Bridged or Parallel 1875 advice / schematic ?

[jimbo1968]

Hi again, I spent some time here a few years ago but I've forgotten everything since then. Hence the noobish questions.

My starting point is that I want to build a stereo power amp consisting a psu box and 2 small amplifier boxes. The amplifier boxes will each have 2 aluminium side panels which are 150x50x10mm (because I've got them cut) and I can't easily connect the 2 sides thermally.

So what's the most powerful amp I can build in this box? It will be used for driving 'normal' 8ohm speakers.

Using both heatsinks is a must I would think, therefore 2 chips (or maybe 4?)

I will solder it point to point.


PS. I picked on LM1875 because it's lower power, but maybe using LM3875 in a low power configuration is an option? Or maybe just build a full power LM3875 and find out later how high I can turn it up?

[danielwritesbac]

Originally Posted by jimbo1968

. . . So what's the most powerful amp I can build in this box? It will be used for driving 'normal' 8ohm speakers. . . .

Bridged doubles the load. Neither LM1875 or LM3875 are rated for much output power with a 4 ohm load.

LM3886TF can do it.
Use Arctic Silver paste to get that plastic coated chip heatsinked. Slightly de-rate the voltage.
Don't put caps larger than 330uF directly at the chip pins, but rather put larger caps at the power supply board, as this approach will decrease the heat.
Use hi-temp caps.
Use also a snubbed 20a rectifier unit (see Mark Houston's project).

Add DRV134 on regs.

That out to be a heck of a lot of power in the little box.
Can your speakers take that much?

[pacificblue]

Most power form a small enclosure calls for class D. Usually power is not a big issue however, except when you talk about PA.

Those side panels will probably achieve a thermal resistance of about 7-8 K/W. The rest of the case will add some heatsink capacitity to that. A 2x15 V transformer will keep you on the safe side with an 8 Ohm load. Depending on how you use the amplifier, you can try slightly higher voltages, and you should prefer the non-isolated packages for improved heatsinking.

Originally Posted by danielwritesbac

Don't put caps larger than 330uF directly at the chip pins, but rather put larger caps at the power supply board, as this approach will decrease the heat.

No. Larger capacitors have lower ESR. Current x ESR = heat dissipation. Larger capacitors decrease the heat in the enclosure, if only by a small amount. The amplifier chip's own heat dissipation is not affected by placing the capacitors nearer to it or further away from it.

Originally Posted by danielwritesbac

Add DRV134 on regs.

For what?

Originally Posted by danielwritesbac

Can your speakers take that much?

Speaker power rating is quite unimportant for most home applications with normal music signals. The important thing is that the amplifier is powerful enough, not to clip during the usual listening levels.

[danielwritesbac]

Originally Posted by pacificblue

. . . No. Larger capacitors have lower ESR. . . .

and thus less able to act as a filter

[AndrewT]

Originally Posted by danielwritesbac

and thus less able to act as a filter

A CR filter benefits from a low ESR.
Less ESR makes the filter more able to act as a filter.

[jimbo1968]

I've done a bit more reading, and realise I just need a simple parallel design because I'm just trying to make use of 2 heatsinks. ie I'm trying to share the heat out, not increase the power.

So now I just need to estimate what sort of cooling I will get from my heatsinks(side panels) and choose an appropriate Vs. I'm thinking Vs=25V which will give a max of 44W from 2x1875 in parallel. I see this agrees with pacificblue estimate.

Can anyone point me to a circuit?
Do I just use the datasheet circuit, with 0R1 resistors on the outputs?

I might just try the chipamp.com kit as dual mono.

[AndrewT]

look at page 2 of the 1875 datasheet.
For 8ohm load (actually 8r0 resistive) and +-25Vdc supplies, the amp puts out ~20W @ ~0.02% distortion.
If one allows the distortion to rise to 1% then the maximum output is 25W.

Your 44Vdc supply (is that single polarity or is it a +-22Vdc supply?)
will give considerably less power.

Paralleling the chipamps will not get you any more maximum power into your 8ohm load.

If you want more maximum output power then you need to bridge these 1875 (not recommended) or buy chips that can run from ~80Vdc supply.

However, paralleling does increase the maximum current output to the reactive load. That can be an advantage when the chipamp has a tiny output current. The 1875 is rated at >=3Apk @ 25degC and typically ~4Apk @ 25degC.

[danielwritesbac]

Paralleling the LM1875 gives it a bit deeper voice (more current?) and brings safety (for longer life) "just in case" somebody connects a 4 ohm speaker or other difficult load (such as an 8 ohm speaker that's "hard to drive").

[jimbo1968]

Andrew, I got it wrong. My Vs=25V was referring to the power dissipation graph on page 4, but I was thinking I'd get 2x 22W as max output. But in parallel they will share the load giving 11W each, and the heat dissipation doesn't drop. So parallel isn't what I want!

Time to read the bridging threads. From memory it takes another opamp to invert one side.

Possibly though I might give up/change my mind and work out how to thermally connect the two sides, then use a LM3886 which I am happy/confident with.

[danielwritesbac]

Originally Posted by AndrewT

A CR filter benefits from a low ESR. Less ESR makes the filter more able to act as a filter.

A capacitor's inductance and potential for ringing, thus de-stablizing the amplifier is a bit closer to the topic of the amplifier's heat output to the heatsink.

And this means. . . Given a limited size heatsink, we should probably go with a traditional amplifier power circuit rather than a heat-inspiring gainclone power circuit.

A bit less noise would be cool, literally.