LM1875 - oh so slowly

[Leolo]

So over a year later, I've dusted off my soldering iron and finaly got my LM1875 chip amp working. Or had it working.

I was doing a test to see how much the chips would heat up, running at full power for an entire album. But I managed to short the chip to the heat sink when I reached in with my temperature probe. TWICE! The second time killed it.

IIRC, there's an LM1875 variant that has a isolated tab. While it might not get the heat off the die as well as the LM1875, surely this is more then made up for by not having to use the little T washers, a thermal pad and 2 layers of thermal gunk.

Anyway, now I have to order another LM1875 (or 2) and get to work on the case. I haven't fully settled on a design yet. In fact, the more I work on my design, the harder it gets to decide. Now I understand why you all go for rectangular metal boxes: they take very little time to build.

Attached is a pic of what I have so far.

Given that I'm going to be using a wooden box, I'm building a metal box within the box for shielding and mounting. Many screws go through the metal into a sheet of wood underneth. The sheet metal is from an old computer case.

Obviously I have to get a power switch and proper plug set up. And the wires between PSU and the power terminal need to go around the heat sink, not over. I'm tempted to go with unshielded wire from input to amps. The cable is stiff and is part of the reason the PCBs are slopping down from the heat sinks. And don't ask me why I mounted the LM1875s so high up on the heat sink. Thinking about it now, it would make more sense to mount it furthur down. But again, I'm not yet conviced 8x3x1/4 inches of aluminum isn't over kill.

I have half a mind to cut the thing in 2 and have PSU seperate from the amp.

One thing that's slowing me down is that I can only work on this at night and it's not recommended to do wood working in pygamas.

[Leolo]

DC offset at first turn on was 1.7mV Given the intructions I had, I was expecting 10-20mV. At first I thought something was wrong and it "wasn't working." But plugging in source and speakers and of course it was working.

Not that I'm complaing about 1.7mV, mind.

[AndrewT]

I have not managed to locate any twisted wire pairs.
Do you believe in twisted pairs for low loop area?
Do you believe that low loop area is irrelevant to performance?

[danielwritesbac]

Originally Posted by Leolo

. . .But I managed to short the chip to the heat sink when I reached in with my temperature probe. TWICE! The second time killed it.. . .

For this reason, my next LM1875 group project will be Parallel LM1875 Single Rail. I haven't worked out the details on it yet, but it will be really easy and it won't use any sort of circuit boards. In the case of the little 5 pin chip, circuit boards aren't making anything easier.

Your photo doesn't show heatsinks, but it does have nice big heat spreaders. The missing item is conduction of the heat into the air. Perhaps some U-channel added vertically or some holes added, could increase the effective surface area. Air is not particularly conductive, so a "slick wall" doesn't make much of a heatsink.

In building your enclosure, you might want to allow for an air intake at or near the bottom of the enclosure. And of course, it could use a hot air output at or or near the top of the enclosure. Allowing cool air to enter and hot air to leave, can save diodes, transistors, chips and capacitors from heat damage. Otherwise you have an oven.

[johnr66]

Mount the chips direct on the heatsink and insulate the heatsink especially if using large supply rails. They need to get rid of a lot of heat.

[Leolo]

Originally Posted by danielwritesbac

Perhaps some U-channel added vertically or some holes added, could increase the effective surface area.

Holes seem counter productive. A 1/2" hole in 1/4" stock is has a surface area of (roughly) 13/16" sq in. But it removes a surface area of (roughly) 6/16". Any other size gains less surface area. In fact, holes bigger then 1" will actually lower the surface area.

U-channels I can get. Would thermal gunk + bolts be better then welded on? I suspect so; the weld would create a small air cap inside.

In building your enclosure, you might want to allow for an air intake at or near the bottom of the enclosure. And of course, it could use a hot air output at or or near the top of the enclosure.

Part of the goal of my aborted experiment is to convince myself that 2x30w chips will produce as much heat as the huge heat sinks I see used seem to imply it will.

But now that I think about it more, the goal of a heat sink isn't to absorbe heat, but to get rid of it. And one needs surface area and cooler air flowing over it to achieve this.

Off to revise my plans. Yet again.

(For reference, I have 27v rails.)

[Leolo]

1. Having a LED on the PSU is useful if you are the type who plugs the wrong plug into the wall socket;
2. No matter how hard you try, eventually ever tool will clutter up on your work surface;
3. My test speakers sound really bad;
4. At loudish volume, the amp is only drawing 0.07 amps from the wall;
5. It turns out that my chip probably wasn't fried.

How do I figure this last one? While I was replacing the LM1875 with the one delivered this afternoon, I noticed one of the traces on the PCB has been fried off. Of course I notice this after unsoldering the chip. So I soldered in a new one and patched the PCB.

I now have quantum LM1875. It is both fried and OK, I won't know until I measure it.

Tomorrow, time permitting, I'm going to do a listening test with better speakers. And some long term heat tests.

Question: is it useful for heat disipation tests to run the amp without speakers? Is that even safe?

[benb]

You have Schrodinger's LM1875, from the same box as his cat. :-)

It's safe to run a solid state amp without speakers, but it's not useful for making it warm up or testing power dissipation. Putting power into the load is what causes the output device to heat up. But you can put a high-power (20 watts or more, depending on the output power this thing will give out) 8 ohm resistor, or whatever resistance your speakers are rated at, on each output in place of your speakers to make the chip "see" the same load as the speakers and warm up with that.

[danielwritesbac]

On Schrodinger's chip, be sure to check for DC offset. The failure mode of chip amplifiers is to send one rail's entire voltage into the speaker, get hot, and yet mysteriously continue playing nicely while the speaker's voice coil burns up.

Also helpful in preventing stability related accidents is to install the RF blocking cap per the National Semiconductor LM1875.PDF datasheet. This part promotes stability and is a standard for most op-amp builds, no matter if the chip is small or large. The tiny little cap is a big head start.

http://www.national.com/ds/LM/LM1875.pdf

[Leolo]

Originally Posted by danielwritesbac

Also helpful in preventing stability related accidents is to install the RF blocking cap per the datasheet.

I don't see a mention of RF blocking in the datasheet. Do you mean C2, the cap between -input and ground? If so, yes I have it.