Feasibility study

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
So I had some free time and decided to come up with a back-of-the-envelope hybrid dipole/monopole design that would deliver almost-constant power output throughout its [full] passband as well as smooth off-axis response, at a reasonable price point. SPL target is at least 110dB at 1m.

Driver selection and estimated optimal crossover points have so far been decided as thus:

(2x) Seas 27TDFC, dipole
SB Acoustics SB17NRX35, HP at ~600Hz, LP at ~1600Hz, dipole
Dayton RS270S-8, LP at ~200Hz, dipole, at least 30" above the floor for full-space radiation
Dayton RSS390HF-4, active LP, monopole

Tweeter - there are two for obvious reasons. I chose the 27TDFC over the 27TBFCG because the TDFC has lower high-order (>=4th) distortion between ~500Hz and ~2500Hz, where the distortion products end up in the most sensitive regions of our hearing. See http://www.zaphaudio.com/tweetermishmash/Seas_27TDFC-HD.gif vs http://www.zaphaudio.com/tweetermishmash/Seas_27TBFCG-HD.gif (I hope Mr. K doesn't mind the direct linking). Between about 550Hz and 2200Hz, the 27TDFC has lower 4th- and 5th-order distortion. Above that, the TBFCG wins out, but we're already beyond ~10kHz where hearing is less sensitive.

Moreover, the TDFC is subjectively more aesthetically pleasing. The top end will probably need to be dropped a bit a la Zaph's SR71, but I think that will be a small price to pay for theoretically better distortion performance. Either one should be viable, though.

I would've considered the D26NC55, but that's been discontinued. The DQ25SC was another option as well, but I don't know how well it performs when crossed below 2kHz. A low crossover point is necessary because the size of the mid determines the minimum baffle width, which must stay low (more on this later). One could use a smaller mid, but I don't want to go that route.

Mid: Again, it must have good distortion numbers in its passband. This time I'm looking at about 80Hz to 1000Hz. Nothing else even comes close in Zaph's tests for reasonable amounts of money save for Vifa's XG18, but that driver doesn't behave too well at high excursions. The AA 6.5 is too good to be true; it's not available anymore. I've high-passed fairly high because I want to minimize the stress on this driver. I've got something bigger right below it, anyway.

Midbass: Selection here was not as rigorous. Not because the performance in this region is unimportant, but because there isn't fantastic relating data for larger drivers like there is with smaller drivers (Zaph et al). I just looked for something with similar sensitivity and enough output at a moderate cost, though the distortion tests at MFK Projects show pretty good performance.

Bass: The Peerless 12" XXLS performs about as well (check diymobileaudio.com), but is more expensive. I don't have WinISD/Unibox right now, but I think ported is the best solution here. Ideally, I'd have three or more additional discrete subs around the room, Geddes-style.

The baffle width will be as low as possible around the mid, going up (not linearly) to about 15" around the midbass. This will allow us to operate mostly underneath the dipole peaks. See http://www.musicanddesign.com/Dipoles_and_open_baffles.html. The power output matching is not ideal, but close: http://www.musicanddesign.com/PowerMatching.html

The crossover will be passive for the upper three drivers and active between the top and the Dayton. Two 41Hz AMP15 units should do the trick for amplification.

Will Zobels be necessary for any of the drivers? Any other suggestions? A system like this would probably run about $1500-$1700 including amplification.
 
So I've changed my mind. List is now:

(4x) Seas 27TDFC/27TBFCG
(2x) Dayton RS180-4
(2x) Dayton RS270S-8
(2x) Dayton RSS390HF-4

Also, I've decided to play through a fully-active system rather than a hybrid. So, now I need to determine the power that goes to each driver. Feel free to let me know if I make any mistakes somewhere. As a reminder, my crossovers are at about 200, 500 and 1800. The mid-bass doesn't do very much, I know, but I don't like having the mid excurse so much.

I start with the power distribution table here. I extrapolate a 50/50 split between the T+M and the MB+B drivers (I'm giving myself some headroom later so the errors won't mean much).

Of the 50% that goes to the T+M drivers, I estimate that 70% goes to the T and 30% goes to the M. 3/4 of 1/2 is 3/8, 1/4 of 1/2 is 1/8. This means that the T output is -4.6dB from the overall output, while the M output is -8.2dB.

Of the 50% that goes to the MB+B drivers, I estimate that 40% goes to the MB while 60% goes to the B. Output for the MB is -7dB, while B's output is -5.2dB.

The approximate sensitivities per W are as follows (from treble to bass): 88, 85, 88, 88. The last one might have a 6dB bump because of the half-space radiation, but I'm not sure if the response graph already includes it. Now I check if we lose any output with SL's spl_max1.xls spreadsheet. Nothing's lost for the tweeters, obviously. The M loses about 2dB on the desired 180mm baffle width, and the MB loses about 2dB on the desired 250mm baffle. The B isn't a dipole, so now the sensitivities per W after dipole loss are:

88, 83, 86, 88

If we use 110dB as a target output, then the required output from each driver is:

110-4.6=105.4dB
110-8.6=101.8dB
110-7=103dB
110-5.2=104.8dB

The increase in output relative to 1W is then:

105.4-88=17.4dB
101.8-83=18.8dB
103-86=17dB
104.8-88=16.8dB

Using the formula P=10^(dB/10) derived from the regular power log formula, we get the following power requirements:

T=54.9W~=75W
M=75.8W~=100W
MB=50.1W~=75W
B=47.9W~=60W

So, our midrange needs more power than the bass does. Whatever. The required voltage comes from

P=V^2/R

Which I do for the one that requires the most voltage. That would be the mid. V is sqrt[100(6)]=24.5V, where 6 is the lowest impedance in its passband.

Here is where I get tired and decide that the VA for the transformer (10-channel amp, anybody?) is simply the sum of all the W multiplied by the safety factor of your choice (in this case let's say 2). VA=1540. For a Class D, I'd drop it down to a factor of 1.25.

Now, can somebody tell me where I went wrong? Do dipoles suffer from the 6dB sensitivity loss from the baffle step? Or are the driver's sensitivities already given for full-space radiation?
 
454Casull said:

Now, can somebody tell me where I went wrong? Do dipoles suffer from the 6dB sensitivity loss from the baffle step?

It´s worse than a single 6 dB step!:eek:

I don´t see where you accounted for the 6 dB/octave dipole loss. if you try to "operate mostly underneath the dipole peaks", your calculation would have to look like this for the SEAS:

Dipole peak at 10k would lead to an effective baffle width of 11,5 cm. That´s just as wide as the tweeter baffle! Because you operate the tweeter in dipole config, your peak sensitivity at 10 k would be 88+6= 94 dB. Down from there you will loose about 6 dB/oct. From 10 k to 1,8 k (stated crossover) this results in ~2.5 x 6 = 15 dB loss. At 1,8 k the effective sensitivity of the SEAS will be 94 - 15 = 79 dB.
Obviously you need to raise the tweeter Xover at least to 2.5 kHz or forget about that "operate mostly underneath the dipole peaks" rule.

"Operating underneath the dipole peak" allows for a 3-octave span for any driver - at max. With four drivers you can span the 10-octave range of 20-20000 Hz comfortly. But you need to distribute the individual drivers ranges more evenly IMHO.
 
Rudolf said:


It´s worse than a single 6 dB step!:eek:

I don´t see where you accounted for the 6 dB/octave dipole loss. if you try to "operate mostly underneath the dipole peaks", your calculation would have to look like this for the SEAS:

Dipole peak at 10k would lead to an effective baffle width of 11,5 cm. That´s just as wide as the tweeter baffle! Because you operate the tweeter in dipole config, your peak sensitivity at 10 k would be 88+6= 94 dB. Down from there you will loose about 6 dB/oct. From 10 k to 1,8 k (stated crossover) this results in ~2.5 x 6 = 15 dB loss. At 1,8 k the effective sensitivity of the SEAS will be 94 - 15 = 79 dB.
Obviously you need to raise the tweeter Xover at least to 2.5 kHz or forget about that "operate mostly underneath the dipole peaks" rule.

"Operating underneath the dipole peak" allows for a 3-octave span for any driver - at max. With four drivers you can span the 10-octave range of 20-20000 Hz comfortly. But you need to distribute the individual drivers ranges more evenly IMHO.
Hm. I just assumed that the tweeter would be firing into half-space no matter what because of the minimum baffle width around the M, and that the rear tweeter wouldn't affect sensitivity.

Checking now - Edge tells me that on a 180mm baffle, a 25mm source (off-center for better response, but this doesn't affect much) doesn't drop until 800Hz or lower. Doesn't this mean that the tweeter doesn't suffer from dipole loss because its radiation doesn't interfere with the radiation from the rear tweeter? But then when I check "Open baffle" the response goes really ugly.

For the other dipole drivers, what I did was use the spl_max1 sheet and avoided the region where the boost got to about 6dB at the top end of the passband. I'll go and double-check the figures... Another way to do it is use d/w=0.5 where for the mid, minimum d=0.09m, so d/0.5=w which allows a max operation of ~1900Hz at the dipole peak. The minimum d for the MB is 0.15m, perhaps more if we want more "meat" around the flange, but it still gives us a potential top end of ~1150Hz. I don't use the 10" that high because stored energy becomes an issue beyond about 500Hz. Also, I was wrong about the figure in an earlier post; the baffle can be up to ~450mm wide for the MB (cross-checking with the SL sheet), which will help its low-end sensitivity before EQ. The midbass' required power is now ~50W. That, or I can drop the crossover to 150Hz instead of 200Hz, but the required power then goes to 100W for the MB (and slightly less for the B).

As for your other comment about the distribution; I would have liked to bump up the mid's low-pass but the RS52, which I would have used, is a sealed-back unit. So the smallest mid I'd use is probably the RS180 as the ~5.5" units don't seem to be that great, what with worse distortion and more restricted backs and all.

I was also wrong about the voltage requirements. It should be the highest impedance, not the lowest impedance in the passband that is used for the calc. So sqrt[100(10)] is +/- ~32V.
 
454Casull said:

Hm. I just assumed that the tweeter would be firing into half-space no matter what because of the minimum baffle width around the M, and that the rear tweeter wouldn't affect sensitivity.

Checking now - Edge tells me that on a 180mm baffle, a 25mm source (off-center for better response, but this doesn't affect much) doesn't drop until 800Hz or lower. Doesn't this mean that the tweeter doesn't suffer from dipole loss because its radiation doesn't interfere with the radiation from the rear tweeter? But then when I check "Open baffle" the response goes really ugly.

I believe you need to decide for yourself first, how much true dipole radiation you really want. Is it just open baffle or do you want the constant directivity a true dipole can offer? On-axis simulations only will tell you only a small part of the truth.
May be my simulations can help to come to a conclusion.
 
Rudolf said:

I believe you need to decide for yourself first, how much true dipole radiation you really want. Is it just open baffle or do you want the constant directivity a true dipole can offer? On-axis simulations only will tell you only a small part of the truth.

Not sure what you mean, but thanks for the link.

Don't think I'll progress from this stage without actually putting metal to wood. Might have to pick up a DCX to make things easier.
 
I wouldn't use the RS180 for midrange purpose. The SB17NRX's wind noise shouldn't be an issue even in a dipole if you cross it above 400 - 600 Hz. I'm listening to both drivers now. Their difference in midrange clarity is night and day to my ear. The RS180 is crossed at 1.6 kHz with 27TDFC and the SB17 at 2.0 kHz with Vifa DQ25SC. If the wind noise is still a concern, another choice would be the Peerless 830883.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.