My open baffle dipole with Beyma TPL-150

[63781]

I have a small company that makes and sells amps based on ICEpower modules, so the amps are from my own production. ICEpower modules are not sold to individuals, only to companies with a partner agreement with B&O.

Most of my amps use the new ICEpower 125ASX2 module, that you can also find inside the Rotel RB-1510 amp....

 

[thadman]

StigErik,

You amaze me. I cannot tell you how refreshing it is to see someone “with means” engineering their own solution to the familiar “loudspeaker problem”. I have traveled to Europe several times and have visited the homes of several individuals who truly value high fidelity sound (something I rarely encounter in the US). These individuals have been very successful in their professional lives and have purchased extraordinarily expensive loudspeaker systems ($>50k). I have a reputation for my interest in loudspeaker design and they have been eager to show off their systems. Although impressive, I always wonder what could’ve been achieved with that same amount of money if a passionate engineer (ie someone who is familiar with the electrical, mechanical, and acoustical constraints) took on the project. You are an inspiration to me. Thank You!

Although I do not possess a disposable income that is comparable to yours, I would like to achieve something similar.

I share your desire for dynamics. I want a system capable of live levels (110dB continuous, 120dB peak) and I also want dipole bass. This presents challenges as I'm a big fan of German electro.

I would be in heaven if I was able to listen to Electrocardiogram by Kraftwerk at reference levels

For the moment, I would like to focus on the dipole LF drivers as I feel this bandwidth encompasses a significant portion of the energy in music.
I'm interested in using four Eighteen Sound 21LW1400 21" drivers for dipole bass below 150Hz in a push/pull configuration (2 drivers per side). I understand they probably aren't exactly ideal as compared to your Beyma drivers, however they are much cheaper. Also, I know you hesitate offering your opinion on drivers you haven't heard, but I feel you are in a unique position since you have used drivers of a similar topology. With that said, I would love to hear your thoughts and will make sure to take them in context.

21LW1400 - Extended LF Ferrite Driver

21LW1400

The 21LW1400 design features include an exceptional displacement suspension system which, in conjunction with a carbon fiber reinforced straight-ribbed cone and the Double Silicon Spider (DSS), produces an ultra-linear piston action, providing full control across the entire working range. The 100mm inside outside copper voice coil, based on our Interleaved Sandwich Voice-coil (ISV) technology, reaches high levels of thermal stability and durability. ISV technology is based on a high strength fiberglass former with half the coil wound on the outside and half on the inside and bonded together using unique high temperature resin adhesives. This results in a balanced linear motor unit which can exert an exceptionally high force factor.
The low distortion and unmatched sound quality of the 21LW1400 has been significantly improved by the Double Demodulating Rings (DDR) embedded in the pole piece of the magnetic structure. These have been designed to dramatically reduce the intermodulation and harmonic distortion while improving transient response at the same time.

Fs=28Hz
Sd=1662cm^2
Qts=.235
Linear xmax=9.5mm
Le=2.85mH
Sensitivity=99dB
VC diameter=100mm
AES power=1400 watts
Peak power=7000 watts
-0.6dB compression @ -10dB power
-1.5dB compression @ -3dB power
-2.2dB compression @ 0dB power

Obviously they will need significant equalization with a Qts of .235, but they appear to be able to handle a significant amount of power, so I am not concerned.

They have quite a high inductance of 2.85mH. The value of inductance does not concern me as I do not believe subwoofers need to be "fast". If a driver is faster, it is reproducing a higher frequency. However, what does concern me is variation of this parameter with displacement. If the value of inductance changes with displacement, significant IMD will be produced due to the shifting of system poles and system Q. For drivers with very low Le (such as my TD15M Apollo drivers), shifting of this parameter may not necessarily be a huge problem as it represents a minor fraction of the total system impedance. However, for drivers with high inductance, it becomes a significant problem.

The question arises, how high can I realistically take the 21" drivers? Baffle width aside, how high would you take your 21" drivers without compromising fidelity? Is 150Hz realistic?

 

[skorpion]

2 AEIB15s in 20x40x15" H-baffles each side will produce your dream. Why cross the river for water ? Just get the MJK MathCad-sheets and model.

/Erling

 

[jerome69]

Yeah, the latest ICEpower modules are the best amps I've ever heard. That Class D is best lest for the bass is probably a myth from the times where Class D designs were very limited in HF extension, that is - back in the 60's......

Yes the measurement of this ICEpower 125ASX2 looks very impressive
I have never seen this with a PWM amplifier
It's a pity, we should cheat to buy this module .

 

[twest820]

That Class D is best lest for the bass is probably a myth from the times where Class D designs were very limited in HF extension, that is - back in the 60's......

Hmm, I dunno. I've looked at a lot of class D datasheets and, while I haven't made a rigorous survey of the measurement conditions used, most exhibit a THD+N hump around 7kHz. And, of the datasheets I've checked, the ones which state the bandwidth usually 20-20kHz. Makes me suspect the hump isn't actually a hump and that the falloff in THD above 7kHz is just a measurement artifact. There's certainly a valid discussion about whether one cares about what happens above 20kHz, but it's enough to make me think twice about putting class D on a tweeter or mid. That hump often peaks above 1% THD and, if you accept THD as a proxy for IMD which tends to underpredict somewhat, it's likely audible. Especially if IMD continues to rise in the highs. Another way of looking at it is the full power bandwidth of the typical class D part is a few hundred Hz. Which, as the amp's not very linear to start with (duh ), isn't particularly attractive; makes me think there's nontrivial improvement yet to be had from getting switching frequencies into the MHz range rather than a few hundred kHz.

There are a few parts which manage to keep the hump below 0.1% where it's likely not too audible even if IMD continues to climb in the highs---the LM4675 and ADAU1592 come to mind---but Hypex's designs are the only exception to this pattern I know of. Pretty flat THD even though the FBP's only 40kHz. (I notice ICEpower doesn't seem to spec THD+N as a function of frequency, but their THD+N versus power figures suggest a hump.)

 

[63781]

ICEpower 125ASX2:

An externally hosted image should be here but it was not working when we last tested it.

 

[twest820]

Yep, that's what I was looking at; 6.67kHz being much higher than <=1kHz is suggestive of the usual hump. And, not having a business, I can't buy one. So it doesn't really matter the performance is the best of any class D spec I've seen---not that Hypex is particularly far behind. But neither Hypex or ICEpower is representative of the majority of class D solutions available; the average class D chip doesn't do a whole lot better than 1%.

 

[DQ828]

ICEpower 125ASX2:

An externally hosted image should be here but it was not working when we last tested it.

I'd definitely use the IcePower modules if I could get hold of them. To hell with the cost.

David

 

[c2cthomas]

Hi Stig - Long time reader of this thread and it has been an educational - mind expanding - pleasure to observe your progress and findings. A GREAT THREAD!!!!!

I was checking out some drivers and was looking at the Beyma TPL 150 and found this new TPLH 150 horn and thought it might interest you. US SPEAKER Home Page Menu - Speaker Cabinets, Guitar & amp; Bass Amps, upgrades and custom design. The world’s widest choice of speaker parts. “If you have listened to live music, you have almost certainly listened to Eminence”



edit - the link will not paste correctly - just look for the TPLH 150 on the site.

 

[jerome69]

The problem with Hypex, I have never seen an IMD graph , they are only good for bass subwoofer ?
IMD for my opinion is more important than harmonic distortion. It gives the real ability of the amplifier to reproduce music in the fullband.

I think a whole part of class D amplifiers has bad IMD measurements, and are not sonically convincing. This ICE amp has very good IMD measurements.

 

[sendler]

Class D vs AB

Hmm, I dunno. I've looked at a lot of class D datasheets and, while I haven't made a rigorous survey of the measurement conditions used, most exhibit a THD+N hump around 7kHz. And, of the datasheets I've checked, the ones which state the bandwidth usually 20-20kHz. Makes me suspect the hump isn't actually a hump and that the falloff in THD above 7kHz is just a measurement artifact. There's certainly a valid discussion about whether one cares about what happens above 20kHz, but it's enough to make me think twice about putting class D on a tweeter or mid. That hump often peaks above 1% THD and, if you accept THD as a proxy for IMD which tends to underpredict somewhat, it's likely audible. Especially if IMD continues to rise in the highs. Another way of looking at it is the full power bandwidth of the typical class D part is a few hundred Hz. Which, as the amp's not very linear to start with (duh ), isn't particularly attractive; makes me think there's nontrivial improvement yet to be had from getting switching frequencies into the MHz range rather than a few hundred kHz.

There are a few parts which manage to keep the hump below 0.1% where it's likely not too audible even if IMD continues to climb in the highs---the LM4675 and ADAU1592 come to mind---but Hypex's designs are the only exception to this pattern I know of. Pretty flat THD even though the FBP's only 40kHz. (I notice ICEpower doesn't seem to spec THD+N as a function of frequency, but their THD+N versus power figures suggest a hump.)

I have tried quite a few different diy amps. TDA7294, Behringer A500, Rega Delta 290, Velleman 50watt module with TIP147, original AKSA, single 3886, Bridged/balanced 3875 BA100, Ucd180, and the previous champ, TDA7250 driving TIPs. In that order of increasing sonic performance. By the time you get back to the A500 it is very disappointing. The 7250 was quite a step ahead of the rest although the BA100 showed how much better the 3875 can be with more than one chip if you have a balanced source such as my direct out DEQ2496. For 3875/86 chip amps, I would consider a BPA200 with four chips a minimum that might get you into the ball park of the class D if you have a nice balanced source to feed it with so you don't need a phase splitter on the input. A modified Sure 2X100 with a cheap smps beats them all. At a lower price. As long as you can get by with a powerful sounding 70 watts. Explosive dynamics and very transparent. I don't build any class A or tubes but have read from another guy with a KSA50 clone and an Aleph that his unmodified classD audio amp sounds better. Most people think of the little Tamp running on 12V when they talk about Class D but the newer, more powerful Class D can be quite impressive when run with a wide open output filter. Based on listening, my $40 TK2050 amp running on a $45 MeanWell SMPS sounds much better than any of my AB amps and better than the UCD180 and is twice as efficient as the AB amps.
__________________

 

[jerome69]

I don't want to say class D is better than an other and vice versa

I think an amplifier whatever his class must be faithful ! And there are not a lot and the best amplifiers I have heard come at a price
This ice power amplifier 125ASX2 seems on the paper very impressive and I believe it is a very good sounding amp.

Every one should find his best sounding amplifier.

We should not forget, the worst element in term of distortion is the loudspeaker

 

[63781]

I'd definitely use the IcePower modules if I could get hold of them. To hell with the cost.

David

Not all that expensive I think. In this amp (see link), there's two 125ASX2 modules, at the price of just 655 Euros:

HFX Online Store

 

[63781]

@Thadman:
That 18sound driver looks pretty good I'd say, should be possible to get very good results from it! The driver itself should have no problems playing up to 150-200 Hz. I used my Beyma 21" up to 350 Hz in an other dipole, and it sounds great even then.

The main limiting factor will be the size of your H-baffle, and the corresponding dipole peak. You should not attempt to cross over above the dipole peak.

My H-baffles are approx 600x600 mm in footprint, and 1100 mm tall. Dipole peak is at 150 Hz, and they do not sound good above 120 Hz or so.

 

[DQ828]

Not all that expensive I think. In this amp (see link), there's two 125ASX2 modules, at the price of just 655 Euros:

HFX Online Store

Thank you

 

[keyser]

(...) the average class D chip doesn't do a whole lot better than 1%.

Really??!

 

[twest820]

Really??!

So what conclusion do you reach after looking through the parts selectors at, say, Analog, Maxim, National, and TI and sampling a few dozen datasheets to get a feel for how the headline THD+N figures correlate to general performance?

To be clear, discussion here on DIYA selects for higher power and performance. So there's a difference between average class D part on the market---which I'd characterize as a 1W 0.5% THD solution---and the average class D implementation discussed on DIYA---which is usually more like 20W clean and 0.01% THD (though the headline wattage is usually an order of magnitude higher since class D parts are often rated at 10% THD). Also, it's something of a moving target; I haven't done a rigorous survey but I'd describe 1% THD as typical of the the older parts on the market and 0.1% THD as typical of the newer ones.

 

[mige0]

2 AEIB15s in 20x40x15" H-baffles each side will produce your dream. Why cross the river for water ? Just get the MJK MathCad-sheets and model.

/Erling

@Thadman:
That 18sound driver looks pretty good I'd say, should be possible to get very good results from it! The driver itself should have no problems playing up to 150-200 Hz. I used my Beyma 21" up to 350 Hz in an other dipole, and it sounds great even then.

The main limiting factor will be the size of your H-baffle, and the corresponding dipole peak. You should not attempt to cross over above the dipole peak.

My H-baffles are approx 600x600 mm in footprint, and 1100 mm tall. Dipole peak is at 150 Hz, and they do not sound good above 120 Hz or so.

Finally had some time left to update my OB / AMT subwofer research :

http://www.diyaudio.com/forums/subwoofers/172553-amt-sub-using-dynamic-woofer.html#post2394749

Bottom line regarding SPL maximization / volume minimization :

take as deep as possible frames (peak at or above XO) and „fill them with woofers“


This rewards with basically two benefits:

1. least attenuation due to dipole seperation
2. if done „right“ some less EQing needed due to less peaky response


Might be, there is even more room for improvement, but as there is no good working simu available for multi-woofer AMT arrangements currently – it will take some further time...

Michael

 

[darkfury]

OB or Baffles side effects

As I know, We need the box for

• enhanced lower response of the driver
• Dynamics
• Linear FR
• and also prevent to propagate the inverse phase from back of the driver to the listening room.

We need the OB/ Baffles to eliminate the

• Box/Baffle sound
• Inner acoustic cancellations

Stig, the inverse phase from back of the driver, is it an issue for this speaker?
Even, the curtain seems behind of the speakers.

 

[63781]

Yes - the out of phase rear radiation is fundamental in how dipoles work, in fact all drivers except the tweeter works in the range where the rear radiation is canceling the front radiation.

The curtains are behind the speakers indeed, more than 1 meter away.