LM3886 based 8 Channel amps for linkwitz Orion

[ST7677]

Hi Pacific blue and Davey, thanks for pointing the mistake in gain calculation. I got it now.
Basically in bridge configuration each IC sees half the load so the gain across each IC doubles and doubles again for the bridge. Total gain here will be 29. Below is the detailed explanation for future reference by some one.

Assume 0.5 Volt RMS is input signal to Non Inv Tml of Non Inv Ampl bridge mode. We assume that Voltage gain of this Non Ampl is 23 dB then its voltage gain will be 14.125. Hence Output voltage of the Non Inv Wing of the bridge will be 14.125* 0.5= 7.063 Volt RMS, which is equivalent to 7.063*(Sq Root of 2) = 9.99 Volts Peak across Speaker Load of 8 ohms ( and will be in same Phase as the input signal at Non Inv Tml of the Ampl.). Therefore its peak output voltage swing across speaker load of 8 ohms will be +/- (Plus Minus) 9.99 Volts Peak.

Now consider the Inv ampl of the bridge. Same 0.5 Volts RMS is applied this time to the Inv Input Tml of the Inv Ampl. which also has a gain of 23 dB and this is equivalent to a gain of 14.125 Hence its output voltage will also be same as that of the Non Inv Ampl and is =14.125* 0.5= 7.063 Volt RMS, which is equivalent to 7.063*(Sq Root of 2) = 9.99 Volts Peak across Speaker Load of 8 ohms.Therefore its peak output voltage swing across speaker load of 8 ohms will also be +/-(Plus Minus) 9.99 Volts Peak. However there is one Important Differance from the output of Non Inv Ampl which is as follows:-

Since the same Input signal of 0.5 Volt RMS is fed to the Inv Input of the Inv Ampl, the output of the inv Ampl will be in anti phase with the signal input to Inv Ampl. Hence output of the Inv Ampl will be in anti phase from the output of the Non Inv Ampl. Hence at any given time if Signal voltage at the Output of Non Inv Ampl is increasing then output of the Inv Ampl. will be decreasing, this makes the Output voltages of the Non Inv and Inv Ampl swing in opposite Direction. This Results in total Peak Voltage swing across speaker load of 8 ohms= +/-(2*9.99 Volts Pk)= +/- 19.98 Volts Peak = 19.98/(Sq Root of 2) =14.128 Volt RMS. Hence total voltage gain will be 14.128/0.5= 28.26 = 20Log(28.26) dB = 29 dB. And power developed across Speaker Load of 8 Ohms= (Sq of 19.98)/(2*8)= (Sq of 14.128)/8 =24.95 Watts when input signal is 0.5 Volt RMS.

However when input signal is 1.0 Volt RMS the RMS Voltage Swing across 8 ohm Speaker will be 2*14.128 = 28.256 Volts RMS . Hence power devloped in it =(Sq of 28.256)/8=99.8 Watts.

Conclusions

Assuming Input Signal is 1 Volt RMS to Non Inv Ampl. As well as Inv Ampl. And Speaker Load = R=8 ohms

So that “V” is the output Signal RMS voltage of non inv ampl in Single.mode If its Voltage Gain is 23 dB=14.125 Then Power developed across “R” will be =(Sq of V)/R.=(Sq of 14.125)/8= 24.94 Watt

In Bridge Mode

Total output Voltage across “R” will be = 2*V =2*14.125 Volts RMS=28.25 Volt RMS Hence Voltage Gain is double of the voltage gain by single mode Ampl. Which is = 20 Log(2V) dB in Bridge mode hence increase by 6 dB.

The power developed by bridge Ampl=(Sq of 2V)/R= 4*(Sq of V)/R = 4*(Sq of 14.125)/8= 4*24.94 Watts. Which in terms of dB increase =10 Log(4)= 6dB. Total Gain =23+6=29 dB.

Hence Power handled by each wing of the bridge Ampl is = 2*(Sq of V)/R=(Sq of V)/(R/2). Hence Each IC in Bridge Mode will see a virtual speaker load=(R/2)=8/2=4 Ohms and its gain (Gain of each wing in bridge ) in bridge mode = 26 dB, which is 3 dB higher than the gain of the IC when operated in Single Mode here it is 23 dB.

Hence it is true that Total Voltage gain of the Bridge Ampl= 29 dB And Voltage Gain of each wing of Bridge Ampl. = 26 dB And when Input signal is 1 Volt RMS then its (of Bridge Ampl ) power developed across speaker Load of 8 ohms will be = 99.76 Watts

Note that Voltage gain of IC in Single Mode is still = 23 dB

 

[AndrewT]

Hence it is true that Total Voltage gain of the Bridge Ampl= 29 dB And Voltage Gain of each wing of Bridge Ampl. = 26 dB And when Input signal is 1 Volt RMS then its (of Bridge Ampl ) power developed across speaker Load of 8 ohms will be = 99.76 Watts

Note that Voltage gain of IC in Single Mode is still = 23 dB

I am not sure I have read correctly how you arrived at this

And Voltage Gain of each wing of Bridge Ampl. = 26 dB

This conclusion seems wrong to me.

 

[labjr]

Or, assume the high dollar amplifiers are partly responsible for the excellent sound of the speakers and make the assumption cheaper amps are not suitable.

I think it would be better to hide/obscure the amplifiers for audio shows. But then you'd spark curiosity from the auditioners about the "mystery" amplifiers and they'd get distracted with that.

It's a real double-edged sword and a no-win situation dealing with audiophiles.

Cheers,

Dave.

Even though they don't work as hard, I still find to hard to believe different amplifiers won't sound different. They sound different in every other setup. And not because they're expensive. But because they're all different technologies and topologies.

 

[AndrewT]

Cordell said something along the lines of:

Amplifiers that are working within their operating limits reproduce the same signal and sound the same.
It's when you take amplifiers outside their operating limits that they sound different.

 

[Davey]

I am not sure I have read correctly how you arrived at thisThis conclusion seems wrong to me.

Yep, that conclusion is incorrect. I think ST7677 is confusing himself with too much calculation.

The voltage gain of each section is 23db. The voltage gain of the two sections bridged together is 29db. Very simple.

Cheers,

Dave.

 

[ST7677]

ok, I get it.

Now Linkwitz recommends same power amp for all channels with same gain for simplicity. However since I am going DIY route, therefore it makes sense to determine how much power each Woofer / midrange / tweeter requires and make the amps accordingly, while still keeping the gains same across. (No point beefing up Tweeter amp).

To find this I looked at the ASP response graph (Active Signal Processor). This graph shows the equalized frequency response curve that will go into the each amplification channel of woofer/ Midrange/Tweeter of the active amp.

I saw a huge boost for the woofer for dipole equilization, So I looked at the chart and approximately determined the ASP equilization at different frequency points. Given that gain on all amps is same, this gives the relative power required (using Power Gain as multiplication factor) for each amp channel when a 0db input to the ASP is provided.

Since the Woofer output of ASP feeds 2 woofers, each having two LM3886 bridged (120W RMS), so one woofer ASP channel output = 240W. Now if I map the woofer amp channel to produce this 240W at 50hz*, then using the power gain values, I can extrapolate, how much watts will be required for midrange and tweeter amp channels based on ASP equalization output.



Amp Power required for Mid / tweeter = 240/3.16 x Power Gain (at respective frequency)

So we see that in above scenario, maximum power required by mid-range is 141W at around 150hz, and 26.3W by tweeter at 4khz.

Based on this I am planning to use
2 LM3886 in bridge mode at 28V rail to give 120W output for each woofer (total 240W for both woofers)
2 LM3886 in bridge mode at 28V rail to give 120W output for the midrange (this is slights lower than max required)
2 LM3886 individually at 28V rail to give 13.2W +13.2W for the two tweeter (It might be possible to use just one LM3886 amp here to power both the tweeters).

IS my approach correct? Does this makes sense?
Are my numbers correct?


* I am not sure if 50hz is the right point to map max woofer power. I have done so, assuming that there is not much musical content below 50hz usually. This can be mapped to a higher or lower frequency also. Mapping to 70hz (where power factor is 1, i.e. woofer curve intersects 0 db, would require about 3 times more power at all levels)

 

[kuldeepsingh]

The calculations looks OK. Anyway you may not be using all the speakers at full power, so for normal listening levels, you can fine tune the gains individually while running the speakers with amplifier.

 

[AndrewT]

What maximum SPL can you predict at different frequencies?
check for three frequencies across each driver,
say 40Hz, 100Hz and 200Hz for the Bass driver.
maybe, 300Hz, 1kHz and 2kHz for the Mid driver
finally, 3kHz, 10kHz, 18kHz for the treble driver.

 

[kevinahcc20]

The calculations looks OK. Anyway you may not be using all the speakers at full power, so for normal listening levels, you can fine tune the gains individually while running the speakers with amplifier.

This won't work for the Orions...to get the intended response from the speakers each amplifier channel regardless of its output capability in watts must be matched in gain to all of the other channels. Otherwise the differing gain settings will alter the balance between the drivers that is set by the response of the ASP.

ST7677 I think you may be drilling too far down in trying to find the exact power demand balance between the Orion channels. Due to the demands of the dipole compensation the ASP response curves will tell you that the woofers need far more power than the mid and tweeter channels. Also the fact that many Orion users have experienced woofer "bottoming" in operation without issues in the mids and tweets reinforces the appetite these woofers have. Your plan to use a bridged pair for each individual woofer, a single amp for each midrange and a single amp for each paralleled pair of tweeters will nicely meet the power needs of each channel. Then carefully gain match all channels (to +/- 0.1dB) and you will get fine results.

 

[ST7677]

ST7677 I think you may be drilling too far down in trying to find the exact power demand balance between the Orion channels. Due to the demands of the dipole compensation the ASP response curves will tell you that the woofers need far more power than the mid and tweeter channels. Also the fact that many Orion users have experienced woofer "bottoming" in operation without issues in the mids and tweets reinforces the appetite these woofers have. Your plan to use a bridged pair for each individual woofer, a single amp for each midrange and a single amp for each paralleled pair of tweeters will nicely meet the power needs of each channel. Then carefully gain match all channels (to +/- 0.1dB) and you will get fine results.

Hi Kevin,
The potential issue with high power amps for mids and tweeters is their low power performance, "the first Watt" cleanliness. It and also increases the risk of damage of mid and tweeters.
On the other hand if the power amps have comparatively high gain, then that will bring up the residual noise of the preamp and ASP, degrading the overall system S/N ratio..
Therefore getting the right power and gain for woofer, mid range and tweeter is important.

That said, I am sure you are correct, just want to convince my self

 

[pacificblue]

Based on this I am planning to use
2 LM3886 in bridge mode at 28V rail to give 120W output for each woofer (total 240W for both woofers)
2 LM3886 in bridge mode at 28V rail to give 120W output for the midrange (this is slights lower than max required)
2 LM3886 individually at 28V rail to give 13.2W +13.2W for the two tweeter (It might be possible to use just one LM3886 amp here to power both the tweeters).

Looking forward to see the heatsinks for that assembly.

 

[AndrewT]

8chipamps rated for 480W per channel in Indian Ta !

I too look to see how it's to be done.

 

[ST7677]

LOL, I am too figuring it out. Haven't reached heat sink part yet. Still with amplifier circuit layout.
The heat dissipation with 28V rail is within 40W per Chip (specification limit). Do you see a problem?

What maximum SPL can you predict at different frequencies?
check for three frequencies across each driver,
say 40Hz, 100Hz and 200Hz for the Bass driver.
maybe, 300Hz, 1kHz and 2kHz for the Mid driver
finally, 3kHz, 10kHz, 18kHz for the treble driver.

Hi Andrew,
The sensitivity of drivers are as following
Woofer - 88.4
Mid Range - 88
Tweeter - 89

The Crossover points are 92hz and 1440Hz

The free space radiation for of one dipole speaker at 1 m will be (from Linkwitz sheet)

1 Woofer @ 120W - primarily operates between 20-92hz
20 Hz - 70 db (each woofer - there are 2 per side)
30 Hz - 80 db
40 Hz - 85 db
72 Hz - 94 db
98 Hz - 98 db
(Below 30 hz woofer is xmax limited)


Midrange 120W (max) - operates between 92-1440hz
98 hz - 99 db
145 hz - 103 db
951 hz - 120 db
1413 hz - 124 db

Tweeter 26W - 102 db (each tweeter - there are 2 per side)

 

[AndrewT]

The Mid Range is showing far higher SPL in the upper frequency range of the pass band.

There is something wrong with your assessment of suitable pass bands for each driver and something wrong with your delivered power to each of the drivers.

 

[ST7677]

A quick question: When input from ASP is shared by two (similar) amps, what is the input V considered?
Eg. say ASP gives 1V RMS output on woofer channel, which is shared between two 100W amplifiers in parallel, then what will be input signal voltage for each amp 0.5V?

 

[kevinahcc20]

Assuming you are referring to the separate amplifiers for two different woofers the input V is the same since you connect them in parallel.

 

[AndrewT]

To do this correctly you must consider emf and the impedances in source and receiver/s.
When it is done that way and you follow the general advice that Zin> 10times Rs you will find that adding a parallel Zin as a second load will only reduce the voltage at Zin by a few percent.

Even connecting just one receiver reduces the voltage given out by the source.

 

[ST7677]

I am planning to use a buffer before the bridge amplifier, so it will not load the source and therefore I am assuming that both amplifiers will share the same voltage as on output of ASP channel.

Here are the revised calculations. Some one who knows this please check for one column and confirm that calculations (till Watts required post equalization) are correct.




And this is what the theoretical sound plot looks like...

 

[Davey]

Analysis paralysis?

ST7677,

You're re-engineering the wheel here and have lost focus on your project. All of this engineering and prep work has already been done by Mr. Linkwitz.

Worrying about voltage gain implications when "Y"ing a low output impedance ASP output into multi-k-ohm loads is waaaaaay down on the worry-about list.

Cheers,

Dave.

 

[ST7677]

Yes Davey, I was getting a similar feeling. Kevin also echoed the same.
Its much more complex than I thought for a non engineer like me to figure without derailing the project.

I guess I will go ahead with 29db gain on all amps without exploring any further on this

Really appreciate the inputs and motivation from all of you here...