The below post is probably the 2,467,149 th time this has been hashed and rehashed, so Im surely NOT doing anything new, but
hopefully its recent and at least marginally helpful to someone.
I'm in the process of also doing whats being discussed on another topic:
http://yabb.jriver.com/interact/index.php?topic=85335.0 What's a little different about my approach is that I have a background as an EE with a huge amount of time listening to
and repairing audio and TV equipment. I've not had a great experience with super high end audio but have heard several really
good systems.
I'm trying to do the same thing as many others, to design the "ultimate high end audio system on a lo-fi budget"
Well, there really is no such thing. The best that most of us can achieve, without spending $50K is, as Monty Python put it, to be content with the following philosophy:
"I dont know much, but I know what I like"
I like rock, disco and LOTS of high frequency response. (I can hear the moans and groans over "disco..")
Being an EE with the ability to design any part of such system, its totally pointless to reinvent the amp or speaker or EQ.
Really good examples of them are everywhere.
The question is, can I rehash this with engineering experience and find anything new. Maybe not. Maybe it's just fun trying.
The center of any audio system is speakers. They convert electrical audio signals to mechanical vibrations.
A good set is easily North of $ 5,000. I'm cheap.
In system design, we start at the output. Yes, actually the speaker is not the end of a hi fi system, the room and
accoustical treatments are, but the loudspeaker is the end of the electrical chain and as far as the listening room or studio
goes, there is such a huge amount of variablenss there I can't even begin to touch that question.
The initial topic was a multi channel audio system, not only L and R channels, but sub channels of L and R that are
divided into frequency bands. Such a design has huge benefits such as eliminating the speaker crossovers and
allowing the use of smaller, lower power and higher accuracy audio amps - there are huge cost reductions in the latter.
As to speakers, as with any other analog component, its dollar vs. audio quality. I'm looking for 'better" audio quality
but don't need, want or necessarily appreciate the absolute top of the line quality because the first thing I'm going to do
when the system is up and running is shove all the EQ setting from 10 Khz and up to the top. So much for fidelity.
There is a huge range of drivers (individual speakers) available from $20 to $1000 +. Their specs vary wildly.
Is there anything to be learned and cost efficiency to be gained by looking at a large number of them and comparing them?
Yes, if I can save $5000 on speakers, I can buy real nice amps for that kind of money!
Analog components have several characteristics:
1. power level.(how high it is) A 1000 watt amp doesnt work well with a 50W driver, or at least for long.
2. bandwidth. ( how wide it is) This is the range of frequencies a component or system can work with.
3. fidelity.( what quality it is) This is a catch all for a dozen tech terms such as total harmonic distortion (THD).
Thats "THD", not "THC" which music lovers have historically associated with loud music...THD is legal in all 50 States
although excess amounts are considered a crime with audiophiles.
Its VERY difficult to optimize all three, so I'm trading off power for fidelity starting with smaller amps, drivers and a smaller listening room.
The beauty of such a multi channel system is that one amp channel is not handling all the audio bandwidth from 25 Hz to 40 KHz.
That is a very difficult and expensive thing at 200 watts per channel. A very cheap amp can produce 50 watts from 50 - 1000 Hz
just to drive a bass driver.
With lower powered amps, the drivers can be lower power. Lower power and higher fidelity go hand in hand.
The goal of driver selection was to have approximately 20-50 watts per driver which adds up to a respectable 120-300 watts
total, thats enough to deafen you in a small room.
So, #1, power, is arbitrarily defined at ?? W (pick a number).
For bandwidth, the assumption is L and R channels with three drivers each, bass, midrange and treble.
The multi channel system assumes those three sub channels are supplied, one to each driver.
Its a total waste of time to then select a driver with a much wider bandwidth than the sub channel amp
is intended to supply.
The bandwidths were somewhat arbitrarily defined at:
bass = 50 - 1000 Hz,
midrange = 1000 hz to 5,000 Hz
treble = above 5,000 Hz
The ranges are fuzzy, they depend on what drivers are available and what their frequency responses are.
Ill gladly take "fuzzy" over the cost and inaccuracy of a crossover network!
Fidelity. Thats not easy to define in just a driver, because its a component that relies heavily on the speaker enclosure.
Fidelity or quality is assumed to be buying from a reputable manufacturer and at least a moderately priced unit.
After that, fidelity was assumed to relate to bandwidth in:
a. driver bandwidth that closely matches the assumptions above for bass, mid and treble
b. FLAT frequency response across the driver bandwidth
c. change in driver response that was less than 3 dB across the bandwidth
"Bandwidth" takes a different meaning with a multi channel system like this. In an ordinary speaker design with crossovers
I'd assume we could use more of a given drivers bandwidth. With this system, I chose drivers that had a very flat response
over some frequency range, with about a 3dB or less change, and not use the remaining bandwidth of the driver.
That's a waste of capacity, but its useless capacity.
With those assumptions in mind, I spent three days searching the web for driver manufacturers and suppliers and there
are hundreds. The first choice, just going to the high end drivers, turned out to not be what I wanted.
Oddly, the final selections turned out to be the somewhat less expensive units and thats good in the contet of a less expensive system.
Whats critical about considering a driver is the graph of its SPL response. Heres one for example from a well known maker, Dayton.
Dayton isn't the quality I'm looking for as my budget is upwards of $5000, but they make decent equipment, a lot of it and its reasonably priced:
http://www.daytonaudio.com/index.php/loudspeaker-components/loudspeaker-drivers-by-type/woofers-1/pa255-8-10-pro-woofer.htmlThis is a 10" woofer, $58. Click on the Spec Sheet link and look at the "Frequency Response" graph on the right.
The ideal response is square. A square, perfectly flat response has a wide flat line across the top and sides that are nearly straight down.
That's impossible in reality.
Nothing in the graph is flat except the roll-off on the left and its going down hill fast. This is a poor frequency response.
The only part of this drivers response that remotely resembles "flat" is from 140 - 400 Hz and thats hardly "flat" with more than a 5 dB change in level.
This is not a driver I want. But sometimes its useful to know what we don't want. While theres no such thing as a driver that only flaty reproduces signals
from 50 - 1000 Hz, I'd prefer to have a response thats very flat in that range and rolls off quickly above and below.
Searching through dozens of driver graphs brings these options:
http://www.seas.no/index.php?option=com_content&view=article&id=113&Itemid=135and:
http://www.seas.no/index.php?option=com_content&view=article&id=118&Itemid=140The first is flat from around 60 - 1000 Hz, the second from about 90 - 1500 Hz. Both have SPL figures (left side of graph)
that vary about 2-1/2 dB across the band. That seems low, but realize its NOT, 3 dB is a doubling or dividing in two, the level.
A good amp might vary a fraction of a percent or dB. But < 3 turns out to be pretty good considering the hundreds of drivers I looked at.
I dont want the second driver as its passband is shifted up, I don't want bass response in the mid-range and the low end of the midrange
was defined at 1000 Hz. I want the bass driver to end at 1000, and the midrange to start there, not mixing the two. That a great theory until
all three are put in the same enclosure, but we have to start somewhere.
The bandwidth assumptions were exactly that, until we start looking at actual drivers. If we find a nice inexpensive unit that has a good bandwidth
and flat response, we can change the frequency definitions to match the driver. Thats the beauty of design, it can be changed.
The midranges are a problem, they are extremely wide -band. I dont want wide band, because I now have three amps that divide the bandwidth:
http://www.seas.no/index.php?option=com_content&view=article&id=101&Itemid=123and
http://www.e-speakers.com/jantzen-audio/jantzen-ja-2006The second one is from a high end audio supplier. Either might work, depending on price, they are both very broad banded
and relatively flat. The Jantzen may have less change in SPL. I'ts whatever you want and want to pay for.
Tweets. I immediately assume a $300 tweeter, cant get much of one for less than that. Tweets have a LOT of bandwidth to process,
and process very accurately. If its < $100, forget it. Try to save on the bass and mid and devote the money to a good pair of tweeters:
http://meniscusaudio.com/images/SATORI-TW29R.pdfI don't want a tweeter to respond to 1000 Hz, but this unit does have a flat response to 40 KHz, about 5 dB change in SPL.
Those are choices with decent claimed specs and moderate price, I've not looked yet at flatness of their impedances. With modern high speed
amps thats not so critical at low powers. The selection could be changed to get a flatter impedance curve. An EQ can be used to change the overall
frequency response, but not the driver impedance so it might be good to select drivers with a flatter impedance response and use the EQ
to deal with the rest.
All this becomes much less meaningful when the drivers are put in enclosures in a room, the enclosure and room determine the overall response,
not the curve of the driver SPL. But its a start to try to get a very flat response along with flat amps.
The graphs of the SPL response were generated in labs with test equipment under controlled conditions and there is ZERO
chance they will have the same conditions and same results in a real speaker design. It's possible to later test an overall system
to see its reponse, then change drivers to fill in a gap somewhere to reduce the amount of EQ needed. But this gives us a starting point
to hopefully not buy overly cheap or expensive drivers and either buy good amps and be disappointed in sound quality, or spend way too much
and not be able to afford sound treatments for the listening room.
Author
Topic: Driver (speaker) selection - sub thread from multiple channel topic (Read 2668 times)