Oh, so yes I'd love to see them. Not saying you haven't done enough work. You've done so much. I can't think of any other DIY guy who's done more.
Unluckilly by lack of space I only run one system ! I have a little Fostex TH-7 headphones, not precise enough in low end, I choose it only for the confort : not a fatiguing headphones ! (cheap too ! good SQ/P).
Not talking to the impact in the low if it was you wanted to say ?! I find the 10F on my main speaker has also impacts in the entire FR but with a little more transcient that many and without being harsch ! I liked more the Vifas than the B80 as well. If TG9 is B driver : it was my second choice. I don't find the 10F has too much treble either, just more resolution here : but maybe it's the same thing ? Climbing treble on the 10F and flater curve on the TG9 ?
Both are good drivers for me in the limit of the test as I never heard both before. I believe maybe I don't like too much the raw paper cones ?
Jay : http://www.diyaudio.com/forums/full...2in-4in-drivers-round-4-a-22.html#post4448327 ! Need your point of view about track 1 & 2 please!
track 1 : do you think reccording engineering of the harmonica is too much climby ?
track 2 : is the piano band large enough ? where in the FR band are the soft piano reverbs in the Church ? Mid (i.e. below 2k Hz)?
of course if you want ! I will try to upload two asc of Tchaikovsky : one for triangle and the other for huge dynamic (may can sature some drivers ???).
@ Mondogenerator : Hi, according to you please, where are compressed in low and end FR most of the discs today ? Any schèmes of dynamic gap or max spl in relation to the FR curve in 24/96 reccording of today (most are 24/96 nowadays ?)
Below 200 hz : our ears compress and get down the mid-bass: is it because most of the discs customers play them on little speaker (bookshelves speakers) ???
Last edited:
I retook the HD data for the 10F, TG9, TC9, B80, PS95 more carefully and in two cabinets: the Dagger and the Nautaloss (larger volume circa 5 liters).
I found out that I had a mistake in my previous HD measurements of the 10F, the mic calibration for SPL level was off about 2dB (too high) so the HD will be too low. My fault for trying to take data at the end of long day and rushing. Anyhow, the HD data for the 10F in the Dagger is still very good, 0.266% THD at 1kHz at 96dB and 0.5m or 90dB at 1m equivalent. What is interesting is that the HD for the TC9 and TG9 is quite a bit higher (about 0.7% for both at 1kHz) but what troubled me was that it started rising below 1kHz. This suggested that the Dagger volume, as many of you suspected, was too small for these higher Qts drivers to have breathing space. I put the drivers in the Nautaloss and retook the data. The THD for the TC9 and TG9 below 1kHz went down significantly, but the THD at 1kHz increased to about 1%. The THD for the 10F also increased to 0.58% (from 0.266%) at 1kHz and everywhere else a little more too. So this says that the Dagger is really well suited to the 10F because I cannot get the HD to decrease any more, and in fact, rises in the Nautaloss. It also says that rear chamber design can make a huge difference in the harmonic distortion. So it is not surprising that Jeshi's measurements show a higher level and distribution of the different components of HD in her EVA cabinet vs my Dagger. I will post the HD plots later - waiting for large REW file to transfer from DAQ pc to workstation for processing and viewing. The B80 exhibits lower HD than the TG9 or TC9 but is not as clean as the 10F. The PS95 had the least clean looking HD plots when driven under stress levels. All data is compared for equivalent of 90dB at 1m, I varied drive voltage to achieve this. The 10F is super efficient, so much less drive is required. This is perhaps why its distortion is so low - the cone doesn't move much. The sensitivity difference would appear to be higher than the manufacturer's spec of 86.6dB at 2.83v.
Oh, and speaking of actual drive voltage - I tried to measure that quantitatively but realize my DMM's are old and plain inaccurate. I have two and the readings differ and it depends on what you measure, a 1kHz sine, white noise, pink noise? I use the generator function in REW. At any rate, I just ordered a new Fluke DMM, no messing around with cheap meters anymore. May take a while to get it. The DMM's I used showed 3.7 vac and 3.52 vac for -14db setting with pink noise. But show 4.74 vac and 4.3 vac for 1kHz sine wave. So what is the drive voltage? I wish I kept my O-scope, sold that many years ago.
More data plots to come - just wanted to let you all to know what is coming.
The different cabinets are showing that the elevated HD for the TC9 and TG9 in the Dagger, may have not given those drivers an even chance against the 10F and B80. They are higher in distortion no matter what though as to be expected for the price differential.
The 10F is about 4 to 6 dB more sensitive than the TC9 and TG9, respectively. It has about 3x lower HD at 1kHz when stressed to equivalent 90dB SPL at 1m, so is noticeably cleaner sounding. Those two things alone, in my opinion, justifies its 5x higher price tag vs the TG9. But can you hear a difference between those levels of HD? Probably yes because they are not flat and uniform in distribution like the 10F.
I found out that I had a mistake in my previous HD measurements of the 10F, the mic calibration for SPL level was off about 2dB (too high) so the HD will be too low. My fault for trying to take data at the end of long day and rushing. Anyhow, the HD data for the 10F in the Dagger is still very good, 0.266% THD at 1kHz at 96dB and 0.5m or 90dB at 1m equivalent. What is interesting is that the HD for the TC9 and TG9 is quite a bit higher (about 0.7% for both at 1kHz) but what troubled me was that it started rising below 1kHz. This suggested that the Dagger volume, as many of you suspected, was too small for these higher Qts drivers to have breathing space. I put the drivers in the Nautaloss and retook the data. The THD for the TC9 and TG9 below 1kHz went down significantly, but the THD at 1kHz increased to about 1%. The THD for the 10F also increased to 0.58% (from 0.266%) at 1kHz and everywhere else a little more too. So this says that the Dagger is really well suited to the 10F because I cannot get the HD to decrease any more, and in fact, rises in the Nautaloss. It also says that rear chamber design can make a huge difference in the harmonic distortion. So it is not surprising that Jeshi's measurements show a higher level and distribution of the different components of HD in her EVA cabinet vs my Dagger. I will post the HD plots later - waiting for large REW file to transfer from DAQ pc to workstation for processing and viewing. The B80 exhibits lower HD than the TG9 or TC9 but is not as clean as the 10F. The PS95 had the least clean looking HD plots when driven under stress levels. All data is compared for equivalent of 90dB at 1m, I varied drive voltage to achieve this. The 10F is super efficient, so much less drive is required. This is perhaps why its distortion is so low - the cone doesn't move much. The sensitivity difference would appear to be higher than the manufacturer's spec of 86.6dB at 2.83v.
Oh, and speaking of actual drive voltage - I tried to measure that quantitatively but realize my DMM's are old and plain inaccurate. I have two and the readings differ and it depends on what you measure, a 1kHz sine, white noise, pink noise? I use the generator function in REW. At any rate, I just ordered a new Fluke DMM, no messing around with cheap meters anymore. May take a while to get it. The DMM's I used showed 3.7 vac and 3.52 vac for -14db setting with pink noise. But show 4.74 vac and 4.3 vac for 1kHz sine wave. So what is the drive voltage? I wish I kept my O-scope, sold that many years ago.
More data plots to come - just wanted to let you all to know what is coming.
The different cabinets are showing that the elevated HD for the TC9 and TG9 in the Dagger, may have not given those drivers an even chance against the 10F and B80. They are higher in distortion no matter what though as to be expected for the price differential.
The 10F is about 4 to 6 dB more sensitive than the TC9 and TG9, respectively. It has about 3x lower HD at 1kHz when stressed to equivalent 90dB SPL at 1m, so is noticeably cleaner sounding. Those two things alone, in my opinion, justifies its 5x higher price tag vs the TG9. But can you hear a difference between those levels of HD? Probably yes because they are not flat and uniform in distribution like the 10F.
Last edited:
That depends... the 10F is the expensive one 😀
Good post X and thanks for going the extra mile (again). It figures the 10F would do better. But good to know the TG and TC drivers could do better in a larger box.
We always knew to test the drivers in the same enclosure could/would be difficult. The 10F is still very happy in a smaller one.
With a firm like Scan Speak I had no doubt you would be getting something extra for the money paid. The market the driver is aimed at is very different from both the Vifa siblings (if we keep considering the long lost family tie). Still the resemblances are there. Think about it, the 10F is in the Discovery line, the budget line of Scan Speak. I'm sure they, themselves, would think more highly about their Revelator and later Illuminator series. Too bad there isn't a 3.5 to 4" full range in that line.
Last edited:
HD measurements in Dagger and Nautaloss
Here are the plots. All at 0.5m, varied drive levels until nominal 96dB was measured at 1kHz - equivalent to 90dB at 1m. The drive levels are listed in the legend as dB full scale in REW's generator. I will provide correlation of drive voltage in volts when I get a good DMM. For now, -14dB corresponds to the two different DMM's I have listed in previous post, they don't agree and the measurement is difficult without an O-scope.
10F in Dagger:
10F in Nautaloss:
TG9 in Dagger:
TG9 in Nautaloss:
TC9 in Dagger:
TC9 in Nautaloss:
B80 in Nautaloss:
PS95 in Dagger:
PS95 in Nautaloss:
Here are the plots. All at 0.5m, varied drive levels until nominal 96dB was measured at 1kHz - equivalent to 90dB at 1m. The drive levels are listed in the legend as dB full scale in REW's generator. I will provide correlation of drive voltage in volts when I get a good DMM. For now, -14dB corresponds to the two different DMM's I have listed in previous post, they don't agree and the measurement is difficult without an O-scope.
10F in Dagger:
10F in Nautaloss:
TG9 in Dagger:
TG9 in Nautaloss:
TC9 in Dagger:
TC9 in Nautaloss:
B80 in Nautaloss:
PS95 in Dagger:
PS95 in Nautaloss:
Attachments
Last edited:
Raaaaaaaaaaaaaah😛 ! I say one wife ! not 25 ! I should try your religion 😀... But fear that who's me in the box with such a load !
I don't know, that is a tough one. Both get the job done and are very competent. One has total mastery in all technical details and dresses in black all the time, is born and raised in DK, the other is a little rough around the edges when flustered and likes to dress in white and is ok eating at fast food place. The one dressed in white was conceived in DK but born abroad in Asia where she spent all her life. The one dressed in black likes more upscale restaurants and you want to show off to all your friends. In the end, it depends, you can love both equally but in different ways. Hmm.... 🙂
With 1.1L, all drivers are underdamped, but 10F benefits the most in term of distortion because this driver requires less volume than the other drivers. After 10F, TG9 and B80 both requires almost same volume, then TC9 which require the biggest volume....
If you increase the volume to 5-8L, 10F becomes overdamped and TG9 and B80 will benefit the most...
But with higher volume, it is fair to all drivers that all are overdamped, than using small volume such as 1.1L.
But don't forget that there is always trade off between distortion and bass "impact". Being slightly underdamped in 1.1L, TG9 appeared to have high distortion BUT very good bass "impact". If you think that the increase in distortion is still way below your threshold then it is better to have the extra bass "impact"...
That's if [you think] the difference in the distortion is audible 😀
Last edited:
5MR450NDY sound clips
Eldam was wondering what the 5MR450NDY sounds like with the latest setup. Here it is with the 350Hz Harsch XO and mounted in the Nautaloss with RS225 woofer for bass. Same sound clips as Round 4. There is a BSC applied to flatten the rising response.
Here is the XO response, the sound clips are of the non-EQ'd response (light blue):
It may sound a bit bright. Let me know if you are interested in the other set of clips with just two gentle PEQ's applied.
Eldam was wondering what the 5MR450NDY sounds like with the latest setup. Here it is with the 350Hz Harsch XO and mounted in the Nautaloss with RS225 woofer for bass. Same sound clips as Round 4. There is a BSC applied to flatten the rising response.
Here is the XO response, the sound clips are of the non-EQ'd response (light blue):
It may sound a bit bright. Let me know if you are interested in the other set of clips with just two gentle PEQ's applied.
Attachments
I'm in the process of getting rid of my first wife. 17 years was enough. Fortunately, with loudspeakers, you can play the field without worrying about legal fees, trial and nasty emails.
She wanted half the value of my Cary Audio tube gear but during negotiation she decided to keep the Onkyo and Yamaha AV receivers. She never liked music anyway.
This weekend I'm building a closet shelf that holds 5 baffles so I can easily swap drivers... unfortunately I have 7 drivers on baffles. I may have to divorce a few.
She wanted half the value of my Cary Audio tube gear but during negotiation she decided to keep the Onkyo and Yamaha AV receivers. She never liked music anyway.
This weekend I'm building a closet shelf that holds 5 baffles so I can easily swap drivers... unfortunately I have 7 drivers on baffles. I may have to divorce a few.
@Eldam,
I'm not sure I follow you entirely, but ill answer as best I can.
Firstly, if we take a CD and replay via 24/96 system and encode to mp3, we don't gain any dynamic range, unless it is artificially 'expanded', nor does upsampling restore what is lost in the cd encoding. (barleywater, ronion, or wintermute should be able to correct me here if I make incorrect statements)
This is what is depressing about CD format. We need far more DVD encoded music CD, even just 2.1 output.
Secondly, from studying Fletcher-Munson curves for dBA and dBC (may have spelling wrong), we can see that both extremes of frequency are elevated as the ear is exposed to greater SPLs. Relatively speaking. So I see it as the mid and increased sensitivity is actually compressed (again, I could be explaining or thinking a little in reverse).
Since each end of the bandwidth we hear is now compressed and higher in relative amplitude (dBC in comparison to dBA) then what affect does that have on spikes at those frequency extremes, and at higher SPLs?
I interpret that they too are compressed, and we are less sensitive to them.
I hope that answers what I think you asked of me.
(any misconceptions of mine, I would be grateful to be corrected on)
Greg
I'm not sure I follow you entirely, but ill answer as best I can.
Firstly, if we take a CD and replay via 24/96 system and encode to mp3, we don't gain any dynamic range, unless it is artificially 'expanded', nor does upsampling restore what is lost in the cd encoding. (barleywater, ronion, or wintermute should be able to correct me here if I make incorrect statements)
This is what is depressing about CD format. We need far more DVD encoded music CD, even just 2.1 output.
Secondly, from studying Fletcher-Munson curves for dBA and dBC (may have spelling wrong), we can see that both extremes of frequency are elevated as the ear is exposed to greater SPLs. Relatively speaking. So I see it as the mid and increased sensitivity is actually compressed (again, I could be explaining or thinking a little in reverse).
Since each end of the bandwidth we hear is now compressed and higher in relative amplitude (dBC in comparison to dBA) then what affect does that have on spikes at those frequency extremes, and at higher SPLs?
I interpret that they too are compressed, and we are less sensitive to them.
I hope that answers what I think you asked of me.
(any misconceptions of mine, I would be grateful to be corrected on)
Greg
The other things I'm wondering is which of these should have the least manufacturing variation? And do we know why the different boxes produce different distortion? Is it shape or rigidity or something else?
Thanks sharing it takes 5 seconds to hear why didn't make it to the finale round with out of box frq response low distortion driver or not, think will need to have frq response somehow tweaked a lot to be more natural sounding.
Regarding cheap DMM precision mostly they not build for AC measurements above 100-500Hz area.
Last edited:
@X,
The only way to measure drive voltage using a DMM is with sine wave at points, and a good true RMS fluke with a good bandwidth (such as the 28 or 87)
For real pink noise efficiency measurement I guess a scope is the only easy way
The only way to measure drive voltage using a DMM is with sine wave at points, and a good true RMS fluke with a good bandwidth (such as the 28 or 87)
For real pink noise efficiency measurement I guess a scope is the only easy way
Last edited:
One amazing result from these comparisons is that many of us will no longer have bright, forward and harsh sound in their homes.
What do you mean by too climby? Things like this is fine with my speakers. If you have a problem with vocal in clip 3, I believe you will have more problem here, as the peak is in the same frequency and higher in amplitude.
What do you mean by large? I don't know what a soft reverb is. But reverberation works in lower frequency. Below 800 Hertz. 200 Hz and 100 Hertz were audible.
HD by itself doesn't matter at these levels. 1kHz mixed with 2kHz at -46dB is indistinguishable from pure tone. Adding in additional harmonics at correspondingly tiny levels makes no change.
IMD is whole other story. Here is another view of burst frequency response plot from my FAST TC9FD thread:
As seen, it has 300Hz, 2kHz, and 3.5kHz components. Voltage peaks of burst signal are 28.9 and -26.1 volts. This asks much more from driver than X's levels used for HD plots. Cursor in this view is at 4kHz with peak -46.2dB to 2kHz peak. Excluding fundamentals, two largest peaks are 1.7kHz at -35dB and-34dB respectively to 2kHz. These have no harmonic relationship to 2kHz and are not readily masked by hearing mechanism. This is readily demonstrated by constructing two tone signal with 2kHz and 1.7 kHz mixed in -35dB. Result is readily distinguished when I listen with earbuds. Presence of the low level 1.7kHz signal adds a grainy fuzz to the sound.
Much of what X's various mountings may show are changes in firmness and continuity of contact between driver flange and baffle. Holding TC9FD in hand and tapping/rubbing finger at different points along flange produce a lot of different hollow resonating sounds. In mounting for this driver 3/32" steel adapter ring was cut and driver epoxied to it. This reduces the difference in sound when tapping on driver flange at thinnest point v tapping on flange around one of the ears with mounting holes, but does not eliminate it.
10f v FU10RB: Tapping about driver flanges of these produce much more uniform results than with TC9. Although both have cast frames, the 10f's is lighter/skimpier with flat flange. The FU10RB frame has heavier flange with tapered thickness. Tapping on 10f at mounting hole produces different sound than at points elsewhere along flange. 10f has mounting holes between basket arms. FU10RB has no readily discernible difference in sound when tapping at different points around the mounting flange, and mounting holes are aligned with basket arms.
Not surprisingly, TC9, 10f, and FU10RB produce different sounds when lightly tapping centers of dust caps with fingernail.
IMD is whole other story. Here is another view of burst frequency response plot from my FAST TC9FD thread:
As seen, it has 300Hz, 2kHz, and 3.5kHz components. Voltage peaks of burst signal are 28.9 and -26.1 volts. This asks much more from driver than X's levels used for HD plots. Cursor in this view is at 4kHz with peak -46.2dB to 2kHz peak. Excluding fundamentals, two largest peaks are 1.7kHz at -35dB and-34dB respectively to 2kHz. These have no harmonic relationship to 2kHz and are not readily masked by hearing mechanism. This is readily demonstrated by constructing two tone signal with 2kHz and 1.7 kHz mixed in -35dB. Result is readily distinguished when I listen with earbuds. Presence of the low level 1.7kHz signal adds a grainy fuzz to the sound.
Much of what X's various mountings may show are changes in firmness and continuity of contact between driver flange and baffle. Holding TC9FD in hand and tapping/rubbing finger at different points along flange produce a lot of different hollow resonating sounds. In mounting for this driver 3/32" steel adapter ring was cut and driver epoxied to it. This reduces the difference in sound when tapping on driver flange at thinnest point v tapping on flange around one of the ears with mounting holes, but does not eliminate it.
10f v FU10RB: Tapping about driver flanges of these produce much more uniform results than with TC9. Although both have cast frames, the 10f's is lighter/skimpier with flat flange. The FU10RB frame has heavier flange with tapered thickness. Tapping on 10f at mounting hole produces different sound than at points elsewhere along flange. 10f has mounting holes between basket arms. FU10RB has no readily discernible difference in sound when tapping at different points around the mounting flange, and mounting holes are aligned with basket arms.
Not surprisingly, TC9, 10f, and FU10RB produce different sounds when lightly tapping centers of dust caps with fingernail.