DRV2603: Missing pulses in drive waveform

Arthur Thompson27

Part Number: DRV2603

Hello,

I have an LRA application that seems to be working fine (it buzzes), but when I inspect the waveforms they're not matching the DRV2603 datasheet Figure 2. The driver seems to be skipping pulses and I was wondering if someone could comment if this is somehow expected behavior. In the attached scope capture we're running a test sequence where we drive the haptic at 25, 50, 75, and 100%.

Thanks,

Arthur

over 5 years ago

0 Luis Fernando Rodriguez S. over 5 years ago

TI__Guru* 84915 points

Hi, Arthur,

Could you share scope capture with the PWM signal at full scale? Is it the first part of your graph above? It seems that this part coincides with the one in figure 2 of datasheet?

In addition, what voltage level and frequency are you using at the PWM? And what is the voltage level at VDD?

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Hi Luis,

At full scale drive, the PWM signal pulses low once per cycle for about 200nsec. The PWM frequency is 30.6kHz and the amplitude is about 3.3V.

I think the part I've circled in purple matches Figure 2 of the datasheet, but everything else seen here in the red differential drive waveform does not.

Edit: VDD voltage is 3.5V

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

Do you get the same results without using a load?

The datasheet mentions that the full-scale output voltage depends of the LRA ( www.ti.com/.../drv2603.pdf ).

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Luis,

Can you please explain if the missing pulses are part of the expected "braking" behavior or related to Back-EMF? I'm getting 6.8Vpp, 2.4Vrms which seems normal. But the actual waveform only hits 6.8Vpp for a fraction of the cycles, highlighted by me in purple. The rest of the time, the peak to peak waveform is smaller. Is this normal?

Thanks,
Arthur

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

Could you tell me what do you get at the output when there's no load connected, please?

As the datasheet mentions, the LRA architecture may affect the output voltage and consequently its waveform.

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Luis,

Here is the waveform at both outputs and the differential output with no load connected. What are you able to determine from this information?

Thanks,

Arthur

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

I'm thinking that probably the output waveform only behaves as the datasheet mentions under the recommended operating conditions. Have you tried to increase the VDD level to 5V and see if there are differences? Also, do you know if there's a change on VDD or EN pins when the you got a wrong waveform?

Best regards,
Luis Fernando Rodríguez S.

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Luis Fernando Rodriguez S.

TI__Guru* 84915 points

Hi, Arthur,

Excuse me, do you have any update on this? Could you take a look at my suggestion above?

Thank you.

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Hi Luis,

I don't see anything about 5V being recommended for VDD. I'm running VDD at 3.5V, which is about halfway in between the 2.5-5.2V recommended operating range given in Table 6.3.

Thanks,
Arthur

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

I agree with you regarding the VDD range specified in table 6.3. However, all the measurements and the plots were taken with some test conditions. The test condition was VDD = 5V in this case. So, I was wondering if this value could make a difference on the results that you are getting.

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Luis,

In which case was VDD set to 5V? Section 6.6, Figure 2 has Vdd = 3.6V.

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

Sorry for my misunderstanding.

You are right, the graph 2 was taken with VDD = 3.6V.

So, I have these additional suggestions:

Could you increase the capacitance at VDD and verify if there are changes at the output? VDD is strongly related to the OUT+ and OUT- behavior, so I just want to ensure that the power supply is not affected in your application.

In addition, which filter are you using for this measurement? As mentioned in the figure 2, the outputs are filtered. Please take a look at sections 7.1 and 7.2 for more details.

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Hi Luis,

I'm using the 7.2 setup for filtering.

For VDD filtering I have the recommended 0.1uF capacitor as recommended in section 10.1. It's located adjacent to pins 7 and 8 as shown in the reference layout. Are you thinking that using much more capacitance (1uF?) would make the circuit perform better? I ask because sometimes drivers do not want to see too heavy a capacitive load for stability.

Thanks,
Arthur

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

I have been doing several tests on our evaluation modules and it seems that the output is strongly dependent of the LRA actuator.

I got the same waveform than described in the datasheet with the LRA of our evaluation module ( www.ti.com/.../slou340b.pdf ).

However, when I disconnect the actuator, the output waveform has a similar behavior that the one you are getting.

Could you provide your actuator characteristics/datasheet, please?

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Luis,

Will send in a PM.

Thanks,
Arthur

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

I just accepted your invitation to the E2E chat. Please feel free to share the details and responses through the chat.

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Hi Luis,

I sent the haptic motor datasheet via PM. You can respond in this forum thread if you like.

Thanks,
Arthur

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

Thank you for sharing the details about the LRA load that you are using.

I didn't find anything on the datasheet that could point us to an LRA problem.

Could you take some capacitance measurements at the DRV2603 outputs? The DRV2603 has auto-resonance algorithms that track the resonance frequency of the device. This algorithm is susceptible to capacitance values at the outputs. So, we don't recommend to have capacitance values higher than 100pF.

Best regards,
Luis Fernando Rodríguez S.

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Luis Fernando Rodriguez S.

TI__Guru* 84915 points

Arthur,

In addition to my last comment, do you have the LRA load fixed or glued on a surface (enough heavy to avoid any LRA movement during the test)?

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

I do not have it fixed down. Would it be best to glue it down to something very heavy to try and replicate the results from Figure 2 of the datasheet?

0 Luis Fernando Rodriguez S. over 5 years ago in reply to Arthur Thompson27

TI__Guru* 84915 points

Hi, Arthur,

Yes, please. Could you try with this? Usually, our evaluation modules and characterization tools have some steel bars were the LRA loads are fixed. The most part of the DRV devices have some algorithms that are sensitive to the actuator state since they require of a feedback. Please take a look at the picture below for details about a normal test with our evaluation modules:

Best regards,
Luis Fernando Rodríguez S.

0 Arthur Thompson27 over 5 years ago in reply to Luis Fernando Rodriguez S.

Expert 1725 points

Thank you Luis, as soon as I glued the haptic motor to a big mass, the waveform started looking correct. I will mark this resolved!

Motor drivers

Motor drivers forum

DRV2603: Missing pulses in drive waveform