The story changes with type of horn. A plain waveguide isn't necessarily supposed to beam more as it gets bigger (although size can affect its performance).
For 500Hz it would be nice to have something towards 70cm across. Since the Dayton tries for narrow dispersion, it will be interesting to see where it flips as it's likely to affect or set your cross.
For 500Hz it would be nice to have something towards 70cm across. Since the Dayton tries for narrow dispersion, it will be interesting to see where it flips as it's likely to affect or set your cross.
Yes. My living room is fairly big and "live" - lots and lots of glass and hard surfaces. The system there uses a large plain waveguide + OB midbass because I wanted even pattern control with not much sound hitting the glass (also partly out of curiosity). It seems to suit the space.
My planned build / test rig will have different needs: it is going to be built into a small, very dead room. Therefore, I was thinking of moving away from plain waveguides, to horns that emphasise sensitivity and on-axis properties over pattern control. Also: I already have several suitable horns 🙂
HF horn with top octave beaming = less need for subtractive EQ in the midband = better final efficiency. This is good for a low powered rig.
* currently half built. The structure is a timber frame braced by large sheets of perforated steel. These panels will have ~20cm stuffed voids behind them. These panels cover most of the side walls and ceiling.
I had already seen those links the first time they were posted, but it is also worth noting that "generally inaudible" doesn't imply that that's always the case. Besides, if one finds IMD to be reasonably audible then it may not be worth using that particular compression driver/horn in those conditions anyway and the links you posted do not change this fact at all, as they only say that these distortions are not generally considered very audible.
Further, I also don't think that the links you posted recommended disregarding these distortions altogether (due to "inaudibility"). So, on the whole, I still think that the idea of these distortions being indicators of "something wrong" is not a bad idea at all.
I've understood it so: that perceived sound quality just doesn't correlate with THD reading. THD can be audible and a problem if playing at the limits of the driver/system. The point is to design the system so that it is not operated it's limits thus making the THD number irrelevant.
I'm very aware of the limited audibility of THD <10% but find it a good indicator of limiting behavior to test capability of devices.
That is logical thing to do but what if you toss the better sounding away because it has bigger number that doesn't correlate with the sound in anyway?😀 For loud applications why not, seems mandatory to find limits. But for home use I'm not sure if the limits are anywhere near? Woofers are probably giving up 10-20db before the compression driver, at least that is the sensitivity difference.
I don't have any experience listening or measuring more than one driver at home so I'm talking without first hand experience on this subject. All that I know I need to pad CD down a lot to meet the rest of the system. Perhaps other drivers sound better or worse at home and measurements can indicate that, then utilizing the measurements are good thing. Since I'm riding on my readings I'd point finger first on too small system (not enough SPL capability) or linear distortions first, or perhaps trying to identify resonances. Resonances should be visible in other plots like impedance or waterfall? Swap waveguide or driver and the graphs change.
Carry on, following with interest.
I don't have any experience listening or measuring more than one driver at home so I'm talking without first hand experience on this subject. All that I know I need to pad CD down a lot to meet the rest of the system. Perhaps other drivers sound better or worse at home and measurements can indicate that, then utilizing the measurements are good thing. Since I'm riding on my readings I'd point finger first on too small system (not enough SPL capability) or linear distortions first, or perhaps trying to identify resonances. Resonances should be visible in other plots like impedance or waterfall? Swap waveguide or driver and the graphs change.
Carry on, following with interest.
Last edited:
Hi Chris,
Yep, I'd like to see such measurements too.
I'd like to even know what "high" IMD means, spec wise.
I've started making some IMD measurement comparisons to try to sleuth why i'm hearing differences between MEHs using different driver complements, but it's still just speculative comparisons. Here's a link explaining.. https://www.diyaudio.com/community/threads/syn-10.383607/#post-6961546
It's all on hold until Spring winds die down for outdoor testing...
@hollowboy, please say if you'd rather not have this type discussion in your thread.
I'm sorry i can't comment directly OT....all my driver experience is with 1.4".
That happens only if you accept / reject a compression driver based on distortion measurements. However, in this thread, distortion measurements were suggested for the purpose of finding out the extent (in frequency) a horn loads a compression driver, in order to decide on an appropriate crossover frequency for the system.
There's also this other point that audibility is about perception by the human ear whereas measurability relates to an instrument picking up responses. Thus, a rising harmonic distortion measurement that may indicate a device limit being approached could still very well be not very audible to human ears.
Ah yeah there is different ways to look at things. In my view it is the size and properties of a waveguide/horn and woofer size one is crossing over to that determines the crossover point, the acoustic radiation of the system. Any driver that fits the horn/waveguide can do it, only thing is that SPL is limited if the driver is not "capable" enough. But home application doesn't really require very mucho I think and only requirement for the driver is that it can do the SPL requirement well enough, perhaps few amplifier watts. Need to dig out some graphs and calculator.
I agree there is very much point to choose best possible drivers, but one could choose by ear and if there is no difference the cheaper usually wins 🙂 Following how the discussion evolves.
I agree there is very much point to choose best possible drivers, but one could choose by ear and if there is no difference the cheaper usually wins 🙂 Following how the discussion evolves.
Last edited:
One more thought on it, assuming we are interested to crossover where HF horn/waveguide directivity matches the woofer (or low horn) directivity, whose response naturally narrows as frequency goes up. If we want to have wide dispersion for the HF horn we have to crossover lower in frequency where the woofer is still with wide dispersion. Unfortunately "loading" on a such waveguide is usually less than same mouth size but lot deeper narrow dispersion device (unless there is some weird tube/extension going on providing extra loading, not too common).
If we want narrower directivity HF then we have possibility to cross higher in frequency, where the woofer beams suitably.
Thought experiment: we have a compression driver with unknown performance and want better sound from it then we can sacrifice wide HF dispersion for more loading and higher xo point to get more SPL capability. The driver can now play at higher SPL for same distortion performance than with wide directivity device. Or the otherway around, there are more drivers that sound fine in narrow directivity devices at home listening levels.
Hollowboy I guess you've gone through the AudioXpress test bench? there is the BMS 5530nd and Faital HF108 tests, both showing about >105db /W/m at 1kHz on narrow dispersion horn. Similar reading as with your link to the eminence driver test, although different waveguide.
This would be 15db headroom over 90db listening level, plenty at one or two meters away at home. Only with 1 Watt, headroom included, pretty amazing how loud they are. Surely passive xo robs some off it and you might like it louder but to me it seems the drivers are fine in your application, just cruising. Both hf108 and 5530nd fall below 100db sensitivity somewhere 700Hz. These kind of levels are very loud. Not sure if this is typical or top performance but I would guess there are many other drivers showing similar sensitivity.
If we want narrower directivity HF then we have possibility to cross higher in frequency, where the woofer beams suitably.
Thought experiment: we have a compression driver with unknown performance and want better sound from it then we can sacrifice wide HF dispersion for more loading and higher xo point to get more SPL capability. The driver can now play at higher SPL for same distortion performance than with wide directivity device. Or the otherway around, there are more drivers that sound fine in narrow directivity devices at home listening levels.
Hollowboy I guess you've gone through the AudioXpress test bench? there is the BMS 5530nd and Faital HF108 tests, both showing about >105db /W/m at 1kHz on narrow dispersion horn. Similar reading as with your link to the eminence driver test, although different waveguide.
This would be 15db headroom over 90db listening level, plenty at one or two meters away at home. Only with 1 Watt, headroom included, pretty amazing how loud they are. Surely passive xo robs some off it and you might like it louder but to me it seems the drivers are fine in your application, just cruising. Both hf108 and 5530nd fall below 100db sensitivity somewhere 700Hz. These kind of levels are very loud. Not sure if this is typical or top performance but I would guess there are many other drivers showing similar sensitivity.
Last edited:
Here are the links
https://audioxpress.com/article/test-bench-the-bms-1-5530nd-high-efficiency-compression-driver
https://audioxpress.com/article/Tes...ession-drivers-coupled-with-LTH102-60-50-horn
THD graphs span only down to 1kHz and both are roughly at similar numbers than higher up in their passband.
I've also checked out the Joseph Crowe multitone measurements Kipman posted earlier. It looks nice way to gauge the headroom and if performance headroom is enough for SPL requirement. Gauging noisefloor between multitone test peaks to figure out dynamic range seems easy and makes sense, evaluating performance between multiple test subjects would work out fine I guess. Does not require naming of the poo between the tones, it is just "noise" and what ever it is 😀 try and keep it below audibility, for example the 1% / -40db suggested
https://audioxpress.com/article/test-bench-the-bms-1-5530nd-high-efficiency-compression-driver
https://audioxpress.com/article/Tes...ession-drivers-coupled-with-LTH102-60-50-horn
THD graphs span only down to 1kHz and both are roughly at similar numbers than higher up in their passband.
I've also checked out the Joseph Crowe multitone measurements Kipman posted earlier. It looks nice way to gauge the headroom and if performance headroom is enough for SPL requirement. Gauging noisefloor between multitone test peaks to figure out dynamic range seems easy and makes sense, evaluating performance between multiple test subjects would work out fine I guess. Does not require naming of the poo between the tones, it is just "noise" and what ever it is 😀 try and keep it below audibility, for example the 1% / -40db suggested
Last edited:
There is relevant discussion on another thread, about max excursion for example
https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6993015
Compression driver capability/distortion talk earlier, Geddes https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6992273 reminds again that at least some B&C drivers were very good among drivers some years back. Can't help but notice they have quite good amount of excursion available in comparison to some other models mentioned. Never thought about it before. Interesting topic
https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6993015
Compression driver capability/distortion talk earlier, Geddes https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-6992273 reminds again that at least some B&C drivers were very good among drivers some years back. Can't help but notice they have quite good amount of excursion available in comparison to some other models mentioned. Never thought about it before. Interesting topic
I'n my case, the LF will be the weak link, but I can use multiple drivers and some boundary loading / room gain to close the gap.
I was thinking of running multiple wall mounted (infinite baffle) drivers per side because why not. I have a bunch of 12" and 15" woofers (leftovers from prior projects / salvage from old PA gear) that I would like to use up. A quad of 15" should give ~103dB midbass, but I don't know where the -3dB point will be, since it will be related to "room gain" from a room I have only half built 🙂
I spent a lot of today rummaging for components and running tests with REW, keeping the mic position + volume the same.
Short version: all the decent drivers I tried had distortion rising at the same frequency. Based on this test, any of them would suit a 1kHz cross.
--
Long version:
I ran the sweeps with REW, about 10dB from clipping. A handheld SPL meter says that's a little over 80dB. Background noise was nearly 40dB (birds, the fans in my PC, traffic rumble).
These tests are not calibrated or ideal, but are enough:
...and this last test seems to be kinda useless (because, at this low signal level, all the results are the same).
Things learned / demonstrated:
1) the HF horn does indeed make a difference.
The Tymphany TDH hits 1% THD at 730Hz on a short horn: P.Audio PH-220 (13cm long).
The same driver hits 1% THD at 585Hz on a longer horn: Dayton h812 (20cm long). No shock here: bigger is better.
The Tymphany is also ~4dB louder and quite a lot flatter on the latter horn. I guess the sensitivity of the Tymphany + Dayton combination is about 105dB.
2) The mid horn is several dB louder than this, and this loudness is fairly driver independent. Swapping from a good 6" to a good 4", the sensitivity 300-1,000Hz stays the same (the HF is a bit different), even though the drivers' free air efficiency differs by 5dB.
One outlier I tried: a free (salvaged) no-name sealed mid. It has a big peak in band, but goes higher than any of the open-backed drivers I tried.
3) All of the brand name compression drivers I tried go well below 1kHz (at this test level) before distortion rises.
Three drivers with 44mm voice coils.
Dayton 250p (small and cheap)
P.Audio BM-D450s (heavy and mid priced)
JBL 2425J (heavy and expensive)
...and all showed the same results as the much smaller Tymphany - distortion shoots up below 600Hz.
They all seem like they'd be fine crossed at 1kHz, or fine until they hit their displacement limits. The Tymphany will, of course, hit that limit well before the others (smaller diaphragm).
I tried a couple of outliers:
a) no-name modern 1" screw-on driver from a cheap 2-way. This was not fine - it screamed like it was being sawn in half.
On the short stock horn it came with, it hit 10% THD well before 1kHz. On the Dayton h812 horn, the distortion was reduced to about 2%, but still kicked in before 2kHz. I guess it has a really stiff suspension / effectively zero excursion. This was good as a sanity check / proof that my distortion tests will spot a truly bad driver.
b) no-name vintage 3/4" exit screw-on driver which I used on its vintage horn (since I don't have a 3/4" adaptor). The horn looks like an Altec 811B (but isn't). The driver has no phase plug - if you shine a light through the bug screen, you can see the diaphragm.
I had to sweep down to 400Hz to see much distortion (about 3%). The JBL was 100% distortion at 400Hz. Not an apples-to-apples comparison, but interesting.
Short version: all the decent drivers I tried had distortion rising at the same frequency. Based on this test, any of them would suit a 1kHz cross.
--
Long version:
I ran the sweeps with REW, about 10dB from clipping. A handheld SPL meter says that's a little over 80dB. Background noise was nearly 40dB (birds, the fans in my PC, traffic rumble).
These tests are not calibrated or ideal, but are enough:
- to tell me relative SPL levels between devices.
- to tell me which driver+horn combo needs the most/least EQ
- to tell me the frequency where distortion goes nuts.
...and this last test seems to be kinda useless (because, at this low signal level, all the results are the same).
Things learned / demonstrated:
1) the HF horn does indeed make a difference.
The Tymphany TDH hits 1% THD at 730Hz on a short horn: P.Audio PH-220 (13cm long).
The same driver hits 1% THD at 585Hz on a longer horn: Dayton h812 (20cm long). No shock here: bigger is better.
The Tymphany is also ~4dB louder and quite a lot flatter on the latter horn. I guess the sensitivity of the Tymphany + Dayton combination is about 105dB.
2) The mid horn is several dB louder than this, and this loudness is fairly driver independent. Swapping from a good 6" to a good 4", the sensitivity 300-1,000Hz stays the same (the HF is a bit different), even though the drivers' free air efficiency differs by 5dB.
One outlier I tried: a free (salvaged) no-name sealed mid. It has a big peak in band, but goes higher than any of the open-backed drivers I tried.
3) All of the brand name compression drivers I tried go well below 1kHz (at this test level) before distortion rises.
Three drivers with 44mm voice coils.
Dayton 250p (small and cheap)
P.Audio BM-D450s (heavy and mid priced)
JBL 2425J (heavy and expensive)
...and all showed the same results as the much smaller Tymphany - distortion shoots up below 600Hz.
They all seem like they'd be fine crossed at 1kHz, or fine until they hit their displacement limits. The Tymphany will, of course, hit that limit well before the others (smaller diaphragm).
I tried a couple of outliers:
a) no-name modern 1" screw-on driver from a cheap 2-way. This was not fine - it screamed like it was being sawn in half.
On the short stock horn it came with, it hit 10% THD well before 1kHz. On the Dayton h812 horn, the distortion was reduced to about 2%, but still kicked in before 2kHz. I guess it has a really stiff suspension / effectively zero excursion. This was good as a sanity check / proof that my distortion tests will spot a truly bad driver.
b) no-name vintage 3/4" exit screw-on driver which I used on its vintage horn (since I don't have a 3/4" adaptor). The horn looks like an Altec 811B (but isn't). The driver has no phase plug - if you shine a light through the bug screen, you can see the diaphragm.
I had to sweep down to 400Hz to see much distortion (about 3%). The JBL was 100% distortion at 400Hz. Not an apples-to-apples comparison, but interesting.
Chris, how are you cooping with your NSD1095Ns?
Some has reported very favourable initial impression sound wise but then abandon them due to sharp/metallic sound characteristics... what is your thoughts here?
//
Some has reported very favourable initial impression sound wise but then abandon them due to sharp/metallic sound characteristics... what is your thoughts here?
//
You're not hoping to get away with not using some degree of equalisation, are you? Some of the anecdotal evidence you'll come across means little because of that aspect.
I would conduct the sweeps at much higher SPL then you will see differences between drivers easier and have more SNR. I usually use 0.283V, 2.83V and a higher voltage that depends on driver capability. My mic will clip >120dB though so this presents an additional constraint.
I'm sorry, did you say you hope someone can tell the difference when they hear a resonant response when they are also listening for higher order distortion? I would hope that a CD doesn't produce higher order distortion and that it won't be audible.
A metallic sound from a resonance can be audible but it doesn't have to be a problem.
A metallic sound from a resonance can be audible but it doesn't have to be a problem.