LM4673 class D audio amp chip

Hi Julian,

Do you have a demoboard or did you build this circuit yourself? ie do all of the connections have proper continuity?

Is the shutdown pin tied to the supply voltage? If left floating, it may leave the amplifier in shutdown mode.

Regards,

JD

I built the board myself, using ground plane construction as i am an rf engineer by trade. The shutdown pin and all others are correct and verified by probing with a cro under a microscope. The strange thing is sometimes it works perfectly at switchon, sometimes not, but I cant put my finger on what is different each time. The metallised pad under the device is not grounded, do you think this is the problem, although there is no mention of the slug in the data sheet. I cant even feel it slightly warm when its working, so its not over heating.

Julian.

Seems to be something to do with the enable line connected directly to the power rail. If I delay the enable, series 22k to 5V, shunt 4u7 to gnd, it enables the device a few 100ms later and all is fine. Shame there is no mention of this in the data sheet.

Julian.

Hi Julian,

My apologies for the delayed reply; the person who is responsible for this part is out of town.

The part should start up very quickly as stated in the datasheet (17us) and I am not aware of any startup issues. I would ensure that the DAP or tab is solidly connected to ground so that the IC has a stable reference upon startup. 

Best Regards,

JD

Julian,

The LM4873 does not need any delay between the supply and SHDN rising.  During bench testing, SHDN is tied directly to VDD, and the device starts up.  The usual suspects here are supply rise time, supply current limit, or ckt design/layout.  Can you send your schematic and layout?  Also, check the device outputs, if the outputs are shorted, the device will go into SHDN mode.

Regards,

royce

The device outputs are not shorted, the psu is capable of 3A. Please find enclosed 2 files, a cct (basically the same as in the data sheet) and the layout.6518.LM4673 class D amp.PDF

Further work has shown this problem is due to power supply ramp up problem. 

With a microswitch in series with the dc output of the PSU, the unit fails to start up, possibly due to switch bounce. The scope plot shows that the dc collapses shortly after power up, then recovers, but the device has not started.

When switching on the supply at the mains, the supply ramps up more gently and the device almost always starts correctly. When touching the red supply lead on to the psu terminal, the device nearly always starts.

It has a 4.7u tant across the rails on the IC. Removing this cap makes very little difference to anything.

Once with the microswitch, the device drew 3 Amps and destroyed itself, perhaps both fets came on together, a kind of stall condition?

Question is what to do now. The unit is powered from a wallplug supply, I cant guarantee how the dc connector of this is plugged into my unit, it might behave as the bouncy switch. Certainly cant risk the device destroying itself.

Blue is the dc supply, yellow is one side of speaker WRT to gnd.

Julian Worskett

Julian,

Two things you could try:

1) Your delaying SHDN rising idea may work here.  Delay SHDN rising until you know the supply is stable.  Just make sure any bounce does not exceed 6V abs max.

2) You could try isolating the amp from the switch through an LDO.  If you can find an LDO with very low dropout voltage, slow turn on time, and good line trasnient suppression that should protect the LM4673. 

Regards,

royce

Royce,

I am having what looks like the same problem.

In my circuit (on a multilayer board with groundplane and decoupling ) the power is enabled by a Vishay SiP32402A slewrate controlled load switch - so should be very repeatable.

(The /SD is tied to the power rail.)

Nevertheless, approximately 1 in 20 times, the device does not start and needs a further power cycle.

If the device could come up with both output 'on', as Julian speculated, that could trigger the thermal shutdown, which would explain what we see, wouldn't it?

Please can you confirm that delaying SHDN would cure the problem. 

Regards,

Geoff

Hi, Geoff,

I apologize for the slow response; Royce has left TI, and I just now came across your follow-on post.

This is very strange, indeed. 

We have sold 10s or 100s of millions of this IC into notebook computers, so I don't think there is an inherent issue with the chip. It seems that both of you have something in series with the power supply which may be causing issues; however, I can't explain it in your case as a slew-rate controlled load switch would seem to be the best thing I could imagine...

Did you get this sorted out, or are you still having issues?

-d2

Have re-visited this product, on the last project I designed it out and used a normal LM4875, couldnt afford the risk.

However, I need now to use this in a new project. Bought some new chips from Farnell in the UK, new layout and guess what? it still has the same problem, sometimes wont power up.

I am using single input configuration, 4u7 and 56k resistors on both inputs.

Thought it didnt like the large value of cap on pin1 to gnd, but changed and made no difference. Tried switching on with and without audio input (1 kHz sine, 100mVpp) but no difference.

Come on TI, any other suggestions? Thanks.