eCAP interrupt latency

Chengcheng,

50ns is is not enough time to handle the PWM change through software.  The context save alone is longer than that.  

You can use the eCAP to measure the signal frequency using an internal 32-bit timer.  HRCAP provides higher resolution on the time-base so period & frequency measurements are more accurate, but it doesn't affect interrupt latency.

To meet your 50ns timing you'll need to use an analogue comparator to automatically trip the PWM. On F28335 the comparator must be external to the device and you'll trip the PWM in cycle-by-cycle mode through one of the trip zone (TZ) inputs.

The F28035 provides more flexibility because the analogue comparators are built-in, and their outputs routed to the PWM trip zones through the digital compare (DC) sub-module in the PWM. Peak current mode control of switching power supplies is done this way.

Regards,

Richard

Hi Chengcheng,

Have you seen the zero-crossing detection software in InstaSPIN-BLDC?  Good example for using the ADC to sense the zero-crossing of the BEMF from the motor's phase voltages.

If you have HALL sensors we also have example software in ControlSUITE that shows how to use the eCAPs for this interface.

Jeff

Hi Richard,

Thanks for your information. It is very informative. Does TI have any example program for peak current control with trip zone (TZ) inputs?

Thanks!

Chengcheng

Hi Chengcheng,

For an example of peak current mode control using F28035 you could look at the 1-day Digital Power Supply Workshop materials.  You can find these at:

processors.wiki.ti.com/.../C2000_DPSWorkshop

Lab 3 introduces peak current mode control using the internal comparators and PWM trip zones. I hope this helps.

Regards,

Richard

Hi Richard,

Thanks for your prompt reply. In peak current control, the PWM output is reset to low once tripped. My scenario is different from peak current control. If instead, I want to set the PWM and start the ePWM counter from zero, how to realize it with trip zone module?

Thanks,

Frank

Hi Frank,

Do you mean you want to reset the PWM counter on an input edge?

Regards,

Richard

Hi Richard,

Yes. I want to reset the PWM counter on an input edge, then the PWM will turn on the switching. 

Hi Rachard,

Thanks! I will also work with F28377, right? I am buying the 28337 lanchupad to do this.

Regards,

Frank

Hi Frank,

Yes, that's correct. The F2837x has a similar DC sub-module, so this is possible.

Regards,

Richard

Hi Richard,

I am coding 28027 now to realize the funciton. As you suggested, I need to force a timebase counter sync. Just realized that DCxEVT1 events can only be defined as one-shot rather than cycle-by-cycle mode. So I cannot refere to the peak current control code, which is using DCxEVT2 events. Any suggestion?

Description of my sencario: I am using DC_COMP1OUT and DC sub-module to generate a trip signal. After the trip signal, the PWM needs to be set with less than 30 ns delay. Meanwhile, the timebase counter needs to start counting to control the pulse width.

Hi Chengcheng,

Yes, you will need to configure one of the DCxEVT2.force events if you want to use cycle-by-cycle mode. You will also need one of the DCxEVT1.sync events to zero the PWM time-base counter on the same condition. The DCTRIPSEL and TZDCSEL registers will allow you to configure the same event to generate both outputs.

Regards,

Richard