Building a six channel amplifier using LM3886

[Pagnol]

Hello,

I am fascinated by the possiblities of a digital crossover network powered by BruteFIR.

My loudspeakers have got three ways each and at the moment I am experimenting with three very different amplifiers in order to power every driver seperately.

Of course I cannot pay thousands of dollars for three equal high-end amplifiers, so I decided to build some myself.

My soldering skill is quite well, but I don't have a grasp of the underlying physics. So there are some questions:

I could easily build three stereo amplifiers using the LM3886 kits from chipamp.com for example. But I prefer putting all the six channels into only one big case.

Is it possible (and recommandable) to put three toroids into one case? Or is there a better solution that I can use without having studied electro-technics.


I'd be very glad, if you answered my question

[Pagnol]

Second question: Say, I use three toroids and put them in a parallel circuit - do I need to limit the switch-on current?

[pacificblue]

Originally posted by Pagnol
Is it possible (and recommandable) to put three toroids into one case?

Possible, yes. Recommendable, so so.

Originally posted by Pagnol
Or is there a better solution that I can use without having studied electro-technics.

Use a single, bigger transformer.

Originally posted by Pagnol
Second question: Say, I use three toroids and put them in a parallel circuit - do I need to limit the switch-on current?

Depends on the transformer size and on the fuse/circuit-breaker it is connected to.
The transformer needs short-circuit and overload protection. If you don't use overload protection fuses on the secondaries, the primary fuse has to serve both purposes. In that case there is no way around a soft-start circuit. Without the soft-start circuit the primary fuse must be big enough to withstand the inrush current, which will make it too big for overload protection.
If you do use overload protection fuses on the secondaries, you will only need a soft-start circuit, if the fuses/circuit-breakers in the distribution box trip due to the transformer's inrush current.

[Pagnol]

If I choose to use one big transformer instead of three smaller ones, I won't be able to use the power supply that is shipped with the lm3886 kit. What power supply should I take?

Thanks in advance.

[pacificblue]

Why not?

[Pagnol]

Well, I thought so. Which transformator to take then? Still the same with around 20V/220VA? I'm quite sceptical.

According to you I can connect as many amplifier boards to a power supply as I want, right? Why do people prefer the relation 1 transformator per amplifier (dual mono)? Searching the forum for this question I found quite paradox answers.

I see, it'd be recommandable to understand the underlying physics before building such an amplifier.

[pacificblue]

Originally posted by Pagnol
Well, I thought so. Which transformator to take then? Still the same with around 20V/220VA? I'm quite sceptical.

Depends on the speakers you drive, how loud you drive them, which IC package you use and the temperatures around the amplifier. To be on the safe side with 4 Ohm speakers go for 2x18 V. With 8 Ohm speakers or the unisolated package you can use higher voltages.

Originally posted by Pagnol
According to you I can connect as many amplifier boards to a power supply as I want, right?

No. With the power supply boards that come with those kits you can feed one or two amplifier channels. The transformer should be rated for more than the combined output power of the connected amplifiers. Rumour has it that an improvement is audible up to the point, where the transformer rating is three times the combined output power. The sonic difference is however much smaller than the price difference. With a six channel amplifier, 2x20 V transformer and 8 Ohm speakers 160 VA should mark the lower limit.

Originally posted by Pagnol
Why do people prefer the relation 1 transformator per amplifier (dual mono)?

In theory the increased ripple produced by any loaded amplifier channel will be reflected on the supply of all amplifiers that are connected to the same transformer. Any changes on the supply lines will reflect at the speaker output, damped down thanks to the negative feedback loop. If you have revealing speakers and good hearing, the remaining small difference could be detectable in a direct comparison. There should be better spatial accuracy and more details. Whether you notice that, depends. And the decision, if it is worth much more money, is a very individual one.

In your case it also depends on the soundcard quality and, how good that software-based filter really works.

[454Casull]

Originally posted by pacificblue

In theory the increased ripple produced by any loaded amplifier channel will be reflected on the supply of all amplifiers that are connected to the same transformer. Any changes on the supply lines will reflect at the speaker output, damped down thanks to the negative feedback loop. If you have revealing speakers and good hearing, the remaining small difference could be detectable in a direct comparison. There should be better spatial accuracy and more details. Whether you notice that, depends. And the decision, if it is worth much more money, is a very individual one.

Isn't ripple mainly (or only) affected by how much capacitance there is? So if you maintain the same amount of capacitance per channel, ripple should not change, right?

[pacificblue]

The formula is Ur = I * dt / C

Ripple voltage increases with current and decreases with capacitance and frequency (dt = 1/f). There is a rule of thumb to use at least 1000 µF per 1 A nominal current, based on that knowledge.

Interesting exercise for the aficionados of the authentic Gainclone configuration with only 1000 µF per rail: find out, how much current can be drawn at low frequencies until you get 100 % ripple. How much undistorted output power does that correspond to?

[kscharf]

The rectifier and filter caps specified on most of these kits are sized to fit a single mono block amp, or two amps in a stereo configuration. If you want to power three or more channels then you will need larger diodes (to handle more current) and larger filter capacitors (for the same reason). You can drive two or three such power supply boards from a single transformer by connecting the ac inputs in parallel using separate wires from each to connect to the transformer.

The is a risk in using a single supply (larger diodes and larger capacitors, or multiple capacitors in parallel) to drive multiple amps of creating a ground loop coupling between the amps. If the power supply is decoupled at each amp with additional capacitors and all grounds are tied to a common star point with separate wires from each amp you should be ok. Separate power supply runs from each amp to the common supply with all connections made to a common point at the supply are also a good idea (DON'T DAISY CHAIN the power wires!!!).