Class D IC running hot with just filter connected?

[peskywinnets]

Any ideas why (no output load, just the two inductors & capacitors connected to pin 6 & 9 as per the datasheet)....

An externally hosted image should be here but it was not working when we last tested it.

(NJU8752V)


put the IC into standby (pin 11 low) ....IC cools rapidly, but out of standby the IC heat ups....I thought class D ICs were meant to run cool?

I'm using 33uh inductors & 0.1uf caps.

PS I don't have the 5-10R resistors on the output pins....as I think they are just there to present some resistance at high frequencies as this IC is meant for a piezo speaker.

 

[peskywinnets]

Nobody?

I'm quite new to class d, so could use a little help here! I've scoped the output with no filter connected (and with the 'in' left side of the input cap grounded)...I see a nice clean square wave @200khz 50% duty cycle (i.e. everything as expected)

I only seek bandwidth up to about 10khz.

How can I progress this so that I can have a cool to touch class D IC?!!

 

[Andrew Eckhardt]

Class D is efficient, not "no loss". Datasheet says 10 mA idle max with 13.7 volt max. That's 137 mW which would make the IC "warm". If your output supply voltage is lower than that and you have more loss then something is wrong. Did you measure the actual supply current?

 

[5th element]

Class D amplifiers have idle losses, ie a queiscent current they run at when doing nothing multipled by the voltage the device is run at.

I had a quick look at the datasheet for the device you're talking about and I couldn't see any where they listed the quiescent current draw for the analogue portion of the device. This makes trouble shooting hard as we don't know how hot it should actually run. Now most Class D parts are fine running hot, this could just be one of them. The package is very small and cannot dissipate that much heat.

If you've got a scope it'd be worth checking the output waveforms for the possibility of any unwanted oscillations, an unstable amplifier will run hotter than a stable one.

 

[5th element]

Andrew, as far as I could see the current listed is for IDD, which I took to mean the digital portion of the chip only, as IDDO would = the current draw for the analogue part.

 

[Andrew Eckhardt]

The control supply probably doesn't use much power at all but must be below 3.6 volts. If the circuit is found to be operating normally, by checking the idle current of the output section, try reducing output supply voltage to 9 volts or so for reduced no signal dissipation, even lower if you don't absolutely need the power. Post supply voltages and currents if more help is needed.

 

[Andrew Eckhardt]

You are right 5th! Along with no idle power dissipation spec, makes it difficult to be confident in the chip and circuit condition. Only dissipation max is given at 450mW max for the SSOP14, which probably would make it "hot". Such small packages get hot fast with fairly little power dissipation. Reducing the output supply voltage should make a noticeable difference in switching loss. Below 10kHz with a piezo load, most of the power loss will be switching loss anyway, changing pretty slightly depending on drive signal.

 

[Saturnus]

Note that this chip is intended for capacitative loads. Piezoes are typical over hundreds of ohms in the audible range, so you shouldn't remove the 5-10 ohm resistor before the output filter. You should replace it with a similar load as the capacitance in the piezo would have been. Typically around 0.10 uF.

 

[godfrey]

With no load and without the 5-10 Ohm resistors, the output filter forms a high-Q tuned circuit with a very low impedance at some frequency. Maybe that's causing instability or some other problem.

 

[DUG]

... this IC is meant for a piezo speaker.

What are you using it for?

 

[peskywinnets]

What are you using it for?

Yes, I realise it's meant for a piezo...but I wanted to utlimatlely attach a 16 ohm inductive speaker that I have (this is mainly a learning/curiousity exercise)...I went for this chip as it least the package pitch is workable! (& the ability to supply a higher output voltage appeals)

I placed some capacitance to simulate a piezo speaker (& put the resistors in)....the IC still got hot.

I'm running the control supply at 3.3V & the output voltage at 9V.

I've clearly a lot to learn here, but can't quite grasp how a class d can be deemed efficient if it's running too hot to touch at idle...IMHO that''s battery power being eaten up as heat (vs a standard class AB at idle....which is cool as can be). Oddly there doesn't appear to be any heat whatsoever on the output inductors at idle? (yet it's only when the inductor/cap of the output filter are connected that the IC gets hot.

the datasheet doesn't mention but the frequency is about 200khz.

I do have a scope & can run any checks you care to suggest (fwiw there doesn't appear to be any nasty transients on the output)

 

[DUG]

Have you set it up as per the intended use and then see if it gets hot?

Start there.

Use a piezo speaker.

Make ONE change and see what happens.

Try the other schematic offered later in the data sheet.

When you change a few things in a recipe, the cake might not taste the same.

 

[Andrew Eckhardt]

Scope the outputs (+/-) after the filter cap and check the ripple voltage at idle. Measure the idle current on the output supply. This will tell us what kind of power dissipation we're talking about. You might want to check the input stage supply current as well.

What inductors are you using exactly (part number)?

 

[phase_accurate]

I'd also suggest to insert the small resistors before the output filter.
Or put a load across the output terminals. Start with something between 33 and 100 Ohms. If it works correctly with this load then try it with your speaker. And because it is class-d doesn't mean that cooling can be completely omitted. Have a good look at the datasheeet to be make sure whether it canbe run like that. And even if it wouldn't need cooling with its intended operation mode doesn't mean it could do without if misused for something else.

Regards

Charles

 

[peskywinnets]

Ok, finally got back to this....so now I know why my Class D IC is running hot....this is seen on one of the output pins ...

(no load, just the filter)

An externally hosted image should be here but it was not working when we last tested it.

or as seen with a different scope timebase...

An externally hosted image should be here but it was not working when we last tested it.

(one of the output pins doesn't have this ugliness on it?)

Shouldn't it be just a flatish line with only a little ripple on it?

When I pull the connectrions on the output pins, I see a perfect 50% duty cycle square wave on each output pin?

Something to do with the filter components perhaps?

 

[Andrew Eckhardt]

(one of the output pins doesn't have this ugliness on it?)

I guess that is a statement. Did you try noting which output is currently defective and swap the inductors to see whether one is "not working" or that the defect doesn't migrate. Can you measure inductance?

 

[peskywinnets]

Yes I tried swapping the inductors (I have a meter to measure inductance, but it's a bit unrefined & doesn't measure lower inductances that well!). As it goes the fault seems to be morphing...as previously the other leg just had a bit of ripple on it, but now it's got a similar waveform (as per the scope trace above)...I'm figuring now this is going to be down to my layout & components used (I'm on breadboard....yeah, yeah I know....but I figured if the output pins scope a nice 50% duty cycle square wave with no signal & no filter connected - and they do scope fine with no filter connected - then there can't be any particularly awful stray capacitance in play!).

I think I'm going to have to do a pcb for this circuit (& perhaps source some better caps)