I actually have one Blumlein recording I did myself with a borrowed AKG 426 B mic. Haven't listened to it for a long time, since the musical content is not all that interesting.... but I do remember that the stereo image was quite good.
I like a 90 degree setup because it provides (of course) a lot more width, almost like surround-sound on good recordings. With the dipoles, depth is not suffering with this very wide setup either.
I like a 90 degree setup because it provides (of course) a lot more width, almost like surround-sound on good recordings. With the dipoles, depth is not suffering with this very wide setup either.
You probably know that Blumlein is meant to be listened to with the speakers at the 90º included angle, but the fact that they are also dipolar creates, to me, a pleasing symmetry in its relationship to the recording configuration, which in my case is a crossed pair of ribbons (Royer SF12 and SF24), almost as close as could be imagined to the inverse of the playback transducers.
I suppose it would be even more true if the speakers were Apogees, but I don't think the difference is significant.
Come to think of it - and it may be coincidental - the most convincing imaging I have ever heard on one of my recordings was with a pair of Martin Logans very carefully set up so as to produce the tightest mono image from mono material. I wonder how much the dipolar character of those speakers contributed to the effect.
The image of the solo piano was utterly detached from the speakers and materialized a few feet behind and between them, with absolutely no suggestion of any sound coming from the speakers themselves.
I suppose it would be even more true if the speakers were Apogees, but I don't think the difference is significant.
Come to think of it - and it may be coincidental - the most convincing imaging I have ever heard on one of my recordings was with a pair of Martin Logans very carefully set up so as to produce the tightest mono image from mono material. I wonder how much the dipolar character of those speakers contributed to the effect.
The image of the solo piano was utterly detached from the speakers and materialized a few feet behind and between them, with absolutely no suggestion of any sound coming from the speakers themselves.
I started to work on Ripol subs and would like to compare with the measurements of yours, Stig Erik.
The problem I have is that whatever I do, I hardly get figures well below 1% (-40dB).
You know I use ARTA not CLIO.
ARTA has a feature to capture THD over input signal for any frequency you like. Simply a sinusoid is fed to the speaker and distortion is then captured. This is repeated several times with increasing level.
Now, when I use that feature, I still get hardly ever below 1% - even at veeeery low SPL levels - which is by no way plausible to me. Basically same behaviour indoor as outdoor by the way.
I have the feeling that those measurements techniques - and math - used here, is not exactly suitable for our purpose - simply as we have the back wave arriving later on - which would impose the urgent need to have a tool available to determine which part of the sinusoid should be windowed.
Does CLIO provide such an option - possibly even with a time plot to check mic signal?
Michael
The problem I have is that whatever I do, I hardly get figures well below 1% (-40dB).
You know I use ARTA not CLIO.
ARTA has a feature to capture THD over input signal for any frequency you like. Simply a sinusoid is fed to the speaker and distortion is then captured. This is repeated several times with increasing level.
Now, when I use that feature, I still get hardly ever below 1% - even at veeeery low SPL levels - which is by no way plausible to me. Basically same behaviour indoor as outdoor by the way.
I have the feeling that those measurements techniques - and math - used here, is not exactly suitable for our purpose - simply as we have the back wave arriving later on - which would impose the urgent need to have a tool available to determine which part of the sinusoid should be windowed.
Does CLIO provide such an option - possibly even with a time plot to check mic signal?
Michael
Last edited:
Very compact ripol sub you have there! Which drivers do you use?
I have a 20+ years old CLIO (DOS version), but it can do gating with sinewave signal, so it should do what you're asking. Not possible to "scope" the signal and determine the delay, but you can easily calculate that I think. There's also "auto-delay", but that dont work all too well.
I have a 20+ years old CLIO (DOS version), but it can do gating with sinewave signal, so it should do what you're asking. Not possible to "scope" the signal and determine the delay, but you can easily calculate that I think. There's also "auto-delay", but that dont work all too well.
Its two 15" Selenium Selenium 15PW5 packed into 38x38x42cm foot print - most compact dipol sub I can think of - my dedicated listening room isn't too big, you have to know...
I'll do two or three each side - if it turns out ok for me.
I contacted Ivo who did ARTA, how best to deal with this situation - will see...
I hate to *calculate* delay
Do you get *much* lower THD than -40dB if you measure at say 0.1 W or even lower ?
Michael
I'll do two or three each side - if it turns out ok for me.
I contacted Ivo who did ARTA, how best to deal with this situation - will see...
I hate to *calculate* delay
Do you get *much* lower THD than -40dB if you measure at say 0.1 W or even lower ?
Michael
Thanks for confirmation regarding "bottom" of THD.
How do you estimate this +6dB respectively +9dB ?
*If* you go by the OB SPL canceling effect, the THD rate - relative to closed box (or infinite baffle) - would depend on frequency response (or more exactly to excursion per SPL) relative to closed box (or IB) - no ? Did I miss something ?
This is spec from datasheet - at 35W given the 10% of 350W AES - though not exactly a very useful plot in the range of 20-200Hz and under unknown measurement conditions.
I mean - even *my* measurements are better at around 20Hz...
Michael
How do you estimate this +6dB respectively +9dB ?
*If* you go by the OB SPL canceling effect, the THD rate - relative to closed box (or infinite baffle) - would depend on frequency response (or more exactly to excursion per SPL) relative to closed box (or IB) - no ? Did I miss something ?
This is spec from datasheet - at 35W given the 10% of 350W AES - though not exactly a very useful plot in the range of 20-200Hz and under unknown measurement conditions.
I mean - even *my* measurements are better at around 20Hz...
Michael
Last edited:
Yes - frequency response will affect distortion. If the response is rising at high frequencies, the higher order distortion components will be amplified. The same goes for the 6 dB per octave loss of dipoles.
My graph below is showing a 6 dB/oct curve, with 2nd and 3rd harmonic of 20 Hz pointed out. As we see, the 2nd harmonic (red dot) is 6 dB higher, and the 3rd (green) ~9.5 db Higher.
My graph below is showing a 6 dB/oct curve, with 2nd and 3rd harmonic of 20 Hz pointed out. As we see, the 2nd harmonic (red dot) is 6 dB higher, and the 3rd (green) ~9.5 db Higher.
Attachments
Good afternoon Stig and everyone.
I must say, Stig, you do very nice work, and you have great taste in drivers! Those are some beautiful looking setups.
I am a little puzzled, however, and in need of education. Open baffle designs are of great interest to me, and something I have been considering and researching on a day-dream level for some time now. It seems reasonable to me, that there are some advantages in removing (at least in part) the enclosure in a loudspeaker system.
However, there are some pictures in this thread of systems completely devoid of front baffles, with the drivers suspended in free-air. Wouldn't this be a rather in-efficient use of the drivers, in the part of the spectrum that moves from 2 pi to 4 pi space? I suspect this configuration would severely limit the use-able low end of a driver. At least on the low-end drivers, wouldn't it be helpful to have some baffle?
JF
I must say, Stig, you do very nice work, and you have great taste in drivers! Those are some beautiful looking setups.
I am a little puzzled, however, and in need of education. Open baffle designs are of great interest to me, and something I have been considering and researching on a day-dream level for some time now. It seems reasonable to me, that there are some advantages in removing (at least in part) the enclosure in a loudspeaker system.
However, there are some pictures in this thread of systems completely devoid of front baffles, with the drivers suspended in free-air. Wouldn't this be a rather in-efficient use of the drivers, in the part of the spectrum that moves from 2 pi to 4 pi space? I suspect this configuration would severely limit the use-able low end of a driver. At least on the low-end drivers, wouldn't it be helpful to have some baffle?
JF
J.R.Freeman,
Have you seen the number and size of the drivers Stig uses? What you are saying is correct; you are throwing away a lot of efficiency. But you get terrific sound quantity in return.
Michael,
After reading a lot about the effect of non-linear distortion on sound quality, I did some tests myself. It appears Geddes is right: quite high amounts of distortion can be tolerated, it's simply inaudible on music.
Have you seen the number and size of the drivers Stig uses? What you are saying is correct; you are throwing away a lot of efficiency. But you get terrific sound quantity in return.
Michael,
After reading a lot about the effect of non-linear distortion on sound quality, I did some tests myself. It appears Geddes is right: quite high amounts of distortion can be tolerated, it's simply inaudible on music.
Harmonic distortion is not all that bad, and levels around 1-3% is totally acceptable in my opinion. Its a good idea to check for THD, as it's an indication other problems with your speakers. Whats much worse is intermodulation... it sounds very very bad.
Dipole design is mainly a tradeoff between efficiency and good off-axis response / dipole radiation pattern. You can't have both.
I do have large H-baffles for the lower octaves, that is 20-100 Hz. The 100 to 500 Hz range is covered by four 8" without baffle. It would be possible to use a small baffle here to increase efficiency without compromizing the dipole radiation pattern, although the gain would not be great.
I agree - I simply have to - as I love OB performance.
So I agree in at least some cases : It *can be* that THD does not matter that much.
But at the same time I strongly disagree on Earls and your generalization that THD *never matters* - even at low figures mind you.
Michael
Last edited:
I did a quick check with Hornresp to get an idea on what s possible with dipole sub.
Results are quite astonishing:
For Ripole - which isn't *that* much different form H-baffle - and a top line 21" it takes 4 double speaker subs to get 120dB free air SPL up from 20Hz.
Even though *in room* performance is a different thing all together, its quite impressive I'd say...
Also I can assure that even low 80-90dB where I measured 20Hz, already were a tactile experience *in room*.
Michael
Results are quite astonishing:
For Ripole - which isn't *that* much different form H-baffle - and a top line 21" it takes 4 double speaker subs to get 120dB free air SPL up from 20Hz.
Even though *in room* performance is a different thing all together, its quite impressive I'd say...
Also I can assure that even low 80-90dB where I measured 20Hz, already were a tactile experience *in room*.
Michael
Last edited:
Kinda of the point, don't you think?
This is the seminal idea -- suspend the drivers with wires or elastic bands -- and use enough of them to to overcome the loss that is incurred, and limit them to a bandpass that does not incur excessive distortion or non-linearities. Baffles vibrate and add diffraction.
Given the appropriate drivers [Xmax, Qts, resonance Hz out of bandpass] and given the right filtering [Active, 24dB LR, equalization to compensate for dipole cancellation] and given enough drivers, the sonic results are spectacular.
-- Charles
This is the seminal idea -- suspend the drivers with wires or elastic bands -- and use enough of them to to overcome the loss that is incurred, and limit them to a bandpass that does not incur excessive distortion or non-linearities. Baffles vibrate and add diffraction.
Given the appropriate drivers [Xmax, Qts, resonance Hz out of bandpass] and given the right filtering [Active, 24dB LR, equalization to compensate for dipole cancellation] and given enough drivers, the sonic results are spectacular.
-- Charles
THD and IMD measure the same thing the same way; THD is IMD in the special case where all tones are at the same frequency. Unsurprisingly, it's been my experience with chips there's good correlation between THD and IMD levels. Data for drivers is harder to come by, but the handful of results I've come across suggest good correlation. Discussion on DIYAudio tends to neglect the DS and Rnonlin methods for estimating percieved audio quality and focus on the Gedlee metric or harmonic masking. As such, it tends to miss the point THD is often the only data one has; there's a very pragmatic problem of how to correlate THD with something like DS. I've had reasonable results from assuming THD = IMD and applying DS to the resulting IMD guesstimate, but the data set I've worked with so far is very limited.
Between that and StigErik's observations about dipole equalization's distortion boosting properties the question of usable xmax is quite a relevant one. The 15PW5 is unusually long throw for a pro woofer and is certainly on my shortlist but based on my experience with Selenium's 8W4P I would guess its clean xmax is around 0.5mm. So I'd not expect a pair of them to have good THD much above 60-70dB at 40Hz. Michael, when you say very low SPL, how low is it and what's the ambient noise floor? In my experience getting good THD data in the low bass is difficult due to limited SnR ratios.
Between that and StigErik's observations about dipole equalization's distortion boosting properties the question of usable xmax is quite a relevant one. The 15PW5 is unusually long throw for a pro woofer and is certainly on my shortlist but based on my experience with Selenium's 8W4P I would guess its clean xmax is around 0.5mm. So I'd not expect a pair of them to have good THD much above 60-70dB at 40Hz. Michael, when you say very low SPL, how low is it and what's the ambient noise floor? In my experience getting good THD data in the low bass is difficult due to limited SnR ratios.
The "long throw" type are the 18" Selenium 18WS600 for example with a "linear excursion range" gap-coil of 20mm versus roughly half of that for the 15" Selenium 15PW5.
Thanks for the paper - seems reasonable that what is just easy to measure (THD IM) not necessarily gets sourted out always the same way by our ear-brain system.
Have to dig a little bit deeper to get the point what the authors think really matters.
----------
Selenium 6W4P has "only" 3.5mm of "linear excursion range" (gap-coil). I did not experience problems on higher SPL though, did you ?
Did you check Doppler intermodulation to keep low with appropriate XO ?
----------
I would not be too pessimistic regarding mudded sound at high SPL form OB sub - But certainly you already have the most experience from listening in that department - hope to get around to do a meaningful plot (in the sense of good measurement practice at least) for THD over signal one day.
Michael
Thanks for the paper - seems reasonable that what is just easy to measure (THD IM) not necessarily gets sourted out always the same way by our ear-brain system.
Have to dig a little bit deeper to get the point what the authors think really matters.
----------
Selenium 6W4P has "only" 3.5mm of "linear excursion range" (gap-coil). I did not experience problems on higher SPL though, did you ?
Did you check Doppler intermodulation to keep low with appropriate XO ?
----------
I would not be too pessimistic regarding mudded sound at high SPL form OB sub - But certainly you already have the most experience from listening in that department - hope to get around to do a meaningful plot (in the sense of good measurement practice at least) for THD over signal one day.
Michael
Last edited: