LM3886 Chipamp Project Resurrected

[slomatt]

Almost two years ago I started building a LM3886 based chipamp, and somehow in the intervening years the project got boxed up and set to the side. Then a few weeks ago I got re-motivated on this project and am now almost finished with it. The power supplies are finished and I'm just buttoning up the amplifier section.

The layout is definitely non-standard and contains two "mono-blocks" combined into three separate chassis. They are "mono-blocks" in the sense that each channel has its own power supply and there are no electrical connections between the channels other than the chassis grounds. Each power supply is in a dedicated case and the two amplifier channels are in a third case.

Here's the original thread related to toroid choice and power supply voltages as relates to speaker impedance.

Toroid question and LM3886 photos

And more excitingly here are some pictures of the progress, additional photos are at the following link:

Blank Photos :: LM3886 Chipamp

Braided power cable for the dc and ground connections between the power supplies and the amplifiers. The wires are salvaged from spare computer power cables (have a ton of them at work).


A completed power supply board with inputs from the toriod on the front and dc outputs on the rear.


Smoke testing a power supply using the lightbulb method.


A/C side of a power supply showing the IEC connector, fuse, chassis ground, switch, and primaries side of the toroid.


The other side of the power supply showing the toroid secondaries and ps board mounted on custom 3/8" delrin standoffs.


The mostly finished power supplies from the top. The enclosures are made of stainless steel which is a huge pain to work with, I recommend avoiding it if at all possible. The sides are solid maple which is also a difficult material to work with. Obviously the U/L would frown on the holes in the top, I'm thinking of possibly putting some metal mesh in to block the holes.


Rear view. I'm debating if I should clear coat the maple or stain it a darker color. Any opinions?


Fronts.


For the LM3886 heat sinks I cut a large computer heatsink in half. It is designed to dissapate a huge amount of heat off of a cpu, hopefully it will keep the temps reasonable with the chipamp. The base is almost 1/2" of solid aluminum and the fins are about 1.5" long.


Mocking up the amplifier chassis. Each board has a dedicated heatsink. All power and signal connections are on the rear panel. More pictures to come as I make progress.


DiyAudio.com was a great resource in researching this project, hopefully this writeup will assist someone else in the future.

- Matt

[lowpoke]

Matt they look great. I really like your enclosures, not the typical approach. Are you impressed with the sound?

[slomatt]

Originally Posted by lowpoke

Matt they look great. I really like your enclosures, not the typical approach. Are you impressed with the sound?

I'm really looking forward to finishing them so I can do a full review. The last time they made any noise was in early 2008 when I was mocking everything up, and I remember being quite impressed with the sound quality at that time.

- Matt

[coolbeer]

I really like your design and the usage of for example the cpu heatsink.
Hopefully the sound will be impressive this time around aswell.

-
Kolbjørn

[slomatt]

The amplifiers are now electrically finished and playing music, all that is left is to clear-coat the maple side panels. So far I've only done initial listening, but I'm very happy with the sound. Here are some more pictures:

Amp boards bolted to heat sinks and input and outputs wired up.


Testing the amp section.


Rear of the amp module.


Inside of the amp module, space is tight but everything fits.


Both power supplies and the amp module.


Top view.


I think the three modules look better stacked, they take a lot less surface space this way. The sides still need to be sanded and finished.




- Matt

[mattmcl]

Looks great. I'm glad you didn't stain the maple, it's such a beautiful tone. I'd just finish them with tung oil, as they won't be handled much.

[slomatt]

Originally Posted by mattmcl

Looks great. I'm glad you didn't stain the maple, it's such a beautiful tone. I'd just finish them with tung oil, as they won't be handled much.

Thanks! I stained a couple test pieces and wasn't happy with the outcome. I was planning to just use a clear poly coating on them and hadn't though of tung oil. Will tung oil yellow considerably over time? Is the benefit that it brings out the grain better? I'll definitely have to do some research.

- Matt

[danielwritesbac]

To last longer, you need big air intake holes underneath the fins of those heatsinks.

Air "inlet"

[slomatt]

Daniel,

Thank you for the suggestion, a set of bottom vent holes are definitely on the list so that convective cooling works properly.

- Matt

[danielwritesbac]

A possibility, but by no means a suggestion to do so, is that there is some tonality control easily available with that kit.

Those 100uF at the amplifier boards, for V+ and V- . . .
You can parallel another pair of 100uF's or. . .
You have the option of replacing the 100uF's with 220uF, 330uF, or 470uF
and
You also have the option of using One 2.2uF, 3.3uF, or 4.7uF big polyester tweeter cap connected from V+ to V- (which is a slight "sink the mids" arrangement and it also blocks noises).

P.S.
The kit's output zobel has the erroneous combination of a small value resistor with a polypropylene cap. That can "dim" your treble, exposing a lot of midrange. You can either increase the resistor value, or its ever so simple to replace that polypropylene cap with a 220nF (0.22uF) or 100nF (0.1uF) green polyester dip cap from your local Radio Shack store (for example). Using an even smaller value is still yet better than omission.
Polyester has a typically high internal resistance, and the national semiconductor datasheet meant to use that type of cap for the speaker output zobel (RC) which is a customary practice. The only way to tell on that particular document is that the resistor figure looks insane. They were counting on the added resistance within an inexpensive polyester cap (the normal thing to use for the amplifier's speaker zobel).
Supporting optional types of caps (like polypro or ceramic) at the speaker zobel would need a larger resistor value, else there is possibly too much load. You can fix this by replacing either the resistor or the cap in the speaker output zobel.