Guitar cab to subwoofer conversion???

[Joe Bee]

Not strictly "music electronics", but I am sure there are people who can help me out here!

I think I posted before about this 1x12" cab that I build for my SLO-Clone, but now that it is finished, with shiny clear coat and all, I noticed I don't like the sound of it at all. Somehow it just sounds very "dead" so I'd rather start over with a new one and use this cab for something else.
It is a closed back cab with a 12" hole, so naturally an active sub comes to mind. Problem is, I never build anything HiFi before, so I don't have a clue where to start. What kind of driver do I need, if I want a nice flat frequency response at moderate volumes rather than an ear-drum splitting "one-note" thud?
And can I just slap together a simple MOSFET power amp, or do I need something fancy with phase control, parametric EQ and whatnot?

I realize that asking you guys to explain subwoofer design in a forum post might be a bit too much, so I would be very happy about links that provide some general introduction to building HiFi cabs and amps, preferably those aimed at low frequency reproduction.
Google just keeps pointing me to those car audio dB shootout freaks and low quality home theater stuff, whereas I am looking for a rather more scientific view at a system that would actually be suitable for classical music reproduction at moderate volumes *gasp*


PS: I have the plans for the Electro Voice TL806 cabinet, and I understand that it was primarily targeted as a midrange PA speaker. Thus the difference between guitar and HiFi cabs cannot be as great as many people belief it is?

[Steve Conner]

It's probably time to go and read up on Thiele-Small parameters and fire up your copy of Excel

A closed-back cabinet has the transfer function of a second-order high-pass filter, so it's completely specified by three parameters: resonant frequency, Q, and midband gain. To get good quality bass, you want the resonant frequency as low as possible, and the Q not too high, usually around 0.7 to 1. To achieve this with a small cabinet, you need a driver with limp suspension, a heavy cone and a big voice coil and powerful magnet. The heavy cone leads to low efficiency, which is the same thing as midband gain.

The only way to screw up a closed-back cabinet is to end up with excessive Q and a high resonant frequency, which gives the booming "one-note" bass you mentioned. However, since it's only second-order, it's hard to make it really sound awful.

Once you add a port, the transfer function becomes fourth-order highpass, and there are two resonant frequencies, and twice as many ways to make a really awful sounding woofer. However, if you add a port to an acceptable-sounding second-order design, and cut the port length by trial and error to get two equal peaks on an impedance analyzer, it should work reasonably well. I think this makes the bass go about a third of an octave deeper, but I could be wrong.

[Joe Bee]

Thanks, Steve!
Yup, seems like I finally have to get to terms with the T/S parameters everybody keeps talking about. Maybe that would even help with guitar cabs in the future, making them less of a trial and error business.

The ideal speaker that you describe sounds pretty much like a car audio speaker to me, they seem to have lots of linear excursion which probably makes for floppy suspension, and they _look_ like they have a magnet about as big as a truck tyre. Too bad it is usually not possible to find the T/S parameters for the cheap ones.
Maybe I should go with an EV guitar speaker, other than the suspension it would also match your description. But then, the price is about an order of magnitude above what I would like to spend.

One thing I came across and that caught my attention though: http://sound.westhost.com/project48.htm This guy describes an active equalization circuit that he uses to make the frequency range linear down to some ridiculously low frequency, like 20Hz. Apparently he is using just an op amp filter to compensate for the 12db/octave falloff of the second order high pass. Since, as I already mentioned, I don't aim for particular high volumes, this seems like a reasonable approach to get low bass from a small enclosure.
I wonder whether this scheme has any flaws other than the terrible efficiency, since I am not sure how far to trust a guy who wont publish his circuits, but wants money for them.

[Joe Bee]

Browsing eBay for suitable drivers I limited myself to companies that also make guitar speakers, so I know their reputation, and I found a rather suitable speaker from Eminence. It's called the "Lab 12" with key parameters
f_s = 22 Hz
Q_ts = .39
V_as = 125 l
it also has a 80oz magnet, a 2,5" voice coil and a rated power dissipation of 400W and a horrilbe efficiency rating of 89dB. I went throgh the calculations and found that in my 26l cabinet, it would give a Q_tc=1 roughly, with an f_res of about 50Hz. The manufacturer even recomends it to be used in a sealed cab of that voulme.

I wont even need a lot of power to drive it: My reference maximum loudness is probably something like 90dB SPL, e.g. my 1W tube amp fully cranked on a 90dB/W/m cab at one meter. Since I will be down some 12dB if I drive the thing one octave below resonance (25Hz is my target frequency), I will need only 16W input. Seems like quite a waste on a speaker rated 400W! Anyway with a 60W amp I will have 6dB headroom, which is probably good.

[Steve Conner]

Yessir, that Eminence speaker sounds good for the job. If you feel the 400W rating is going to waste, just add an active equalizer to force the rolloff down another octave, and you'll have an excuse to deploy 12db more power! :-) I think the circuit to do this is called a Linkwitz transform equalizer.

I always suspected that car audio woofers have somewhat stiffer suspension, to help stop newbies from destroying them with poorly tuned ported enclosures. But noobs can wreck anything! I've seen footage on YouTube where a guy takes a subwoofer driver out of its enclosure altogether and plugs it into a wall socket.

BTW, Rod Elliott is God and can do anything he wants!

[Joe Bee]

Whoops, sorry about dissing Rod Elliott, I didn't come across his stuff before, so little did I know that he is in fact a divine entity. Probably it was his wrath that got me outbid by more than 3dB on that Eminence speaker, I am starting to fear that I will have to settle for something 10".

I did a lot more reading on the Linkwitz transform stuff, and the more I understand it, the more I like it. At first I thought that all it does is boost the lows to compensate for the roll-off, and heck, that is in fact all it does. But looking at it from the control therory point of view, with poles and zeros on the S-plane reveals the real magic, allowing you to choose not only Fs but also the Q at will and thereby taming the boomy-ness of a cheap, high Q driver.

Anyway, I am now starting to build an amplifier for the thing, which turns out to be harder than I imagined. For maximum simplicity I just hooked up an STK402 integrated, stereo, Class AB amp, but for some reason it puts a large DC component on the speaker, and generates massive 100Hz hum from my badly fitered power supply. Maybe I should build something from discrete components, but my only experience with audio electronics is with tubes, and I don't think a power amp with only four active components is possible in solid state - or is it?

[Steve Conner]

Hi Joe,

The old Hitachi MOSFET amplifier comes pretty close to the "ideal" of an amp with four components.

http://www.cpemma.co.uk/map_100w.html

Maplin sold it as a kit for many years, I still have mine :-) Though it's only about 100W, it managed to shake a light fitting off the ceiling at a jam session once, when used for bass with a Peavey 4x10.

The only drawback is that the MOSFETs need to be lateral types. Switched-mode power MOSFETs won't cut it. The Hitachi 2SJ50 and 2SK135 are long out of production, but Exicon and Magnatech make suitable replacements. They are expensive, but that's the price you have to pay for the simplicity of the rest of the amp.

The other transistors can probably be any fairly high voltage BJTs, such as the MPSA42/MPSA92, or 2N5401/2N5551.

The Linkwitz transform is a nice piece of math, isn't it? Before you get too excited, remember that the resistance of the voice coil changes a lot with temperature, changing the transfer function of the speaker, so the transformation won't be perfect.

[Joe Bee]

Steve,
that link looks quite interesting, it is a shame the IRF... Hexfets cannot be used, since I have about 100 of them flying around. But since I got the integrated amp to work, thats what I will use. For some reason all the components that were indicated in the data sheet were really neccessary, even though I didn't understand their purpose and therefore left them out. I mean, why would the voltage divider for the negative feedback go into a 33uF electrolytic instead of straight into ground?

I also realized there is a rather large community on the web around the "gainclone" / "chip amp" which is based on the National Semiconductor LM3886 power OpAmp. They seem to work without the funny extra components that my STK402 needs, so I ordered a couple of those.

If I took the time to build my own amp from discrete components then I would really like to make it Class D. The advantages in power supply and heat sinking requirements seem like a great bonus over any conventional topology. It also does not appear so much different from driving a SS tesla coil, but at the end of the day I guess I have to agree with Rod Elliott that "although very attractive, Class-D designs are not very DIY friendly". Getting the feedback right could be a major P.I.T.A., at least I would have no idea where to start.


Edit: Steve, since you are answering all the questions I nevered dared to ask about audio anyway, maybe you can have a shot at this one thing that has been bothering me.
As a first approximation, speakers can be taken to have a linear input/output relation, so if you pump in 10dB more power, you get out 10dB more sound. But once you start paralleling speakers, this relation seems to fail, and adding more speakers while keeping the input power constant (and just distributing it differently) increases the SPL by a considerable amount. I remember having read somewhere that replacing a particular 12" cab by a 4x12" would increase the sensitivity from 100dB to 110dB and thereby boost the efficiency from 3% to 30% or something like that. How is that in line with some rather basic principles such as energy conservation?
The only explanation I could imagine is that changing from a point source to a "wall of sound" would change the radiating pattern and get more sound at the microphone / audience, since SPL is not measured with an integrating sphere.

[Joe Bee]

I am back! And I have a few nice pictures for you:

First the subwoofer, it turned out quite well and it is now living its life in my old mans living room. It's mainly used as a piece of furniture there, but that's OK since it wasn't really build to rock anyway.

But at the end I needed a new guitar cab for my Soldano clone, and this time I opted for an ultra-compact design. The box is only 13x13" and 6" deep with an open back. It still sounds remarkably good with an unbelivably powerfull bottom, which is probably due to the speaker I used. It's a 10" pro audio speaker from the 80s with a cast aluminium basket and a voice to match the infamous EVM L12, albeit in a smaller package.

Oh, and the skin was kind of an experiment, I took some ordinary fabric, glued it on with wood glue and then soaked it with clear coat. I am going for a look similar to Steve Vai's Jem with the flower fabric finish, just without the flowers :-) I may take a few more coats though, it's unbelivalbe how much the fabric soaks up!