Mini DIY

[johnaec]

Mine are finally all trimmed down, (for some reason, yours were much easier, Zilch - I think my pieces may have warped a little bit sitting around so long waiting for assembly). They're now ready for black paint on the fronts, finishing, and filling out. 'Coming soon...

John

[Zilch]

Well, you're gonna hafta go into production on cabs, John, 'cause I have two spare sets of woofers, tweeters, and crossovers, and plenty friends and relatives wantin' their own "Minis" now.

[I COULD rework these AR4xs, I guess, but they're not mirror-imaged, and they lack Mini-panache.... ]

[Zilch]

Comparison of nearfield LF response of the two crossover topologies.

HTI-8 18 dB/octave filter buries the 3.5 kHz peak -30 dB.

So much for L100 midrange voicing, tho.

Well, maybe there's a wee bit left -- still got the midbass bump....

AutoEQ says I'm 4 dB too "hot" at 800 - 1 kHz, bottom.

Zobel tweaking time, maybe.

[Purty close, tho, +/- 2.5 dB overall. Normally, I'd quit, but I can hear the difference on these.... ]

[Earl K]

Hi Zilch

HTI-8 18 dB/octave filter buries the 3.5 kHz peak -30 dB.

- I haven't been paying close enough attention I guess / but what is a HTI-8 network ? & where is the schematic ?

Maybe Earl'l tell us what's going on there.

- I'm lost in the ozone here / hard to offer any insights , since I don't even know the topic .

- The above curve looks " nice " . Where did the 3K peak go ? ( Or was that the question that I was supposed to guess an answer to ? )

- Anywho,,, as the AM series of networks clearly indicate / one can use a series LCR ( strapped in parallel ) after the main LC elements. In that position ( posterior position ) , one will get the expected "notching" . The "R" value is dictated by 2 things; the impedance one finds in that part of the circuit ( I call it the "loop" impedance / it needs to be measured or calculated ) and the amount of desired attenuation. Using the simplified db 20 Log formula for impedances ( which you gave me ) / one can get a good ballpark "R" figure to start tweaking with ( or one can guess at it easily enough ) .

[Zilch]

- I haven't been paying close enough attention I guess / but what is a HTI-8 network ? & where is the schematic ?

Heya, Earl.

HTI-8 is an 8" woofed JBL Consumer in-wall (i.e., closed box) home theater speaker. The similar Pro version is Control 128W, which uses the same crossover network, but has slightly different woofer parameters. I linked to the schematic here:

http://audioheritage.org/vbulletin/s...0&postcount=97,

and built up a pair to install at #105.

It looks like standard three-pole plus LF Zobel topology to me; I was hoping you could offer insight as to the humps and dips both sides of the 2.5 kHz crossover frequency there: Chebychev?

In any case, in switching to 2.5 kHz crossover point from 3 kHz and using steeper slopes, this buries the 3.5 kHz bump in 116(H-1) response effectively without a notch filter, so I'm now looking at what additional tweaking to consider. It's playing about 4 dB too hot at 800 Hz - 1 kHz. See GEQ curve after AutoEQ, above.

I've got the fixed HF padding dialed in pretty well, looks like. There's just 1.5 dB excess with the adjustable pad full on, leaving a little headroom for an absorbent listening space. I may give it just a bit more.

*****

Jeez, these sound good.

I'm playing them with LE14H-3 subs right now....

*****

Edit: Zobel tweak'll do it, looks like (blue). Just hafta make an "empirical" determination of the optimum values, now.

[Wearin' out the T-nuts on these little boxes tryin' all this stuff.... ]

[Earl K]

Hiya,

It looks like standard topology to me; I was hoping you could offer insight as to the humps and dips both sides of the 2.5 kHz crossover frequency there: Chebychev?

- Oh,,, the peaking . Anytime I see that, I assume the filters' poles have come closer together than what a typical Butterworth transform calls for .
- IOTW ; a bump filter has been created.
- In a 2-pole setup , one can't bring the poles closer together than an octave spacing before peaking/resonance/gain starts to occur.
- ( Bessel = 1.5 octaves / LR = 2 octaves in pole spacing ).
- I had wanted to talk about this in my thread called "Bump Filters" ( or something like that ) but it didn't go anywhere. All the old stuff that uses 2307 horns have significant "Bump Filters" in them / artificially supporting the lower end response of those weenie horns ( or so say my calcs. ) . This does have relevance to you guys wanting to "squeak" out a bit more lower end response from those PT wavequides ( even the 2344/4430 "cheated" a bit in this regard ) .

- I've got to go now , it's BBQ season .

[johnaec]

OK - first little teaser stuff. Cabinet fronts are finally painted, but I still haven't decided whether to stain the cabinets or leave them natural for that "modern" look. They'll have JBL Blue grilles either way.

I'm going to try the 6.5" 706G-1 LF speakers from the L1 series first. I attached a couple pics showing the difference between these and the 115H-1's, which are my other option. As seen in the pics, the 706G-1 has a ribbed paper cone, and its magnet is also about 1/8" deeper than the 115H-1, (magnet diameters are the same). Both speakers use the same mounting and bolt holes.

The last pic shows one of the cabinets with only the drivers mounted, (excuse the dust!). I still need to mount the crossovers and binding posts Zilch provided, install the port tubes, attach the foilcals provided by JohnW, and assemble and mount the grilles, (and decide on the finish).

I'm going on vacation starting next Thursday for about a week and won't get anything more done on them before I return - 'too much other stuff I need to do before leaving. 'Just wanted to provide an update.

John

[Zilch]

OK - first little teaser stuff.

Way COOL, John!!

[Minis are proliferating.... ]

[Zilch]

Top:

1) Nearfield LF response before tweaking (red).
2) Zobel compnents adjusted per sim, above (green). It did just what the Spice sim said, between 400 Hz and 2 kHz.
3) Full range response afterward (violet).
4) For those thinking to use 2118 (orange) before tweaking, but it's a "J" so ignore the low pass response at the high end there.

Second:

1) Full range RTA response 4 Pi freefield (violet).
2) Same, 2 Pi, about 4" from wall (white).

So, useable bass response (-10 dB) is good to ~25 Hz.

[It's down there, but it ain't real HUGE. ]

Remaining 3.5 kHz peak is apparent, but it's clearly not (much) coming from the woofer anymore. :dont-know

[I'll see what AutoEQ says after installing the mods in the second speaker....]

Edit:

Third: AutoEQ result after adjusting the Zobel on both. Now +/- 2 dB. It sees the remaining HF peak as 3 dB between 3150 Hz and 4 kHz.

Bottom: RTA with AutoEQ (red) and with no EQ (orange), raised 10 dB.

I'm about to invoke the DIY Credo on these: "Hey, that sounds good enough to me. My system is done dude!"

[Just 'til I figure what else to try, of course.... ]

[Zilch]

My 8" woofer, closed-box Minis went to Mr. Widget's for an audition today. Results, playing freespace:

1) Major points for cuteness.

2) Bass has good tonality.

3) A midrange peak.

4) 2404H tweeter is a bit harsh.

Overall: Fun....

[Mr. Widget]

My 8" woofer, closed-box Minis went to Mr. Widget's...

Zilch had asked me to comment, but I was AWAL earning a living... or attempting same.

We put the little Mini's on makeshift stands next to a pair of stand mounted mini-monitors that I use in my "upstairs" system. We fired them up and it was surprising how tuneful the bass was. I doubt it would satisfy any fans of organ music or contemporary electronic music, but for "normal" music the low end sounded flushed out, had a nice snap to it, and sounded "right".

I thought the mids were a bit harsh and in your face. They were also not terribly revealing or detailed... more or less what you would expect using a doped cone 8" woofer up to 2-3KHz.

I have never been a fan of the 2404 and sure enough the VHF was missing, there was a lack of a sense or air or space, and what was there was a bit hard and abrasive. Comparing it to an aluminum domed direct radiator tweeter in my mini-monitors, even the designer commented on the harshness.

Since "cute" is the primary design goal, the option of changing the tweeter which could remedy most of the system's sonic short comings isn't an option. It'll be nice to see them in walnut with some JBL blue grilles.


Widget

[Zilch]

Thanks for your review, Widget!

I'll work more on the mids; Audivex gave me some suggestions today.

But we're SOL on the tweeter 'til I find a small Ti dome or re-entrant Mylar to mate with the 2404 horns.

[Zilch]

O.K., fixed the midrange hump per Audivex suggestion:

1) Untweak the Zobel

2) Add notch filter

Tried parallel notch first, but additional inductor impedance in series with the woofer pulled down the entire midbass response. Sounded like L100. Kinda nice, actually, but puts the woofer more at risk.

Series notch (in parallel with the woofer) levels out the response (blue). Zobel's gotta stay, looks like.

See post #110, above. Same game; bigger and lower than what I could achieve tweaking the Zobel, is all....

[Zilch]

Johnaec provided an additional closed Mini box for testing of 8" woofers. Kinda backwards working with a fixed box size, but that's the project.

Here's the nearfield LF response of the variety I have tried here:

1) 116As and 708G-1s (Org, Lt. Grn)

2) 116H-1s (Red, Grn, Gry,) LE8s (Blu, Vio,) and LE8T-Hs

3) D216s, 2118J, Control 128W and HTI8

Face-mount D216s were, uhmm, problematic.

The in-box numbers (measured with WT2) tell the tale:

Fc, Qtc

116A = 71.51 Hz, 1.03
116H-1 = 74.28 Hz, 0.94
708G = 69.26 Hz, 0.84
LE8T-H = 82.20 Hz, 0.95
LE8 = 85.77 Hz, 0.90
D216 = 110.33 Hz, 0.57
2118 = 111.23 Hz, 0.70
128W = 73.79 Hz, 0.82
HTI8 = 72.51 Hz, 1.00

I am reporting the measured Fs and Qts of the woofers mounted in the box as Fc and Qtc here. It takes less than 5 minutes to do the measurement on each using WT2.

Dickason characterizes Qs in the vicinity of 1 as having "a warm, if somewhat robust, quality that audio marketing people describe as 'saleable.'"

"Lower Q values around 0.8 sound more detailed (due in part to improved transients) and, by comparison, somewhat shallow. Qtc = 0.5 is usually regarded as excessively taut and overdamped. Some authorities, however, still consider this value (0.5 - 0.6) to be optimal."

Use 708G-1 or 128W for the somewhat flatter bass response of Qtc just above 0.8. LE8(T), too, but they don't play as low, by ~10 Hz. See the first set of curves comparing 708G-1 vs. 116A for the difference between Qtc 0.8 and 1.0 presented graphically. Yes, the 116A's 2 dB "hump" at 100 Hz. in this alignment matters, and yes, the difference in bass response is really, really audible. Think L100.

My personal main pair are currently running 116As with the notch filter posted above to tame the mids (400 Hz - 1.2 kHz) a bit. They've got good extended bass with 2 Pi boundary reinforcement; no sub required. Note that these have the highest Qtc of all those tested. Minis should play "Big," I say.

Zilch's DIY 8" closed-box mini builder recommendation (bottom): HTI8 with rubber-surround polymer-coated aluminum cone, available lately in pairs from Harman on eBay for under $100 including crossovers ($549 on their website, $579 list.) L-pad down the HF ~12 dB for 2404H(-1). You get two bonus Ti Lam Dome tweeters (Made in China,) and plastic in-wall frames to play with.

Harman lists them with a space before the "8", so search for "HTI 8". There's typically two one-day auctions running continuously. Check completed auctions to see what they've been going for....

[Note: HTI8 requires a slightly larger than JBL standard baffle cutout, and mounting bolt locations are different as well. Pay particular attention to achieving a good seal to the limited available flange area between screws without warping the stamped frame; Parts Express #269-300 sealing caulk will do it.]

[Zilch]

May be hard to find HTI8 in all of that, so here it is separately:

1) MLS nearfield response of two HTI8 woofers, blue and light blue. Unit #2 on JBL factory HTI8 filter, red. Midfield response at 1M, grey and orange. Filter knocks the 5 kHz peak down 30 dB.

2) Same midfield on filter, but measured sinusoidal instead of MLS. Could use a notch between 400 Hz and 1 kHz, a familiar tweak from above.

3) RTA of full system response with 2 Pi boundary reinforcement. Less boomy than 116A version, but still good down to 35 Hz and below.

[Not that I mind Mini boom, of course.... ]

Edit:

4) With notch filter, nearfield (red) and midfield (orchid).

Peak at 300 Hz is today's room thing, looks like.

[What Zilch did today.... ]