DRV8301 / TPS54160 higher voltage then calculated

Lukas Palkovic

Prodigy 10 points

Other Parts Discussed in Thread: DRV8301, TPS54160, TMS320F28052F

Hi all,

I use DRV8301 and there is a problem with its integrated buck converter.

It is based on TPS54160.

The main problem is that I set 3.18V and the output is 5.35V.

The internal reference voltage is specified as 0.8V, but I measure 1.34V on Vsense pin.

There are some other facts that seam strange:

The switching frequency is 2100kHz, though set to 1100kHz (by 100kOhm resistor).
The PWRGD is tied to output by a pull-up resistor. I measure 2.4V.
The output voltage is stable with changing load (20mA or 500mA).
The output voltage changes slightly with input voltage
5.35V for 6V input,
5.44V for 12V input
5.24V for 30V input

Does anyone have a clue what could be wrong?

Can this be caused by a wrong board design or quality of components?

over 10 years ago

0 JohnTucker over 10 years ago

TI__Guru**** 191445 points

Poor layout and bad quality components can cause a variety of issues. In your case, the results seem consistent though. How many circuits exhibit this problem? If it is just one I suspect the device is damaged.

0 Lukas Palkovic39 over 10 years ago in reply to JohnTucker

Intellectual 420 points

Thank you for your answer.

I have 3 circuits with the wrong output voltage.

I measured only one in detail. The other two have slightly different layout.

I tried to use the output inductor and capacitor from the original TI board. The behavior is the same.

We have more boards also with ICs from 2 different suppliers, so we'll try other ones too.

I hope this time the schamatic will display correctly:

Lukas Palkovic

0 Lukas Palkovic39 over 10 years ago in reply to Lukas Palkovic39

Intellectual 420 points

So I tried another boards having DRV8301 from a different supplier.

Please see also the layout of the converter.

0 JohnTucker over 10 years ago in reply to Lukas Palkovic39

TI__Guru**** 191445 points

It may in fact be a layout issue. I can't really tell too much from the parts you show. Can yu pleas supply a little larger area with the top and bottom layers in the same size so I can superimpose them. Even better would be your PCB file.

0 Lukas Palkovic39 over 10 years ago in reply to JohnTucker

Intellectual 420 points

Please see the Eagle files:

buck_converter.sch

buck_converter.brd

0 Lukas Palkovic39 over 10 years ago in reply to Lukas Palkovic39

Intellectual 420 points

and gerber files, in case it's easier to check thesebuck_converter.zip

0 Lukas Palkovic39 over 10 years ago in reply to Lukas Palkovic39

Intellectual 420 points

Can anyone check the layout, please?

I checked the output signal on oscilloscope.

The PH and output signal are in the following picture. The frequency si set by 100kOhm to 1100kHz.

I changed also the frequency to some 2000kHz, the picture follows:

0 JohnTucker over 10 years ago in reply to Lukas Palkovic39

TI__Guru**** 191445 points

I can't download any of your files. I can tell you the switch node waveform looks very abnormal.

0 Lukas Palkovic39 over 10 years ago in reply to JohnTucker

Intellectual 420 points

OK, I'll insert simplier files:

0 JohnTucker over 10 years ago in reply to Lukas Palkovic39

TI__Guru**** 191445 points

I am not surprised you are having trouble with that layout. You pretty much have to follow the layout guidelines closely. The 4.7 uF input cap should be placed directly at the PVDD pins and connected to top side ground copper. It needs to be a ceramic cap as well. The circulating current path of L1, D1 and C16 should be as tight as possible with wide traces or even better copper fill areas. I really do not recommend any routing under the IC on the top layer. It should be dedicated copper pour for GND. The inductor footprint looks small. Is it rated for your current?

0 Lukas Palkovic39 over 10 years ago in reply to JohnTucker

Intellectual 420 points

Thank you very much, John.
Size matters, every mm2 counts. We did not follow it exactly, but tried to have as short wires as possible. Also for the L1 current path - 7mm. If I compare it with the design of TI board, the our paths are much shorter. I agree the shielding of a GND copper pour is missing.
The converter is used to supply power only for 1 MCU, we calculate with 200mA current, though it will be less. The inductor has 0.4A rating and 1.4 Ohm and is made for such purposes. We were testing other inductors we had as well.

I am afraid the recommended design would stretch the board a lot and it is not the way. In our case we'll probably look for a more compact separate DC/DC or maybe even linear regulator and find a space there.

0 Alexandre BALDINO over 8 years ago in reply to JohnTucker

Intellectual 520 points

Hi John,

I've the same phenomenon, my design should output 15V, but I have 25V.

Vsense is around 1,34V.

I've 5 boards, one of them has this problem, so I can't consider a layout origin.

The DRV8301 is linked to a TMS320F28052F. I've damaged the piccolo when I put the oscilloscope probe on the DRV8301's DC_CAL pin. I suppose I've done a short-circuit between DC_CAL and GVDD, as I didn't put a resistor on the DC_CAL signal between the piccolo and the driver in order to protect the microcontroller.

Before this bad handling, the board worked very well.

So first I thought the DRV8301 was damaged, and then I understood it was the microcontroller. That's why I changed the two devices.

The driver seems to be well-soldered, except I haven't soldered the thermal pad for convenience.

Can it be the cause of my problem (I'm afraid it is)?

Thank you,

Best regards

0 JohnTucker over 8 years ago in reply to Alexandre BALDINO

TI__Guru**** 191445 points

So originally, your bad circuit had the correct output of 15 V, but went to 25 V output after mis-probing the DC_CAL pin?

0 Alexandre BALDINO over 8 years ago in reply to JohnTucker

Intellectual 520 points

No, I wanted to explain you why I changed the driver.
Before misprobing, the board worked correctly, with an output voltage of 15V.
Then I changed the DRV8301, but didn't solder the thermal pad.
Now I have an output of 25V.

Alexandre

0 JohnTucker over 8 years ago in reply to Alexandre BALDINO

TI__Guru**** 191445 points

I would check for solder defects then re-solder the IC including the thermal pad.

0 Alexandre BALDINO over 8 years ago in reply to JohnTucker

Intellectual 520 points

Hi John,

I've changed again the DRV8301, and soldered the thermal pad.

Now it works!

Comment: I'm surprised that the thermal pad is not connected to the ground analog pad (pin number 28) on the component.

Thank you for your support

Best regards

Alexandre

Power management

Power management forum

DRV8301 / TPS54160 higher voltage then calculated