not have sufficient back EMF fields to push the operating point of the structure below the knee. They are essentially stable regardless of input signal. The short gap-long coil speakers are the ones that have a problem. A 2235 can take a hit of up to 3 dB if a big enough hit of current takes place. 1.5 dB to 2 dB is more common. The effect does not get better or worse with time, it solely depends on how much current is driven through the coil. The more current, the more field. Once the field is bigger than a certain number, some amount of demagnitizing occurs. It is perminent (until externally recharged) and will only increase if a larger sustained current hit occurs.

Therefore, if you have a qualifying alnico woofer and you have played it loudly you have some damagging. You can have the unit recharged and it will be fine until you play it again. Exceed the critical level and it will start happening. If you never do, it won't ever demag. Most of these designs trace back to the 50's and 60's where 15 - 30 watt tube amps were the rule. They didn't have the current capability to hurt anything. With the advent of big solid state amps, the current levels went up and the problems started to surface.

Most of the qualifying 4" motors will loose 1 - 1.5 dB unless they are pummeled. Some of the older 3" with really short magnets, like the 2213A and 123A will typically be around 3 dB down. They go really easily. The old decade woofers (116A and 127A) only had to see an amplifier in the room and they got really nervous. FYI, the new 1500Al used in the S9800 can take continued pulses of 5000 watts and loose no more than 1%. The test can only be done a few times before the coil is destroyed, but the magnetic assembly is totally stable.

http://audioheritage.csdco.com/vbulletin/showthread.php?10280-Recharging-Alnico-Ferrite-magnets&p=110522#post110522

http://growabrain.typepad.com/photos/uncategorized/wild_kid.gif

[1audiohack]

Yes, from best to worst. It's not my held theory but I have read it. This could very easily go somewhere I didn't intend and I just don't want to go.

I am looking to quantify if possible, to show and compare with hard numbers and graphs, what it is that I discern by listening in hopes of getting the absolute most out of what I have and can afford.

I also have a bent for absolutely having to know how things really work. Sometimes it's a curse, I am not an engineer or scientist and as such have the expected limitations. A problem with our hobby is anecdotal information is prevelent, but real knowledge is often much more rare than many are willing to admit. Some answers are not easily found in the general literature so I search still. I had hoped that somehow there was a solid answer to the original question and that I had just missed it.

Hopefully I will have something usefull to add in the future, for now I have far more questions than answers so I'll pipe down.

Thanks all,
Barry.

[JeffW]

Don't stop on my account, you never know what'll pop out of a discussion. The way I look at it, JBL wouldn't waste a chance to make a 476Be with something other than neo if something else would have been better. When you make a $$$$ compression driver with a pretty small production number, cutting corners on magnets doesn't really make any sense. The D66000 weighs over 300lbs each as it is, surely a few more pounds of magnet wouldn't have made them scrap the design.

It doesn't really matter to me in a practical sense, I'm stuck with neo-based drivers now anyway. They'll just have to be good enough :D

[1audiohack]

Oh I'm not going to stop. I'm just getting started.

[Mr. Widget]

I think the design integrity and quality of manufacture are more important than the material the magnet is made of... as Ragnar Lian and GT suggest in the post above some of the short comings of ferrite can be over come by clever engineering as can the short comings of alnico... even the inherent design of the driver affects this, (GT's comparison between the alnico mags used in the underhung vs. overhung designs)... another data point, TAD uses the same diaphragm in the 4001 and 4002 compression drivers... the 4001 is alnico and significantly costlier than the 4002 which uses neodymium. So, is neodymium only meant for a lighter and cheaper design in TAD's belief? No, their TOTL 4003 is also a neodymium driver... I think it is safe to say that simple ratings of different drivers based on their magnetic materials is a pointless exercise.


Widget

[pos]

The 2452 would also be an interesting case to study: it has the same flux density and specifications than the 2451 (also neo) but half the weight!

[Robh3606]

There is one thing which has always been apparent and is again shown by some studies with these and some other compression drivers- neodymium magnet structures always skimp on the magnet. There are several reasons for this but suffice to say neodymium is never worth the money...so far. These neo transducers seem to almost always lead to much less than ideal results than their ceramic magnet cousins and could lead to false conclusions about driver usefulness and behavior. Use the ceramic for more accurate understandings until the neo magnet type are of the same or better performance than there presently superior cousins.

Don't put too much weight in some guys statement on the Internet. There is nothing wrong with Neo, Alnico or Ferrite that a good transducer design cannot work around. You need hard data and it would be very foolish to think JBL or Tad don't already have it. I have read many times that Neo is Superior to alnico and ferrite. From my point of view I don't care as I have drivers using all three materials that all "sound" quite good in actual use.

Rob:)

[4313B]

Don't stop on my account, you never know what'll pop out of a discussion. The way I look at it, JBL wouldn't waste a chance to make a 476Be with something other than neo if something else would have been better. When you make a $$$$ compression driver with a pretty small production number, cutting corners on magnets doesn't really make any sense. The D66000 weighs over 300lbs each as it is, surely a few more pounds of magnet wouldn't have made them scrap the design.

It doesn't really matter to me in a practical sense, I'm stuck with neo-based drivers now anyway. They'll just have to be good enough :DYep.

While there is an obvious advantage to "AlNiCo done right" in the case of the 1500AL, 1500AL-1, and 1501AL, JBL must not have thought that the same advantage existed for the 476BE. The only cause for pause seemed to be a four slot phase plug versus a five slot phase plug. We all know how that worked out - we have a four slot phase plug. While the 476BE doesn't have the mass that the 2441 sports, it is definitely not merely a scaled up 435BE.

The 2452 would also be an interesting case to study: it has the same flux density and specifications than the 2451 (also neo) but half the weight!If I remember correctly JBL stated that the 2452 was simply a 2451 manufactured with lighter, less expensive materials. Same MSRP though. ;)

[1audiohack]

I think it is safe to say that simple ratings of different drivers based on their magnetic materials is a pointless exercise.
Widget

I absolutely agree.

[1audiohack]

You need hard data and it would be very foolish to think JBL or Tad don't already have it.Rob:)

I agree completely again.

So, riddle me this gentlemen, I have measured the D16R2441's in all four pairs of the test drivers and the SPL variations and the impedance peak magnitude at and below 700 Hz is best behaved in the 2441, a little worse but line on line in the 2445 and 2446, and markedly worse in the 2450. Is it irrational to wonder why?

[yggdrasil]

I agree completely again.

So, riddle me this gentlemen, I have measured the D16R2441's in all four pairs of the test drivers and the SPL variations and the impedance peak magnitude at and below 700 Hz is best behaved in the 2441, a little worse but line on line in the 2445 and 2446, and markedly worse in the 2450. Is it irrational to wonder why?

What about the size of the rear chamber?

[Charlie4350]

I agree completely again.

So, riddle me this gentlemen, I have measured the D16R2441's in all four pairs of the test drivers and the SPL variations and the impedance peak magnitude at and below 700 Hz is best behaved in the 2441, a little worse but line on line in the 2445 and 2446, and markedly worse in the 2450. Is it irrational to wonder why?

could it be the 2450's throat starts out at 1.5"?

[Robh3606]

Is it irrational to wonder why?

No but I wouldn't hang my hat on it being the Neo. There are simply to many other variables.

Rob:)

[JeffW]

The 2452 would also be an interesting case to study: it has the same flux density and specifications than the 2451 (also neo) but half the weight!

Not that JBL hasn't had typos in their spec sheets before, but I'm not seeing the "half the weight" part. Both are listed by JBL as 4.5 kg, but the wording is a little different.

2452 (http://www.jblproservice.com/pdf/Discrete%20Sales%20Models/2452HJ.pdf)

2451 (http://www.jblpro.com/pages/pub/components/2451.pdf)

[4313B]

I agree completely again.

So, riddle me this gentlemen, I have measured the D16R2441's in all four pairs of the test drivers and the SPL variations and the impedance peak magnitude at and below 700 Hz is best behaved in the 2441, a little worse but line on line in the 2445 and 2446, and markedly worse in the 2450. Is it irrational to wonder why?Now put the diaphragm back in the 2441 and see if you get the same impedance curve that you got the first time...

[1audiohack]

No but I wouldn't hang my hat on it being the Neo. There are simply to many other variables.

Rob:)

I really didn't think I had.

[1audiohack]

Now put the diaphragm back in the 2441 and see if you get the same impedance curve that you got the first time...

I did, I always A B A.

[4313B]

I agree completely again.

So, riddle me this gentlemen, I have measured the D16R2441's in all four pairs of the test drivers and the SPL variations and the impedance peak magnitude at and below 700 Hz is best behaved in the 2441, a little worse but line on line in the 2445 and 2446, and markedly worse in the 2450. Is it irrational to wonder why?You put the diaphragms in the cores and adjust them to get the best sonic performance and then live with the resulting impedance curves. They are all going to vary. The primary focus is on sonic performance, there should be no objectionable resonances across the intended bandwidth. Once you get the drivers to that level then you start matching impedance curves if so desired. You'll need a bunch of drivers to get reasonably matched pairs. A micron shift in a diaphragm results in a different impedance curve as well as a potential resonance point.

[pos]

Not that JBL hasn't had typos in their spec sheets before, but I'm not seeing the "half the weight" part. Both are listed by JBL as 4.5 kg, but the wording is a little different.

2452 (http://www.jblproservice.com/pdf/Discrete%20Sales%20Models/2452HJ.pdf)

2451 (http://www.jblpro.com/pages/pub/components/2451.pdf)

The spec sheet for the 2452 is bogus.
The diameter measurement are also all wrong.
My 2452 is ~14cm in diameter and weigths less than 2.5kg

[Robh3606]

I really didn't think I had.

Sorry didn't mean to imply you had, 4313B has a very valid point in as they do change when installed as an example. Not getting into different horns and so on. Why don't you post what you are measuring and on what horn??

Rob:)

[pos]

The spec sheet for the 2452 is bogus.
The diameter measurement are also all wrong.
My 2452 is ~14cm in diameter and weigths less than 2.5kg
confirmation from the engineering sheet:
http://www.audioheritage.org/vbulletin/showthread.php?21706-2452h-sl&p=215553&viewfull=1#post215553

2kg it is

They managed to keep the flux density the same while deviding the weight by more than 2 !
I also like the fact that the wire plugs are directly fixed to the diaphragm: clever design, no more wire inside the back chamber, and a more direct connection!

They came back to the older design for the 476Be/476Mg though...
Maybe the longer copper sleeved pole piece and (larger magnet it implies) could not be installed in the new design? (as well as these new "tuned acoustic vents" under the surround)

[1audiohack]

When I first started this study my reasons for the original question was driven by one observation and a theoretical wonder.

First, one reads all the time that you often listen to these large format drivers at home driven at milliwatt levels and that any decent amp will be capable of making your ears bleed. I find this to be untrue. I don't drink and I'm not deaf, but,, when listening above background levels, a Crown D45 rated at 20 watts per channel into a 16 ohm load will clip easily, a D75 rated at 25 watts per channel into a 16 Ohm load will run them OK, barely if you get aggressive.

Proff. Marshal Leach Jr makes a good point in one of his papers about designing for maximum sensitivity rather than efficiency since power is so easy to come by these days. I haven't had the chance to measure any of the new horn / driver combinations but I do wonder what the combined operational impedances are for these. I need to get out the Pearson current monitor and O scope and see if I can figure out where all the power is going.

Second, I wondered if the difference in series L of the different impedances would make a notable difference in the very high frequencies. After measuring and listening for a couple of months now I decided it doesn't matter, none of these combinations I have here run high enough clean enough to make me not want to use a tweeter anyway. The TruExtent diaphragms are the best behaved all around I'm sure. The real issue now is that the Be's behave well enough upstairs that improper tweeter integration would be a spoiler where damn near any tweeter anywhere in the room would benefit some of the other combinations.

Looking into some of the newer super cool drivers that one can actually find measurements for like the 435Be and the 2435Be I found something I have never before noticed JBL do, they leave the fundamental and distortion curves in place on the sheet. I also noticed that all these measurements were done with TEF, the same system I use. I had measured the 435Al's and noted the divide between the fundamental and 2nd and 3rd harmonics as the cleanest driver I had yet measured, when I saw the JBL measurement set up was identical to mine and the graphs looked almost identical to what I was getting, I felt that at least I was looking at truly comparable data.

I want to thank all who shared, and if there is anything I might be able to add please ask. I think I have at least figured out what ranges to use the available materials in if the goal is minimum phase operation and best signal to noise. I know none of this is new, it's just some things aren't easily found in common literature and it's just plain fun to explore and learn.

All the best,
Barry.

[Mr. Widget]

I don't drink and I'm not deaf, but,, when listening above background levels, a Crown D45 rated at 20 watts per channel into a 16 ohm load will clip easily, a D75 rated at 25 watts per channel into a 16 Ohm load will run them OK, barely if you get aggressive.

I need to get out the Pearson current monitor and O scope and see if I can figure out where all the power is going. If you are using a CD horn the HF roll off is likely quite considerable... the required compensation lowers the driver's practical sensitivity significantly.


Widget

[speakerdave]

. . . . I have measured the D16R2441's in all four pairs of the test drivers and the SPL variations and the impedance peak magnitude at and below 700 Hz is best behaved in the 2441, a little worse but line on line in the 2445 and 2446, and markedly worse in the 2450. Is it irrational to wonder why?

This may be germaine, or maybe not, but I have a recollection that in the literature touting the large format Altec drivers--the 288 family--one of the arguments was that for the smaller 1" compression drivers the lower body mass gives rise to resonance issues in the lower frequencies.



. . . . So, is neodymium only meant for a lighter and cheaper design in TAD's belief? No, their TOTL 4003 is also a neodymium driver... I think it is safe to say that simple ratings of different drivers based on their magnetic materials is a pointless exercise . . . .

Yes, and though the 4003 uses neodymium, it also weighs 30 lbs, about the same as a 288, 4001 or 2441.