A Subjective Blind Comparison of 2in to 4in drivers - Round 4

[xrk971]

I would expect SL to laugh at our method of evaluation, and lecture us on the lack of validity of any subjective test. Then he might point out the weak links in this particular subjective "experiment". (sorry x)

Round 5, if there is one, really needs to be objective, as in no opinions, just scientific facts. FR we know, step response would be nice (i saw it posted for two of the drivers). Differential gain would be nice (how gain varies across positions of the diaphram - overload compression and such; does it soft clip or hard clip mechanically? symetrical? What percentage of the mechanical dynamic range is fairly linear?). Ringing from gaussian envelop tone bursts would be nice (the right way to do a CSD - Linkwitz talks about this on his website). Harmonic and I.M. distortion over frequency at several levels would be nice. A test for "resolution" of the cone material too (?). Some of these tests are somewhat redundant with each other. Just brainstormin' here. I'm sorry I don't have the ability to contribute any of these things with what I have at my disposal.

That's a lot of work Bob. Remember I do this as hobby on the side - it would be a full time job to do what you describe. The progenitor to these "Subjective" threads was the "Objective" thread that caused a lot of heated debate and the thread was shut down. I think we need to keep it subjective otherwise perspective may be lost again and it's work and not fun anymore.

I plan on adding HD measured at equivalent to 90dB at 1m.

 

[BYRTT]

Excellent choice BYRTT!!
If I would get an 8" or 10" woofer for a FAST I think I would also get one of the discovery woofers. And I know Troels loves the ScanSpeak Discovery line (like his most recent 4way). I would basically build this speaker but with the 10f/8414 switched to the 10f/8424 and no tweeter
ScanSpeak-3W-Discovery

Thanks hope same cone material pay back in some way, wesayso got a pair 10F too and suggested 10F to have some wave guide ala for tweeters so seems a bigger cooperation started when X suggested he can 3d print.

Other way around excellent taste you pick TG's 3w Discovery, that one X and i had turned around tens of times half a year back and think X was very close import a kit driver/XO from Jantzen Audio shipped to US. X was all in for the 3 way i had FAST in mind as you, and have a laugh look the edited picture below i did 23 of Marts : )

 

[xrk971]

Ha ha ha - nice photoshop skills there Byrtt. I forgot all about that picture. It still looks good. Something about black MDF as a baffle material. That ATC look is timeless.

I may have to ask Jessman to make that box for me yet.

 

[jeshi]

The spread is pretty ambiguous. From a quick glance it looks like it confirms what many may suspect; vifa/typhany/scan/SB all know what theyre doing.

that is because they are all designed by the same engineers (scanspeak or ex-scanspeak engineers).
http://www.diyaudio.com/forums/full...2in-4in-drivers-round-4-a-25.html#post4448894

There was a bit of convoluted history in there (History of Scan-Speak).
http://www.scan-speak.dk/news/20090421a.pdf
and at some point for a short time (2006-2009) Tymphany did own ScanSpeak, but ScanSpeak drivers were always manufactured in Denmark. Then in 2009 there was a split where the Vifa/Peerless brands and designs remained with Tymphany in china and Tymphany sold off ScanSpeak (high end designs, danish engineers, Denmark manufacturing) and it came under new Danish management. Also the TC9/TG9 dates back before the split, so they were almost certainly designed by ScanSpeak engineers.

But the ScanSpeak Discovery 10f/8424 is a newer design appearing after the 2009 split so it can be considered the evolution of the TG9. But all Discovery drivers are manufactured in Denmark.

And as an aside, when Tymphany aquired ScanSpeak in 2006, some engineers were not happy and left, forming a new company called SBAcoustics. SB Acoustics :: Home

another part of the history I was a little unclear about until I read these articles, was that Vifa bought scan-speak in 1989, so for many of us, from 1989-2006 vifa/scan-speak were one company

 

[Bob Richards]

That's a lot of work Bob. Remember I do this as hobby on the side - it would be a full time job to do what you describe. The progenitor to these "Subjective" threads was the "Objective" thread that caused a lot of heated debate and the thread was shut down. I think we need to keep it subjective otherwise perspective may be lost again and it's work and not fun anymore.

Yes, it would be a LOT of work. Luckily, I think any of these drivers would be fine. Past a certain point, the recording quality will be a limiting factor. My latest batch of speaker systems sound so good to my ear, that I honestly don't care if something could be better. Sorry if that sounds arrogant. Now I've got to give more time to the allegedly pending earthquake, the Fukushima radiation, global warming weather issues, rents getting ridiculously high, forest fires, etc. etc.. Where will it be safe and affordable to continue my audio engineering madness?

 

[xrk971]

Yes, I have a reference all horn-loaded system that sounds da'best too but can't seem to stop building more speakers.

 

[Godzilla]

>> Hi Godzilla,
>> Thanks - I missed your post. Glad you are enjoying the TC9's in your OB's. They sound
>> wonderful with a good woofer supporting the low end. What is your crossover point and
>> are you running active or passive XO?

Here is the simulated response (using XLBaffle) on a 19"x19" baffle about 20" high (on top of H-frames). Looks like there's a natural rolloff just below 200Hz. This way I don't have to use any circuits - which I prefer whenever possible. The H-frames are connected to a typical sub amp and blended into the mains to taste. Everything sounds great.

If you look carefully the TC9's relatively high Qts of .89 adds a bit of a bump around 200Hz giving it a bit of a fuller sound than the other drivers. I've been swapping the Vifa, Tangband and Eminence 12LTA for months. They all sound very good. Clearly the Vifa at under $15 is the value leader and it's certainly not trounced by the other much more expensive drivers.

 

[Jay]

Round 5, if there is one, really needs to be objective, as in no opinions, just scientific facts.

Is it possible to write a program that will rank the drivers and to be installed in each "voter"'s computer?

The program will read the frequency response from the DSP output, then measure the plus minus sound pressure deviation from XO point up to 10kHz, or is it 6kHz? then put the result in Var_FR double precision variable.

Then do certain algorithm to pack all the variables related to impulse and step response into a single double precision variable Var_IR.

Then do another algorithm to get similar variable for the other parameters.... The winner will be based on the maximum number of k1*Var_FR + k2*Var_IR + ... + k(n+1)*Var_x. To calculate the values for the constants ki we can use voting/polls.

FR we know, step response would be nice (i saw it posted for two of the drivers). Differential gain would be nice (how gain varies across positions of the diaphram - overload compression and such; does it soft clip or hard clip mechanically? symetrical? What percentage of the mechanical dynamic range is fairly linear?). Ringing from gaussian envelop tone bursts would be nice (the right way to do a CSD - Linkwitz talks about this on his website). Harmonic and I.M. distortion over frequency at several levels would be nice. A test for "resolution" of the cone material too (?).

 

[Jay]

Clearly the Vifa at under $15 is the value leader and it's certainly not trounced by the other much more expensive drivers.

Much more expensive drivers may probably have much lower Qts. You cannot put them as is in an open baffle like the cheap Vifa.

 

[Godzilla]

>>> Much more expensive drivers may probably have much lower Qts. You cannot put them as is in an open baffle like the cheap Vifa.

You can and I do. But I recognize the advantage of higher Qts on open baffle when reaching for lower octaves. Remember, I'm using the baffle size to create the low frequency rolloff. Making the baffle 17"x17" creates a rolloff beginning around 250Hz. Smaller baffles produce higher rolloffs without concern of Qts.

Even doubling the size of these baffles will not produce more bass with the Eminence and TB. The higher Qts Vifa may extend a bit lower. Regardless, they all need bass support.

 

[xrk971]

Looks like there's a natural rolloff just below 200Hz. This way I don't have to use any circuits - which I prefer whenever possible.

That's good, but you subject the full range driver to unnecessary cone movement in its attempts to create bass, which creates higher levels of distortion. It sounds a lot cleaner and more clear when hi passed filtered.

 

[xrk971]

Waterfall Plots (CSD)

Here are the measured CSD plots for the drivers. Settings for the CSD are 31 slices, 5ms range, 5ms window, 0.1ms risetime.

PS95-8:

TG9FD:

TC9FD:

10F/8424:

B80:

The TC9FD looks pretty good.

 

[Barleywater]

Can't wait to see what you're planning with these. The Neo magnets are something else compared to the TC9. You could build a magnetic hovertrain with them .

With 10f I plan to start by measuring it as naked dipole for polar response and burst testing to see how low it can go without gross IMD.

Target is LXmini with DSP to get cardioid down to 500Hz, hopefully a bit lower depending on IMD.

Previously I've used Seas Prestige FU10RB H1600-04, SL's choice, with DCX2496 for crossover, EQ, and timing. First I cloned response of LXmini using 3" diameter rear damping chamber as laid out by SL. Then I optimized with 4" diameter rear damping chamber to reduce and smooth rearward pattern; enabling speaker to be placed closer to front wall. Either implementation sounds very good, but power handling is no better than my Pluto clones with Peerless 830970. With optimal listening triangle in listening space for Pluto I feel imaging is better with Pluto, than with LXmini clone with same listening setup. Growing the listening triangle the LXmini comes into its own for imaging, but SPL is not quite enough. Pluto with same bigger listening triangle suffers for a little more SPL, but room artifacts impinge with a certain sameness added to phantom image.

Imaging with Cardioid as sum of monopole and dipole speakers: using Vifa NE180W-04 as full range is amazing. Suppression of sidewall and front wall reflections seem to be big player, but sweet spot is too small due to HF beaming. With Vifa driver cardioid response down to about 200Hz is achieved, once again with SPL too limited. I could limit Vifa down to 500Hz, but seems a waste of driver potential, and beaming HF would remain.

10f has much lower Qts than FU10RB, but claims somewhat greater linear excursion, and almost double mechanical excursion limit. Burst testing will reveal how much of the excursion potential is useful.

Fallback is using 10f in FAST/Pluto that should outperform 2" Peerless and TC9 in this role. I haven't sealed the FU10RB in a FAST setup.

 

[Bob Richards]

>>> Much more expensive drivers may probably have much lower Qts. You cannot put them as is in an open baffle like the cheap Vifa.

You can and I do. But I recognize the advantage of higher Qts on open baffle when reaching for lower octaves. Remember, I'm using the baffle size to create the low frequency rolloff. Making the baffle 17"x17" creates a rolloff beginning around 250Hz. Smaller baffles produce higher rolloffs without concern of Qts.

Even doubling the size of these baffles will not produce more bass with the Eminence and TB. The higher Qts Vifa may extend a bit lower. Regardless, they all need bass support.

I know Qts is important info when designing a closed box volume or ported box stuff, but I'm not convinced that it plays a big part in the case of open baffle. The air impedance is much higher and the voltage source poweramp tells the cone exactly where to be with substantial muscle. With open baffle I think cone surface area and Xmax will be the dominant issues to worry about. "Baffle step" with open baffle requires a 6dB/octave slope (not shelf) for a flat on axis response. My main system is a tri-amp'd open baffle with active EQ and 4th order LR crossover. It measures flat from 20 - 20kHZ at the listening position.

 

[xrk971]

Harmonic Distortion at low SPL (74dB at 1m)

This probably is not much use as the levels are so low that drivers are in general very clean. If I have time I will remeasure a few like B80 and 10F at 90dB and 1m. For next round I will keep this in mind and collect it.

PS95-8:

TG9FD:

TC9FD:

10F/8424:

B80:

 

[Bob Richards]

Previously I've used Seas Prestige FU10RB H1600-04, SL's choice, with DCX2496 for crossover, EQ, and timing.

Gary Pimm at Welcome to Gary Pimm's DIY Page (a local guy here in Portland who works at Tektronix and has designed many very high end audio amps and speakers) put up a webpage years ago (not sure if it's still up) about the rather poor analog electronics in the Behringer DCX2496 (before the ADC and after the DAC). The input section has a non-biased polarized electrolytic cap in series with the input lines. An unknown opamp drives approx. a 300 ohm load. The circuit topology is low end (DAC puts out a balanced signal, but circuit uses a crappy differential receiver circuit). He claims that after re-designing all the analog circuitry, it sounds significantly better.

 

[vulejov]

This probably is not much use as the levels are so low that drivers are in general very clean..

THD+CSD very similar to Klang+Ton mesurements..
Three best drivers still are 10F, B80 and TC9.. exactly in the same order.. nothing has changed..
B80 has only one problem, poor 3rd harmonic..
Here is B80 with AL170, 3rd order xover..

 

[BYRTT]

.....Here is B80 with AL170, 3rd order xover..

Thanks keep and eye round five X announced FR88EX and new US developed 3,5" driver produced by Alpine.

 

[xrk971]

I just finished taking data for the Alpine for Round 5. Just the TB left to go.

Drivers so far may include:

1. Peerless NE65W, aluminum cone Nd motor
2. Tectonic Elements TEBM46C20N-4B paper cone BMR Nd motor
3. Vifa TC7FD00-04 paper cone ferrite motor
4. Fountek FR88EX aluminum cone Nd motor
5. Fountek FR58EX aluminum cone Nd motor
6. Alpine 3.5in OEM full range for Jeep Wrangler, poly carbon fiber cone Nd motor
7. Tang Band W2-852SH paper cone Nd motor

That's a lot of drivers and may be too much for easy voting. I might trim this down to essential drivers.

 

[vulejov]

..keep and eye round five X announced FR88EX..