Looking good!
Some of them, but probably not all. The easiest way to find out is to add damping material to the box and see which wiggles change.
The xover has some built-in compensation for the upper bass-reflex peak, which also affects the shape of the curve at the impedance minimum between the peaks. I would therefore use the curve measured without the xover to check the tuning. It might be best to listen to different port lengths and do some impedance measurements in order to understand how the tuning correlates to what you hear.
Some of them, but probably not all. The easiest way to find out is to add damping material to the box and see which wiggles change.
The xover has some built-in compensation for the upper bass-reflex peak, which also affects the shape of the curve at the impedance minimum between the peaks. I would therefore use the curve measured without the xover to check the tuning. It might be best to listen to different port lengths and do some impedance measurements in order to understand how the tuning correlates to what you hear.
Looking at your "no crossover' curve again, there are two interesting features.Well finally I was able to tame the measuring device
Now I have this two measurements with and without crossover. I didnt add the stuffing yet. If I understand those little bumps are standing waves? But other than that do these graphs look like they should? There is some difference in tunning frequency between the two should I tune with or without crossover?
thanks
View attachment 846781 View attachment 846782
1. The bass reflex peaks are not very high.
2. There is no steady impedance rise at high frequencies, but rather a broad bump, and then another rise.
Did you maybe connect/include the tweeter in this measurement? It would be better to just measure the woofers alone.
Elsinore db 1 watt @ 1 meter ?
Searched through Joe's site, see him mention easy to drive, and High sensitivity vs. electrical impedance/phase response but nothing that I can see there or here mention a number...
Im very surprised how much sound the Elsinores put out at a very low volume... Back in the day, we had the "loudness" button. Helped with the low end while music was low. No need for that with these speakers....
That said, my uneducated and unconfirmed hunch is efficiency is not always linear among speakers. Where only 1 watt may be measured, a more detailed curve should be somewhere in the ether and this is truly the information I hope to have, but will at least settle for the generic sensitivity rating...
Ultimately, I would like this information to help me in choosing my last amp! (I plan on building myself, but hope to choose a design as amazing as this one so it can be my last)
Searched through Joe's site, see him mention easy to drive, and High sensitivity vs. electrical impedance/phase response but nothing that I can see there or here mention a number...
Im very surprised how much sound the Elsinores put out at a very low volume... Back in the day, we had the "loudness" button. Helped with the low end while music was low. No need for that with these speakers....
That said, my uneducated and unconfirmed hunch is efficiency is not always linear among speakers. Where only 1 watt may be measured, a more detailed curve should be somewhere in the ether and this is truly the information I hope to have, but will at least settle for the generic sensitivity rating...
Ultimately, I would like this information to help me in choosing my last amp! (I plan on building myself, but hope to choose a design as amazing as this one so it can be my last)
Look at the dB scaling of the frequency on the left of the response, that is the voltage sensitivity, 70-80-90 etc.
Edit: That graph is actually 15 degrees off axis, so on axis the gentle slope is not there. A sensitivity rating is subjective, but I would say 92dB-SPL. Please note that speaker manufacturers almost routinely inflate this figure, the graph I presented is based on measurement that has been calibrated, and not inflated.
Last edited:
What I do before I build an amp is to connect an oscilloscope to the output of the amp I'm using. Watch the levels for a while to gauge the power I need.
Forget it. Once you finish building an amp, you'll find this other design which looks so much more interesting... I stopped worrying about getting the "best" amp.
Looking at SPL graphs and SPL numbers is fine, but you always need the drive voltage/power and information about the measurement conditions. For instance, a "92 dBSPL" number alone is meaningless when you're interested in sensitivity. You need something like "92 dBSPL @ 2.83 V(rms) @ 1 m @ 2pi" or "92 dBSPL @ 1W @ 1 m @ 4pi".
I know this feeling well. However I've also started getting quite deliberate with my speaker designs and subsequently have found that the 1W/10W/100W configuration I use has been allowing me the versatility to continue to make new speaker designs for some years now. YMMV
I thought that was obvious when I said it was calibrated. I also said "voltage sensitivity" and it would be referring to the standard yardstick which indeed is 2.83V/1M.
But in fact, the measurement was actually taken at 2M and to adjust for the difference, 5.65V was used (+6dB) and then it shows up the same as per 2.83V/1M, but I wanted to model the speaker at 2 metre and not 1 metre. The reason is obvious, not too many people listen to speakers at 1 metre.
I always do the measurements and computer modeling at 2 metre. To make that possible, it is done outdoors on a non-windy day.
It's not a standard. It's just what some people use. Others will use other drive levels. 1 Watt. 1 Vrms. Whatever. Without specifying the drive level, the SPL information is meaningless.
2pi or 4pi? That makes a huge difference.
I don't want to sound annoyed, but "voltage sensitivity" of 2.83V has been the standard for as long as I remember and I have been designing and researching loudspeakers for fifty years and I also happen to know a fair few rather famous loudspeaker designers and they all understand what the standard is.
Yes, but you don't seem to be aware that it is a combination of Nearfield and Farfield measurements. The Fairfield contains the phase information (phase reference) and the Nearfield measurements (note that it is plural) are always 2Pi. But the Nearfield measurement does not tell you the "voltage sensitivity" - that can only be done Fairfield. This method allows you to get a pseudo-anechoic frequency response - in a real anechoic chamber, you get a 2Pi response (but it would have to be huge). With vented enclosure, you will need to sum the responses of the vent and that of the cone driver, this is then merged with the Fairfield measurement.
I can recommend Joseph D'Appolito "Testing Loudspeakers" as a good read.
PS: Below is a 6th Octave RTA measurement measured at the seating position at several metres, the sub-1KHz shows the effect of room modes, but the sub-100Hz is definitely 4Pi, other than the dip around 60Hz, a room mode too.
Last edited:
As Toole says, measure in the farfieled and adjust the measure to a 1m distance.
2m may not even be in the farfield but unless you have an anechoic chamber or can get the box 10-some meteres in the air, hard to get further out and get reasonable resukts at LF.
dave
Joe, I also don't want to sound annoyed, but... well. I am not just "not aware of the combination of nearfield and farfield", but I am not aware by what you mean by "it". What is "it"?
All I was saying is that if someone asks for the sensitivity of a speaker, it's wrong to respond with just an SPL number without specifying (1) the voltage or power level used during the measurement, (2) at what distance the measurement was taken, and (3) if the SPL value reflects 2pi or 4pi. I believe you as an expert should have no objections to this.
Don't mean to step in the middle of something, but i do think it's fair to simply say "X"dB sensitivity.
1 watt @ 1 meter is fully implied .... it's just been too much of a standard for so very long.
Perhaps space need to be given but that's it imo.
I also measure far field, and simply do the math to bring it back to 1 watt @ 1 meter.
(for me far field begins at 3x longest baffle dimension)
1 watt @ 1 meter is fully implied .... it's just been too much of a standard for so very long.
Perhaps space need to be given but that's it imo.
I also measure far field, and simply do the math to bring it back to 1 watt @ 1 meter.
(for me far field begins at 3x longest baffle dimension)
Thanks for this perfect example! How do you know it's 1 W? Could well be 2.83 V instead.
What's wrong with giving complete, meaningful information?
1 watt / 1 meter is so heavily implied that the voltage used (against given nominal impedance) doesn't need stating.
You know, ......2.83v for 8 ohm, 2 v for 4 ohm, etc.
Personally, I agree with you and your appeal for full specifics.
I literally abhor the 1 watt / 1 meter industry tradition.
I'd prefer a simple 2.83 v / 1m reference, no matter the nominal impedance.
And i also hate just eyeballing a response curve to specify sensitivity.
I like to use 2.83 v pink noise with xovers, EQs, processing, etc, in operation, and measure the SPL.
I use about a one minute time average of both the pink noise rms voltage, and SPL flat, to get real world, no BS, sensitivity.
I've occasionally measured average current at the same time to determine a real world nominal impedance.
I wish it were common practice for specs....
You know, ......2.83v for 8 ohm, 2 v for 4 ohm, etc.
Personally, I agree with you and your appeal for full specifics.
I literally abhor the 1 watt / 1 meter industry tradition.
I'd prefer a simple 2.83 v / 1m reference, no matter the nominal impedance.
And i also hate just eyeballing a response curve to specify sensitivity.
I like to use 2.83 v pink noise with xovers, EQs, processing, etc, in operation, and measure the SPL.
I use about a one minute time average of both the pink noise rms voltage, and SPL flat, to get real world, no BS, sensitivity.
I've occasionally measured average current at the same time to determine a real world nominal impedance.
I wish it were common practice for specs....
Last edited by a moderator:
I always thought I did.
The equivalence between 1 Watt and 2.83V is well understood, but it is predicated on the notion that the load is 8 Ohm and resistive. Only then is it 1 Watt dissipated as heat (or 99% of it as drivers are mostly less than 1% efficient).
Using efficiency as the baseline is difficult and hence sensitivity is used rather than efficiency. Not a perfect solution because the impedance can be very low and hence a lot of heat. It is possible for a speaker to have high voltage sensitivity, but such a low impedance (like an impossible 1 Ohm) and yet be terribly in-efficient. Nothing new about that.
So what we have is really a standard of "voltage sensitivity" based on 2.83V, hence we have a standard that is accepted by virtually everyone. But it is also then well understood that you take a look at the load and make a judgment there. Not ideal, but that is what most do.
So I am a little surprised that there is even any controversy. In fact, in my mind there is no controversy at all, just accepted practice. Not perfect practice, though.
So when I say that the Elsinores are around 92dB sensitivity, I am of course referring to 2.83V relative to 1M, but then the graph shows a fairly flat 6 Ohm and I think most will be pleased by that. Note the current phase angle is flat too, so Equivalent Peak Dissipation Resistance (EPDR) is benign.
I think that "full disclosure" has been satisfied. No point not to, right?
But I don't think too many are using it.
I am not a specs person, I am a measurement guy.
I some aspects we are in furious agreement. Except I don't go tilting at windmills. This is not an area that I want to be a crusader. But I would like to be a crusader on a different topic:
My efforts is more towards the aspect of proportional current causes change is the dB-SPL of the driver. My hypothesis is that we listen to the current of the amplifier and not its voltage (voltage only creates a potential for current) and hence all amplifiers just different types of current delivery systems. I have already shown to a number experts and physicists that an amplifier can produce two 2KHz sine waves simultaneously and 30 degrees apart, you have to ask the question, which sine wave are you listening to? One represents the voltage of the amplifier and the other is the current of the amplifier.
I don't really ever measure SPL sensitivity as a spec. Because the response is not totally flat, and you have different responses off axis, and really power response into the room also comes into it, so I agree with John Atkinson that even coming to -/+1dB accuracy is difficult. So manufacturers tend to go with optimistic figures that I largely ignore. I suspect you do as well.
So I calibrate the microphone, the whole measurement system, with a driver that I have set aside as my standard. It is one of the originally made Vifa XT25 tweeters (made in Denmark, not China).
Take a look at this, my 91dB reference, 2.83V @ 1M, I think you might approve:
Pretty neat, eh?
I hit that 91dB and I am good to go. It normalises all my acoustic measurements. I just measure the response and the load and let the reader interpret them. I believe this is more correct that trying to do pink noise tests to determine SPL sensitivity. I use pink noise for RTA 6th octave in-room measurements at the listening position, like the one I showed a few messages back. They are not calibrated and I don't need them to be. I want to see the behaviour in the room as I move the microphone around.
I think my method is highly correct and an improvement over what many others do. Hope you agree.
At least you can see that I am being highly systematic and also believe I have set a standard for myself.
PS: To answer your question earlier, my calibrated measurements (XT25) are always 2Pi and my RTA are 4Pi.
Last edited: