DRV8850: Needs reset

Small correction, its oscillating about 5khz

David,

I am a little confused.  You first mention that you have to remove power for device to work again.  This would imply an over-voltage fault (section 7.3.9.4 in datasheet).  

But then you mention the output is switching at 5kHz which may be either an over-current fault or over-temperature fault.  Both auto-recover with no need to remove power.  Do you have a way to check temperature of case using either a thermal camera or thermocouple on top of case?

Regards,

Ryan

Hello Ryan, yes this is confusing. We've done more testing. Removing the battery does not always fix the problem, and sometimes the problem resolves by itself. So we think resetting may have been coincidental.

We have found that not all motors do this. Another motor will work perfectly. The good/bad motors test identical on an independent circuit. But these motors have very low resistance, less than 1 ohm. it could be the driver is mistaking it for a short circuit. We have very short wires and the driver is close to the motor. The motor terminals are only about 1cm from the driver outputs.

We thought that over current would just turn the device off. But if you say a 5khz frequency could be from over current, then we should assume this is an over current situation. Its not likely over temp, because this happens when the driver is cold from a fresh start.

Is there any way we can get around this? Does the VPROP resistor affect the current handling, or just related to the measurement? Perhaps using larger SR resistor to reduce the current spike?

If nothing we can do with the driver configuration, then we may have to adjust our motor winding to reduce the resistance slightly. Or possibly add a resistor in the circuit. We designed this system to get maximum torque from a single Lipo battery and very small motor. Maybe we went just a little too far.

Upon further testing we found that the stall current of the motor is higher than we thought. We don't have great equipment for this, but seems like its probably 12 amp at 3.3V with the motor locked. This was not an issue with the previous motor driver. But the DRV8850 seems to have an extremely fast protection trigger. I think the spec sheet says 1uS. Is there any way we can work around that?

Yes, we are using a ramp up with PWM, but does not make any difference. Even at 10% the driver fails instantly.

We did some testing comparing the motor power and speed with the DRV8850 and just straight from battery power. There is a marked difference in power. Much more than to be expected from the rdsON. So either the driver does not operate as expected, or the motor is just too large. But we only have 2 A running current.

David,

Do you have a current probe to monitor the current and check when the driver disables?  If you also scope one of the outputs, it should reset at ~4ms if it is an over-current event.  

What slew rate setting are you using?  

Regards,

Ryan

We don't have a current probe. But we can monitor battery voltage with a high speed analyzer, which gives us a very good indication. There is nothing that correlates to the 4mS over current event. At startup, the driver will oscillate at about 5khz for the duration of the programmed on time, which is about 100mS. We've tried various SR from 1k to 100k, but seems to make no difference at all.

In this example, we are monitoring battery voltage. You can see when the driver is on/off by the voltage drop. In this example, you can see we are ramping up the PWM from 10% to 100% in about 30mS. But after about 14mS, the driver starts to oscillate at about 5khz. Its roughly a 20% on duty cycle. Its rather random. Sometimes it will oscillate immediately, even at 10% duty. Sometimes never.

Now zooming in just after that 14mS period, you can see the oscillation. You can also see the PWM interlaced on top of it.

Based on the low power, even when functioning, I'm guessing this is related to the gate voltage of the drivers. They are not high enough to fully turn on the high side N channel MOSFET. But its hard to guess, because we can't monitor whats happening in the chip.

David,

Hard for me to see in plots.  What is the supply voltage on the DRV8850 dropping down to?  

Regards,

Ryan

In that example, it dropped to about 3.85V. But it really doesn't seem to be related to voltage. Same problem with fully charged or discharged battery.

David,

Do you have a high power supply to use instead of the battery?  That would help rule out the voltage drop as the issue.  Also, what bulk decoupling capacitance do you have on VCC?  With these high currents, I would suggest 220uF or higher.  That would help avoid the voltage drop outs.

Regards,

Ryan

Hello Ryan,

We could test with a higher voltage, but there really is no point, because it needs to function on battery voltage. That is what the IC is specifically designed to do. The VCC range on the chip is 2.0 - 5.5V. We are squarely in the center of that range, even with the instantaneous voltage drop.  We get same issue from 3.5 - 4.2V.  As for decoupling capacitors, we have tried very high values, up to 1000uF, with no effect on voltage drop. A battery is essentially a huge capacitor, so the additional bulk capacitor has very little measurable effect. Even though our stall current is 12A, that is absolute worst case. The driver is failing at far lower currents. And it is not failing due to current overload, because that would be confirmed by the 4mS recovery. It is oscillating, which would imply inadequate gate voltage to the high side.

"If" we were drawing more than 9amp, even for 1uS, for sure the battery protection would kick in, which it is not. We have tested this motor in direct drive with a battery that has a single 8205S MOSFET, which should be considered grossly inadequate for this application, but the motor has full power.

To be honest, we are at the point we need to abandon the DRV8850 unless there is some revelation we have missed.

The only thing I can think of that would be causing such a disparity between published specs and actual performance is that maybe we have counterfeit or defective product. These particular ICs we are using for testing were purchased about 1 year ago. I rather doubt these chips are being counterfeited because its rather new and not common.

David,

Went back and scanned the post and you mentioned that is depends on motors.  Some work and others have issues.  We have seen before that when brushes wear out on a brushed motor that it can cause intermittent shorts on the outputs.  This may be causing the issue you are seeing.  On the "good" motor, everything seems to work fine?  

One thing that would help if this is the case is some inductance between the outputs and motor.  A test if nothing else would be to put 10uH of inductance in series with the motor.  

After this, I may be out of ideas. :)  I appreciate your patience.

Regards,

Ryan

Well, the other motors fail rather quickly also. These are new motors, definitely not worn out. They work perfectly outside of the DRV8850. We definitely won't have space in our device to add a 12 amp inductor. There is no point in trying countermeasures that can't be used, and also not even recommended in the specifications.  I think maybe TI should revisit the specs on this driver because it absolutely does not perform.

David,

Got it...I was just going off your earlier post where you said other motors worked perfectly.  Seems that has changed as well.

Thank you for your feedback.  I think the only thing we haven't ruled out is thermal effects.  Do you have anyway to measure chip temp?  

And I know your patience is wearing thin, but have you tried on an EVM.  In stock and $29 if you are curious.  

https://www.ti.com/tool/DRV8850EVM

Regards,

Ryan

Its definitely not a heat issue because it can happen immediately from a room temperature system. Our testing of the chip compared to other discreet MOSFET and direct battery power is showing a significant reduction in power to our motor. This is even when it appears to be working correctly. So even without the output issue, this chip is not going to work for us.

David,

Ok.  I wish I could offer some alternatives, but this device is pretty unique with the voltage range.

Regards,

Ryan