diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Pass Labs (http://www.diyaudio.com/forums/pass-labs/)
-   -   What watt? Why do we need all that power? (http://www.diyaudio.com/forums/pass-labs/65336-what-watt-why-do-we-need-all-power.html)

hilbert 3rd October 2005 01:53 PM

What watt? Why do we need all that power?
 
Dear Mr Pass! (or anyone else who might know the answers)

I'm new to this forum, and I might not have browsed previous messages well enough, so please excuse any repeated questions.
My knowledge of electronics is fairly limited, but I am a doctoral student in acoustics, so I know a bit about complex impedance and such. At the department I have built a low frequency wall of 15-inch woofers in a 4x4 matrix. I use a Lab Gruppen 1.4 kW power amp to drive it and it goes straight down to 12Hz. Or, to cite the honourable Mr. Pass: -"MuuuuHahahahaha……".
Now, in the specs it says that the amp gives 58 Vrms peak, over 8 ohms load. Loudspeaker impedance is rarely just resistance, but if it was, 8 ohm gives something like 7.25 amp, which gives about 420 W fed into the load. A loudspeaker is to the amplifier a complex impedance that depends on the radiation impedance of the woofer and tweeter, the crossover and so on. Is it common to have so low impedance that one actually needs an additional 100 W? What happens then to the loudspeaker? My 16 woofers can dissipate a lot of power due to their numbers, but for an ordinary loudspaker, isn't there a risk that something goues up in flames? Bob Carvers Sunfire subwoofers use a way too small box and a long stroke woofer. The small box means huge loss in efficiency, so he feeds some kW's into it as compensation. It gets really hot (I managed to trigger the thermal protection), but it's kind of built for it. Ordinary loudspeakers are not. In the Aleph 1, there are, if I understand it correctly, 40 IRF244's to get enough power out of it. I have just bought a subwoofer that needs an amp, and I was thinking of building one based on the Zen and Aleph articles by Mr. Pass. The subwoofer is kind of similar to a Sunfire in the sence that it has a T C Sounds long stroke 12-inch woofer in a relatively small box. Now, do I really need some kW's or is it enough with say 150 W like in the Aleph 1? Instead of using 40 IRF244, can one use fewer high-power FETs like the STE180NE10/STM (http://www.st.com/stonline/products/...re/ds/6872.pdf), given that they sound good?
I was given a 500VA toroidal transformer with 2x38V secondary for free, and It would be nice if it would be suitable for a amplifier for my subwoofer, but I guess it might be a little too weak?

Regards

/Anders

Nelson Pass 3rd October 2005 09:10 PM

Re: What watt? Why do we need all that power?
 
Quote:

Originally posted by hilbert
Is it common to have so low impedance that one actually needs an additional 100 W?

What happens then to the loudspeaker?

In the Aleph 1, there are, if I understand it correctly, 40 IRF244's to get enough power out of it.

I have just bought a subwoofer that needs an amp, and I was thinking of building one based on the Zen and Aleph articles by Mr. Pass.

The subwoofer is kind of similar to a Sunfire in the sence that it has a T C Sounds long stroke 12-inch woofer in a relatively small box. Now, do I really need some kW's or is it enough with say 150 W like in the Aleph 1?

can one use fewer high-power FETs like the STE180NE10/STM, given that they sound good?

I was given a 500VA toroidal transformer with 2x38V secondary for free, and It would be nice if it would be suitable for a amplifier for my subwoofer, but I guess it might be a little too weak?

1) It is not uncommon for speakers to dip to low impedances,
so an amplifier with a high current capacity is sometimes a
necessity.

2) The presumption is that a speaker which demands high power
will accept it without damage. This is not always the case.

3) Later Aleph 1.2's used 24 devices per channel, and they
idled at about 25 watts each.

4) Neither the Zens nor the Alephs were actually famous for
deep and powerful bass. Their attractions were elswhere.
You might think about a version of the A75.

5) You can use higher power Mosfets. In the case of the part
you mention, it is rated at twice that of a 244, so you might
think of running it as high as 100 watts of dissipation if you have
a good heat sink.

6) I'm sure you could build a fine amplifier with that transformer.
Running balanced (bridged) you could maybe get a kilowatt in
peaks.

BWRX 3rd October 2005 09:19 PM

Re: What watt? Why do we need all that power?
 
Quote:

Originally posted by hilbert
I was given a 500VA toroidal transformer with 2x38V secondary for free, and It would be nice if it would be suitable for a amplifier for my subwoofer, but I guess it might be a little too weak?
I realize this is the Pass Labs forum but your transformer would be perfect to use in a power supply for a UcD400.

Get some nice caps for cheap from Sander, some good rectifier bridges, other necessary parts and you've got yourself a power amp :)

GRollins 4th October 2005 04:18 AM

There are more aspects to this than a leopard has spots. Some random thoughts:
--Low frequency wattage requirements are absolutely huge compared to midrange and high frequencies. It quickly gets worse if you use eq, a feedback network, or any of the other tricks people have tried over the years in order to get the carpet rippling.
--RMS or average wattages are one thing. Peaks are another. A good rule of thumb is to assume a 10:1 ratio between the regular listening power and the spike you'll see when someone does something rude...like hit a drum.
--The Carver sub plays tricks. So do the variations on the ELF concept, etc. See my first point, above.
--Class A is somewhat less important at 20Hz than it is at 1kHz.
--Few, if any, of these long throw woofers are anywhere near flat. They're modified boom-boom car drivers. Expect a very strong peak in the 80-100Hz region and be prepared to use eq to knock it out. Don't tell me what the simulations say, measure it.
Sixteen 15" drivers ought to be able to move a decent amount of air. I use twelve 12" drivers, not for volume (although I did cause some hairline cracks in the ceiling once), but because it greatly reduces distortion at normal listening levels.

Grey

AuroraB 4th October 2005 07:33 AM

As you probably know, this also largely depends on which driver you choose....the average HiFi bass driver of today lists an effeiciency or sensitivity of 85-90 dB SPL, whereas a lot of the pro drivers ( JBL, Pioneer, etc) will give you typically 96-100 dB for 1W/1m.... something to think about. ( listed levels not valid at sub frequencies , of course, -- just a guide line..)

I am personally fumbling around with parts and PCBs for several of the published Pass products, and I have the utmost respect for NPs work and maybe in particular his commitment to the DIY crowd (all my hats off!-- I actuallly have three!) .. NPs "open door" policy towards the DIY community is quite unrivalled on the High End designer side...

However, if your target is a subwooofer amp, the PassLabs forum may not be the right place to be, depending on which drivers you choose. Most amps in this section, except for the X-ed versions, have a fairly low efficiency, even if they are excellent amps.
A 2 x38V trafo will give you a standard class B amp of appx 200W +, which should be more than enough for a decent size driver and sub cabinet-- take note of the word "decent", - 'cause if you're trying to squeeze 110 dB at 20 Hz out of a shoe box, you will more than likely have to add an extra 0 behing the 8 on that trafo rating.
Use, say a 15" pro driver, a decent cab' - and you're back in NP land ;)

hilbert 4th October 2005 12:59 PM

Thank you all for the sharing of knowledge!

Luckily I have access to a lot of mesurement equipment, so I will do as suggested and measure the properties of the woofer. I guess I can hook up the 1.4 kW amp to investigate impedance variations, especially with rude doings like stomping a kick drum. Isn't there a recording of 1812 with real cannons...? Then I'll know what power I need.

Have I understood it correctly that by using class A design, you avoid distortion due to the slight difference between the amplified positive and negative signal halves? Then I guess, as you say, I might as well go for a "AB" design and a low pass filter, because the distorsion will be of higher frequency than the original signal? But a push-pull design need a matched pair of MOS-FETs, doesn't it? By just browsing my local online electronics store, I find that there's plenty of N-channel devices, but very few P-channel ones, however they do have both the IRFP240 and the IRFP9240. Could I then use a whole bunch of these devices in parallell, like in the Alephs, to get enough output power?

AuroraB 4th October 2005 01:16 PM

If you plan to use the amp as a true sub-amp, there should be no need to sacrifice the efficiency of a straight class B for running AB or true A, as the distortion of the loudspeaker itself will most likely be a lot higher then the amp.
If by push-pull amp, you mean a bridged amp, you actually have a two channel amp driven by a balancing network to produce the push-pull action. In this case , and at these power levels, you will probably need several matched pairs, at least two, for each half of the bridge...

The IRFP pairs are good for PassLab clones, but I think your "local" supplier ( Elfa-?? ) also stocks the Toshiba 2SK1530/2SJ201 VFET pair, which is good for other designs. However, there is a rather hefty markup on the price of both types, - to my opinion.

hilbert 4th October 2005 09:27 PM

Oh, yes, Elfa it is, and the price is accordingly high.
I guess I could find another supplier almost anywhere in the world that would be a lot cheaper, including p&p to Sweden.
By the way, I long ago bought some parts from a british supplier that specializes in out-of-production electronics parts, so if anyone really needs to find some of the original Zen or Aleph parts that perhaps are otherwise hard to come ny, the place to go might be: http://www.dialelec.com

Well, in the spirit of Mr. Pass. I'll use whatever I can get my hands on.

GRollins 5th October 2005 12:25 AM

It's not the Thiele-Small specifications of the driver that are in question, it's the performance of the speaker as a finished system. All the T-S specs in the world will not tell you that the cone is breaking up. All the simulation programs do is present an idealized graph based on the T-S numbers fed into the formulas. They say nothing about such real world phenomena as non-linear suspensions or loss of flux due to heating of the magnet structure.

Grey

PRR 8th October 2005 07:04 AM

> There are more aspects to this than a leopard has spots

Indeed.

> I know a bit about complex impedance and such.

Forget the electrical impedances. Assuming no stupid tricks in a passive crossover, a woofer looks like a 6Ω resistor in series with the reflected mechanical load. The real impedance magnitude will never be less than 6Ω. In typical subwoofer duty, the mechanical resonance will raise the impedance, but you ignore that because most analysis assumes constant-voltage drive, and there will be some significant range where impedance is not much higher than 6Ω. You may be safe calling it ">8Ω"; that's what we usually do. FWIW, the maximum impedance is rarely over 50 ohms, and then only at one or two narrow bands; however impedance may be over 10 or 16 ohms over most of a subwoofer's working range.

If you build an effective (big!) horn, the transformed acoustic resistance will swamp much of the mechanical reactance. If cone and throat areas are similar, acoustic resistance reflects back to the electrical side as something like the electrical resistance: the 6Ω coil looks like 12Ω as long as the horn loads the cone fully (and up to the mass breakpoint). This will also happen in large arrays: put your 4x4 15" array in air or a vacuum (or connect every other driver out-of-phase), the impedance in the 40Hz-150Hz range will double when air loads the array.

> Low frequency wattage requirements are absolutely huge compared to midrange and high frequencies.

NO! This only seems true because we (or our housemates) insist on using ink-dinky toy speakers. 50Hz is a 20 foot wave. Poking at the air with a 1-foot disk will NOT make much 20-foot slosh. Inefficiency is certain. Efficiency will be roughly the square of (cone-size/half-wavelength). One foot cone at 50Hz is about (1/10)^2 or 1%. (I am ignoring the slight narrow gain possible with bass-reflex: a Fifteen in a vented box can do 2% at 50Hz.)

If we stop fooling around, bass is easier than higher frequencies. Mass is a non-issue below 50-200Hz; coil mass kills any attempt to make middles or trebles. Take Hilbert's 4x4 15" array and put it in a doctoral student's apartment. At 50Hz, (5/10)^2= 25% efficient. Since hi-fi acoustic power is rarely over one Watt (maybe less in a small apartment with neighbors), 4 electrical watts should keep up with any mid-high system. At 20Hz, (5/25)^2= 4% efficient. Now we might need 25 electrical watts.

OTOH, a ten-inch trying to make 20Hz is (0.5/25)^2= 0.04% efficient. To make one acoustic watt reference, we need 2,500 electrical watts.

A Ten realistically doing one acoustic watt to 50Hz needs 400 electrical watts.

We have a choice: sixteen 15" speakers and 4 watts, or one Ten and 400 watts. Today, the cost of electronic amplification is much lower than the cost and space of large driver arrays. Even I have switched to a big amp and little woofers. But if you change your spots, big amps are quite unnecessary (though a larger apartment may be needed).

> depends on which driver you choose....

Only as a second-order. Driver SIZE determines low frequency efficiency. When you have enough size, other specs say how well the specific driver fits that theoretical limit.

> for an ordinary loudspaker, isn't there a risk that something goes up in flames?

The old E-V 30-inch woofer with styrofoam cone could catch fire. Paper is tougher to light. Used to be that the coil glue would melt, the coil winding shot off both ends like an explosion in a Slinky factory. Glue got better so then coils expanded from heat until they rubbed the pole pieces. One company puts Teflon on the poles so it can rub without sounding awful. (Heat dissipation improves markedly when rubbing starts, so this condition is semi self-limiting.)

The real reason we can run near-KiloWatt amps is that the peak/average ratio of most music is over 10dB. 400 Watts on peaks is 40 Watts average. The thermal time constant of a large voice coil is many seconds, enough to integrate the boom-da-boom. 40 Watts is not a large power in a 1.5" large-magnet coil, with modern glues.

Of course a single transient can over-travel the speaker. I've ruined my share of woofers that way. But modern heavy-duty woofers are very cleverly designed. The apex of mega-woofer tricks is enough suspension travel to let the coil mostly leave the gap (force goes to zero) and suspension strong enough to halt the cone inertia and bounce it back toward the working zone. They will survive really insane amounts of power more or less gracefully and with a low death rate.

> Have I understood it correctly that by using class A design, you avoid distortion due to the slight difference between the amplified positive and negative signal halves?

No. A push-pull AB, B, or even a Class C amp can have equal (and equally distorted) positive and negative signal halves. That isn't a real problem, nor an annoying one, and Class A isn't necessarily better for that.

In the ideal Class B amp, at low signal level the device current is low, tending to zero. All devices (tubes, BJT, FET) give zero gain at zero current, and low gain at low current. So gain is small for soft sounds, normal for louder sounds. A Class B amp typically sounds OK when played slightly below clipping, but gets raspy at low levels.

The fix is to run a small current at idle, and use an amplifier topology where the gain equation looks, not like Rl/Re, but Rl/(Rl+Re). Re is the effective resistance of the device, and if we make Re<<Rl, the gain approaches unity, with very small shift for large change in Re. Example: a BJT at 100mA has Re about 0.3 ohms. 8/(8+0.3)= 0.96. If peak current is 3A, Re drops to 0.01 ohms. 8/(8+0.01)= 0.999. 0.999/0.96 is only a 4% shift of gain for a 30:1 shift in device gain and Re. This gets smaller in push-pull, smaller yet if we add resistors to limit the drop of Re at high current, and when you wrap a little NFB around it gets good enough for almost any audio purpose. (The hidden flaw is that when one side rises from 100mA to 3A, the other side drops from 100mA to 0.00000A, and when we come back near zero output current the hard-off side has to be dragged or kicked back into life. This is swamped by the still-on side, but sophisticated AB audio amps often try to hold the "off" side "slightly on" at all times.)

The Class A amp idles at half the peak current. Change of Re is small and cancels, though added resistance does not help. Neither device ever turns off (if it does, it is a hot AB, not a true A). And Class A can work with just one device (B and AB audio amps must use two devices).

On paper, an optimized AB amp shows lower THD than an A amp. But the AB will throw higher harmonics than an A amp, and probably more IMD, so A has virtue even though it is very hot/costly.

For a subwoofer: Fletcher-Munson prove that you won't hear soft bass, so low-level distortion per se is not at all an issue. An under-biased "AB" BJT amp may however throw 9th and 11th harmonics of inaudible bass decay, putting trash up in the midrange where you WILL hear it. A subwoofer amp must not be nasty, but does not have to be near as good as your mid/high amps. Class AB showing a trace of heat at idle is plenty good.


All times are GMT. The time now is 12:14 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2