LM3886 at 4 Ohms?

[phatt]

I've been messing with LM3886 power chip and run into a maybe not so small issue.

It seems these chips suffer with low impedance loads, i.e. 4 Ohm loads. The issue seems to be even moreso with higher supply voltages, so thought I'd ask greater minds for advice.

Without looking hard enough I purchased 2 kits which I assumed where LM3886 powered BUT are instead LM3876 chips and did not notice till after sale that there maybe issues with 4 Ohm loads.
A note on the paper work clearly states that IF you wish to use 4 Ohm speakers you need to use LM3886 instead. Also adding that even with the LM3886 to keep supply rails below 30 VDC.

The 2 short form kits are from Jaycar, KC5150 (Silicon Chip Mag, March 94)

It seems I've got everything against me,
As I've got a 36-0-36 VDC supply and I need to drive a 4 Ohm speaker. Bummer! 

I'm adding what I think shows the current rising dramatically as it approaches 4 Ohms.
(It's lifted from page 13 on the LM3886 spec sheets)

Obviously this is not good?

I do however have an old stereo transformer which does deliver 30-0-30VDC and I assume that will drop under load. The power amplifier section of the stereo unit is dead, not worth fixing so might be worth the effort to pull it down for parts.

The old amp has four 2Amp fuses on the main supply rails so looks like it would go close to delivering enough power for 2x LM3886 chips.

The new build will be for a small 2 channel pa mixer circuit I'm building which will be used for voice and will not need super low doof doof bass response so that should help relieve some need for massive supply grunt.

So I'm thinking this might just work but as I'm not sure of actual current ratings of either the transformer nor the lm3886 chips I'm askin for help.

The transformer in the old stereo is actually a C section design and I believe these are very efficient designs, only surpassed by those donut types.
Thanks all,, Phil.

[J M Fahey]

+/- 36V is way too much for any of them, so forget it.
+/- 30V, which will drop under load, is fine.
A somewhat undersized transformer, in this case, is not a bad thing.

Problem is, LM38x6 is internally protected (good ) and you can't "cheat" it (better ) , so higher PSU voltage actually means *less* power, go figure.

Heatsink them well, because hot chip= even lower power.
They have something called SPIKE protection.

[phatt]

Thanks, Makes a lotta sense. :tu:
I think I've been living in a false sense of reality because if you remember my subwoofer posting which I remade into my lunch box Amp

http://www.ssguitar.com/index.php?topic=1672.0

It ran a LM3886 from 36 volt rails and drove a 4 Ohm speaker. I actually belted guitar through that very hard for long periods with the 4 Ohm speaker and it has survived but hey it cost me nothing.

But this time I'm trying to build something that is reliable and do not wish to tempt fate.

Another recent observation was a quick peek inside a dB powered speaker the 10 inch model.
Sure enough two LM3886 chips. Ok driving 8 Ohm speaker but the rails are definitely
36VDC.      One wonders why they push the limits of the chip?

Even more scary was when I realized the filter caps are rated at only 35Volt. Yikes!
(Trying to add pics of dB)
Phil.

[phatt]

Adding more db internal pics,, note the transformer voltages.
Phil

[Roly]

Silicon Chip Mag

Well there's yer problem right there.   :trouble

dBspeaker
Always beats me how designers expect a heatsink to get rid of heat inside a box, e.g.;



If you suggested to them that they wrap it in styrofoam they would say you were mad, yet they think the inside of a speaker box is "free air".   :duh

[Enzo]

Roly, I see a heatsink on the left, bolted to the chassis, making the chassis itself a heatsink.  On the right, I see a shield  sticking up behind the preamp on the front panel.



The purpose of a heatsink is to lower the temperature of a component on it.   I see small three-leg voltage regulators with their own little heat sinks inside closed boxes all the time.   That heat may not escape the chassis directly, but when it spreads it out, the temp at any point is lower, and that is what we need from the sink.

[Roly]

This is nominally an 85 watt amplifier, and they even made one the same as a nominal 100 watter, with waste heat a far cry from a 3-pin regulator.

The amp is (I'm ashamed to say) an Aussie made early 1970's horror called a Savage (no relation at all to current American amps of that name), the thin (1/8th inch) ali L-bracket is bolted against a steel chassis painted inside and out.  In later builds they mounted this bracket on standoffs which only made matters worse.

This was only one of many things that was wrong with this particular amp which included raging optimalistic interpretations of semicon datasheets.  It's quite clear what the "designer" hoped would happen, and had the L-bracket been 5+mm and the back plate been a large lump of black anodized ali they may have had a chance, but painted steel... and in a sealed box, no chance.

I was once lumbered with making a shop full of these just basically reliable, requiring a heavy redesign of the power amp, after the company went rapidly broke.  Some owners resorted to drilling holes in the wooden sleeve case or back coz you can just imagine the internal temperature after a couple of hours with the equivelent of a 60-80 watt light globe running inside; and not helped by positive temperature compensation due to using diodes instead of emitter resistors.   :duh

They also suffered some of the most extreme damage I've ever seen in an amp, the sundry driver transistors literally vapourised leaving melted leads and sooty footprints to show where they had been.  I jest not;



I only know of three survivors, none unmodified.

The full 'orrible saga can be found here.

After I put this page up I was contacted by the tech who was brought in to try and sort out the mess when the "designer" decamped on a pile of dead returns, and after modifying dozens of them he told me that even several decades later the mention of the name Savage still gives him the jeebies.

By an astonishing stroke of luck I have managed to find the original datasheet for the heatsinks I used in my Twin-50, and a 4 inch length of black anodized Mullard 55D has a thermal resistance of close to 1ºC/W with fins vertical in free air (which mine are).



For a 50 watt amp with 1ºC/W for a good heatsink, 1.52ºC/W chip-to-case resistance for a 2N3055 (while derating 0.657W/ºC above 25ºC_case), with a mica washer that amounts to around 3ºC/W giving 150ºC T_junction above ambient at full output.  Add a summer stage ambient under lights of maybe 40ºC and we are getting uncomfortably close to Tj(max) of 200ºC.

Comparing this with the Savage one can only guess what its thermal resistance to ambient must have been, but very much higher in any case, and at 85 watts they didn't stand a chance.