Altec 9844-8B

[Skywave-Rider]

Zilch,
Here’s what I plan to do tomorrow, let me know if I’m missing something:

All tests nearfield and with 902 on dummy load. Tests 1& 2 with inductor at N-800-F output only, tests 3 & 4 from both x-over input and output.


1. 20mH shorted, both woofers playing.
2. 20mH in line, both woofers playing. (Video at change from short to in line.)

3. Helper alone, coil at N-800-F input.
4. Helper alone, coil at N-800-F output.

I think I should dummy load 16 Ohms at main woofer for #s 3 & 4, yes?


Unless I missed something, next one will be with M-19Z.

[Zilch]

Yup.

You're s'posed to listen some, too, you know.

[I claim us measurers listen more and with greater precision than listeners do.... ]

[Skywave-Rider]

Yeah, thanks.
I will this time.

[Skywave-Rider]

I was bumped yesterday and am on standby today. Worse comes to worse, will go Weds....

Might be for the best because what I thought was fiberglass in my throat turned out to be a bad cold*, so I’ve been moping around recovering for almost a week -- I feel OK now, but my ears are still stuffed a little. And since I missed the listening test last time, have to do a good one this time.

So I turn my attention to a couple of points. First, when I look at the Valencia thread, jackgiff posts these FR readings:
http://audioheritage.org/vbulletin/s...&postcount=802

Comparing his BMS 4550s to my Altec 902s:
http://audioheritage.org/vbulletin/s...&postcount=146

I do not see the roller coaster in the HF coming from his 4550s on 811B horns. Mine look scary. On some of Zilch’s Clio graphs I think I see camel humps, but it is primarily on the 802 measurements.
http://audioheritage.org/vbulletin/s...&postcount=126

So, is this possibly the N-800-F ‘s effect on the 902 HF? Or is it inherent in this driver/horn combination.

Zilch, you remarked that you see the undulations at the Lab and attribute it to the 811B horn. When you see that, is a 902 bolted on?

Maybe this is moot until the Z-19s?

The other thing, about the dummy loads I’ve been using: This is a newbe question I know, but since I’m substituting resistances for nominal impedance, what relevance do those 8 or 16 ohms have? Am I right in thinking that if I want the crossover to function as it would when the actual driver is in place, then there is no substitution for the actual driver with its frequency dependent complex impedance?

So, If I have that assumption right, then when doing tests like we are doing, is there a way to insert a fixed resistance which is a number more effective than nominal? For example, if I want to look at the 414s upper bass FR near the crossover point as a focus, would it be better to look at the 902s impedance at about that frequency and substitute that value for the dummy load? Then the result would be good for that narrow range only. So I’m thinking a non inductive pot would be useful.

Or is the nominal figure really some complex mathematical average of the entire impedance curve the driver exhibits? So perhaps the 8 or 16 Ohm dummy is the best substitute.

And all of this assumes that the filters do interact.

I can feel you rolling your eyes....hehehe.

Getting antsy, sorry for the delays.



*Or a combination of both; pretending to be a lab-coat guy here, gotta watch the assumptions.

[Zilch]

My most expert analysis of the roller-coaster thing at this juncture is, "I don't know," and I can only point to some of the same indications you have observed. I see it with Gary's 802-8G tangerines on 811 and 511, and I see very similar in the CBS Labs measurements of Model 19 from 30 years ago.

In my own situation, I'd be tempted to point to the suboptimal conditions under which I measure stuff, but I can put BMS drivers on PT waveguides and get stunningly flat response curves at any time, no sweat. Jackgiff has repeated that result with his Santana project.

As you've seen, I can also get respectably flat results with BMS on 811 and 511 without doing anything in particular to address individual response anomalies. Pi-speakers gets similar with an Eminence driver, often posted by others for comparison. I've never measured 902s; best I can do until then is assume they are similar to Gary's refurbed 802s, but I don't think there's enough evidence pointing to the Altec drivers as the source to condemn them.

I'm ignoring that in your measurements for now because, worst case, the response undulations are on the order of +/- 2.5 dB. The pattern is regular, and it's coming from something, somewhere; it's not a random error. There are several members here who can look at that and tell us what it's about, and I may be PMing them for their expert opinion if it persists under other conditions. I may also be proposing that we swap drivers for reciprocal testing at some point.

Regarding dummies, yes, using a resistor closer to the actual impedance of the woofer is probably desireable, and there are ways to construct a more accurate diriver simulator than a simple resistive load, but with parallel crossovers such as we're dealing with here, the filters do not interact but in detail, so it doesn't much matter.

The primary purpose is merely to approximate the loaded condition, and at the level of resolution we're able to measure, a fixed resistance of the nominal value is most likely good enough. As you suggest, one could use the driver in a second system isolated apart, closed in a neighboring room, for example, for an ultimately accurate "dummy," if desired.

RTA inherently provides a smoothed metric of what's going on, and is very useful in assessing that in large scale. Yes, we can get more accurate with greater resolution using alternative measurement technologies, but would likely be applying mathematical smoothing to make equivalently useful determinations at this level.

[I presume M19Zs are nearly ready for prime time by now.... ]

[Skywave-Rider]

Near field of helper woofer. 902 on dummy load. Main woofer on dummy load.

Listening planned for weds, current setup allows for normal woofer operation and 20mH in line via patch point insertion.

M-19Z should be this weekend. Hope to have seawolf’s Hafler pre amp by weds so I can easily do source switching and, actually, adjust volume like a person.

Edit: I just realized I forgot to short the main woofer for the helper woofer "alone" tests.

[Zilch]

Arbitrarily picking -10 dB as criterion, the "Helper" is working primarily below 200 Hz, as we saw in the video. We could narrow its band of influence by making the filter 2-pole. Last time I tried that, my network impedance went to about 1 Ohm, as I recall, tho. That would take some further research.

I'm having difficulty seeing much of a difference above that, not what I'd expect to see from summing, at least. It's clear the lowpass filter is behaving differently, 8 versus 16 Ohms. Do we have a new video? I think we could see the differences more clearly there.

Yeah, shorting the main would make results on the helper more meaningful in comparison to earlier measurements, but I think we've got a good handle on what it's doing, and it's what we want. The time's better invested in listening at this point, and then moving on to the new crossover measurements. Gonna be some surprises there, I betcha....

Edit: Here's some results I'm currently getting with a different driver on 811B horn. The roller-coaster is considerably in evidence. M19 crossover highpass filter is Cyan:

[Skywave-Rider]

Life has intervened, but I'm back on track. One M-19Z complete, the other in progress, please forgive the "tribute" Zilch. Crossover features construction site "sidewalk" lumber and a ton of lead. Not featured: hot glue on components

[Zilch]

Looking good!

I decided to leave my vintage M19 crossovers alone, and just build this version new from scratch myself, so parts are ordered.

I've had a couple of inquiries; it'd be good if I was working with all new parts, as well.

[This had better work. My "I am a LAMER" bumper sticker looms in the background.... ]

[Skywave-Rider]

Thanks.
OK, but no matter what the result, this is the bumper sticker:

[Skywave-Rider]

OK Zilch, here goes....

Measurements made at 3’ and slightly off axis, like last time.
Nearfield at 3’’ between both woofers and ports.

Unfortunately, I forgot to bring my dummy loads, so I couldn’t do any single driver measurements.

In the L pad photo, the left pot is MF, right, HF. Yes, I suck at woodwork.

Big surprise to me is the narrow passband in the MF control. I thought it would be broader or even shelving-like in its effect, with the HF control working on the very upper end. Is this result expected? I went through my wiring to check for screw ups. There is one in the pic I posted last time, but I caught that early on.

I left one 9844-8B with original N-800-F and rolled both into a rectangular room for listening impressions. I fed both sides mono signals and played one at a time. I determined speaker placement options and several listening positions through the Room Mode Calculator. Program material included music and voice (speech.)

First, the good news:
The VHF gain, clarity and control via the L pad is exceptional. Now, this is like adding a tweeter. The top octave is even “airy.” HF transients sound clear, clean, lifelike. The N-800-F, in comparison, is not as extended sounding and might accentuate c. 8K where the 19 does it higher.

And the other news:
The mids have a few problems. The 19 honks out at about 6-800 Hz sounding like a narrow kind of forwardness making voices hard and fortissimo operatic passages difficult to play near “loud.” In comparison the N-800, much maligned by my initial impression, is smooth. Though It may be forward, it is not peaky sounding.

There is also a lower mid entanglement, 200-350 Hz approx., which makes bass instruments for example, hard to follow. Though the original network is far from perfect, it offers more clarity in this range.

Overall the new crossover sounds louder, but I’m assuming this has more to do with the pronounced mids as stated.

2.5 Mode: (These impressions apply to both crossovers, though I feel it sounds better in the N-800 cabinet.)
This gives a thump at about 80Hz. It’s exactly what Zilch predicted would result, a mode allowing the monitors to be used in a midfield position, without much reinforcement from wall boundaries. The los don’t seem to extend much, if at all, just congregate into a bump at 80Hz which works well on some material. On other material it can muddy LF clarity. During my listening session I used it more than I did not.

This mode engaged also removes some 200 or 250 Hz. Noticeable when playing electric guitar music. That’s where the body of the electric is, and I suppose the more paper the better when you’re emulating a 4x12 cabinet. However, shoveling some of the lower mids into the bass makes the mids leaner and clearer on other material.

Although I haven’t gotten into port tubes yet, I imagine these cabinets will never provide forceful low end. But I’m not expecting that either.

I have to head off to bed, I’m sure I’ll have more to say.

[Zilch]

Okay, DROP the scissors and step BACK from the project cart!

The new N19Z crossovers have clearly assumed control of the HF, and your measurements indicate that the adjustability is very much according to the design sims. You apparently have sufficient driver response and headroom to achieve full VHF extension to 20 kHz, and adjustment range suitable for balance with either one or two woofers running. To use it most easily, adjust the MF control for optimum balance with the woofer(s), and then adjust the HF control for flattest VHF response.

Look again at this sim. There's shelving, but it is short-lived, up to 3 kHz only, above which frequency the HF control adjusts the slope of the compensation. If you crank the mid control up, it's only the shelf that gets boosted, and I certainly see that in your "MF max" measurements. I'll confirm once mine are built:



Optimally adjusting the response appears somewhat difficult to do, as the roller-coaster ripple remains, well, "overwhelming." We now pretty much know it's not the crossover doing that. It'll be interesting to see what happens once you install BMS drivers; I'm still not entirely convinced there isn't something else going on we haven't figured out yet. :dont-know

We really need to see the response without the woofers running to know how the basic highpass filter is operating, and where. Dummy loads aren't essential for that, if your HF and LF filters are separately accessible; simply disconnect the input to the LF. I always build with separate input terminals to the two sections to facilitate this sort of testing or use, and simply bridge them for normal combined operation. Dummys on the output work too, as you know from experience, and probably offer a somewhat better approximation, at least from the standpoint of amplifier loading.

That's half of sorting out the midrange issues. The other half is what the woofers are doing. The lowpass filter in N19Z is simply that of Model 19 (2.7 mH and 21 uF). I used that as a starting point, since we're presumably operating at 1.2 kHz now, rather than the original 800 Hz, or wherever it actually was. See the sim, bottom, below; your midrange "honk" is clearly shown in 16-Ohm mode, red curve. As previously observed with N800F, this filter behaves very differently with 8- vs. 16-Ohm loading.

From your listening evaluations, it would seem that the lowpass in N800F (3.5 mH, 10.5 uF) may actually be more appropriate at one impedance or the other (2.0 vs. 2.5 operation). Compare the sims for both filters to see what's going on. I've reposted the N800-F immediately below. From looking at them, it seems your LF nearfields are done with the new crossover? If so, it looks fine with both woofers running, but you may want to try the N800-F LF in combination with N19Z HF. We can easily tweak the lowpass according to your findings.

In any case, it's now clear what "Helper" mode is doing. With just one woofer running, the response falls ~3 dB above 100 Hz, and everything below that, relatively speaking, is boosted accordingly. I maintain the thesis that the LF response is also more extended in 2.5 mode than it would be with one woofer running alone, other factors being equal, according to the rules of low frequency summing and mutual coupling, as outlned by Eargle and Foreman in JBL Professional Sound System Design Manual here:

http://www.jblpro.com/pages/tech_lib.htm

Still have a big hole at 3K15.

Are you SURE there's no EQ or anything operating in that DEQ2496?

[Chas]

Guys, very impressive to see an 806 extend to 20KHz. Maybe I should have kept mine....

[Zilch]

Guys, very impressive to see an 806 extend to 20KHz. Maybe I should have kept mine....

Can't get there with 806, nope. These are 902s....

[Chas]

Can't get there with 806, nope. These are 902s....

Ha! Okay, now it makes more sense!