PFC implementation regarding spra902.pdf Application Report

Ant,

Usi is the controller output after saturation. Ui(n) is the same controller output before saturation. Therefore Epii measures  the difference between the saturated and normal output. This difference is used with Kcorri to unwind the controller from its saturated state.

We also released a PFC Evaluation kit last year. You can get the s/w and documentation by installing ControlSUITE from TI website. After you install controlSUITE, this PFC folder will be under ControlSUITE\development_kits\ILPFC_v1.0. You can also order the PFC kit from eStore.

Shamim

Hi Shamim,

Firstly thank you very much for your answer, and nice to know the Application Report author heard about my question.  I have to admit I have a fairly good understanding of the PFC operation now, but the control algorithm is proving difficult to implement, so hoping I can ask for a little more of your time?

As per my previous post I am using the basic TMS320F28027 processor.  I have not used Q15 notation or Assembly, so I don't have the best implementation for a PFC.  I new nothing on the subject of PFC's 7 weeks ago, so would implement this differently in the future.  However I am quite sure I am close to having something that works, albeit a very noisy current waveform.

I have all the rectified voltage samples and calculation of the average voltage working fine, as well as the a calculation for the line frequency.  The PWM frequency is now set to 24kHz, and I trigger an ADC read on every third event, this gives enough time to run the code inside the ADC interrupt without any overrun (checked this by polling a GPIO on/off to confirm stable timing).

I am trying to develop a second PI control algorithm based directly off your application report, and have the following so far.  Bearing in mind I am not using Kcorri and relying on floats!  Therefore I am unsure how to implement Usi (if I can at all).

	if (Slow_Start == 1)
	{

		Err_V = V_Ref - Vout_Volt;				// Ref voltage - Instantaneous Output Voltage
		V_control = Kp_V * Err_V + Iv_nminus1;	// Proportional voltage constant * Voltage error + Iv_nminus1
		Iv_n = Iv_nminus1 + (Ki_V * Samp_Freq) * V_control;	// Iv_nminus1 + (Integral voltage constant * Sampling Frequency=0.000125) * V_control

		I_Ref = (V_control * Rect_Volt) / ( Vavg * Vavg);	// This is effectively A*B*C from the application report

		Err_I = I_Ref - I_Rect;					// I_Ref - Instantaneous current sample
		I_control = Kp_I * Err_I + Ii_nminus1;	// Proportional current constant * Current error + Ii_nminus1
		Ii_n = Ii_nminus1 + (Ki_I * Samp_Freq) * I_control;	// Ii_nminus1 + Integral current constant * Sampling Frequency=0.000125) * I_control

		Duty_Cycle = I_control * PWM_getCmpA(myPwm1); // I_control * latest duty cycle

					// limiting for testing will extend limits as test progress
					if (Duty_Cycle > 1800)
					{Duty_Cycle = 1800;}
					else if (Duty_Cycle < 400)
					{Duty_Cycle = 400;}
		
		// interleaved design so 2 PWM's to update
		EPwm1Regs.CMPA.half.CMPA = Duty_Cycle;	// Update PWM1
		EPwm2Regs.CMPA.half.CMPA = Duty_Cycle;	// Update PWM2
		
		// Store values
		Ii_nminus1 = Ii_n;
		Iv_nminus1 = Iv_n;
		
		//d_nminus1 = I_control;

	}	

I have a 2nd algorithm which produces the oscilloscope traces shown in my previous post

	if (Slow_Start == 1)
	{

	Err_V = V_Ref - Vout_Volt;
	V_control = V_control_nminus1 + Kp_V * Err_V + (Ki_V * Samp_Freq - Kp_V) * Err_V_nminus1;
	I_Ref = (V_control * Rect_Volt) / ( Vavg * Vavg);

	if (I_Ref > 0.7)
	{I_Ref = 0.7;}

	Err_I = I_Ref - I_Rect;
	d = d_nminus1 + Kp_I * Err_I + (Ki_I * Samp_Freq - Kp_I) * Err_I_nminus1;

	Duty_Cycle = d * PWM_getCmpA(myPwm1);

				if (Duty_Cycle > 2400)
				{Duty_Cycle = 2400;}
				else if (Duty_Cycle < 10)
				{Duty_Cycle = 10;}

	EPwm1Regs.CMPA.half.CMPA = Duty_Cycle;
	EPwm2Regs.CMPA.half.CMPA = Duty_Cycle;

	Err_V_nminus1 = Err_V;
	Err_I_nminus1 = Err_I;
	V_control_nminus1 = V_control;

	d_nminus1 = d;

	}

I appreciate you are busy and understand if you don't have time to look into this further, will leave the post as unanswered for now and change it to answered after the weekend.

Best regards,

Ant