Random shut down on DRV8829 driving a dc motor.

Hi Dave,

It is my impresion at this moment the problem you are observing is thermally related. If the fault is non latching, then it is clearly not an OCP. Do you see the nFAULT line asserted during this faulting period?

To gain more confidence on the previous statement, however, it would help to take a look at the Gerber files. You specify they have a good thermal relief. Actually, the term "thermal relief" scares me a little bit as it implies the power pad is connected through some small traces to the remainder of the copper structure. If my assessment is correct, then we have a poor thermal path which will increase system's thermal impedance. Can you corroborate what is the case?

Other things to check is how well soldered the device is to the Power Pad. In essence we need at least 50% solder coverage between the device's Power Pad and the board's Power Pad opening.

To answer your questions:

1. Yes, the DRV8829 can be PWM'd on either PHASE or ENABLE input pins. There is no preference on our side, but there will definitely be a preference on the customer's side as the algorithm to control the DC motor load will be completely different on their side.

2. There is no suggested PWM frequency, but I would like to express my personal preference for something like 20 KHz to 30 KHz. In theory you can still go up to 100 KHz, but you must understand there will be a penalty in switching frequency losses, whereas you will really not gain much in terms of motion quality or audible noise components. Hence, there is really not much of an advantage to use anything larger than 30 KHz and if this can be avoided, I would recommend it. It seems you are using 10 KHz, so this is perfectly fine as well.

3. Using ITRIP current regulation on a DC motor only makes sense if you are trying to control torque. If you are applying your own PWM, I would try to stay away from the internal current regulation engine as this will impose a PWM on top of your PWM, which in my opinion is ill fated. If you want to use the internal current regulation engine to control speed or torque, you will need to disable your own PWM control and close the loop on the analog input VREF, alone. I have never tried it, but in theory it should work.

4. I am not aware of any sensitivity to noise that you should look into, but like with anything in this field, good layout practices are always recommended. For example, we like to see the GND plane to be as solid as possible. We also like to see the SENSE resistor to be as close to the SENSE pin as possible and with a nice trace connecting both elements. The idea is not only to minimize resistance on the path, but also inductance. No need to mentions caps, fat traces to limit resistance, fair amounts of copper to aid thermal energy released out of the chip, and so on.

Hope this info helps. Best regards,

Jose Quinones

Jose -

I have attached an Xray image that was take from a suspect assembly yesterday.  This shows the PCB thermal relief (looks good) as well as the soldering.  Since I am not skilled in looking at this xray image, can you (or someone at TI) comment on this image ?  Does it appear that the heat slug is soldered correctly ?  If not, do you have an example xray image of what it should look like ?

Hi Dave,

It definitely looks like there is some solder between the device's Power Pad and the board. I would imagine this should be enough to tackle a 2W power dissipation inside the device, given the amount of copper extending outside of the device is enough to keep it cool for prolongued periods of time. Can we take a look at the Gerbers? If they are available, you can send them to jquinones@ti.com.

One last question. You mentioned 2.5A average. Any chance the motor is actually asking for much more than this for some period of time? Say it is running at 2.5A continuous for 4 minutes, but then starts to require 4A for a few seconds. Such a condition would need to be taken into consideration. Is it possible to get a good picture of wha the power profile looks like? Thanks!

Best regards,

Jose Quinones 

Sorry for the long time lag on this response.  I was not able to get the complete gerbers from the customer.  This design went from DRV8828 to a DRV8829 (higher current).  The original board was laid out with the DRV8828 foot print and thermal recommendations.  Did you know the DRV8829 Thermal pad is much larger ??  Customer is making PCB changes and will be testing shortly.  Will advise on the results.

Now, onto the PowerPAD data sheet issue -

Both DRV8828 and DRV8829 have a PWP (R-PDS0-G28) package call out, but have different thermal pad dimension.  If a customer puts the DRV8828 pad dimensions in their CAD library, they may not get it correct for future boards and future parts.  I would like to suggest that TI add some additional call out so that designers will quickly pick up the difference.  Maybe something like PWP (R-PDS0-G28) (Thermal A).  The datasheets are correct, but could be confusing as customers switch between different DRV88xx devices.

Regards,

Dave Johnson

May 7th Update -

Power pad changes do not seem to make any difference.  If it helps we are applying the PWM signal to control the chip (and motor) to the PHASE pin so 50% is idle / no motor movement,  0% is full power in one direction and 100% is full power in the other direction.  The issue we are seeing is the NFAULT line would activate if we were moving the motor and hit the current regulation in the chip (set by the ISEN, VREF, and I0-I4).  We have I0-I4 all pulled high with a 3.32K (100%) and the decay mode was also pulled high with a 3.32K (putting the chip into fast decay mode).  

We switched the DECAY line on one of the chips we have been having issues with from pulled up (fast decay) to pulled low (slow decay) and we no longer see the issue.  I can now drive the chip into the set current limit and it will hold there for long periods of time without throwing the NFAULT line like it was previously.  

Are we not using the decay mode correctly ? 

Are there conditions under which the fast decay mode will cause excess current to flow in the part and thus tripping the OCP?  Is there interaction between  fast decay mode and the current limit feature.

Any guidence / feedback is appreciated.

 Dave

Hi Dave,

I am afraid I will not be able to offer a conclusive answer until I see the Gerber files. I can tell you, however, this does not look like OCP and that there is no way to generate OCP by selecting a particular DECAY mode. But if what you are saying is that on SLOW DECAY the fault goes away and with FAST DECAY the fault appears, then this points to switching losses and hence thermal problems. In FAST DECAY you switch all four FETs and on SLOW DECAY you only switch two FETS. This only applies to current regulation, however, and since you are PWMing the PHASE pin this should not matter what the DECAY pin is set to. In essence you are always in FAST DECAY. 

I do want to point something out, though. You should not use current regulation while you are PWMing a control line unless this is for stalling control and current limiting. If you allow current regulation to kick in, then you will have a PWM on top of a PWM. In essence, this is the same as a loss of control as the PWM you are using to control motion performance no longer applies. If you are controlling motor speed/direction by PWMing the PHASE pin (or even the ENABLE), then I would make ITRIP larger than the current you will ever see or accept the loss of control when the current regulation kicks in.

We can use that last behavior to point out to thermal problems again. As you describe, the problem appears when the current regulation is engaged. Since your PWM now has another PWM on top, chances are the switching losses increase all of a sudden, bringing in new power dissipation and increasing temperature on the device.

On another note, the fact that DRV8829 works better than DRV8828 also points out to thermals. DRV8829 has better RDSon than DRV8828 which is why it can do much more current. But if with DRV8829 and FAST Decay you are still experiencing the problem, this tells me your layour does not have enough copper to tackle this application. Such a small change should not cause so different behavior.

If I recall from previous communications, the amount of current you want to drive requires top notch layout, so it is crucial I take a look at the art work in order to determine whether the required current is manageable or not. Although the device is rated at 5A, this only applies if the thermal impedance as offered by the layout is good enough. Feel free to send the Gerber files to jquinones@ti.com and I'll be glad to take a look as well as make suggestions on what could be going on.

Hope the info helps. Best regards,

Jose Quinones