thoughts on hi-vi b3s line array /w single tweeter

Status
Not open for further replies.
Jump to Latest
I

Ignite

Member
Joined 2001
#1
I'd like to build a set of line arrays on a student budget, so I am looking at probably an array using hi-vi b3s or possibly the tangband w3-871s's. Now I know comb filtering will be an issue with 3" drivers in the upper range (starting around 5-6kHz as far as I know?) so I'm considering using a single tweeter to fill in the upper range, mostly because I cannot afford a complimentary array of exotic planar tweeters. This causes an issue with efficiency since the array would loose 3db as distance doubles while the tweeter will loose 6db.

I'd probably use ~18 of the hi-vi b3s'es per side, and place the tweeter in the middle. I am interested in input on whether this is a good idea or not.

Also very much interested in input on xo frequency/point and tweeter selection.

Thanks
 
#2
The 3 dB versus 6 db power drop with distance is not a problem IME in a home listening situation. If you were designing a full range system that had to equally illuminate audience participants in a movie theater in the front, middle and rear rows, then this would be an issue.

My arrays employ a single tweeter of high efficiency centered in the array.

You can build some fine line arrays spending even less for your drivers than those drivers you mentioned. Just happened to notice your comment about 'student budget'. Since each driver in the array works less hard by a factor of how many drivers are employed, cone excursion and distortion is reduced. There are some very fine sounding speakers made in China that are available as buyouts from people like PartsExpress that sound really good if the power into them is limited to say 1/20th of nameplate rating, and also if you keep the bass out of them. One 4" FR speaker from PE has enjoyed a lot of experimentaion amongst enthusiasts, the action being generated in part by the price which was just 49 cents each at one point. I don't know if these are still available but there are threads in these forums about these drivers. They are known as NSB for No Stinking Badges, a reference allegedly to a customs importation problem and the removal of factory identification markings to alleviate this problem. if my info is correct.

There are other sources for suitable low cost drivers as well. You need to actually test samples of cheap drivers though so don't just buy a case lot of the first ones you find that are priced right.
 
#4
generally this is NOT a good idea, however this can be "worked around" - as seen here:

http://www.geocities.com/dmitrynizh/labaffles.htm

The "point" of this design is not to achieve a perfect summed acoustic center "point-source" well away from the listener, but rather to create a line that will suffer less comb filtering at the expense of line length at a modest distance from the listener. Additionally, the eff. gain should be much more "linear" across the driver's bandwidth.

BTW, "Zaph" seems to really like the Hi-Vi driver..

http://home.new.rr.com/zaph/audio/index.html

http://home.new.rr.com/zaph/audio/audio-speaker18.html

http://home.new.rr.com/zaph/audio/minitest/

You will still need a powered sub that goes pretty high in freq. (but thats fairly easy to do today with plate amps and decent drivers..)
 
#6
Yeah, I love that driver.

While not quite a line array, I have messed around with a four B3S + single tweeter design. I haven't built it but I have rough models and drawings. Basically, it's a 2.5 way design.

It's still a point source rather than a line source. Obviously there's no need for power tapering when the outside woofers serve .5 BSC duty.

What's neat about a design like this is that it's a super-slimline design that looks elegant in most rooms. It won't suffer from the nearfield/farfield level difference between the tweeter and woofers like a design with many woofers flanking a single tweeter.

Maybe I'll actually build it someday, if I can find the time. (heheh, yeah right) Given the bandwidth of this driver, the woofer-tweeter crossover could be a nice sounding shallow slope.
 
#7
Ignite said:
I'd like to build a set of line arrays on a student budget,
Click to expand...

I'm trying to do the same for a friend of mine who is still studying and lives in a Student Dorm.

I was planning to start a separate thread on this, but I think this thread is the best place to do it, since I think our goals are very similar. I hope this is not considered thread jacking, more something like thread 'expanding'
😉

Here are the design considerations and specifications I have typed the last few days:

An as sensitive as possible speaker who is composed of minimum component *variëty* (for ease of construction and low price), and who will give as full range as possible at 10 watts RMS, without reaching the excursion limits of any of its drivers at this
specification. Total price of 2 finished speakers should be less than 50€.

Since this will almost certainly require active equalization, a high sensitivity is my main goal, as to be able to EQ (on the PC) a flat response. For a high sensitivity there are 2 options: horn-loaded and a line-array. Horns are very difficult to construct and multiple horns would be nessesairly for a high bandwith. Looking at price and ease of construction, a line-array with multiple cheap extended range drivers would be the best choice for my design.

If it is to be an array, then it can be used as a small PA for presenting the annual quiz of the Counsel of the Student Dormatory. The design of the array itself will be optimized for this purpose. 12 Drivers in one speaker should give a 10dB increase in efficiëncy which is adequate and economical from #drivers/gain point of view. Also 24 drivers in total is a good number to reduce the shipping costs to an acceptable portion of the total cost. Power handling for a single driver should then be 1watt RMS so one speaker can handle 10 watts.
I have found out bij Googleing that for speech intelligibility the array should cover 250hz-4000hz. If the soundspeed in air is 340m/s, then for 4000hz the wavelength is 8.5 cm. (full wavelength) If for the design to work as an array the maximum distance between the point sources aka driver centers can be no more than 1/2 WL, then even 7cm diameter drivers should go up till (almost, a 7cm driver is not a true point source) 10000hz, ensuring the speechrange will be radiated according to line array theory. The lower freq limit of the LA is defined bij the length of the array. Earhight while sitting on a chair is 1.25m, and thus the lower limit is 136Hz, also outside the defined speech range.

Now the LA-specifications are defined, lets look at the normal (horizontal) radiation behavior. For good sound quality (and my friend who I'm designing this speaker for also expressed his concern for this), the high freqency range should extend high enough. I'd therefore like it to extend till 15000hz. Obviously this requires an as small as possible driver. This again compromizes low frequency responce, where I'm aiming for an extension till 40-50hz if possible.
The ideal driver should therefore be oval in shape. The smaller cross-section can be coupled together for the LA effect, and the taller cross-sections will give a larger surface area to get as much displacement as possible. Given the 8.5 cm LA limit from above, and assuming a 3/4 ratio for the oval shape, the largest possible driver is 8.5cm/11,5cm. Again, smaller drivers for better HF are preferred.

Now I have come to a point where there are 2 possibilities to cater for the low frequency responce, and this is where I would like some brainstorming from you guys: bass reflex and hornloading the back of the array.
I know some things about horns, since I'm still busy creating a spreadsheet according to the Leach Model, and I have modeled a few horns in HornResp. But for bassreflex I don't have a feeling what is possible or what to expect.
Summing up 12 7cm circular drivers, gives one equivalent 24cm driver, but still with 1mm Xmax! (assumed 1mm xmax for the drivers). Finding parameters to model 12 extended rangers in Hornresp is difficult, but with some educated guesses about BL factor etc ... I found out that Hornloading is a good way to reduce excursion at lower frequency.
The lower the tuning of the horn, the lower the excursion, however that backhorn should be fitted in the max 1.25m height, so there will obviously be some lower limit. I haven't looked in the lowest freq one could expect at max excursion at 10watt RMS yet, but I expect it should be under the 250hz-limit of the line array. At 1 watt I hope to get as low as 40-50hz.
For Basreflex I have no idea how much the excursion will be, and how low it can be tuned, so hopefully someone can give me some hints. I also worry about how much SPL is possible at 1 watt. If there is lets say 10dB less from 300-50hz, then a horn would
obviously be better for downfill.

Some toughts about Fs Vs F3: i think it was Thiele who stated that drivers should only be operated one octave under their Fs and still have acceptable quality. However, since with a backhorn the loading and excursion should be a lot more civilized than in BR, I think Fs is not an issue for that. In BR I could lower the Fs a bit (and smooth out the High end - remember I'll use an an Eq anyway to extend HF responce, so some sensitivity loss up there is not a real problem) by spraying some sort of waterproofing spray in tiny layers on it, but i doubt that would be nessesairly.

So to summarize the driver I'm looking for:
- 1 watt RMS
- Oval shaped
- Max size 8.5cm/11.5 cm
- As much Xmax as possible 😉
- Around 1€/piece.
- Low Fs
- Preferably available/sold in Belgium.
This is very close to Parts Express' famous 4" Extended Ragers at 0.69$, but I think it should be possible to find similar ones that are more to my specs.

A driver I found (available in numerous shops that sell electronic parts here in Belgium) is the SP-170. I googled an even found a price quote for it here:

http://www.shop3000.de/product_info.php?products_id=2850

The price is a few times higher than I'm looking for, but it's also the largest possible driver I would use, so I think I can find a smaller model at lower price.

Ideas anyone?

thx

Cordraconis
 
#8
Zaph said:
Yeah, I love that driver.

While not quite a line array, I have messed around with a four B3S + single tweeter design. I haven't built it but I have rough models and drawings. Basically, it's a 2.5 way design.

It's still a point source rather than a line source. Obviously there's no need for power tapering when the outside woofers serve .5 BSC duty.

What's neat about a design like this is that it's a super-slimline design that looks elegant in most rooms. It won't suffer from the nearfield/farfield level difference between the tweeter and woofers like a design with many woofers flanking a single tweeter.

Maybe I'll actually build it someday, if I can find the time. (heheh, yeah right) Given the bandwidth of this driver, the woofer-tweeter crossover could be a nice sounding shallow slope.
Click to expand...

Another nice thing you have mentioned elsewhere is the high Qts relative to the fs. Its concievable (especially with lower dampening amps like most tube amps) that the response will be relativly flat to fs in a dipole design that has a minimum baffle width. If you go with the curved array then you might be able to get the response to extend fairly flat to 6kHz or more (depending on the curve and the distance), which would provide a natural low pass on the array. The only crossover component then could be as simple as a small value cap on the tweeter.
 
#9
A couple of things:

Scott I'm not sure where you got the info, but Dmitri's concave lines still suffer from suppression of the high end and they are concave for imagining purposes. He listens in the near field and the drivers are equal distances from his ears eliminating any time difference.

Second, 10 1 watt drivers can't handle 10 watts unless you wired them all in parallel, causing an impedance problem. The series/parallel wiring configuration necessary results in a lower power handling than the sum of all the drivers. eg Four ten watt drivers wired in series can handle only 10 watts because the 1st driver can handle only 10 watts and it has to go thru the first driver to get to the second one.

A single super tweeter can be very effective with a line array. You just need a tweeter that has an efficiency in excess of the total array. They put an LPad on the tweeter to adjust it to your listening distance.
 
#10
johninCR said:
A couple of things:

Scott I'm not sure where you got the info, but Dmitri's concave lines still suffer from suppression of the high end and they are concave for imagining purposes. He listens in the near field and the drivers are equal distances from his ears eliminating any time difference.

Second, 10 1 watt drivers can't handle 10 watts unless you wired them all in parallel, causing an impedance problem. The series/parallel wiring configuration necessary results in a lower power handling than the sum of all the drivers. eg Four ten watt drivers wired in series can handle only 10 watts because the 1st driver can handle only 10 watts and it has to go thru the first driver to get to the second one.

A single super tweeter can be very effective with a line array. You just need a tweeter that has an efficiency in excess of the total array. They put an LPad on the tweeter to adjust it to your listening distance.
Click to expand...


I got the info from the site..

Additionally though I happen to have the exact same fullrange drivers and have similarly experimented with them (and have done measurements, though quite some time ago).

The suppression is due to "summing" problems related to acoustic center of a point source and freq., and is an artifact of all SUMMED array lines generally above 400 Hz (though driver dependent). (By summed I'm refering to the interaction of output between two or more point-source drivers - something like a planar tweeter with little or no vertical output when used in a vertical line array does NOT have this characteristic).

This array graphically shows what happens when line has no eq. to compensate:

http://www.partsexpress.com/projectshowcase/Kuze3201/Kuze3201.html

Here Dmitry subjectivly describes the effect:
"I added 2 more speakers and confirmed what I have already heard with 4-element array: high-frequency roll-off."

Unfortuntly his reason: "lobing" (at least as a measure of cancelation from comb filtering or "interference" as he put it) is NOT what is causing the reduction. Again, you can actually see on the Kuze Array what comb filtering is doing to the response (the ragged response as freq.s increase, i.e. the "unsmoothed" response).

At least to me it is quite clear that he is CURVING the array in an effort to increase the linear response as freq.s increase without eq. (i.e. nothing to do with imaging).

Moreover IF his listening distance is 4-8 feet away from the speaker then it is very unlikely that the drivers perfectly sum at his position. (and I've done this and measured this myself..) Instead the "effect" is more like that of a shorter vertical line with much closer acoustic centers. As a result, the freq. loss is still present BUT it begins higher in freq. (which allows for the super tweeter.) In otherwords though he talks about JBL's concave curving - he hasn't achieved it to be equidistant in his design (which would be a summed point source, not a line source).

As to powerhandling - I'm not to concerned with that as long as gain is increased significantly. Also the powerhandling limit as specfied by the manufacturer usually has more to do with excursion vs. spl than VC heating (i.e. blowing out a cone vs. burning up a VC..). Certainly though distortion could be higher for that driver.

There are some serious problems with just a supertweeter in a normal line array (that are quite audible if extending below 10kHz).

1. where will you put it? If you put it in the center of the line you are "breaking" the line for a certain freq. and above which will increase comb filtering radically. The only acceptable solution in this instance is to extend the freq. response of the tweeter below that of the point that comb filtering starts (usually with a steep crossover for both the highpass and lowpass). Placing it above, below, or to the side of the line will still have the problems in #2 (-as you will with a curved array)..

2. SPL matching (of the line and the point source) will be dependent on distance because the array suffers spl losses theoretically half that of the point source super tweeter as distance increases. Practically speaking you'll need a super tweeter with an output that is 6-9db greater than that of the line with an attenuator to adjust for listener distance.
 
#11
johninCR said:

Second, 10 1 watt drivers can't handle 10 watts unless you wired them all in parallel, causing an impedance problem. The series/parallel wiring configuration necessary results in a lower power handling than the sum of all the drivers. eg Four ten watt drivers wired in series can handle only 10 watts because the 1st driver can handle only 10 watts and it has to go thru the first driver to get to the second one.
Click to expand...

That is incorrect.
 
#12
Scott,

Thanks for the detailed theoretical info. I've talked to Dmitri quite a bit and his findings coincide with my real world results. He calls his position near field because he has a big room and I forget if it is a 9 or 12 ft listening distance. A flat array didn't work for him because of the driver to ear distance differences. It impacts the imaging which is quite apparent when you listen to a flat array (without power tapering). You can hear the timing difference and it is most apparent in that you can hear the drivers at the ends of the array because their sound arrives last.

He tried a continuous curve but the sweet spot is way to small and he settled on curving in steps.

I've only used flat arrays and it's been my experience that the effects of comb filtering, lobing, etc. are not that great. They come through sounding like just a general, but significant, reduction in HF responce. A supertweeter fills that in nicely. Yes it needs to be more efficient than the array and it needs to be adjustable.

I believe the overall reduction in HF response isn't so much due to comb filtering as it is due to the driver spacing causing the HF to decrease in magnitude at 6db per doubling of distance not 3db since the HF isn't behaving as a line source. This also greatly reduces the effects of comb filtering. In essence, yes comb filtering is attentuating specific frequencies but the volume of the entire spectrum above that initial comb filter point is greatly attentuated since it is not a line source above that frequency.

For me a single super tweeter with an Lpad is an easy and effective solution. It helps that the cheap 4" drivers I use extend nicely to just above 10khz without a harsh cone breakup.

Also, the hole in the array just isn't audible, at least for me.

A line array is just a different kind of speaker with advantages and disadvantages, like any speaker has. If you are looking for a speaker for light acoustical music where you have intimate listening sessions and want to hear every nuance and detail, then a line array isn't what you want. However, if you want front row, center stage seats with a big dynamic sound and a very large sweet spot using speakers that fill your room more evenly with sound, then I highly recommend a line array.
 
#14
I believe if you wire 4 drivers in series, they will each only see 1/4 of the power because as a group they have 4 times the impedance. If you then wire 4 of those groups in parallel, the impedance presented to the amp drops back down approximately to that of the original single driver.

This is a simplification based on a speaker driver as a purely resistive load, but I believe it's still a useful rule of thumb.
 
#15
From a constant voltage source, four identical drivers connected is series draw 1/4 the current of a single driver because the resistance is 4 times greater. The voltage drop across each of the drivers is 1/4 of the applied voltage.

So, with 1/4th the current through the drivers and 1/4 the voltage dropped across each driver, each driver is dissapating 1/16 the power.

Edit:

Actually, my answer is incomplete but it's late and I'm too tired to continue. If nobody has filled in the blanks by the time I log back on tomorrow, I'll finish it up.
 
#16
Bill Fitzpatrick said:
From a constant voltage source, four identical drivers connected is series draw 1/4 the current of a single driver because the resistance is 4 times greater. The voltage drop across each of the drivers is 1/4 of the applied voltage.

So, with 1/4th the current through the drivers and 1/4 the voltage dropped across each driver, each driver is dissapating 1/16 the power.

Edit:

Actually, my answer is incomplete but it's late and I'm too tired to continue. If nobody has filled in the blanks by the time I log back on tomorrow, I'll finish it up.
Click to expand...

I think the remark about power handling was directed at my design (which i will move to a separate thread - it doesn't really belong here, sorry for that 🙁 ), so I will answer it.

12 drivers of 8 ohm can be connected in 3 clusters of 4 drivers in parallel. One cluster has an impendance of (8/4=) 2 ohm. These 3 clusters in series makes a total impendance of 6 Ohm.

If we want to put 10 watts in the array, then the voltage is
the Square root of (6 Ohm * 10 Watts) = 7.75 Volt over the array.

The voltage over one cluster is calculated by the Voltage Divider formula, so we divide the impendance of one cluster (2 ohm) by the sum of all the impendances of the other clusters (2 Ohm + 2 Ohm + 2 Ohm) and multiply this with the total voltage drop across the array (7.75 Volt). Hence 7.75 * (2/6) = 2.58 Volt across each cluster. (The 2 other clusters are identical to this one.)

The parallel wiring of the 4 drivers in a cluster makes that the drivers in that cluster (and thus each of the 12 drivers in the array) has 2.58 Volt across it.

The power dissipated in the driver is (Volt*Volt)/Resistance.
So (2.58V*2.58V)/8 Ohm and thus each drivers sees 0.83 Watts of power. I didn't calculate it till now, but my gut feeling of 1 watt/driver wasn't that far off 😀

One last remark: if one of the clusters has more drivers than the other clusters (= different impendance), then the Voltage Divider will give different voltage drops for each cluster, and it is obvious there will be differences in power between the clusters (and thus drivers) in the array. This is called "power tapering", but I'll not elaborate on that further.
 
#17
OK guys I'm sorry if I'm being a simpleton here, but electronics always frustrated me way back when in school.

I understand the impedance part of series vs parallel and the power disipated by a driver makes sense. Bill says voltage is constant, so OK I get that must be typical. That leave us with current being the variable.

Now lets take it to the extreme. Let's say we have 10,000 drivers that each truly can handle a maximum of 1 watt. I can understand how wired in parallel they could handle 10,000 watts because each only gets their little piece. On the other hand, I don't understand how if you wire them in series the same is true. How can the first driver possibly handle that much current going through it?
 
#18
Now lets take it to the extreme. Let's say we have 10,000 drivers that each truly can handle a maximum of 1 watt. I can understand how wired in parallel they could handle 10,000 watts because each only gets their little piece. On the other hand, I don't understand how if you wire them in series the same is true. How can the first driver possibly handle that much current going through it?
Click to expand...

You are making an assumption that isn't true. You are assuming that Current is constant. Try assuming that Power is constant. Power is Current times Voltage.

If you took all 10,000 in series, the current going through all the speakers would be the same current, and would be the same as 1 speaker to get 1 watt. The voltage across each speaker would be 2.83 volts, assuming 8 ohm speakers. The power amp would need to output 28,300 Volts into an 80,000 ohm load in order to accomplish this.

To Contrast that, 10,000 in parallel, the power amp would only need to put out 2.83 volts, but 10,000 times the current into the load.

HTH

Doug
 
#19
Because it doesn't.

Put all the drivers in series in a black box and then mesure the huge impendance. Put a volage across it and you'll see you get a VERY low current. This will be the maximum possible current ANY driver can have.

Or try to look at it this way: if you put in a lot of electrons (current), then the same amount of electrons has to come out or else you should have a place where they all pile up. 😀 So if you only put in a certain amount of current, no more current can happen at a driver or this would mean you have the other drivers actually *delivering* current for the others.

Think about a river diverging into a lot of other rivers and then again coming back into one. It's impossible for any of those smaller rivers to have more water than the source river. If any of those would have the same amount of water, then the others are dry and then you only would have one river.

🙂
 
#20
OK guys one last time. Regardless of whether there's an amp that can handle it. In my example, can 10,000 watts be put through the drivers in series? Why wouldn't the first one just burn out immediately?

The river example is exactly how I look at it. 10,000 little rivers parallel can handle it, but how can you jam the same water through the 10,000 little rivers joined end to end?
 
Status
Not open for further replies.
Top Bottom