I have an old thread about musical spectrum here:
Spectrum of Musical Genres
While it doesn't tell you the power requirements for each band, it does show average levels at different frequencies. Maybe you'll get some help from that. Just remember that you have to cover the peak value in each band if you don't want to clip.
Spectrum of Musical Genres
While it doesn't tell you the power requirements for each band, it does show average levels at different frequencies. Maybe you'll get some help from that. Just remember that you have to cover the peak value in each band if you don't want to clip.
Not sure what you mean. My thinking: One 2A3 for tweeters, one 2A3 for midrange, one class-D for midbass, plate amps for powered subs.
High output impedance is typical for SETs, yet I understand likely a non-issue with tweeters and probably a minor thing with midrange.
Thank you Pano.
Interesting read. You have me thinking.
To simplify, let me take the red "rock" line in your first chart, and assume: subs doing 20-60Hz, midbass 60-300Hz, mids 300-2kHz, treble above 2kHz.
Since the red line is generally decreasing I take the max SPL for each of the 4 ways:
Subs: -36dB peak
Midbass: -36dB peak
Midrange: -42dB peak
Treble: -50dB peak or so
What does this mean from a power requirements across the spectrum? Here's a thought, for your feedback: subs and midbass peaks would become my reference. If my RMS C-weighted SPL is 95dB, should I add additional 15dB SPL for peaks? So midbass and subs peaks would be 110dB peaks?
So 110dB is the reference, and the midrange peaks are 6dB SPL lower than the reference, so 104dB, and treble is another 8dB below, so 96dB.
Continuing with the line of thinking that a 2A3 is used for the midrange and another for the tweeter, and the tweeter has higher sensitivity than the midrange, then the midrange amp-sensitivity becomes the bottleneck. If I wanted to run the 2A3 to up to 2W to keep distortion down during peaks, I would need at least 100dB sensitivity in the midrange (to get to 103dB SPL peak). Hence one 8PE21 wouldn't suffice and reinforces the attractiveness of double 8PE21.
What do you think of the rationale?
I think it's a pain in the *** to try to fit a square peg in a round hole.
A bit more: Music and the Human Ear
Last edited:
Just proceed with caution
Most likely, a 2a3 amp will work quite well for the Beyma tweeter. I wouldn't buy more than one amplifier, to start with. You could experiment with using it on the mid-range, after you have established it works well on the tweeter. This, of course, assumes you have a few other amps on hand, such that you can mix and match. An audio buddy of mine has a pair of Bottlehead 2a3 monoblocks. Uses them on small cone speakers, which, admittedly are much less efficient than the 8PE21. But, even at moderate levels, he's running into clipping 50% of the time. He's a swell guy, but he just can't hear clipping.
Yet another buddy of mine uses a 45 amp, directly to his tweeters. A 45 amp won't even hit one watt, even with a tail wind, but his tweeters are 108db/w.
He never clips.
For an experiment in my system, since I was dubious about a 5 watt class A transistor single-end being enough power, he brought over a LED power meter. We hooked it up tot he output of said amp, driving my TB1772 loaded into he big red horns. (an HONEST 99db/w). We did hit the 2 watt light once, but bout near drove us out of the room; it was so loud !
Thus, you just have to try, and see (that means listen).
Best wishes. Over & out.
Most likely, a 2a3 amp will work quite well for the Beyma tweeter. I wouldn't buy more than one amplifier, to start with. You could experiment with using it on the mid-range, after you have established it works well on the tweeter. This, of course, assumes you have a few other amps on hand, such that you can mix and match. An audio buddy of mine has a pair of Bottlehead 2a3 monoblocks. Uses them on small cone speakers, which, admittedly are much less efficient than the 8PE21. But, even at moderate levels, he's running into clipping 50% of the time. He's a swell guy, but he just can't hear clipping.
Yet another buddy of mine uses a 45 amp, directly to his tweeters. A 45 amp won't even hit one watt, even with a tail wind, but his tweeters are 108db/w.
He never clips.
For an experiment in my system, since I was dubious about a 5 watt class A transistor single-end being enough power, he brought over a LED power meter. We hooked it up tot he output of said amp, driving my TB1772 loaded into he big red horns. (an HONEST 99db/w). We did hit the 2 watt light once, but bout near drove us out of the room; it was so loud !
Thus, you just have to try, and see (that means listen).
Best wishes. Over & out.
Somewhere in that thread I posted peak values, those are what are important to avoid clipping.
I agree with what was said above. A 2A3 on a tweeter/horn that is >96dB is going to be plenty. You can figure midrange from there looking at peaks. Bass generally needs more power because of lower sensitivity and high peaks. But midrange can be just as hot as the bass. Big peaks are frequent in the midrange.
I agree with what was said above. A 2A3 on a tweeter/horn that is >96dB is going to be plenty. You can figure midrange from there looking at peaks. Bass generally needs more power because of lower sensitivity and high peaks. But midrange can be just as hot as the bass. Big peaks are frequent in the midrange.
I was looking at the graph on post #23 with red line showing average and yellow showing peak, for rock. Here.
From my post #44 above:
Subs: -36dB peak
Midbass: -36dB peak
Midrange: -42dB peak
Treble: -50dB peak or so
From 250 to 1000Hz I see yellow peaks at 400 and 500Hz at -42dB, so I'm taking those as reference peaks for midrange.
I was probably making 2 mistakes in post #44, here for feedback:
- I was treating these -42dB as RMS and then adding 15dB headroom for peaks. But these were already peak dB. So I shouldn't add 15dB for headroom.
- I'm realizing I really don't know what -42dB, -36dB and other "-" dB levels in that graph mean. I was assuming they are relative dB SPL vs a reference set to 0. But how should I translate from these values to my listening at 90dB RMS SPL C-weighted? Could you please explain?
I apologize that the graph can be a little difficult to interpret. It's meant to show an overall trend or tonal balance. Think of the levels as relative to each other.
What I see in the peak graph is from 32 to about 500 peaks are at the same level. From 500 to 2K they decrease a little, then start rolling off, being maybe 40dB down at 20kHz.
I'd say don't sweat the details when choosing the amps. Any amp that plays from 20Hz-2kHz is going need the same peak voltage. Above 2kHz peak levels will decrease a good bit. Of course you are taking into account the sensitivity of your drivers in each range so that you can hit the same peak SPL. You have already figured that out.
Average power in those amps will be different, but the peaks aren't all that different from 20Hz-2kHz. A super tweeter won't see much level at all. I once had a Fane bullet tweeter that played only above 7kHz and it didn't take much to get it where it needed to be.
What I see in the peak graph is from 32 to about 500 peaks are at the same level. From 500 to 2K they decrease a little, then start rolling off, being maybe 40dB down at 20kHz.
I'd say don't sweat the details when choosing the amps. Any amp that plays from 20Hz-2kHz is going need the same peak voltage. Above 2kHz peak levels will decrease a good bit. Of course you are taking into account the sensitivity of your drivers in each range so that you can hit the same peak SPL. You have already figured that out.
Average power in those amps will be different, but the peaks aren't all that different from 20Hz-2kHz. A super tweeter won't see much level at all. I once had a Fane bullet tweeter that played only above 7kHz and it didn't take much to get it where it needed to be.
When I put an amp in my system I watch the speaker input on the screen for a while, a day or two of challenging tracks. The needs are then pretty obvious.
I built a 6V6 SET for my waveguided compression tweeters, with worst case scenario: the option to cross as low as 200Hz to a 98dB sensitive midrange in quarter space at the most. If I remember correctly It was well under 50mW most of the time in the original configuration and I was going for at least 300mW (it probably does a little more than that but not heavy quiescent conditions) I wanted a useable operating point at least. I suspect a 2A3 has 10dB on what I was aiming at. I was going to go 45 but I never really saw the need. I use it crossed at between 550-650 at the moment. I've never heard it clip in several years of use.
Did you read the "Peak Power" table in Ludwig's page I linked to in post #45? It's an easier way to understand (obviously the dataset is a little different), as long as you can work around the fixed crossover frequencies (300 Hz and 3000 Hz).
Last edited:
I hadn't looked into it, but did so last night and found it super-interesting. The whole thing, not just the power % in frquency ranges!
The peak power table is quite telling indeed. My listening habits are closer to Talking Heads and Diana Krall (and maybe more bass heavy material), and the xo slopes I envision are closer to 4th than 1st order, so this provides a good, albeit wide starting point: 35-89% of peak energy in the midrange 300-3000Hz!! So reinforces the importance to focus in the midrange driver-amp as my envisioned xo points are around 300 and 1600Hz.
God food for thought. Thank you!!
Here's maybe another way of thinking about power needs...
for a piston source, displacement needed for equal spl increases 4x per octave decrease...
i'm not good at physics, but i think that translates into a 4x power per octave decrease...assuming equal efficiency....someone pls feel free to correct me...
So, ignoring the efficiency of whatever driver sections we are using to split up the spectrum, and ignoring the distribution of energy in the forms of music we like, it seems to me power needs go up 4x per octave decrease.
Which means the bottom end of the spectrum will always need the most power.
When i add my personal preference for a downward sloping frequency response; and enjoying rock, hip-hop, edm, etc; and wanting to reach as low in freq as possible.....
well, i keep finding the subwoofer power needs dominate...oh so greatly....as in way, way above mid-range needs.
for a piston source, displacement needed for equal spl increases 4x per octave decrease...
i'm not good at physics, but i think that translates into a 4x power per octave decrease...assuming equal efficiency....someone pls feel free to correct me...
So, ignoring the efficiency of whatever driver sections we are using to split up the spectrum, and ignoring the distribution of energy in the forms of music we like, it seems to me power needs go up 4x per octave decrease.
Which means the bottom end of the spectrum will always need the most power.
When i add my personal preference for a downward sloping frequency response; and enjoying rock, hip-hop, edm, etc; and wanting to reach as low in freq as possible.....
well, i keep finding the subwoofer power needs dominate...oh so greatly....as in way, way above mid-range needs.
Equal efficiency, equal power. The question is, where does the power need to go when actual music is playing?
I guess another question gets begged...
What kind of actual music?
Pano's spectral breakdowns were really cool. It's clear i think that hip-hop or EDM are going to need more bottom end power than other types.
Imo though, in thinking about power required across the spectrum...it's less about the type of music and its spectral content, and more about the ability to faithfully play whatever signal we put to the system.
Here's some power observations that kinda cut through the analytical content chase ime...
I use 4-way active designs that have an amp for each section.
Typical xover points at 100Hz, 600 Hz, & 6.3kHz, divide the spectrum from bottom to top into roughly: 2.5 octaves (sub), 2.5 octaves (mid), 3 octaves (HF) , and 2 octaves (VHF).
Each amp has continuous readout meters for voltage, current, and power; all switchable between rms and peak.
If you watch the meters on about any type music, and especially music with strong bass content,
it's clear the sub amp produces far more power than the mid,
the mid plenty more than HF,
and HF quite a bit more than VHF.
In fact, VHF gets almost no power relative to the sub, even after adjusting for the sensitivity difference.
As of course, sensitivities aren't equal. At 2.83v to all, sub is around 101dB, mid 102, hf 110, and vhf 108.
But i think these relative sensitivities probably apply to most speakers in general, than not. Which makes the broad meter observations fairly valid for other speaker setups through the spectrum.
So in the original thread opening post, where a question was posed if the following makes sense: "the amount of power needed for 320 to about 2kHz is the same as from 2kHz to 20kHz"
I'd say 320 to 2k will use quite a bit more power, despite spanning a little less octave width. ........ fwiw
Mark,
In the context of my system, you are arriving at the same answer: the mids driver-amp sensitivity-power relationship will be my bottleneck. I also have 4-way active with powered subs, 400W for 2x10" midbass, looking into 2A3 for mids and another for tweeter. The 400W would drive me out of the room with those drivers. The 102 dB sensitivity tweeters have headroom with 2A3. The midrange does not. Hence the need to look carefully into this link, and thus my other thread on MTM with dual high efficiency 8" mids.
This is helpful for focus/prioritizing.
In the context of my system, you are arriving at the same answer: the mids driver-amp sensitivity-power relationship will be my bottleneck. I also have 4-way active with powered subs, 400W for 2x10" midbass, looking into 2A3 for mids and another for tweeter. The 400W would drive me out of the room with those drivers. The 102 dB sensitivity tweeters have headroom with 2A3. The midrange does not. Hence the need to look carefully into this link, and thus my other thread on MTM with dual high efficiency 8" mids.
This is helpful for focus/prioritizing.
If the given frequency is dropped down by an octave, the required cone excursion increases by a factor of 4. If the cone area is doubled, the excursion is cut in half. Thus, if we do both, the excursion for each driver now only doubles. These are design considerations only.
The amount of acoustic energy, as in density, is greatest in the octave 250-500. This is not the same as the consideration for power requirements. To reproduce acoustic density requires a different set of considerations than that of simply "just power". Since this thread is actually just about POWER, we need to differentiate between SPL sensitivity and that of efficiency. For simplification here, we are only going to discuss efficiency considerations for 100Hz and lower. More on that in a moment.
In discussing the 8PE21 operating as low as 300Hz, we observe the X-max of 1 mm. This limits the max SPL to 114db. The SPL claim on said driver is 98db/w. The difference between 114 and 98=16. The power ratio of 16db = 39.8 That's the most amount of power needed for said driver, no matter how high the power rating. 114db is VERY VERY loud and in hearing damage territory. (This assumes a one meter distance from a speaker). Doubling the distance drops the spl by -6db.
There's also a difference between frequency-dependent potential maximum power needed and that of any given program material. This is where the density comes into play, AND said density is all different for each type of music. The 250-500Hz "rule" is that for orchestral music. Your rock bands require a different set of rules.
An illustration is offered here for efficiency and operation of 100Hz and down.
A one cubic foot box can only be as much as 5% efficient at 95Hz. This is in consideration of system efficiency and is the relationship of the given driver and required box volume. The driver requires the box volume, but you can not increase/decrease the box volume in hopes of gaining efficiency, based on the given driver.
In order to maintain a reference level of 5% efficiency at 40Hz, our box size now becomes 16 cubic feet. Good luck with that. This is why most bass systems are of lessor efficiency. So it will fit through the door.
2a3 amp on the tweeter is fine. Not much energy up there. I measured the power consumption on mine at VERY loud levels and it was 320 milliwatts.
2a3 amp will probably not run an 8inch cone to full dynamic potential.
The amp seller will tell you that it will work fine, because he wants to sell you the amp. I don't have an agenda to sell anything. I'm only here because I love audio and am willing to put the time in to share my knowledge and experiences.
From my perspective it's less about power than about headroom. You want to be able to play at your chosen volume without clipping.. What are the peak voltages you need for each diver to play at that volume without clipping?
Power is of secondary importance, you might need to know it for sizing the PSU and the heatsinks.
Power is of secondary importance, you might need to know it for sizing the PSU and the heatsinks.
I think that too Pano, that peak voltage / headroom is typically what's the limiting factor in faithful, unclipped reproduction. (or even live music).
All the meters I watch are nearly always set to read peak.
That said, power and peak voltage capability are inextricably linked I think, mainly due to marketplace convention.
Over the years I've often heard the expression "all we really need is a 100 watt amp, with 20dB of honest headroom."
I guess FTC regulations, stipulating continuous RMS ratings, pushed competition to achieve the highest power rating possible, kinda orphaning headroom.
I see most proaudio amps typically have no more than 3dB headroom, and who knows how long a burst cycle that 3dB is even good for.
Here, maybe excellent audio amps beat out the pro world...dunno..
all I know is I don't expect to find an amp where the voltage implied from its continuous rating and it's true peak voltage capability are very far apart.
Which means imo, we almost have to look at power as a surrogate for deteriming peak voltage capability.
I mean, whose putting out specs that show either burst power vs time, or better, burst voltage vs time??
Anyway, as your voltage test shows, most typical listening is likely to be headroom constrained before power constrained, simply cause most listening isn't done that loud.
The one thing i think matters least in all this is the spectral composition of the type music, or the spectral density per octave region.
Personally, I want a system that simply stays linear with whatever music i throw at it, at whatever volume I choose.
Like said earlier, I find that it takes increased power, or increased peak voltage, however you want to look at it, as frequency decreases.
Vastly so, ime looking at muti-ways meters
And theoretically:
Setting aside efficiency, i read pistonic displacement (Sd x excursion) has to increase 4x per octave decrease for equal spl across the spectrum.
Am i missing something, or isn't that an astounding 10^4th, 10,000 to 1 ratio, over a ten octave span ?
And doesn't that displacement difference imply a direct power difference?
All the meters I watch are nearly always set to read peak.
That said, power and peak voltage capability are inextricably linked I think, mainly due to marketplace convention.
Over the years I've often heard the expression "all we really need is a 100 watt amp, with 20dB of honest headroom."
I guess FTC regulations, stipulating continuous RMS ratings, pushed competition to achieve the highest power rating possible, kinda orphaning headroom.
I see most proaudio amps typically have no more than 3dB headroom, and who knows how long a burst cycle that 3dB is even good for.
Here, maybe excellent audio amps beat out the pro world...dunno..
all I know is I don't expect to find an amp where the voltage implied from its continuous rating and it's true peak voltage capability are very far apart.
Which means imo, we almost have to look at power as a surrogate for deteriming peak voltage capability.
I mean, whose putting out specs that show either burst power vs time, or better, burst voltage vs time??
Anyway, as your voltage test shows, most typical listening is likely to be headroom constrained before power constrained, simply cause most listening isn't done that loud.
The one thing i think matters least in all this is the spectral composition of the type music, or the spectral density per octave region.
Personally, I want a system that simply stays linear with whatever music i throw at it, at whatever volume I choose.
Like said earlier, I find that it takes increased power, or increased peak voltage, however you want to look at it, as frequency decreases.
Vastly so, ime looking at muti-ways meters
And theoretically:
Setting aside efficiency, i read pistonic displacement (Sd x excursion) has to increase 4x per octave decrease for equal spl across the spectrum.
Am i missing something, or isn't that an astounding 10^4th, 10,000 to 1 ratio, over a ten octave span ?
And doesn't that displacement difference imply a direct power difference?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.