DRV8244S-Q1LEVM: failure on evaluation board

Hey Rob, 

So when IN1 and IN2 both go HIGH at the same time in PWM mode as you show, then if DRV_OFF is a 0 then both outputs would go HI-Z.  This could cause VM pumping, and an overvoltage event is likely what damaged the device.  See this article, The art of stopping a motor – VM pumping.  I also suspect this because your VM voltage is 30V and the part is rated for 40V, you will need to be a little careful about avoiding VM pumping or add a voltage limiting circuit.  

You can also use the calculation tool "Full Bridge Driver Junction Temperature Estimator" on the device page to estimate device temperature to rule that out.  I ran it with 5A at 4V/us slew rate and that should overtemperature almost instantly, but at the 43V/us should be fine indefinitely. 

Regards,

Jacob

Hey Jacob,

Thanks for your quick response. I tried to simulate the pumping on PSpice, but do not see this occuring. (probably due to ideal components in PSpice??) I tried several overlap times (both IN1 and IN2 High), with different SRs. What I do see is several current  in I(U1:VM) spikes up to 5A when the outputs are switching, as if both FETs in a half-bridge are ON simultaneously. Could this also be a reason that the device is damaged?

I’m now thinking that PH/EN mode is better suited for our application. If I keep EN/IN1 high, and PWM (20kHz) on PH/IN2, I should get the alternating (0V – VM) OUTs, just like I want. In simulation, I also see current peaks in I(U1:VM), especially when OUT1 is rising and OUT2 is falling. Any thoughts on this?

This is my simulation model and results:

Another question, why is IPROPI negative in the simulation model? Is this a model failure?

Thanks in advance,

Rob

Hello Rob,

Can you remove the PI filter and see if the current spikes are still there on Vm as well as can you check the voltage on the node after the PI filter and see if this node gets pumped with the existing configuration:

Rob Boesten said:

I’m now thinking that PH/EN mode is better suited for our application.

I agree PH/EN might make things simpler and in that case, you have to just worry about sudden PWM duty cycle changes for VM pumping. Give me some time to look into why IPROPI is negative.

Best regards,

Keerthi Kumanan

Hi Keerthi,

Thanks for your answer. First simulations show that also without PI filter the (sporadic) spikes are present. This concerns me because they seem like shoot-through currents, which will damage the device. Can the load of about 12mH and 1.1DCR be a problem? I'm not able to do work on this project this week, but will come back to you next week.

Thanks in advance,

Rob

Hi Rob, 

Can you try a lower PWM frequency and a higher duty cycle and see if the sporadic spikes are still there and or how they change. 

Best,

Keerthi 

Hi Rob,

Any updates.

Best,

Keerthi

Hi Keerthi.

I tried to simulate with lower PWM frequency (10kHz, which is definely not our desired frequency), and higher (80%, I(L1) = positive) and lower (20%, I(L1) = negative) duty cycles. In both cases, I see the spikes occurring, see images below:

To be completely clear, this is my setup:

First, I think it is remarkable that the spikes (shoot-through??) only occur at the end of discharging the coil. (in this case also IPROPI is negative)

Second, why are the spikes occuring sporadiccally?

Thanks in advance, Rob

EDIT: The maximum step size in the simulation profile seem to have an influence on the spikes in I(U1:VM). The simulations above were done with 1us max. step size.

See below an compare between max. step size 1us (left) and 0,1us (right), with PWM frequency = 1Khz, 90% dutycycle.

See below a “zoom-in” from the right image above:

Hi Rob,

Have you gotten a chance to try it out on the bench and capture the same waveforms. Can you also describe your entire setup, so I can try to replicate it on my end.

Best,

Keerthi 

Hi Keerthi,

I did not try the PH/EN mode on the bench, only in PSpice simulation. Since the simulation is not showing good results, i was hoping for a solution, before i build a test-setup with my second evaluation board (DRV8244S-Q1LEVM). I'm thinking that if there are shoot-through currents (over 21A) like in the simulation, the DRV8244 will be damaged again, right?

Best,

Rob

Hi Rob, 

Give me some time to look into this and consult a more experienced member and I will get back to you by the end of this week.

Best,

Keerthi 

Hey Rob,

There will not be any shoot through conditions on the output FETs as the device has a built-in automatic dead-time generation.  

I see you're using a 30.7V supply, there is very little headroom before the absolute max value for VM is exceeded with VM pumping up.  I don't think the SPICE model will show VM pumping  unless it has an accurate model for the BDC motor itself since VM pumping comes from back EMF.  So I don't think the simulation will help a ton here

I see you're using GUI version 0.5.0, this isn't the latest version for a Rev A EVM (MD047A).  Recommend using latest version, at time of writing that is 1.0.1:  https://dev.ti.com/gallery/view/MotorDriversBSM/DRV824x_DRV814x-Q1EVM-GUI/  You'll also have to perform a firmware update via the new GUI.  

Regards,

Jacob

Hi Jacob,

Thanks for your answer. I could simulate VM pumping up using a non-sinking voltage supply (using diode). In your first answer you mentioned a voltage limiting circuit, do you have any recommendations?

Thanks in advance,

Rob

Hi Rob, 

What is your application and are you going to be using a battery in your application and if so what type? you might also want to consider the DRV890x family of devices, which has an integrated overvoltage protection.

Best,

Keerthi 

Hi Keerthi,

our application is a proportional fast moving solenoid which is precisely position-controlled in an automotive environment.  We want to use the DRV824x family to replace our existing H-bridge with drivers, in order to simplify the design. In order to move the valve correctly, we need to change duty cycle fast, which leads to VM pumping up. We are looking into serveral options to handle the VM pumping. We will not use a battery for this.

Best, Rob

Hi Rob, 

I will look into this and give you an update Monday.

Best,

Keerthi 

Hi Keerthi,

any updates?

Rob