JBL Lancer 77's I Am Impressed!

We're having a similar discussion in an AR2ax tweeter upgrade thread right now. It's often either cabinet modifactions or custom flanges to make modern tweeters fit, mechanically, but the REAL issue is performance objectives.

http://audiokarma.org/forums/showthread.php?t=132936

If the desire is to restore the original sound, use the original or known equivalent components. If the objective is to experiment or upgrade, recognize that you're doing a speaker redesign, in fact, and approach it from that perspective using the appropiate toolbox.

I suspect everyone recognizes by now that, generally speaking, I don't give one whit about restoring original performance unless that's the specific intent. In fact, whenever I look in detail at vintage designs using contemporary tools, they rather suck, and they sound bad, too, by comparison.

With respect to tweeters, specifically, I've looked at a Tang Band (for AR4x,) a Fostex, and now a Morel, and have prior experience with JBL 035tiA, Crites, Eminence, and others. Bottom line: I can make any of them sound good or great, even, but it's not a slam-dunk plug-'n'-play retrofitting modern tweeters into these vintage systems, rather, each of them is a different, and more comprehensive, loudspeaker design project.... :yes:
 
Sidebar: Quality L-Pad Impedance

PE's 8-Ohm, 100W L-Pad.

1) Baseline using Mills 8-Ohm resistor as load, and,

2) With Morel DMS-37.

Settings by the clock:

Red = 5 - No attenuation
Org = 3
Grn = 1
Blu = 11
Vio = 9
Cyn = 7
Gry = Off, max attenuation.

The impedance "seen" by the filter @ 2.3 kHz is somewhere between 6.6 and 10.22 Ohms with the DMS-37, depending upon the setting.

With the purely resistive (well, for the most part,) fixed resistor, it varies between 8.10 and 10.71 Ohms....
 

Attachments

  • DMS-37 Impedance Baseline.jpg
    DMS-37 Impedance Baseline.jpg
    48.4 KB · Views: 16
  • DMS-37 & L-Pad Impedance Study.jpg
    DMS-37 & L-Pad Impedance Study.jpg
    57.1 KB · Views: 20
PE's 8-Ohm, 100W L-Pad.

1) Baseline using Mills 8-Ohm resistor as load, and,

2) With Morel DMS-37.

Settings by the clock:

Red = 5 - No attenuation
Org = 3
Grn = 1
Blu = 11
Vio = 9
Cyn = 7
Gry = Off, max attenuation.

The impedance "seen" by the filter @ 2.3 kHz is somewhere between 6.6 and 10.22 Ohms with the DMS-37, depending upon the setting.

With the purely resistive (well, for the most part,) fixed resistor, it varies between 8.10 and 10.71 Ohms....

What you found regarding the so-called constant (not) impedance of affordable L-pads is the norm IME. I pulled three 8 ohm L-pads out of my junk box and made some quick meas. One was old Japanese unit salvaged from a 60s speaker, one the PE 50W unit and the third a Fostex 50W unit with supposedly better constant impedance performance. Load was Dale NI resistor and all meas with DC ohmmeter. Results:

Rotation____Jap____PE____Fostex
CCW_______6.9____7.5____7.6
1/4________9.1___10.0____10.0
1/2 _______8.8____9.5_____9.5
3/4_______ 7.6____7.9_____7.9
CW________8.1____8.1_____8.1
Max________9.3___10.0____10.1
Min________6.7____7.1_____7.1
Range______2.6____2.9_____3.0

As you can see, results are similar to yours and certainly not as constant as one is lead to believe. Also interesting is that the Fostex unit, for which one pays a premium, is no better than the 1/2 cost PE...in fact they are EXACTLY the same unit in every respect. The Fostex does come with a nicer escutchen and knob and the packaging is neater but...

In the overall scheme of things the deviations from 8 ohms don't seem to be life threatening but if for whatever reason, precision is in order, a switch type attenuator looks to be the only reasonable alternative.
 
Thank you, Steve, for reporting your similar findings. :thmbsp:

What are the implications in actual use, then? The reason I did this little investigation was because I could see the response varying with the attenuation, and worse, that mostly occurring in the typical adjustment range. I did not expect that, and wanted to know what was going on.

The min and max I found with the DMS-37 certainly impacts the actual two-pole filter I'm using with it, below. The filter "Q" goes up by 3 dB at the 2.3 kHz crossover frequency, and more, slightly higher, at ~4 kHz. ACK!

So, what's to be done? Switched resistor or autotransformer attenuation? I'm not above using either, but in this case, there's several dB of sensitivity I can give up, so it's resistive isolation between the filter and the L-Pad, with the result shown in the second graph, where the variability is now reduced by half, and the effects of inductive reactance of the driver itself also substantially mitigated.

I'm calling it "Good enough" for now -- schematic below.

Maybe we should go into business making L-Pads that actually maintain a constant impedance.... :yes:
 

Attachments

  • L77 Filter DMS-37 Impedance Variation.jpg
    L77 Filter DMS-37 Impedance Variation.jpg
    35.8 KB · Views: 39
  • DMS-37 & L-Pad Impedance Study 2.5S 20 PF.jpg
    DMS-37 & L-Pad Impedance Study 2.5S 20 PF.jpg
    52.2 KB · Views: 29
  • DMS-37 L44 Circuit.jpg
    DMS-37 L44 Circuit.jpg
    15 KB · Views: 103
Thank you, Steve, for reporting your similar findings. :thmbsp:

What are the implications in actual use, then? The reason I did this little investigation was because I could see the response varying with the attenuation, and worse, that mostly occurring in the typical adjustment range. I did not expect that, and wanted to know what was going on.

The min and max I found with the DMS-37 certainly impacts the actual two-pole filter I'm using with it, below. The filter "Q" goes up by 3 dB at the 2.3 kHz crossover frequency, and more, slightly higher, at ~4 kHz. ACK!

<snip>

Yet another case of "never assume". What I find disturbing is the fact that the response variations introduced by the commercial Lpad imperfections overwhelms the effort I've expended in evening out impedance variation in the HF section of my Carmel project.


<snip>
So, what's to be done? Switched resistor or autotransformer attenuation? I'm not above using either, but in this case, there's several dB of sensitivity I can give up, so it's resistive isolation between the filter and the L-Pad, with the result shown in the second graph, where the variability is now reduced by half, and the effects of inductive reactance of the driver itself also substantially mitigated.

I'm calling it "Good enough" for now -- schematic below.

Maybe we should go into business making L-Pads that actually maintain a constant impedance.... :yes:

I really believe that the switched version of an Lpad is the only practical alternative. Although the transformer is in vogue at the moment, because it actually transforms impedances while transforming levels, it only worsens the non-constant impedance situation. It seems that the mfg of true constant impedance Lpads of the continuously varible variety is fiendishly complicated given the non-linear resistors reqd. Easy with composition resistor materials but not so easy with high power WW materials.

BTW...Curious how you came to use the Morel DMS-37 HF driver (shielded version of the MDT-37) in a vintage speaker upgraded. Wasn't someone recently pondering the cost/benefit aspects of using a $65 part in an upgrade project? ;)
 
BTW...Curious how you came to use the Morel DMS-37 HF driver (shielded version of the MDT-37) in a vintage speaker upgraded. Wasn't someone recently pondering the cost/benefit aspects of using a $65 part in an upgrade project? ;)
It was the waveguide, of course! :p:

Well, the element of constant directivity evident in the response curve in comparison to most dome tweeters, and really to determine for myself whether the Morels are "all that." As I understand it, the purpose of domes is max dispersion, which is inconsistent with my view of how HF should be done, i.e., uniform power response.

The whole tweeter thing is daunting -- there are literally hundreds to choose from, and not much in the way of information other the opinions. Seems like Parts Express took down their own measurements of their offerings, which looked very much like the results I get in actual use, i.e., not one heck of a lot like the manufacturers' published curves, which may be the reason, actually.

I chose the shielded version because the alnicos in L77 (and others) are inherently shielded, offering the prospect of use in proximity to CRT TV, which seems an appropriate application. Even though it's all going planar and mooting the issue, I suspect our big old Sonys will be around a bit longer.

With respect to the economics, y'all convinced me, is what. I don't even have any L77s, but my new PE "AR4x project" cabinets are now in, and I'm deciding what load to put in them; the Morels are a top candidate. I'd put them in the L44s they're right now playing so splendidly with, but they won't fit. I'm leaning toward using the little Tang Bands in those. :thmbsp:

[The original AR4xs are getting HORNZ! Heh, heh.... ]
 
Yet another case of "never assume". What I find disturbing is the fact that the response variations introduced by the commercial Lpad imperfections overwhelms the effort I've expended in evening out impedance variation in the HF section of my Carmel project.

I really believe that the switched version of an Lpad is the only practical alternative.
There's documentation of the "backwards" fixed L-Pad I implemented here somewhere on line, but finding it among my bookmarks has thus far been fruitless. I tried both ways, and this produced the better result. Perhaps you have some information on this approach? :dunno:

You can see I'm now operating around 7 Ohms nominal. If I were willing to give up a bit more sensitivity, I could move that up closer to 8 Ohms by changing R2 to 3.5 or 4 Ohms, and reduce the variability somewhat more. That'd require reworking the filter again, though, so I'm leaving it be for now.

Whether it's 7 Ohms or 8 Ohms is somewhat of an academic issue. With just the driver, it was 6.5 or 6.6 Ohms, depending upon which one of the pair I used.... :yes:
 
L77 Baseline Measurements

Test samples courtesy AK member Mxlews.

1) Gated sinusoidal full-range frequency response, two units, set to Med attenuation.

2) Windowed MLS quasi-anechoic high frequency response of Unit #1, showing the three switched attenuation settings.

3) Same for Unit #2.

4) Unit #2 with driver phase reversal.

LE10A woofers run full range, an apparent sub-optimal arrangement according to these results:

http://manuals.harman.com/JBL/HOM/Technical Sheet/L77 Lancer ts.pdf

Useable bass response (-10 dB) to ~40 Hz.

LE20-1 tweeter performance is rolled off above 10 kHz, as expected and documented earlier in this thread.

Foxtex or Morel? Woofer response measurements may dictate the choice....
 

Attachments

  • Mxlews L77.jpg
    Mxlews L77.jpg
    17 KB · Views: 131
  • Mxlews Baseline #2 Phase Rev MLS.jpg
    Mxlews Baseline #2 Phase Rev MLS.jpg
    47.7 KB · Views: 74
  • Mxlews Baseline Attn #2 MLS.jpg
    Mxlews Baseline Attn #2 MLS.jpg
    51.4 KB · Views: 57
  • Mxlews Baseline Attn #1 MLS.jpg
    Mxlews Baseline Attn #1 MLS.jpg
    50.1 KB · Views: 78
  • Mxlews Baseline FR SIN.jpg
    Mxlews Baseline FR SIN.jpg
    53.2 KB · Views: 97
A Challenge:

Morel DMS-37 acquires a bunch of issues once mounted on the L77 baffle in place of the stock JBL LE20-1. Despite incorporating a "waveguide," being recessed 1" and centered a mere 2.5" from that edge, reflections, diffraction, time-alignment, and phase interference play havoc with the response. First curve shows performance with the stock crossover, Med attenuation. While the HF is extended, the consequences of bandwidth overlap remain prominent.

Using the 2-pole crossover developed above for this Morel improves the situation somewhat, though the LE10A woofer evidences a 5 dB notch at 1.4 kHz, and the tweeter now has one at 3.2 kHz. Second curve. Even with these anomalies, the system sounds much better than the original. Note that installing this lowpass filter on the woofer has flattened its rising response, substaintially improving the bass and the overall balance.

The crossover region is still a puzzling mix of cancellations and reinforcements; the drivers are sometimes summing and sometimes nulling. So, I move to a 3-pole crossover to bring order to the chaos, third curve. The drivers are summing properly. The sound is similar, but less sensitive to listener position.

For anyone interested in comparing the behaviors two crossovers, they are superimposed in the fourth curve.

Finally, fifth curve, frequency response of Mxlews's two LE10A (alnico magnet) woofers and a pair of LE10H-1s (ferrite,) the factory recommended replacement in L77, which would exhibit even more bandwidth overlap with the tweeter. Both driver types have rising response, 5 dB between 100 Hz and 1 kHz.

Bonus curve, last, 3-Pole filter works fine with LE10H-1, too. Note again orderly summing of driver responses in the crossover region. The crossover has assumed control of L77, and it sounds mighty fine with this Morel tweeter.

Schematic here:

http://www.jblproservice.com/pdf/Control Contractors In-Wall Speakers/Control 128W.pdf

As built here:

http://audioheritage.org/vbulletin/showthread.php?p=116250#post116250

Can it be tweaked better for L77? Yeah, later, maybe.

On to Fostex, next.... :thmbsp:
 

Attachments

  • L77 LE10A & LE10H-1.jpg
    L77 LE10A & LE10H-1.jpg
    56.6 KB · Views: 76
  • DMS-37 LE10A 2- & 3-Pole.jpg
    DMS-37 LE10A 2- & 3-Pole.jpg
    57.8 KB · Views: 59
  • DMS-37 LE10A 3-Pole.jpg
    DMS-37 LE10A 3-Pole.jpg
    54.3 KB · Views: 60
  • DMS-37 LE10A 2-Pole.jpg
    DMS-37 LE10A 2-Pole.jpg
    54.6 KB · Views: 65
  • DMS-37 L77 Stock SIN.jpg
    DMS-37 L77 Stock SIN.jpg
    54 KB · Views: 67
  • DMS-37 LE10H-1 3-Pole.jpg
    DMS-37 LE10H-1 3-Pole.jpg
    53.6 KB · Views: 67
  • DMS-37 in L77.jpg
    DMS-37 in L77.jpg
    25.5 KB · Views: 107
Trifid - Another Round:

Triangulate flange soft dome Fostex FT207D is also a mechanical drop-in to the existing L77 LE20-1 tweeter cutout. I was able to locate it an additional 1/4" away from the cabinet edge/lip.

It plays better than the original tweeter using the stock crossover, first curve. Driver bandwidth overlap generates now familar phase interference issues, but the overall system response is, well, "listenable."

Fostex specifies a recommended 2-Pole network, the performance of which is shown in the second curve. The phase issues are somewhat mitigated, as is the woofer's rising response, to a degree, by the addition of a lowpass filter on the woofer, but it's clear that adjustments of that filter specific to LE10A might improve the crossover.

Said adjustments made and the results shown in the third curve, providing the best measuring and best sounding system response of these attempts. Rising woofer response is further mitigated making the unique passive radiator bass rich and full. Drivers are summing predictably in the crossover region. If these L77s were mine, they'd be keepers with this tweeter and filter combination.

Finally, a 3-Pole filter shows promise for producing an even flatter response, but this Fostex tweeter is out of headroom for response shaping; it'd need at least 3 dB greater sensitivity for this approach to work optimally.

In summary, then, tweeter upgrades are not a "plug-and-play" proposition. Manufacturer's specifications and published response curves may best be viewed as comparative guidelines only; when it comes to actual performance in an existing system, other factors prevail, not the least of which are driver, baffle and network variables.

Yes, any of scores of choices will work as replacements, and modern tweeters can certainly upgrade vintage system performance, but achieving an optimum outcome may require comprehensive system redesign....

Edit: I've put some additional effort into tweaking the highpass filter Q to improve the system response more, but what I have seems to be the best compromise without complicating the circuit with notch filters. Circuit, sim, and system frequency response, last three images.

Sim says I'm crossed electrically at 2.55 kHz; acoustically, pick your pleasure.

Circuit is six components, and tweeter polarity is inverted. Leave out the L-Pad if you like the sound of the measured response, +/- 4 dB from 40 - 22+ kHz.... :thmbsp:
 

Attachments

  • FT207D L77 3-Pole.jpg
    FT207D L77 3-Pole.jpg
    53.4 KB · Views: 105
  • FTD207D Fostex XO Mod LF.jpg
    FTD207D Fostex XO Mod LF.jpg
    55.1 KB · Views: 112
  • FTD207D Fostex XO.jpg
    FTD207D Fostex XO.jpg
    54.5 KB · Views: 106
  • FTD207D L77 Stock.jpg
    FTD207D L77 Stock.jpg
    57.1 KB · Views: 131
  • Fostex and Filter.jpg
    Fostex and Filter.jpg
    44.4 KB · Views: 318
  • FT207D in L77 Circuit.jpg
    FT207D in L77 Circuit.jpg
    15.1 KB · Views: 598
  • FT207D in L77 Sim.jpg
    FT207D in L77 Sim.jpg
    40.3 KB · Views: 96
  • FT207D 2-Pole.jpg
    FT207D 2-Pole.jpg
    51.5 KB · Views: 130
I have three "loose" LE20-1s here; all say "8 Ohms" on their foilcals.

Here's the results of quick runs on WT2. I'll do more definitive curves using CLIO in comparison to the Fostex once they arrive.

Re = 3.5, 4.0, 4.0 Ohms
Fs = 1450, 1200, 1300 Hz
Z 2kHz = 5.0, 4.7, 4.8 Ohms
Z 2.5kHz = 4.4, 4.8, 4.9 Ohms

Body diameter = 2.65"
Flange O.D. = 3-13/16"
Mounting hole cutout in S99 = 3"

I'm not making it up when I say manufacturers played fast and loose with "nominal" impedance specs in the olden days. :p:

The L77 filter is 6 uF and 0.4 mH. What's that calc as frequency @ 5 Ohms?

i finally removed the tweeters from my L77's. the darned things have an 16 ohm rating. would that mean that the fostex build that this thread pertains to would not be a go in the pair i have. by the way one tweet id doa.:tears: i was watching one on ebay and noticed the differance and it got me wondering.
thanks.
 
i finally removed the tweeters from my L77's. the darned things have an 16 ohm rating. would that mean that the fostex build that this thread pertains to would not be a go in the pair i have. by the way one tweet id doa.:tears: i was watching one on ebay and noticed the differance and it got me wondering.
thanks.

To the best of my knowledge, all LE20s are nominal 8 ohm, actual ~5 ohm regardless of what the label says. If your L77s are nominal 8 ohm (I've never seen a 16 ohm version but I haven't seen everything...) the Fostex discussion applies to your situation.
 
i finally removed the tweeters from my L77's. the darned things have an 16 ohm rating. would that mean that the fostex build that this thread pertains to would not be a go in the pair i have?
Nope, you're swapping out the tweeters and building new crossovers. They're all nominal 8-Ohms, anyway, what Steve O said. Measure their DCR to confirm.

Byy the way one tweet id doa.:tears: i was watching one on ebay and noticed the difference and it got me wondering.
thanks.
If you're refurbing for resale, you want the correct LE20-1s in there. If you want mighty nice-sounding speakers, do the Fostex thing, above.... :thmbsp:
 
16 ohm LE20-1's

To the best of my knowledge, all LE20s are nominal 8 ohm, actual ~5 ohm regardless of what the label says. If your L77s are nominal 8 ohm (I've never seen a 16 ohm version but I haven't seen everything...) the Fostex discussion applies to your situation.

you are correct about the 5 ohm thing on a supposed 8 ohm speaker, also about 5 ohms on a 16 ohm speaker. got a couple of pics here for you. if i can get the system to work. thanks.
 

Attachments

  • 2008_0215jbl0001.JPG
    2008_0215jbl0001.JPG
    117.5 KB · Views: 111
  • 2008_0215jbl0003.JPG
    2008_0215jbl0003.JPG
    119 KB · Views: 102
used your design, results are great!

Thanks so much for this detailed post. I have a set of Lancer 77's I bought in the early seventies. I had the woofers refoamed, built a crossover from your schematic, and installed Fostex tweeters. The sound is better than ever, clear and crisp in the midrange. There are several recordings I have that have choral music, and I can now understand the words they're singing. Makes a huge difference.

The real kudos come from my teenage son, who takes great pleasure in showing his friends how great they sound. They are all impressed.

Thanks again for the schematic - wouldn't have know where to start without it.
 
Back
Top Bottom