TPA6120A2 quiescent power - getting warm

Hi, Marc,

This part has insanely high bandwidth. Are you sure it's not oscillating?

And, if you run it from +/-15V, then it's 0.9W which is quite a lot.

How warm is "quite warm"?

-d2

Hi Don,

Unless it's oscillating at >500MHz, I would have caught it...  

quite warm is 56C in open air, with chip bonded to about 2 sq in of copper or so via the central heat pad and footprint from TI...  with a chunk of Al attached to the top of the package as well, that goes down to 49C.  Investigating anodizing the heatsink for better dissipation.

Sounds like this might be 'normal'?

thx

M

Also I have a question about your power calculation.  Data sheet shows 15mA quiescent for each channel with +/-15V supplies.  So wouldn't that be the sum of currents from +15V to GND and from -15V to GND?

thx

M

Hi, Marc,

That's how I would interpret it as well.

-d2

that's how I got to 0.45W... do you concur?

regarding temperature - does that seem normal?

thx

M

I tested the TPA6120A2 EVM with +/- 15V supplies and measured 20mA of current for 0.6W of power dissipation at idle. The part got slightly warm but not as hot as you had described. That part on the EVM has the TSSOP with PowerPad package. The QFN package will get significantly warmer due to the smaller size which may explain the discrepancy.

Can you confirm which package you have?

If you measure the supply current and it is grossly different from 20mA that can be a good way to tell if the part is oscillating. (Note: 30ma is the specified maximum idle current. 20mA would be more of a typical spec)

Regards,

Matt

Hi Matt,

Yes, I'm using the TSSOP with PowerPad.  And it got quite hot on the EVM board as well.

Have you measured the actual temperature, after it's been running for say 5 minutes?

thx

M

Hi Marc,

I tested on the EVM using an Fluke IR temperature gun with the following results:

Idle no load: 29 deg C

Idle with 32ohm loads: 29 deg C

1V RMS input one channel only: 31 deg C

1V RMS input both channels: 37 deg C

With both channels running with a load, the part does get pretty warm but not quiate as warm as you suggested.

Regards,

Matt

 

Hi Marc,

I let the part run for 30mins with the temperatures stated above.

I am willing to bet that the issue is in the DC coupling since everything else is the same.

Any offset difference between the source and the input of the TPA6120A2 will cause significant power dissipation. Whichever offset is higher, will attempt to dump current into the other causing one to source current and the other to sink it. Therefore, both interfacing devices will dissipate energy.

Although this may not be an issue for the output device, the input to the TPA6120A2 most likely will not like to source or sink DC current and can cause internal biasing issues further adding to heat dissipation.
Try adding DC blocking cap. This should take care of the issue.


Regards,
Matt

Hi Marc,

Common mode ranges generally refer to the AC characteristics or the signal swing level between the inputs before non-linearity takes into effect due to the signal swings proximity to the supply rails. It does not account for DC offsets between coupling devices together. In most circumstances the DC offset must be 0V. This is done by adding a DC blocking cap as to isolate the DC levels between 2 components or by insuring that the DC offset is identical.

If I understand your setup correctly, you have a 2.5V DC output offset from the OPA1664. However the TPA6120A2 input will be centered around ground when using symmetrical power supplies. Therefore, the OPA1664 will be sourcing current into the input of the TPA6120A2 at idle and drive current into this input to bring the TPA6120A2 input node up to 2.5V. This is what will cause extra idle power dissipation in the TPA6120A2 as well as biasing problems.

(As a side note: this will reduce the common mode positive signal swing range by 2.5V since the input is sitting at 2.5V closer to the +15V rail at DC. This will increase the common mode negative range by 2.5V)

You can correct for this by adding DC blocking capacitors between the OPA1664 and TPA6120A2 or bias the OPA1664 output around ground using symmetrical supplies. This should fix the issue!

Matt

Hi Matt,

I'm assuming the TPA6120A2 is essentially an op-amp, and thus conforms to normal op-amp theory.  To wit: the voltage of the +/- input nodes doesn't really matter as long as it's within the common mode range.  What's important is that the sum of currents in and out of the nodes totals zero.  With the differential input circuit I have, since both inputs are at the same voltage, the unit is quiescing in a normal range.  I measure the nodes at 1.30V, well within the common mode range.  And the output is about 12mV.

Just as an experiment I removed the input resistors R19, R20, R24, R25 and powered up the circuit.  Then, the +/- inputs were also around 12mV.  The  TPA6120A2 reached the same 50°C temperature.  So, DC offset is not the issue.

However while measuring the  TPA6120A2 EVM for temperature, I noticed that, unlike my layout, all the NC pins are attached to a *huge* ground plane.  This is likely the explanation for the lower temperature.  The EVM board is probably 3x the area of my entire PCB, and the heatsink area is way way bigger.

So I'm thinking that my problem is simply that I need to provide more heat sinking.  Thoughts?

Best,
Marc

Hi Marc,

Sorry for the delayed response, I am transitioning into a new role at TI. I have discussed and passed this issue to a colleague within the team to get this straightened out for you.

Off the bat the lack of heat sinking the NC pins may be casing a temperature rise. I assume that the PowerPad is grounded to a large plane for heat dissipation?

Look for a follow up from the Audio Team.

Regards,
Matt

Hi, Marc,

I apologize, your issue got overlooked.

Did you get this resolved? If not, what do you still need help with?

-d2