Hi there, Happy New Year, and thanks for a great thread.
I sit only 5 feet (1.5m) away from my speakers. They are Snell Type J/III, sensitivity rated at 91dB/1w/1m (anechoic).
I burned a CD of the test tones, and played the first 120 Hz track.
At a volume setting that would usually disturb my neighbors*, I measured 2V AC on my old Micronta FET input analog multimeter (AC = 10k ohm/V). Is my meter reading peak volts or rms volts? I assume peak, but I dunno...
To figure the equivalent power output in watts, it's Vsquared/R. So I multiply 2 * 0.707 to get the rms voltage, which is 1.414. 1.414 squared = 1.9994. Then divide by 4 ohms (approx. DCR of woofer).
Let's call it 2V rms / 4 ohms = 500mW
For 22dB headroom on top of that, multiply by 12.5, so 6 watts required per side, right?
Well, my trusty push-pull 2A3 amp delivers 6 watts rms per side. I guess I have what I need, just barely.
I've also been mucking around with a pair of Klipsch RF-3, which are rated at 95dB/1w/1m (anechoic). I'll see if I can't test with those too.
BTW, Klipsch owners... Klipsch rates their speaker sensitivity as an "in-room" spec. I read that one should subtract 3dB from their rating to get the "anechoic" sensitivity rating. Klipsch quotes the RF-3 sensitivity at 98dB, so that's how I got 95dB.
--
* - I live in a condo; it's way too easy to disturb my neighbors.
I sit only 5 feet (1.5m) away from my speakers. They are Snell Type J/III, sensitivity rated at 91dB/1w/1m (anechoic).
I burned a CD of the test tones, and played the first 120 Hz track.
At a volume setting that would usually disturb my neighbors*, I measured 2V AC on my old Micronta FET input analog multimeter (AC = 10k ohm/V). Is my meter reading peak volts or rms volts? I assume peak, but I dunno...
To figure the equivalent power output in watts, it's Vsquared/R. So I multiply 2 * 0.707 to get the rms voltage, which is 1.414. 1.414 squared = 1.9994. Then divide by 4 ohms (approx. DCR of woofer).
Let's call it 2V rms / 4 ohms = 500mW
For 22dB headroom on top of that, multiply by 12.5, so 6 watts required per side, right?
Well, my trusty push-pull 2A3 amp delivers 6 watts rms per side. I guess I have what I need, just barely.
I've also been mucking around with a pair of Klipsch RF-3, which are rated at 95dB/1w/1m (anechoic). I'll see if I can't test with those too.
BTW, Klipsch owners... Klipsch rates their speaker sensitivity as an "in-room" spec. I read that one should subtract 3dB from their rating to get the "anechoic" sensitivity rating. Klipsch quotes the RF-3 sensitivity at 98dB, so that's how I got 95dB.
--
* - I live in a condo; it's way too easy to disturb my neighbors.
Last edited:
Normal multimeters whether analogue or digital always read (or should I say are calibrated) to give an RMS value.
Try this to get a figure for impedance (post #16) ...
http://www.diyaudio.com/forums/multi-way/203497-4-8-ohms-what-does-mean.html#post2841318
Nope.
For every +10dB, you multiply the power requirement by 10. So here you'll need 150-200x the power.
I suspect there's something amiss here - how reliable is your meter on music?
I don't see how any hard and fast rule can be drawn about sensitivity of speakers in-room vs anechoic.
Sensitivity is a function of frequency response, and this can vary so heavily room-to-room that saying take away X decibels for all rooms can't possibly be right.
Chris, who is slowly getting around to measuring his system.
Hmmm...
OK, let me try again. On a pair of Snell Type J/III, with the 120Hz test signal playing through the CD and amps, I measured 2Vrms AC across the speaker terminals. 6 ohm *nominal* impedance. Did that following Pano's method, as best I could.
What power output does that equate to?
To find the actual impedance at 120Hz, I could...
____________________________
Re: Klipsch sensitivity ratings -- I got that from some advertising of theirs, which I can't find right now. It said they measured their sensitivity in a way that more closely matched typical room usage (or something like that). You can ignore that if you like. I could be totally wrong.
Last edited:
OK, I guess I used the woofer's DCR for the "R" part of the equation. I gather that is not useful.
It looks like 666mW is the answer, at least at midband.
So, does that mean that in my particular situation, with my near-field listening position, that 9.1 times that measured power would be adequate for handling musical peaks at that volume setting?
I find that my available 6 watts per channel is right on the edge of adequate. I have a 10 watts per channel PP EL34 triode amp that can play just a tiny bit louder, but I don't like its sound as much as the PP 2A3's.
Since the woofer is ported, I figure the impedance goes up at the two peaks just under and just over box resonance. Maybe 20 ohms? So, 2*2 = 4, divided by 20 ohms = 200mW at those frequencies. I figure the impedance would go up around the crossover region too. I believe that's 2.3kHz in this system...
--
The speaker is voiced by the designer to take account of these impedance variations when driven by typical amplifiers that met the designers criteria.
You can ignore the power consumed/demanded at the various peaks and troughs, the frequency response rules as far as your ears hear the output.
But the amplifier does not ignore these impedance changes.
It must be able to drive a resistor, where the phase angle is zero. It must be able to drive a series pair R+C where the slope is falling and the phase angle is not zero and it must be able to drive a series pair of R+L where the slope is rising and the phase angle is not zero. The further the phase angle from zero the more the amplifier output current is out of phase with the output voltage.
Further the amplifier must remain comparatively unaffected by the back emf fed into the NFB loop from the reactance of the speaker.
What was 6ohm? was it 6r0 for DCR, or was it an average impedance for pink noise over the whole audio spectrum, or was it an arithmetic average of the 4ohms and 8ohms, or was it some other impedance?
You can ignore the power consumed/demanded at the various peaks and troughs, the frequency response rules as far as your ears hear the output.
But the amplifier does not ignore these impedance changes.
It must be able to drive a resistor, where the phase angle is zero. It must be able to drive a series pair R+C where the slope is falling and the phase angle is not zero and it must be able to drive a series pair of R+L where the slope is rising and the phase angle is not zero. The further the phase angle from zero the more the amplifier output current is out of phase with the output voltage.
Further the amplifier must remain comparatively unaffected by the back emf fed into the NFB loop from the reactance of the speaker.
What was 6ohm? was it 6r0 for DCR, or was it an average impedance for pink noise over the whole audio spectrum, or was it an arithmetic average of the 4ohms and 8ohms, or was it some other impedance?
Okay, here we go.
Details of my system as it's set up are in my signature.
I ran Money as loud as I'd want to listen to it, then did the 120Hz test, using a Fluke meter (ought to be reliable). Came out at 1.3v for the Fostexes, 0.1v for the subwoofers - the crossover isn't far from 120Hz, so thought I might as well measure it. Turns out negligable.
Taking 1.3v RMS,
1.3^2=1.69
1.69/8=0.21w average power to the Fostexes.
Now, as many of you will know, the FE126s (and others) have a prominent upper midrange: sensitivity ranges from low 90s in the lower mids, peaking at 100dB at 7kHz.
I have installed a notch filter to tame this peak by ~4dB, leaving a sensitivity in the mid-90s.
Listening distance of maybe 2m, calculated SPL (taking 95dB@1w) = about 88dB.
Peaks of 102dB, 4w per side used out of maybe 15w.
Subwoofers aren't even trying.
Chris
Details of my system as it's set up are in my signature.
I ran Money as loud as I'd want to listen to it, then did the 120Hz test, using a Fluke meter (ought to be reliable). Came out at 1.3v for the Fostexes, 0.1v for the subwoofers - the crossover isn't far from 120Hz, so thought I might as well measure it. Turns out negligable.
Taking 1.3v RMS,
1.3^2=1.69
1.69/8=0.21w average power to the Fostexes.
Now, as many of you will know, the FE126s (and others) have a prominent upper midrange: sensitivity ranges from low 90s in the lower mids, peaking at 100dB at 7kHz.
I have installed a notch filter to tame this peak by ~4dB, leaving a sensitivity in the mid-90s.
Listening distance of maybe 2m, calculated SPL (taking 95dB@1w) = about 88dB.
Peaks of 102dB, 4w per side used out of maybe 15w.
Subwoofers aren't even trying.
Chris
I only ever tested the amplifier into a resistor dummy load.
A series pair R+C - would that be 8 ohm dummy load in series with a capacitor? How to choose the value of C?
Series pair R+L is dummy load resistor in series with inductor? How to choose the value of L?
I guess I could look that up, search on these forums maybe...
The amp does not have a global NFB loop.
I was splitting the difference between the nominal 8 ohm impedance quoted by the mfg and the 4 ohm DCR of the speaker as measured across its terminals.
-
Hi,
the test signal was from memory recorded at -16dBfs.
The maximum signal from that source will be your measured 0.21W + 16dB or about ~8.4W
0.21W fed to a 95dB/W @ 1m speaker will give about 82dB at 2m.
Add on 16dB for a maximum signal from that source, leads to a maximum SPL of ~98dB.
If two speakers are in use then maximum SPL will be ~ 3dB more than that.
Peak transient SPL could be a further 3dB above that, resulting in an absolute maximum peak SPL of 104dB @ 2m
That is not quiet by any definition of loudness assessment.
An 8W amplifier with your speakers, using that source and that gain structure will be loud enough for most listeners.
the test signal was from memory recorded at -16dBfs.
The maximum signal from that source will be your measured 0.21W + 16dB or about ~8.4W
0.21W fed to a 95dB/W @ 1m speaker will give about 82dB at 2m.
Add on 16dB for a maximum signal from that source, leads to a maximum SPL of ~98dB.
If two speakers are in use then maximum SPL will be ~ 3dB more than that.
Peak transient SPL could be a further 3dB above that, resulting in an absolute maximum peak SPL of 104dB @ 2m
That is not quiet by any definition of loudness assessment.
An 8W amplifier with your speakers, using that source and that gain structure will be loud enough for most listeners.
the measured DCR=4r0 probably indicates a 5ohm bass/mid driver+crossover combination.
The majority of SPL from conventional music will come from that bass/mid driver.
I would assess the speaker impedance for typical music SPL as a 5ohm speaker.
This according to my method of designing for reactive speakers is not as onerous as designing for a 4ohm speaker.
Your 4 to 8ohm speaker impedance stated by the manufacturer in effect is "safe" value to use for amplifier selection, provided we use the method I explained earlier, i.e. select an amplifier for the lower of the two impedances since much of the current demand of the speaker load is determined by the Bass/Mid driver.
Thanks for all this.
The mfg stated impedance is 8 ohms.
So I should use the 4 ohm tap on the amp's OPT secondary?
-
Thanks guys for the tests. Nice to see what you got.
Of course this test measures voltage, not power. We have to calculate the power from voltage and load. For amps i prefer to think in terms of voltage, but that's not how they are spec'd.
From the recent results I'd say that Chris needs about 8V RMS to get the minimum 16dB of headroom from loud to peak. Rongon would need about 13V RMS. That's an approximation, of course, as we don't know the actual levels of the tracks they used for setting "loud" by ear. But it's a good approximation.
You can figure backwards from the amp's rated power how much voltage it can supply.
Of course this test measures voltage, not power. We have to calculate the power from voltage and load. For amps i prefer to think in terms of voltage, but that's not how they are spec'd.
From the recent results I'd say that Chris needs about 8V RMS to get the minimum 16dB of headroom from loud to peak. Rongon would need about 13V RMS. That's an approximation, of course, as we don't know the actual levels of the tracks they used for setting "loud" by ear. But it's a good approximation.
You can figure backwards from the amp's rated power how much voltage it can supply.
Hi Andrew, I think you've missed something - you gain 6dB for the other speaker: +3dB on double cone area, +3dB because you double electrical power.
... and the 120Hz was 14dB down.
Chris
PS - Pano, you're about right: the Tripath amp is running off a 12v SMPS (which happens to run closer to 13v).
Chris, with a Tripath amp running off 13V you can probably hit 11 volts (peak) before clipping. The chips generally get to within about 2 volts of Vcc before clipping. Depends on the chip, but that's close.
My guess would be that you are close to clipping at the loudest peaks, but probably won't notice it in normal use.
My guess would be that you are close to clipping at the loudest peaks, but probably won't notice it in normal use.
"The stereo speakers are too far apart to gain anything of significant in SPL."
Conservation of energy requires that two more or less identical speakers reproducing the same signal and receiving the same amount of power in a typical listening room WILL produce an spl that is a clearly perceptable (about 3db) as being louder than only one speaker because approximately twice as much acoustical energy is being delivered to the room. To convince yourself of this simply have someone turn one of the speakers off while listening to some music.
"you gain 6dB for the other speaker: +3dB on double cone area, +3dB because you double electrical power"
I think this is double counting. There is no inherent advantage with respect to spl gained by doubling the cone area, provided that the cones are too far apart to couple (increasing radiation resistance), cone motion is linear and they are being operated well below thermal compression. If the amount of power delivered to the first speaker were increased by the amount that would have gone to the second speaker, and the speakers were operating well below significant thermal compression, then the spl would be more or less the same as with half the power going to each of two speakers. Two watts into one speaker should provide about the same spl as one watt into each of two (+/- identical) speakers.
Regards,
Bob
Conservation of energy requires that two more or less identical speakers reproducing the same signal and receiving the same amount of power in a typical listening room WILL produce an spl that is a clearly perceptable (about 3db) as being louder than only one speaker because approximately twice as much acoustical energy is being delivered to the room. To convince yourself of this simply have someone turn one of the speakers off while listening to some music.
"you gain 6dB for the other speaker: +3dB on double cone area, +3dB because you double electrical power"
I think this is double counting. There is no inherent advantage with respect to spl gained by doubling the cone area, provided that the cones are too far apart to couple (increasing radiation resistance), cone motion is linear and they are being operated well below thermal compression. If the amount of power delivered to the first speaker were increased by the amount that would have gone to the second speaker, and the speakers were operating well below significant thermal compression, then the spl would be more or less the same as with half the power going to each of two speakers. Two watts into one speaker should provide about the same spl as one watt into each of two (+/- identical) speakers.
Regards,
Bob
Hi Bob,
Your post would imply that placing the two speakers next to each other will result in a 3dB increase in sound level, as they are now acoustically coupled.
People are in bed here, so I'll leave this quick test in your capable hands.
Alternatively, with an SPL meter running, disconnect a speaker and see what the level drop is.
I'd be interested to hear of your results: I'll be trying the above experiments tomorrow.
For now, goodnight
Chris
Your post would imply that placing the two speakers next to each other will result in a 3dB increase in sound level, as they are now acoustically coupled.
People are in bed here, so I'll leave this quick test in your capable hands.
Alternatively, with an SPL meter running, disconnect a speaker and see what the level drop is.
I'd be interested to hear of your results: I'll be trying the above experiments tomorrow.
For now, goodnight
Chris
Bob,
As long as the speakers are within 1/4 wavelength distance between the two there will be a 6 dB increase going from one to two, 3 dB for doubling cone area, 3 dB for doubling the power, exactly as stated.
Room modes and different distance between the two speakers and the listener (or test mic) will result in fluctuations from the ideal, but one can easily see the 6 dB level difference between one and two speakers tested outside, as in the chart below.
Art
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.