Hi all,
Now that I'm happy with my system even if still in progress (OB 4 way+2 sealed subs), I want to optimise the power needs and rebalance everything.
And maybe test different amps later, I'm planning on Neurochrome for different reasons (fun with first diy amp, and lower noise floor).
Current power (considering 1% thd for ease):
I did it at -30dbfs baseline for easy calculation, got of course very low voltages but it was not a surprise as I started with big amps for flexibility.
Then calculated the power needed for -0db:
Considering that I can reduce headroom a bit since it's only few super loud parties a year, or when I'm gardening, I might be ok with 20-24db headroom, so all these numbers halved at least. I want more than 12db headroom to manage all kind of recordings, or bad tv stuff.
But these numbers already are low enough to find other kind of amps.
The pivot point seems to be the midbass, everything depends on them.
I can reduce drastically the power if I find the mechanical limit there.
To do that I need to get a ballpark idea of the max power to reach Xmax, or even Xmech that is a lot bigger on the SB.
They are crossed at 50hz but sometime I set them free.
I don't have winisd anymore, my parallel desktop is out and no pc around.
So I played a bit on loudspeakerdatabase with a faital hf510 that is very close to the sb18sw450.
After changing xmax to 11mm, in the biggest box available:
Here's where I'm a bit lost, the conversion to 4ohms once two woofers are wired in serie:
And if so I can safely keep the current midbass amp, just being carefull when full range and play with much less power for the rest.
Do you see any mistake or miscalculation here?
Do you know a better way to find out about the midbass limits?
Now that I'm happy with my system even if still in progress (OB 4 way+2 sealed subs), I want to optimise the power needs and rebalance everything.
And maybe test different amps later, I'm planning on Neurochrome for different reasons (fun with first diy amp, and lower noise floor).
Current power (considering 1% thd for ease):
- 60w on tweeter
- 150w on high mid
- 150w on low mid
- 700w on midbass
- 125w on sub (I sure need more power there and I'm waiting for a T.amp E800)
I did it at -30dbfs baseline for easy calculation, got of course very low voltages but it was not a surprise as I started with big amps for flexibility.
Then calculated the power needed for -0db:
- 5w on tweeters
- 5w on high mids
- 60w on low mids
- 360w on midbass
- no test on sub, probably need 500w based on levels in minidsp
Considering that I can reduce headroom a bit since it's only few super loud parties a year, or when I'm gardening, I might be ok with 20-24db headroom, so all these numbers halved at least. I want more than 12db headroom to manage all kind of recordings, or bad tv stuff.
But these numbers already are low enough to find other kind of amps.
The pivot point seems to be the midbass, everything depends on them.
I can reduce drastically the power if I find the mechanical limit there.
To do that I need to get a ballpark idea of the max power to reach Xmax, or even Xmech that is a lot bigger on the SB.
They are crossed at 50hz but sometime I set them free.
I don't have winisd anymore, my parallel desktop is out and no pc around.
So I played a bit on loudspeakerdatabase with a faital hf510 that is very close to the sb18sw450.
After changing xmax to 11mm, in the biggest box available:
- it crosses Xmax with 12 volts at around 20Hz
- it crosses Xmech with 40v below 20hz
- hard to simulate with crossover but once high passed it would probably not cross Xmax even with 40v, for 120+db per woofer.
Here's where I'm a bit lost, the conversion to 4ohms once two woofers are wired in serie:
- w/o crossover: 12v would give 18w per woofer at 8ohms, but 144w per side once wired in serie for 4ohms.
- w/ crossover: 40v per woofer, that's 200w at 8ohms and 800w per side once wired in serie for 4ohms. I don't even know if the NC500 can deliver 80v.
And if so I can safely keep the current midbass amp, just being carefull when full range and play with much less power for the rest.
Do you see any mistake or miscalculation here?
Do you know a better way to find out about the midbass limits?
Last edited:
Two 8 Ohm woofers wired in SERIES gives a 16 Ohm load (not a 4 Ohm load) as seen by the amplifier. Wiring the two drivers in parallel will give a 4 Ohm (Nominal) load.
I find it easier to think about it in terms of power. This makes it somewhat easy to figure out the SPL gains as well as the amplifier POWER requirements. Each driver will have some voltage imposed on/across it by the amplifier connection. For a single driver, all the amplifier power is delivered to that single driver. Easy peasey. For two drivers, no matter whether they are connected in parallel or series, the power delivered by the amplifier is distributed across both drivers - each driver will receive half of the total power. But the amplifier will see a different load for each electrical configuration (series or parallel) and usually it is well known how the maximum amplifier power scales with load impedance. Typically, when the impedance is low, the max amplifier power will be current limited and when the impedance is high the max amplifier power will be voltage limited. Consider your case of two 8 Ohm drivers in parallel. The load impedance seen by the amplifier is 4 Ohms nominal. What is the max power that the amplifier can deliver into 4 Ohms? Each driver will see half of that. What is the SPL from the two drivers? Calculate the SPL for a single driver given the amount of power it gets and then double that (add 6dB) to get the output for two drivers. That's it! Easy. For the series connection the load presented to the amplifier will be 16 Ohms. Look up the maximum power that the amplifier can deliver into 16 Ohms - each driver will see half of that. Use that power level to calculate the output for a single driver. The total SPL will again be twice that of a single driver. You just calculate the single driver SPL for the power input it sees and then add 6 dB to get the total SPL for the pair.
You can use this approach to calculate the SPL produced by any number of drivers. The relationship between number of drivers (each supplied with the same amount of power) and total SPL is:
SPL_total = 20*LOG10 [ N * 10^( SPL_single/20 ) ]
where N is the number of drivers being used together and SPL_single is the SPL from one driver. This equation is valid when all drivers are located "acoustically close" to each other, which will insure that the outputs add coherently. The same amount of power must be dissipated across all N drivers. More info here:
http://www.sengpielaudio.com/calculator-coherentsources.htm
Two identical coherent sources add to +6dB SPL... check the link in my post above for more/useful info. Here is an example: we all know that putting two drivers in parallel you get +6dB compared to a single driver connected to the same amplifier, right? This is because when connected in parallel each driver see the same voltage from the amplifier, and with the same voltage imposed across it will draw the same power. So you have two drivers, each drawing the same amount of power (assume our hypothetical amp can deliver the necessary current and does not sag). We have the case that N=2 and by the equation I posted above the SPL_total is +6dB compared to the single driver getting the same amount of power. What is implied in this situation is that the sources are coherent, that is the phase angle at the listening position is the same, or nearly the same, for both. Two incoherent/uncorrelated sources add to +3dB. An example of this would be two random noise sources. These add to a +3dB SPL.
Driver sensitivity would be included in the modeling of a single driver to get the single driver SPL that I mentioned. It is very easy to model a single driver in whatever configuration (e.g. OB) but doing so for multiple drivers is not always possible in the modeling software. So I suggested a way to get the SPL for multiple drivers from the SPL of a single driver based on whatever max amplifier power is available per driver.
Driver sensitivity would be included in the modeling of a single driver to get the single driver SPL that I mentioned. It is very easy to model a single driver in whatever configuration (e.g. OB) but doing so for multiple drivers is not always possible in the modeling software. So I suggested a way to get the SPL for multiple drivers from the SPL of a single driver based on whatever max amplifier power is available per driver.
Last edited:
I see now that the figures where there:
Then calculated the power needed for -0db:
Then calculated the power needed for -0db:
- 5w on tweeters
- 5w on high mids
- 60w on low mids
- 360w on midbass
Thanks CharlieLaub, my bad I keep mixing serie and parallel, these words just don't record in my brain. So yes of course they are wired in parallel and then 4 ohms for the amp.
I see that I kind of took it the other way around, not starting based on max SPL but more on the Xmax « safe » spl, and then power associated.
The driver sensitivity is 99db, but that’s at 300Hz and I cut it before that, around 150Hz if I remember, so in my case it's probably more 95db.
Still pretty good, 95db at 1 watt, that means 119db with just 64 watts, and 125db with 128 watts for 2 in //.
I know bass is less dangerous or even perceptible than mids, but I can't see myself needing that spl, even at 3meters of them.
I redid the simulation as I’ve made a mistake, forgot to check but the 18hf510 is 4ohms, and not 8ohms like the 18SW450
With 8ohms entered, of course all voltages go up (I would prefer to keep it to watts too, but the simulator needs a voltage value for the power).
Now:
Looks like this is the answer, I need to reduce midbass power by a lot too.
Of course all this is not very precise, and reaching Xmax is not the end of the world since Xmech is around 30mm.
I could also use the compressor, or wire them in 16ohms.
I will just do some test with another amp that is 2x125w at 4ohms, might not even hear the difference.
I see that I kind of took it the other way around, not starting based on max SPL but more on the Xmax « safe » spl, and then power associated.
The driver sensitivity is 99db, but that’s at 300Hz and I cut it before that, around 150Hz if I remember, so in my case it's probably more 95db.
Still pretty good, 95db at 1 watt, that means 119db with just 64 watts, and 125db with 128 watts for 2 in //.
I know bass is less dangerous or even perceptible than mids, but I can't see myself needing that spl, even at 3meters of them.
I redid the simulation as I’ve made a mistake, forgot to check but the 18hf510 is 4ohms, and not 8ohms like the 18SW450
With 8ohms entered, of course all voltages go up (I would prefer to keep it to watts too, but the simulator needs a voltage value for the power).
Now:
- based on the draft simulation with 40w at 8ohms one woofer will reach Xmax at 20Hz unfiltered, so two woofers would need 80w to reach Xmax too, and for 120+db.
- I extrapolate based on the draft simulation that with 200w at 8ohms one woofer will reach Xmax at 50Hz if filtered, so two woofers would need 400w to reach Xmax too, and at insane spl.
Looks like this is the answer, I need to reduce midbass power by a lot too.
Of course all this is not very precise, and reaching Xmax is not the end of the world since Xmech is around 30mm.
I could also use the compressor, or wire them in 16ohms.
I will just do some test with another amp that is 2x125w at 4ohms, might not even hear the difference.