Not really, was just throwing it out to see where the discussion went. If you look at his part 2 article he restated an effective 2x, so he seems happy with it as a rule of thumb.
you can easily drive yourself mad trying to work out what power you need for domestic active systems tho! Just thought those experienced in the art might chip in to help people on that path. I am a lost cause due to weird speakers I should note so all rules of thumb other than 'built what you can afford' are out.
you can easily drive yourself mad trying to work out what power you need for domestic active systems tho! Just thought those experienced in the art might chip in to help people on that path. I am a lost cause due to weird speakers I should note so all rules of thumb other than 'built what you can afford' are out.
You mean this bit? It does not look like a statement of a rule to me.
In general, my experience is DIYers make level management much harder than necessary. In pro audio typically every track has sticky peak metering by default and tools like Reaper are easily configurable for RMS. Makes it simple to run a few test tracks through prospective XO+EQ configurations and see where the peaks will fall. I usually watch things with an oscilloscope too, which many folks don't have, but an audio interface and some type of RTA serves the same purpose and is inexpensive compared to the total cost of most projects.
Rod Elliott said:
In general, my experience is DIYers make level management much harder than necessary. In pro audio typically every track has sticky peak metering by default and tools like Reaper are easily configurable for RMS. Makes it simple to run a few test tracks through prospective XO+EQ configurations and see where the peaks will fall. I usually watch things with an oscilloscope too, which many folks don't have, but an audio interface and some type of RTA serves the same purpose and is inexpensive compared to the total cost of most projects.
i did say 'rule of thumb', not 'rule'. At least on this side of the pond they have completely different meanings!
Which is fine if you build the crossover and eq before the amplifiers, or have a brace of mule amps you can use, but for the impecunious, or just starting out DIYer they do not have this benefit so need pointers. Makes it easy for people to get scared off active, or build a lot more than they need.
Which is fine if you build the crossover and eq before the amplifiers, or have a brace of mule amps you can use, but for the impecunious, or just starting out DIYer they do not have this benefit so need pointers. Makes it easy for people to get scared off active, or build a lot more than they need.
Umm, Reaper is a free trial download with a low cost personal use license and no output device is required, much less an amp. Several similar tools exist at comparable price points and share the lack of hardware requirement. There's also no need for a brace of mules; used with a little intelligence one or two channels generally does the job just fine.
Much of the discussion here on DIY Audio is devoted to intentionally massive overengineering in projects. I would venture to say people like building more than they need, so it's unclear if that constitutes much of a problem. There are those of us who enjoy more exact sizing and triamp with composite channels at BOM under 10 pounds a channel without any trouble.
As for pointers, there's a couple forums already dedicated to this with extensive discussion over the years. Last I dropped in on the multiway forum refactor thread there was a fair amount of advocacy for an active subforum.
Much of the discussion here on DIY Audio is devoted to intentionally massive overengineering in projects. I would venture to say people like building more than they need, so it's unclear if that constitutes much of a problem. There are those of us who enjoy more exact sizing and triamp with composite channels at BOM under 10 pounds a channel without any trouble.
As for pointers, there's a couple forums already dedicated to this with extensive discussion over the years. Last I dropped in on the multiway forum refactor thread there was a fair amount of advocacy for an active subforum.
ok, so what is reaper and what does it do and where do I get it? (you made me post a link to ESP so I'm just returning the favour).
We are in danger of a tail spin. I posted the ESP wet finger as a way of discussing that you don't need 4 channels at 60W/ch to drive average 2 ways in an average living room to sensible levels (for given values of average and sensible). You appear to be in violent agreement with that?
Which forums, where? How would someone considering active know where to look?
We are in danger of a tail spin. I posted the ESP wet finger as a way of discussing that you don't need 4 channels at 60W/ch to drive average 2 ways in an average living room to sensible levels (for given values of average and sensible). You appear to be in violent agreement with that?
Which forums, where? How would someone considering active know where to look?
Click Forum in the menu bar at the top of the page and choose Source & Line -> Digital Line Level or PC Based. You'll notice crossovers are mentioned in the one liner descriptions for both. Reaper is distributed from http://reaper.fm/; you may also find an intro to DAWs helpful.
My present amp channels current limit at about a third of an amp, which is enough to drive the dipoles to where I'm about to reach for my musicians plugs. Three and six ohm drivers nominal, so 160 to 300mW RMS max per channel. More here and here.
My present amp channels current limit at about a third of an amp, which is enough to drive the dipoles to where I'm about to reach for my musicians plugs. Three and six ohm drivers nominal, so 160 to 300mW RMS max per channel. More here and here.
Calculation and simulation? YES! I wholeheartedly agree. PSUD2 is a nice tool. However, it does have a few major shortcomings that make it not so useful for power supply design for class AB amplifiers. The major one is that it only supports resistive and constant current loads. That's a reasonable model for a class A output stage, such as those common in SET tube amps (PSUD2's target audience). However, a class AB output stage is about as far from a constant current load as you can get.
A class AB output stage, when driving a sine wave into the load, will draw half-sine current pulses from the supply. The two supply currents are 180 º out of phase. This means the average current drawn from each half of the supply is actually considerably lower than quick back-of-envelope math would suggest. It also means that if you use PSUD and quick math to estimate the supply ripple, you'll end up with a rather pessimistic estimate of the supply ripple and an over-designed power supply as a result.
If you're going to push the limits on the power supply, i.e. do the proper engineering on it, I suggest setting up a simulation in your favorite circuit simulator and working out the details there. It's relatively straight forward to set up a current source load to draw half-sine pulses + idle current.
One common pitfall is to start with the RMS power draw calculated from an equation and saying [transformer VA] = [RMS power draw]. However, the transformer delivers the charging pulses needed to replenish the charge on the supply caps. This is an impulse load. Using VA = PRMS will lead to an under-sizing of the power transformer of about 1.5x.
I go through the math and simulation on my Taming the LM3886 - Power Supply Design page. If you're looking to push the limits of physics, I suggest taking a look there first.
Tom
Last edited:
If one is unwilling to spend a few minutes to enter filters on channel strips a number of speakers---such as the LXminis being discussed in the Mod build thread---are sold with preconfigured MiniDSP units or similar. Beyond that, XO+EQ is considerably easier in DSP than line level analog or with passive components. It's rather difficult to implement an active system if one's unwilling to accept the learning curve to at least minimal proficiency in the area.
Quite frankly, the greater difficulty is folks tend not to do the measurements to be aware of how little power is used in most home audio configurations. Hence the steady demand for oversized amplifiers when best available data indicates most folks could run a single chipamp per speaker on ±12 and do fine in the wattage department. The recommended ±28 for the Mod covers pretty much everybody.
This is not to imply there aren't individuals who'll benefit from the 4-5dB equivalent offered by biamping or from additional wattage in the Parallel. But for the majority of folks the primary benefit of the Mod in this direction is it's a low cost, high quality, compact amp channel which considerably simplifies bi/tri/quadamping due to the high CMRR and PSRR providing excellent rejection of channel to channel crosstalk mechanisms. This is actually what motivated the design of the Mod in the first place---as a bit of historical trivia I did the original Mod layout on a 90 x 60mm 0805 surf mount PCB with an LM4782 for compact triamping (three THAT 1200s and an LME49710 and LME49720 as the control devices for the channels). Then Tom and I got to talking and he picked up the overall topology and commercialized it.
For someone who has never used a DAW the learning curve is a tad steep is all I am saying. Otherwise we are still in violent agreement and you have stated the exact reason I bought 4 boards.
Of course in my case my lack of experience in another area (speaker design) means that a 2.1 way is likely to become a 3.1 way, but that is a story for another day
Of course in my case my lack of experience in another area (speaker design) means that a 2.1 way is likely to become a 3.1 way, but that is a story for another day
Eh, unless you've got something specific, nah. Yes, there's a nontrivial learning curve to understanding filters, XO, EQ, audio routing, and taking measurements. Very little of it is DAW specific and all non-DAW solutions I've used over the years or am aware of range from somewhat to significantly worse. Personal preference and one's particular cases can skew this to some extent. But meaningfully better---as opposed to just tweaking selection of tradeoffs---ranges from rather difficult to impossible to achieve.
Feel free to PM for things 3.1.
Feel free to PM for things 3.1.
"This bit" was an interesting read. I'll be using MiniDSP too (nanoDigi is the digital version I have)
Rod seems to put a lot of calculation into getting gains correct when biamping. Understandably one would not want to have dips and peaks in the frequency response due to not properly calibrating amps and speaker sensitivities.
For me I'm planning to just get "in the ball park", then I'll measure with a calibrated mic, and correct as needed. Getting it flat is one thing, but in the end, I plan to spend some time and energy on building bass traps, and reduce the ringing in the room as much as possible. I'll use a waterfall graph and REW to help me. I hope to achieve superior holographic like imaging, and improve stability of instrument placement on the sound stage by spending energy here.
To try it get back on topic, after reading Rod Elliott's article I'm even more sure that Mod86 will give me as lots of headroom and also be very quiet to improve dynamics.
One other thing I wanted to mention. When figuring out how much amplifier power one will need in an active system, don't forget that a Linqwitz Transform in a sealed enclosure will eat power. The speaker cone has to travel a lot farther to achieve the lower frequencies, the higher power needed effectively reduces the sensitivity if the speaker. I found Jeff Bagby's excel sheet to be good in showing this.
Cheers,
AlexQS
Not usually all that much power, however. Whilst the equalization boost in a Linkwitz transform tends to be high it's applied over a small bandwidth. The result is the integral of the program power spectral density times the transform's gain doesn't move a whole lot. This is particularly true if one's keeping compensation low enough xmax related distortion terms are minimally audible.
As an example of deep bass correction consuming around four times the power of a typical Linkwitz transform, my three way dipoles are crossed LR6 at 200 and 1700Hz with the subs EQ'd flat to a bit under 40Hz. Peak levels to the subs are typically 5 to 6dB lower than those to the mids despite the subs having about 6dB higher drive and peak EQ boost close to 20dB. As usual, the amplifiers for all three channels share the same rails. Result: maximum power before clipping is limited by the mid channel.
As an example of deep bass correction consuming around four times the power of a typical Linkwitz transform, my three way dipoles are crossed LR6 at 200 and 1700Hz with the subs EQ'd flat to a bit under 40Hz. Peak levels to the subs are typically 5 to 6dB lower than those to the mids despite the subs having about 6dB higher drive and peak EQ boost close to 20dB. As usual, the amplifiers for all three channels share the same rails. Result: maximum power before clipping is limited by the mid channel.
Rod's explanation of the benefits of 'Bi-amping' is the best I've seen but its still riddled with misunderstandings that horny-handed electricians have about evil speakers.
There are a whole bunch of issues eg
- peak-mean of music, ancient & modern
- distribution of peak levels with frequency
- is the arrangement less likely to clip?
- is the clipping audible?
- perhaps most important of all, "is the clipping objectionable?"
I've spent a LOT of time listening to stuff with amplifiers set up to detect & hold transient clipping in 'bi-amped' speakers .. starting from experiments with scopes set to XY mode.
Rod is wrong about a couple of points eg
There IS an advantage with simply separating the Bass & Treble xovers and driving each section with a separate amplifier. [*]
The Audible advantage with the correct xover frequency is 6dB. ie it sounds like an a 100W amp rather than 2x25W. This isn't subtle. It's the audible effect of his Fig 3B
The 'best' xover frequency is around 1kHz. Like most electricians, he has rather naive ideas about psychoacoustics, speakers & peak distributions in music.
But xovers at 2.5kHz and above hardly gain any advantage from 'Bi-amping', regardless of electronic xovers or not. Better to use the 2nd amp in bridge with the first, perhaps reducing the rails (especially with LM3886) to stay within the SOA. ie use a single bigger amp rather than 2 amps if your xover is above 2.5kHz.
Electronic xovers allow a couple of extra dB of capability to be squeezed out of the treble but the whole speaker has to be designed to take advantage of this.
With a funky passive xover at 2.5kHz and above, you can still use Electronic EQ to get this advantage with a single amp. Rod doesn't take into account unit responses and efficiencies so his talk about Linkwitz bla bla is meaningless. It's the acoustic response that has to be Butterworth, Linkwitz bla bla
_____________________
The other 'xover' worth making electronic with separate amps is of course below about 70 - 50Hz to 'stereo subs'. This isn't from music distribution or how amps behave.
It's to do with making speakers go below 50 - 70Hz with substantial output compromises so much other stuff that handling stuff below 50 - 70Hz separately with dedicated electronics will give better sound.
____________________
To find out whether YOU need all this BS with YOUR speakers and YOUR music, do the test in
test-how-much-voltage-power-do-your-speakers-need.html
No need for a DAW etc. Just a voltmeter.
I'm sure you're right if these folk listen only to modern 'music', with 1dB dynamic range and peak/meantwest820 said:
But if you make your own recordings and record traditional music, like a good small unaccompanied choir or Beethoven piano, you may find, like me, that 2x200W 8R is often clipped.
I'm happy with 2x200W cos the clipping is not usually audible or objectionable but some of my own recordings clip 2x500W at quite reasonable levels at home with 90dB/1m/2.83V speakers.
[*] This simple arrangement doesn't get you the full 6dB audible advantage but is still worthwhile if you have a spare stereo amp and suitable speakers.
Last edited:
The big advantage of Bi-amping is that you have better options to control interaction between the driver and amplifier. Once this is taken into account, the sound cleans up and you get more resolution, darker background, better depth presentation in band.
Not sure really how many watts you need, but with small speakers, getting up to 90db listening levels is not a problem. Big mystery is how the drivers respond linearly at higher SPL levels. I would make this first priority since this really had a big impact when SPL levels get higher. Most driver used in the home start bottoming between 12~25 volts of amplifier output, and goes beyond the general linear range before that. This will cause great misperception during listening. If you look at the spectrum in real time, this occurs in the low frequency region generally below 60Hz.
Not sure really how many watts you need, but with small speakers, getting up to 90db listening levels is not a problem. Big mystery is how the drivers respond linearly at higher SPL levels. I would make this first priority since this really had a big impact when SPL levels get higher. Most driver used in the home start bottoming between 12~25 volts of amplifier output, and goes beyond the general linear range before that. This will cause great misperception during listening. If you look at the spectrum in real time, this occurs in the low frequency region generally below 60Hz.
Exploring the Zobels and Thiele network seem to be giving me mixed impressions.
The Thiele network in the version 2 Mod is different from the one in the data sheet. I did not notice until I started to play around with these. Sound different, but really do not know what to make of it yet. I am getting a meter with lower inductance measurement capability to see what inductance value I actually put into the amp.
The driver side Zobel seems to depend on what kind of amplifier output you have. Seems that no specific trend can be sorted out at this point.
Amp output Zobel is interesting, and seems to effect sound more. Up to now, I find using R-C-R to be most neutral, and I will also try this on the Mod 86 V2 when I put it together.
The Thiele network in the version 2 Mod is different from the one in the data sheet. I did not notice until I started to play around with these. Sound different, but really do not know what to make of it yet. I am getting a meter with lower inductance measurement capability to see what inductance value I actually put into the amp.
The driver side Zobel seems to depend on what kind of amplifier output you have. Seems that no specific trend can be sorted out at this point.
Amp output Zobel is interesting, and seems to effect sound more. Up to now, I find using R-C-R to be most neutral, and I will also try this on the Mod 86 V2 when I put it together.
If your per channel definition is normalized with Rod's total power definition what you're calling 6dB is what he calls 3dB. This doesn't appear to be a misunderstanding on Rod's part.
Perhaps you've overlooked Rod's paragraph beginning with "The table assumes equal efficiencies for the bass and mid+high drivers. Should they be different, then a correction factor must be added in."?
If the priority is obtaining maximum peak power from the available rails, yes. If one's interested in avoiding damping factor reduction from passive crossovers, not so much.
I've not had access to a copy of Fane's book Loudspeaker Enclosure Design and Construction Rod's reader appears to have taken the power split table from. So don't have insight into what it may or may not be modeling. In regards to peak levels my experience agrees with yours, with the optimum XO frequency being significantly higher than the table indicates. In terms of RMS considerations for amplifier heatsinks and driver cooling the table looks about right to me. (In a rule of thumb general sense these sorts of things necessarily involve, anyways.)
Dynamic range is 9-14dB typical and exhibits flat historical trend; refer to discussion of Emmanuel Deruty's loudness war analysis showing an increase in clipping earlier in this thread. The assumption I do not record and am therefore unfamiliar with the dynamic range differences between raw and mastered material is also incorrect.
The speaker voltage poll has been linked several times here and in related threads, including recent posts in discussion with Bill above. All evidence I'm aware of supports the notion most folks listen to recordings over a fair range of genres made across several decades. Ergo, the assertion the peak level reporting therein isn't representative of your personal musical preferences does not seem well supported. It would appear more likely you're indicating a preference to listen approximately 20dB louder than most folks.
Also, whilst the assertion the speaker voltage poll can be completed without active XO is certainly correct, it's of no great relevance to the preceding discussion of assessing level requirements for multiamping. The point there is since folks often find it difficult to do the maths to translate pure tone data or power spectral density integrals into per channel power requirements direct assessment of a configuration of interest can be more attractive. (It's also desirable to validate the maths are correct if they were done.)
Sure. But even if one's scrupulous about holding drivers in their optimum frequency ranges directivity mismatches are going dominate summation error in most speaker designs. If the drivers are pushed to the edge of their working range where post-XO EQ becomes meaningful for evening out the sum it's my experience directivity problems become so pernicious one's better off junking the design.