Ok That explains. So basically due to thermal heat dissipation limit the 50W @8 ohms is not equal to 100W @ 4 ohm but only 68W @ 4 ohms, therefore when the amp is used at 4 ohms the voltage rail is reduced to manage the heat, reducing the amp output.
Thanks man! That explains your point that 2 chips in parallel will provide more sound output than a single chip, 1.9db to be precise.
SL does not recommend 50W at 4 ohms, he recommends 50W at 8 ohms (and his calculations show that he assumes the amp can deliver 100W at 4 ohms, he is taking Vpeak=28.2). So that way the parallel amp at 100W@4ohm only meets SL's minimum spec.
Yes, but there he is comparing 2 woofers put in parallel on a 180W amp with 12 amps as current limit, so the amp becomes current limited above 60hz.
He suggests that an amp with 200W at 8 ohms can drive the woofers in parallel.
The BPA200 amp will be 225W (continuous) and 295W (burst) at 8 ohms and at 4 ohms it is 335W (continuous) and 450W (burst).
Therefore with a strong power supply to match, I presume the 4 LM3886 BPA200, will be significantly more powerful (both voltage and current wise) than the 180W amp considered by SL.
ok, I did further modeling of this. This time including Orion ASP equalization and the actual Speaker impedance curve over frequency range for each speaker. I compared scenarios of each speaker being powered by a single Amplifier, a Bridge Amp and a Parallel amp.
And the best combination to power these speaker was found to be following:
Each woofer will gets powered by a Bridge amp running at 24V rail producing about 37W average (74W max) and requiring 2.9A max current (per amp). The Woofer offers <8ohms impedance above 50 Hz, however since ASP equalization is -ve db, therefore current requirement and heat dissipation stays in control. At the same time woofer impedance increases significantly below 50Hz, that's where Bridge amp is able to provide a higher voltage swing. Since speaker excursion is Voltage limited, therefore bridge offers higher output in sub 50hz space till restricted by max excursion, above 50Hz it does not hit the current / thermal limit so output is same as what parallel amp (at 40V rail) offers. Overall Bridge amp runs 6deg C higher than Parallel amp, but still within limits.
Each mid-range will get powered by a single amp running at 35V rail producing about 23.8W average (54W max) and requiring 2.9A max current.
Two tweeters (on one side of speaker) will run in parallel, powered by a single amp running at 21V rail producing about 21W average (27W max) and requiring 3.3A max current. Two tweeters in parallel offer 2.5 ohms resistance, however since the ASP equalization is very -ve therefore current and thermal requirements are within limits.
All amps at all intended frequencies are within their heat dissipation limits and when set for 26db gain, with 1V input calculate to produce about 90db freespace SPL (+/- 5db) in woofer, mid range and Tweeter approx.
Finally for the first time I have started seeing a somewhat flat speaker output across entire audible range
And the best combination to power these speaker was found to be following:
Each woofer will gets powered by a Bridge amp running at 24V rail producing about 37W average (74W max) and requiring 2.9A max current (per amp). The Woofer offers <8ohms impedance above 50 Hz, however since ASP equalization is -ve db, therefore current requirement and heat dissipation stays in control. At the same time woofer impedance increases significantly below 50Hz, that's where Bridge amp is able to provide a higher voltage swing. Since speaker excursion is Voltage limited, therefore bridge offers higher output in sub 50hz space till restricted by max excursion, above 50Hz it does not hit the current / thermal limit so output is same as what parallel amp (at 40V rail) offers. Overall Bridge amp runs 6deg C higher than Parallel amp, but still within limits.
Each mid-range will get powered by a single amp running at 35V rail producing about 23.8W average (54W max) and requiring 2.9A max current.
Two tweeters (on one side of speaker) will run in parallel, powered by a single amp running at 21V rail producing about 21W average (27W max) and requiring 3.3A max current. Two tweeters in parallel offer 2.5 ohms resistance, however since the ASP equalization is very -ve therefore current and thermal requirements are within limits.
All amps at all intended frequencies are within their heat dissipation limits and when set for 26db gain, with 1V input calculate to produce about 90db freespace SPL (+/- 5db) in woofer, mid range and Tweeter approx.
Finally for the first time I have started seeing a somewhat flat speaker output across entire audible range
Last edited:
think i'll chip in a little.
first i built myself a 10 channel gainclone based on the inverted lm3886 design. power is provided by 500VA toroid mounted in another case with 20,000uf smoothing caps. then i have another 60,000uf smoothing cap inside the amp case. lm3886 is running on a snubberized unregulated 35V voltage. preamp is a simple 5 channel buffer with lm317/lm337 power supply at 12V.
(pardon the messy wiring as it is my first try at multichannel past 3 channel config. had to experiment with the grounding)
after running it for a few month and after getting quite comfortable with more than 5 channels i started working on 6 channel lm3886 gainclone based on the myref rev. c.
active crossover is a linkwitz-riley 24db/octave and output buffer and level equalization follows the design proposed by Red Elliott Sound Product project 09.
after that is built i started an even more complicated project. a 12 channel amp also based on the myref rev.c and with a 5 channel active crossover.
even after having a testbed to test and refine my knowledge, i still have quite a problem with the grounding. had to redo the grounding a few times before reducing the hum to a minimal. even now i still can't make the amp dead silent, there's still some audible hum if you put your ears next to the speakers.
my advice to you is to start small. no point in trying a major and complicated design when you don't have any past experience with something simple. no matter how you plan it, you will get into some problem.
and speaking from experience, it's better to build it rather than keep planning it and not even starting the build. the hardest part is to complete the build and finalize it into a case.
your build is a major undertaking. and if a problem arise from such a complicated project, it will be really hard to troubleshoot.
An externally hosted image should be here but it was not working when we last tested it.
first i built myself a 10 channel gainclone based on the inverted lm3886 design. power is provided by 500VA toroid mounted in another case with 20,000uf smoothing caps. then i have another 60,000uf smoothing cap inside the amp case. lm3886 is running on a snubberized unregulated 35V voltage. preamp is a simple 5 channel buffer with lm317/lm337 power supply at 12V.
(pardon the messy wiring as it is my first try at multichannel past 3 channel config. had to experiment with the grounding)
An externally hosted image should be here but it was not working when we last tested it.
after running it for a few month and after getting quite comfortable with more than 5 channels i started working on 6 channel lm3886 gainclone based on the myref rev. c.
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.
active crossover is a linkwitz-riley 24db/octave and output buffer and level equalization follows the design proposed by Red Elliott Sound Product project 09.
after that is built i started an even more complicated project. a 12 channel amp also based on the myref rev.c and with a 5 channel active crossover.
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.
even after having a testbed to test and refine my knowledge, i still have quite a problem with the grounding. had to redo the grounding a few times before reducing the hum to a minimal. even now i still can't make the amp dead silent, there's still some audible hum if you put your ears next to the speakers.
my advice to you is to start small. no point in trying a major and complicated design when you don't have any past experience with something simple. no matter how you plan it, you will get into some problem.
and speaking from experience, it's better to build it rather than keep planning it and not even starting the build. the hardest part is to complete the build and finalize it into a case.
your build is a major undertaking. and if a problem arise from such a complicated project, it will be really hard to troubleshoot.
Andrew, single LM3886 is capable of delivering only 33W at 50hz with 32V rail, (woofer impedance drops below 4 ohms therefore it is not advisable to run rail V higher than 32V.) whereas the bridge setup would give 77W at 50hz with 25V rail.
While the output of single IC is acceptable over 50Hz (bridge is 33% higher), however below 50 Hz single IC looses out big time (output is much less than 50% of bridge setup). Orion users have substantiated this point the bridging woofer amps gives much better low end.
On tweeter I can possibly use one IC per driver, and use the same PSU as of Mid range. There it's a choice of one shared PSU between Mid and tweeter vs one shared amp for the two tweeters in parallel.
Since two tweeters in parallel can easily be handled by one IC (at 21V rail), I believe the first watt cleanliness would be better with one IC powering both the tweeters at a low voltage rail.
Pascal, I agree with your suggestions. I am going to start first by building the single chip amps and then progress towards the bridge setup at the later stage. I am almost at the verge of ordering components now (in Feb), will do lot of practical trial and testing and finish my amps probably by Jun
Then finally assembling of speakers will start...
Last edited:
how large your case is gonna be? i'm only using a 500VA toroid (single voltage) and that takes a large chunk of the case space. an e-i core transformer is even larger. and you said about custom wounding the transformer in which are usually done with an e-i core instead of a toroid. never found any custom wounding service for a toroid before. at least not locally to me.Andrew, single LM3886 is capable of delivering only 33W at 50hz with 32V rail, (woofer impedance drops below 4 ohms therefore it is not advisable to run rail V higher than 32V.) whereas the bridge setup would give 77W at 50hz with 25V rail.
While the output of single IC is acceptable over 50Hz (bridge is 33% higher), however below 50 Hz single IC looses out big time (output is much less than 50% of bridge setup). Orion users have substantiated this point the bridging woofer amps gives much better low end.
On tweeter I can possibly use one IC per driver, and use the same PSU as of Mid range. There it's a choice of one shared PSU between Mid and tweeter vs one shared amp for the two tweeters in parallel.
Since two tweeters in parallel can easily be handled by one IC (at 21V rail), I believe the first watt cleanliness would be better with one IC powering both the tweeters at a low voltage rail.
Pascal, I agree with your suggestions. I am going to start first by building the single chip amps and then progress towards the bridge setup at the later stage. I am almost at the verge of ordering components now (in Feb), will do lot of practical trial and testing and finish my amps probably by Jun
Then finally assembling of speakers will start...
planning to fabricate a proper PCB for all this?
No document, just calculations this time
For single amp
32V rail /1.15 regulation - 2.8V clipping voltage = 25 V peak
Woofer Impedance at 50hz = 8.7ohms
Power output = ((V Peak /1.414)^2)/woofer impedance
= ((25/1.414)^2)/8.7
= 36 W
Why 32V rail, because between 70-80hz, the woofer imedance is between 5.2 - 4.6 ohms, and the average heat dissipation reaches 40W (max limit for LM3886)
For Bridge amp, max rail V can be 25V and at 50hz, the clipping V is 2.5V, so available V Peak is 38.5.
ASP equalized signal with 26db gain at 50hz is 36.39V (within the available V Peak limit).
Power output = ((V Peak /1.414)^2)/woofer impedance
= ((36.39/1.414)^2)/8.7
= 77 W
Last edited:
If I put one amp for each tweeter separately and run them at 35V rail (same as midrange, by sharing the PSU), then each amp will produce on an average 32W heat dissipation in the required application as the tweeter impedance drops down to 5 ohms from 2khz - 5khz .i.e. combined the two amps will emit 64W heat.
Where as if I run 2 tweeters in parallel from one amp at 21V rail. The combined tweeter impedance drops to 2.5 ohms, and the average heat dissipation will be 30W. This is less than half of above, i.e. it will require less heat sink.
So with single amp
Adv: Low capacitance (<1500uF) PSU, Cleaner 1st watt output, Compact (less heat sink, one amp circuit)
Dis Adv: separate PSU with 21V rail
and with individual tweeter amp
Adv: Can share the Midrange PSU at 35V rail
Dis adv: More heat sink, two amp circuit, extra speaker cables and tweeter protection circuit
Sound Quality wise the single amp wins the case, at the same time it is more compact also (if I take two extra secondary from the Transformer, then only bridge and few filter capacitors are required)
HI Andrew,
The formula is taken from the LM3886 data sheet only, Pg 20, Formula 7
Max Supply = (V Peak + V Clipping)*(1+regulation %age) / 10% supply V variation
i.e.
V peak = ((Max Supply *10% supply V) / (1+regulation %age)) - V Clipping
My formula
(32V rail /1.15 regulation - 2.8V clipping voltage = 25 V peak)
is same, except that I have intentionally, not taken supply voltage variation in this calculation yet.
The heat dissipation is also calculated using the formulas in AN1192 note.
Only differences are
1. Instead of assuming one fixed speaker impedance, I have taken speaker impedance at some 20 data points in the intended frequency range
2. I have taken 1V peak signal as input to ASP and then used the equalized output of the ASP at each of the 20 data points (for each speaker) above to calculate the actual output. The lower of peak voltage swing available and ASP equalized Voltage swing (at 26db gain) is taken to compute the realistic output.
3. Using above two, amplifier Output Watts, Current and thermal dissipation at each of those 20 data points are computed for Single, Parallel and Bridge amps to determine the most suitable configuration and ideal voltage rail.
Last edited:
Obviously the amp is for music.
The 1V peak test signal is only indicative and for design purpose. I understand that in real life there will be a wide variation in the frequency spectrum based on the type of music and the volume control will determine the input V levels.
I could as well increase the volume and use 1V RMS (0 db) as input signal, however irrespective of the input level, the relative output and heat generated across the amps will be in same ratio.
Last edited:
What exactly is the point of comparing the heat dissipation of 2 amps that deliver ~90 W into 5 Ohm each with the heat dissipation of 1 amp delivering ~60 W into 2,5 Ohm?
If you use two amps at 21V you get the same total heat dissipation. The heatsink can be smaller, because a) the heat transfer area is doubled and b) each IC must only be derated for half the power dissipation.
The point is that both tweeters together require only about 28 watts (at certain music level, here taken as 1V peak), therefore thermally it is much more efficient to get this from a single amp at lower voltage (than two amps at higher voltage), considering that both amps are capable of providing this.
Why I have taken 1V peak as input to ASP, because at that level with recommended 26 db gain, the woofer channel becomes voltage limited to produce flat response below 45 Hz (even with a bridge amp).
I could have run two amps at lower Voltage also (say 21V), but then what's the point using 2 amps when 1 is sufficient (even current wise).
The idea Andrew suggested was to use same PSU (35V rail) for tweeter also. If I take 35V PSU for two separate tweeter amps, then total heat dissipation goes up. If I use only one amp, then a separate PSU will be required to provide 21V rail.
Of course, I am assuming I will be able to get two custom made torroids with 3 pair of secondaries each to give output rails of 25V, 35V and 21V for Woofers, midrange and tweeters respectively.
One torroid for all right side speakers and one for all left side speakers.
The Torroid will be common and each channel will have it's own 2 secondaries, bridge rectifier pair, filter / reservoir Capacitor, etc.
I intend to provide minimal capacitance on Tweeter channel (a la gainclone), low moderate capacitance (snubberized) on Midrange and heavy capacitance (snubberized) on Woofer channel.
If custom torroids don't work out than I might still go with using 2 amps for tweeters sharing the 35V rail Mid range PSU.
Last edited:
i agree with andrew on this.
when dealing with an active system, you're separating the music passage into separate frequency. so each channel will only be seeing a specific range. the tweeter amp will be running very cool compared to the channel driving the bass drivers.
i say go with the simpler arrangement and run the tweeter amp at 35V together with the midrange. i'm running mine at 35V for all channels and most of the heat is from the bass channel
i cross my tweeter at 3.48khz, midbass at 286hz
Hi Paskal,
I have already incorporated only the frequencies that the tweeter amp is going to see based on the crossover from ASP ie. 1.4khz to 20khz.
And that is the reason where while 90-20hz will get 154W max (97W avg), the 1.4khz-20khz only gets 28 W max (21W avg). So if you see the tweeter amp is already running much lower compared to the woofer.
Also if you see, Orion has a much lower cutoff for tweeter than most other designs.
I tried measuring levels in various music and broadly on an average I can say that same levels are needed from 40hz to 15khz. That said some musical passages do have higher energy in 30hz range, or 15+khz range.
I have already incorporated only the frequencies that the tweeter amp is going to see based on the crossover from ASP ie. 1.4khz to 20khz.
And that is the reason where while 90-20hz will get 154W max (97W avg), the 1.4khz-20khz only gets 28 W max (21W avg). So if you see the tweeter amp is already running much lower compared to the woofer.
Also if you see, Orion has a much lower cutoff for tweeter than most other designs.
I tried measuring levels in various music and broadly on an average I can say that same levels are needed from 40hz to 15khz. That said some musical passages do have higher energy in 30hz range, or 15+khz range.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.