Trying to design the world's simplest sub

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have a system with an Arcam AVR300 amp, running two very old but modded Monitor Audio Monitor 9 speakers, which I have converted to run bi-amped, and love to bits. More importantly so does my wife, so much so she'll let me get a sub if we keep them. They are a two way stand mounted black box design, with 6.5" mid-bass units, giving bass down -3 db at 55Hz. See here for a little review.

Anyway, they are a little dry in the bass, and what with being used for films and CDs with bass that didn't exist in sources from 1991, I'd like to build a sub to help them along. As the Arcam amp will digitally filter out the LF channel at any crossover from 40 to 130 Hz, I'd like to see if I can build a very simple sub, with a mono amp like the ones from BK electronics, this one being a candidate, Modules. I like the idea of something that will very smoothly roll off down to the point where room gain kicks in, my room being 3.2m wide, 2.4m high and 10m long. The lowest modes should be (according to Room Mode / Standing Wave Calculator) 17.2, 34.4, and 51.6 and so on down the length, and then multiples of 54 across and 72 vertically. I feel that if I can control both the cross over frequency and the gain of the sub, and play some sweeps via the computer, I'll hope to fit the sub in as well as I can. I listen to a lot of two channel music, from Suzanne Vega and Gillian Welch to Cookie Monsta and Foo Fighters, and we watch films too, so I want a sub to work for all applications.

At the other end of the spectrum, I have no idea what the low frequency content is of films is going to be in the future, but from what I have read, I understand that there are already titles out there with big 8.5Hz pulses in (Black Hawk down, see The New Master List of BASS in Movies with Frequency Charts) so if I'm going to build something, I'd like to make sure it will produce bass as low as practicable in the room. I'm not trying to make it flat, I'd just rather it didn't drop off suddenly. I would also like to try to make sure it will not get blown up. I'm not worried I'll blow it up, but I have kids who will find out how, unless I make it impossible at design time (I have learned this philosophy when making furniture).

So, I've been trying to find the ideal driver and cabinet combination for a sealed sub, not too big, with decent volume at allowable cone excursions. This seems like a topic that has been asked about a lot so rather than just ask for general advice, I have tried to already answer the questions I had when I had when I started researching this, and I'm posting here in case I've got it wrong, in the hope someone will notice before I spend a load of cash on something that will fail. I'm going to run through the various options I have considered in a chronological fashion, highlighting the questions I have come across on the way.

So, the first thing I did when I started looking at this was to google 'Best Sub Driver' and start reading what I found. I pretty soon came across this site, which was useful, The Subwoofer DIY Page and led me to download and install WinISD Pro from LinearTeam which allowed me to try modelling some driver and cabinet combinations. I have Windows 7 32 bit. and until I set it to run in Win 98 compatibility mode, it crashed every time I tried to do anything. I also found these pages http://www.diyaudio.com/forums/multi-way/62383-peerless-xls12-vs-scanspeak-21w8555.html and The Ariel and the ME2 discussing the Scanspeak 21W 8555 8" driver and and 10" 25W 8565 drivers. So, I modelled these using the recommended flattest profile with Qtc = 0.7, giving a 32 litre box for the 8555 and a 118 litre box for the 10" 8565 and looked at the transfer function. All fine, the bigger one goes lower.
An externally hosted image should be here but it was not working when we last tested it.


Next, I looked at the cone excursion, and raised the signal level from the initial 1W until the very end of the low frequency plot at 10 Hz was just below the limit for the drivers.

An externally hosted image should be here but it was not working when we last tested it.


This showed that for the 8" Scanspeak 21W 8555, the drivers would max out at 11.8 W, and the 10" Scanspeak 25W 8565 would begin to exceed the linear region at 5.7 W. I know that playing at 10 Hz is not what these drivers are designed for, but if you look at the curves, the values are fairly flat from 30 Hz to 10 Hz, and if you were to up the power to find where they reach xmax at 30 Hz, it is 16 W and 12 W respectively, so no great change. Fianlly, I plotted the overall Sound Pressure Level.

An externally hosted image should be here but it was not working when we last tested it.


This brings me to my first unanswered question: How loud does 97 db at 100Hz or 86 bd at 20 Hz sound. I have no idea, yet.

These sims didn't convince me that these two drivers were perhaps what I wanted for music and home theatre frequencies, so I next looked at the bigger Scanspeak drivers, the 25W/8567-SE, but the only difference in results was that the cone extension would got to 7.5 mm instead of 6.5 as for the 25W 8565, but still only a marginal difference. Next, I tried the 10" Scanspeak Discovery 26W/4558T00 and the 12" 30W/4558T00. These both have linear extension to 12.5 mm, and this seems to make a massive difference.

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


So, the 10" Discovery 26W/4558T00 goes to 157.5 W before going non-linear at 10 Hz, and the 12" 30W/4558T00 goes to 102.9 W. The 12" driver is louder, even given the power difference. Also, these figures are for boxes of volumes of 24 litres and 55 litres respectively, which seems fine for the average living room.

Next, I looked at the 10" Dayton RSS265HF-4 and the monster 15" Dayton RSS390HF-4. The ideal box for the 10" driver is a mere 31 litres, but the 15" beast requires 160 litres. That isn't so likely to impress my wife, but it isn't crazy exactly.

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


As you can see, the 10" Dayton driver appears to have more efficiency from 40 Hz up, but is almost the same as the Scanspeak 26W/4558T00 below. The Daytons both handle about 160 W before going non-linear, so the same as the Scanspeak 26W/4558T00. The 15" RSS390HF-4 shows very high efficiency at low frequencies, going much louder than the 10" or 12" drivers.

Next up, I tried the Monacor Stage Line SPH-390TC and SPH-380TC.

An externally hosted image should be here but it was not working when we last tested it.


Obviously, these need a bass port, they are very efficient, but not what I'm looking for.

Finally, I looked at the 12" Tangband WQ-1858, the 10" Seas L26RFX-P, and the 10" Peerless 830669. These work in boxes of 82, 51 and 148 litres respectively, to give varying outputs in terms of power, but quite similar overall sound levels. The Peerless is about one third the cost of some of the others, and the Seas unit is also a lot cheaper.

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


The Tangband WQ-1858 appears to have a very flat response down into the low end in the recommended cabinet size, which might be a useful thing as I don't know where the crossover will be until I've built it. The measured response at http://www.europe-audio.com/document.asp?document_id=5601&link=datasheets\TBS\WQ-1858.pdf also looks quite flat in this region from 20 to 100 Hz. The Seas looks reasonable, and is not expensive, and the Peerless seems about the same, although both seem less flat about the possible crssover region.

So, I'm wondering if I should got for the 12" Tangband which is very flat, costs £170 ish and needs 80 litres, or the 10" Dayton RSS265HF-4, which will work in a 30 litre box, and only costs about £140 or so.

My final topic to ponder is what protections are built into the amp, and if I need to work out a low frequency filter to stop the cones getting damaged should the rest of the system misbehave. I have read about it being hard to execute suitable passive filters at such low frequencies, but I don't know if this is necessary for this application.

Anyway, I'm still researching this but I hope to order the bits as soon as I'm happy I'm buying the right things. All suggestions gratefully recieved! :)
 
Last edited:
Just another Moderator
Joined 2003
Paid Member
The reason for the TangBand looking much more flat is it's low voice coil inductance. You have modeled including the effect of le by the looks of things.

The sacrifice with the tangband is much lower efficiency.

Just a suggestion, I would try comparing the curves in a normalised way (perhaps with le not moddeled) so you can at least make a direct comparison of the rolloff of the drivers.

As for the question how loud will 20Hz be at 86db I can't really answer except to say, you won't be able to hear 20Hz, you can feel it (if it has enough power) but not hear it.

You could try modelling the 15" driver in a smaller box just to see how it goes.

Tony.
 
Fanatic
Joined 2009
Paid Member
I agree with turbodawg.

Unless you post size and price limitations and goals there is not much else to recommend really. Increased size will go usually lower in frequency and/or have higher efficiency. I would look at other solutions than closed boxes unless there is a special reason why you want closed box subs.
 
As for the question how loud will 20Hz be at 86db I can't really answer except to say, you won't be able to hear 20Hz, you can feel it (if it has enough power) but not hear it.

Tony.
Tony,

20Hz at 86db sounds the same loudness to normal ears as 30 dB at 1000 Hz, quite quiet.
A 5 dB change at 20 Hz sounds twice/half as loud, while it takes a 10 dB at 1000 Hz to sound twice/half as loud.

Low frequency perceptions are different for different people, what may be imperceptable to you at 20 Hz might be clearly perceived, or even sound annoyingly loud to others.

That said, conversational level is around 70 dB, for 20 Hz to sound that loud requires about 95 dB, which subjectively would sound almost 4 times louder than 86 dB.
In other words, 86 dB is not going to cut it for T-Rex footfalls and helicopter chugging;).

Art
 
Go ported.

Lots of large drivers (and lots of power) is the requirement for loud, low bass from sealed cabinets. Ports allow more output for any given bass driver, at the expense of a larger cabinet.

If its HT, aim for, say, 95dB at 30Hz or lower.
Make sure you add an infrasonic filter - anything you're likely to build isn't likely to play 10Hz loud and clean.
 
The reason for the TangBand looking much more flat is it's low voice coil inductance. You have modeled including the effect of le by the looks of things.

The sacrifice with the tangband is much lower efficiency.

Just a suggestion, I would try comparing the curves in a normalised way (perhaps with le not moddeled) so you can at least make a direct comparison of the rolloff of the drivers.

I have modelled the Tangband, Scanspeak 12" and the Dayton 15" with Le set to zero, and it does flatten all the curves out to zero on the transfer function graph at higher frequencies. Is it possible to achieve this in practize by using a Zobel network to compensate for the driver coil inductance?

An externally hosted image should be here but it was not working when we last tested it.


As for the question how loud will 20Hz be at 86db I can't really answer except to say, you won't be able to hear 20Hz, you can feel it (if it has enough power) but not hear it.

You could try modelling the 15" driver in a smaller box just to see how it goes.

Tony.

Thanks, I gave it a go, but it bumps up in the 80-100 Hz region, and I'd like to try to keep things flat in this region to make it easier to integrate the sub with the existing speakers.

Hum. I would suggest a high excursion 10" or 12" in a ported box.

The right long throw 10" in about 2.5 ft^3 with a low tune will give considerable low end output while still sounding excellent on music.

Dayton Audio TIT280C-4 10" Titanic Mk III Subwoofer 4 Ohm 295-414
TC Sounds Epic 10" DVC Subwoofer 293-656

Thanks again, but I am a bit wary of going the ported route. I have a wooden floor, fairly bare plastered stone walls with the odd glass covered print on, and a very live acoustic. If you clap you have half a second to listen to it. It is my understanding from the reading I've done so far that ported designs do resonate and have a slower decay than sealed systems, and with such a live room, I don't want to add anything that will delay transients or add to the echo, so I'm concentrating on a sealed system. I had a look at the two drivers you mention in WinISD, but they do seem to need ports to reach down.

I agree with turbodawg.

Unless you post size and price limitations and goals there is not much else to recommend really. Increased size will go usually lower in frequency and/or have higher efficiency. I would look at other solutions than closed boxes unless there is a special reason why you want closed box subs.

Thanks, I have had a chat with my wife about sizes that would be acceptable, and it seems we are really looking for something of 100 litres or less. The total budget could be up to £400 or so for the whole thing, so maybe £200 or just over for the driver, the 200W amp from BK is £90 delivered and £100 to build the box. So far, the only one I've had to rule out on size is the 15" Dayton, the cabinet size of 160 litres to get a maximally flat response is too big. My favourite so far is the Scanspeak 12" 30W 4558. This is supposed to be the world's simplest sub, and I'm thinking what could be more simple than a 12" driver in a 55 litre box?

Another thing I've realised I don't know is the characteristic of the filter the Arcam amp uses for its crossover. I've emailed them to ask, just out of curiosity. I did find review that said it sounded best at 90 Hz while giving no details of the speakers or sub used. Hmm, I may have to check that myself.
 
Tony,

20Hz at 86db sounds the same loudness to normal ears as 30 dB at 1000 Hz, quite quiet.
A 5 dB change at 20 Hz sounds twice/half as loud, while it takes a 10 dB at 1000 Hz to sound twice/half as loud.

Low frequency perceptions are different for different people, what may be imperceptable to you at 20 Hz might be clearly perceived, or even sound annoyingly loud to others.

That said, conversational level is around 70 dB, for 20 Hz to sound that loud requires about 95 dB, which subjectively would sound almost 4 times louder than 86 dB.
In other words, 86 dB is not going to cut it for T-Rex footfalls and helicopter chugging;).

Art

Thankyou so much, this is so useful! To get the maximum from the T-Rex footfalls scene is exactly what I want from this in terms of extension and volume. If I can get that without sacrificing timing, then I'll be very happy.

Go ported.

Lots of large drivers (and lots of power) is the requirement for loud, low bass from sealed cabinets. Ports allow more output for any given bass driver, at the expense of a larger cabinet.

If its HT, aim for, say, 95dB at 30Hz or lower.
Make sure you add an infrasonic filter - anything you're likely to build isn't likely to play 10Hz loud and clean.

This is definitely an area I need to look into, thank you. Do I have to go active for this component? I have built ZRA's and circuits for work, so I can do this, but if I can keep it simple I'd rather. It won't really be the 'World's Simplest Sub' if I have to build in a 5th order Tschebyscheff cut off...:( There may even be a subsonic filter in the output of the amp, and I hope the Arcam guys can tell me if so.

Thanks for all the replies, I'm looking at finish at the moment, and working out if I want it finished in wood veneer, gloss white or gloss black. I've also got to pick a volume control, an RCA socket and I'd love to fit a proper old school needle to the front to show the power. The BK amp has an output for their LED dial giving up to 775 mV, but I'd like a serious old school analogue dial. Will I be able to get one from an ancient all in one Akai unit like my Dad had, with a brushed metal finish?
 
Thanks again, but I am a bit wary of going the ported route. I have a wooden floor, fairly bare plastered stone walls with the odd glass covered print on, and a very live acoustic. If you clap you have half a second to listen to it. It is my understanding from the reading I've done so far that ported designs do resonate and have a slower decay than sealed systems, and with such a live room, I don't want to add anything that will delay transients or add to the echo, so I'm concentrating on a sealed system. I had a look at the two drivers you mention in WinISD, but they do seem to need ports to reach down.

I'm sorry, but you have either read something incorrect or have misunderstood what you have read. It is a complete myth that sealed designs (always) sound better. The designs you are currently entertaining will not sound any better than a proper ported design on music, and they will have mediocre extension for HT.

Are you performing time correction on this system? Because that is pretty much the only way to get transient correct bass.
 
Last edited:
I'm sorry, but you have either read something incorrect or have misunderstood what you have read. It is a complete myth that sealed designs (always) sound better. The designs you are currently entertaining will not sound any better than a proper ported design on music, and they will have mediocre extension for HT.

Are you performing time correction on this system? Because that is pretty much the only way to get transient correct bass.

Thanks Turbodawg, I may well not have the full picture yet, and if this thread can help me to correct any misapprehensions I have picked up before I set out to build something, I'll be very grateful to all involved!

I can delay each channel individually in the amp, but this might be more for video sync than for audio purposes.
 
Here's a random WinISD comparision of group delay I pulled off the net:

s42qsj.png


Let me assure you that the 1ms difference at 50Hz will not be audible. 5ms at 30hz will not be audible, nor any of the huge spike below that.

What will be completly audible is if the subwoofer is correctly time aligned to the mains.

I'm sorry to say, there are many poor subs and poorly integrated subs out there, many of which are sealed.
 
Last edited:
Just another Moderator
Joined 2003
Paid Member
As for the question how loud will 20Hz be at 86db I can't really answer except to say, you won't be able to hear 20Hz, you can feel it (if it has enough power) but not hear it.


OK maybe I made an uniformed gross generalisation ;) I was under the impression that you cannot actually "hear" 20Hz, I personally can detect it (above a certain threshold) through the unpleasent pressure I feel in my ears, but I would not be able to say ha that's a 20Hz sine wave. I can start to actually hear it when the volume is cranked up but that is various resonances (and perhaps some 40Hz second harmonic) when my crappy speakers start to rattle due to the pressure ;) Maybe some people can actually hear a PURE 20Hz sine wave, a quick google search suggests this is so. I stand corrected!!

Tony.
 
The designs you are currently entertaining will not sound any better than a proper ported design on music, and they will have mediocre extension for HT.

I'm looking at the curve I posted for the WinISD model of the Scanspeak 30W. This shows -3 db at 40 Hz, -6 db at 30 Hz and -12 db at 20Hz. Should I worry that this will be insufficient for my application? I'm not looking to shake stuff off the walls here, so it may not get used very loud, I just want some solidity and space down in the low end.

I have read some posts suggesting that at the low frequency limit there could be up to 12 db of room gain per octave. HTGuide Forum - Guestimating room gain Linearteam - For everyone wanting to predict "Cabin Gain". However, if this starts at below the lowest mode, 17Hz, it's not going to make a lot of difference.
 
Going for active circuitry is always a requirement with dedicated subwoofers. Adding in a 20Hz low-cut is just a case of another TL072 with some caps and resistors.

Here's some reading for you - scroll down for the stuff related more to what you're after, especially the bit about Linkwitz Transform.

DIY Audio Articles

Chris

Thanks, I am waiting to hear back from Arcam if they have a high pass filter in the LF output for the amp. The bit of the circuit is here,

An externally hosted image should be here but it was not working when we last tested it.


but I have not figured it out yet. I guess if there is not then I'll have to add one.

That's a great link, thanks for that!
 
Just another Moderator
Joined 2003
Paid Member
I have modelled the Tangband, Scanspeak 12" and the Dayton 15" with Le set to zero, and it does flatten all the curves out to zero on the transfer function graph at higher frequencies. Is it possible to achieve this in practize by using a Zobel network to compensate for the driver coil inductance?

The driver won't necessarily behave exactly as it does in the sim, you may find that the natural tendency to roll off due to the VC inductance actually combines with a natural tendency for the driver to be more efficient at those higher frequencies so it will even out (it might not either!)

the main reason I suggested turning it off was so that it was easy to line up the "zero db" point so you can get a direct comparison between the extension each driver has, however as you are now questioning, are you going to have enough spl at those low frequencies with 6db or more down at 20Hz...

If you do decide to remain with sealed you probably need to choose the driver with the highest power handling and do a linkwitz transform on it. but take all this I'm saying with a grain of salt as I have no experience with high output down this low!

Tony.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.