This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

UCC28019A: Non-Sinusoidal Current and abnormal duty cycle at full load.

Part Number: UCC28019A
Other Parts Discussed in Thread: UCC28019, , UCC28180

I am using the UCC28019A for a design that has 185Vac-305Vac input range and 600Watt @ 450Vdc output.

I used the Texas Instruments  design calculator SLUC117E to calculate the recommended component values.

The circuit design works perfectly with loads up to 400Watts however in the 500-600Watt load range the AC line current becomes non-sinusoidal.

More Specifically the AC line current waveform appears to be clipped at the peak, however there is no change in output voltage it holds steady at 450Vdc.

Using a current probe I checked the boost inductor current waveform and found that at the peak of the AC line where this phenomenon occurs the switching duty cycle is inconsistent.

When the circuit is working properly the duty cycle will vary gradually in synchronization with the shape of the AC line voltage, however in this case the duty cycle varies significantly from one switching cycle to the next, but only in the area around the peak of the AC line.

Also this issue exists over the entire input voltage range from 185-305Vac.

If I reduce  Rsense from the recommended  value of 100mOhms to 50mOhms the issue disappears.

However this is not an acceptable solution because it increases the SOC and PCL thresholds by a factor of 2 which makes the boost inductor design a problem.

Using the TI Tina Reference Circuit model for the UCC28019 I found what I believe is the cause, i.e. the voltage on Pin2 (Vicomp) is too high.

At  a load of 600Watts the simulation indicates that the voltage on Pin2 reaches its maximum limit and results in clipping of the sinusoidal waveform at the peak exactly where the inconsistent switching duty cycle is occurring in my prototype circuit. 

In an attempt to verify the simulation results i added a 4.3Vzener in series with a 5K resistor from Pin2 to Pin1 to prevent the Pin2 voltage from reaching its maximum limit.  This did indeed eliminate the inconsistent duty cycle issue at the peak of the AC line however it also resulted in crossover distortion of the AC line current waveform.

Can you please give me some guidance regarding how to make adjustments to component values so that Pin2 voltage (Vicomp) does not max out too soon?

Also upon review of the UCC28019AEVM User's Guide, page5, the schematic shows  unpopulated components R17 & C9 which connect to Pin2, would adding these components be a possible solution? If so what are the associated design guidelines.

Best Regards,

Jeff Ferrick

  • Hello Jeff,

    The UCC28019A (and also the UCC28180) has a non-linear current amplifier whose gain is dependent M1,M2 coefficients controlled by the voltage error amplifier.
    You can see a block schematic of this in Section 7.2 of the data sheet.
    M1 is the current loop gain and M2 is the ramp slope.

    You can use equation 76 to calculate the M1M2 product.
    You can then calculate the value of M1 when the power demand is greatest.
    If M1 is too high the current amplifier will saturate and that is what is occurring in your application.
    The current amplifier saturates at a value of ICOMP = 5V

    The only way to prevent saturation is to lower the ISENSE signal  and this means reducing the RSENSE resistor value.

    If RSENSE is set too high you will get a current waveform that looks like this :

    Regards

    John

  • John,

    Thank you for the prompt response.

    I reviewed my calculations and it does not look like M1 or M2 are too high.

    For my circuit design; M1 = 0.41 ,  M2= 0.62, Vcomp = 3.75

    For the TI reference design; M1= 0.484, M2= 0.764, Vcomp= 4.0

    The value for the current sense resistor I am using ( Rsense = 0.100 ohms), corresponds to an Isoc threshold current of 6.6amps peak and I am experiencing this issue at a threshold of approximatly 5 amps peak inductor current.  

    If the current amplifier is saturating under these conditions wouldn't  that mean that it would never be possible to reach the Isoc threshold and consequently the SOC  function could never work?  Also can you please reconfirm the saturation voltage for the current amplifier ( V Icomp, Pin2 ), Section 6.1 of the datasheet specifies 7.5V  vs   5V from your response above ?

    Also, the current waveform in your response (above) is the opposite of what I see with my prototype.  In your example the current abruptly increases at the peaks of the  AC line and in my scenario the line current is clipped (flat) across the peaks of the AC line.

    Jeff.

  • Dear All here!

    I think I'm facing the same issue that Jeff has. First I suspected poor assembly and bad wiring, so I rebuilt my whole circuit with a little better routing of the power/signal GND trance and especially the Isense signal line, so by now there must not be any ground bouncing or anything noise related problem. To be honest, the first build wasn't so bad regarding the GND and Isense traces, so it was only a try.

    When I calculated an Rsense=0.146 ohms for my Vin range, I set the PCL -> Ipeak(*1.25) around 7Amps which is way higher than I really need, since my target is only 400W @ 230VAC +/-10%, and I don't care about 85VAC input. The inductor and stuff in the main circuit is suitable for this high current, so I keep this setup and run the tests. Vout setpoint is 390V, full load resistor is 380 ohms. At light load and full load the Vout keeps at 390V, but the input current is not good near full load. I hear the bad noise from the inductor caused by inconsistent switching, so I took a closer look on the waveforms. And I see that inductor current starts "clipping" at 3-3.5Amps with 2.5ms/div on the scope. Then I took an even closer look on the waveform (25us/div with a time delay to see the peak current) and realized that there were two ~70% duty cycle ON and after a complete OFF cycle. So at the current peak the duty cycle pattern is 70%-70%-0%. With the 0% duty the current falls a lot, and then the two 70% tries to rise it up again, but then at the peak the 0% comes in again and keeps up till the descending side of the half sine.

    So, the waveform in my and in Jeff's situation seems to be the same, but not like the one in John's answer.

    I did a lot things to cure this problem but nothing helped. In the end I resigned to descrease Rsense to 0.11ohms by adding more parallel elements in my shunt and the problem started to disappear but still not the best. It smelt it was like false triggering of PCL.

    An exact ICOMP "saturation" is not mentioned in the datasheet, only an absolute maximum is defined, but besides there is a "can go to 7.5V +/-6% due to internal drive circuitry" text which doesn't mean that this is the normal maximum output of the amplifier.

    Later at the "6.5 Electrical Characteristics" there is a test condition with ICOMP=6.4 regarding to minimum duty cycle (Dmin at PWM section). So, it seems that there shouldn't be an ICOMP saturation at 5V, otherwise the actual datasheet on the site is bad or not up-to-date.

    Also an interesting thing about the problem is when I apply a lighter load, for example 560 ohms, there is no problem even at 110VAC input, but in this case the current is higher than it was with the 380 ohms load at 230VAC input - I checked this with the first 0.146 ohms Rsense. So, this can't be a false PCL triggering (because of noise), maybe it is really a problem of the control loop but Jeff and I , we are not out of the specs of the datasheet.

    I attach some pictures, sorry for the quality but I only had phone photos of the scope screen.

    First, there is a far view of the inductor current when the problem occours.

    Current peak is ~3A.

    And the switching is like this at the top (not made on the same day but in the same situation):

    And now after take the ICOMP voltage saturation account I made a measurement on pin 2 (ICOMP) with the 0.11 ohm Rsense which is still too high but the inconsistent current waveform has a narrower width. It seems that that amplifier really saturates at 5V. Then the datasheet is bad when states 6.4V ICOMP for Dmin. So is this mean that we can't protect our circuit for overcurrent if we want higher power than 350W (design example)? Datasheet states >2kW power with this IC.

    Regards,

    Gyula

  • Gyula,

    Thank you for posting the oscilloscope pictures.  The results you have documented are virtually identical to mine.

    Jeff.

  • The current amplifier has an absolute maximum voltage of 7.5V and has  has a maximum linear range to 5.6V.

    If you check the value of ICOMP at the peak of your waveform you should find that the amplifier is saturating.
    The only way to prevent this saturation is by reducing the current sense resistor which lower the value of ICOMP.
    The UCC28180 has the same issue but the calculator was modified to calculate the correct Rsense.
    The UCC28019A calculator has not yet been modified.

    You should be able to see on your current waveforms that the PWM is at a  max and this is what is giving the flat top shape to the current.

    I think the waveform I attached was not as severely in saturation as these other waveforms.

    In any case you are correct in your observations.

    Regards

    John

  • John,

    Thank you for confirming the linear operating range for the current amplifier section.

    Is there any way to calculate Vicomp (pin2) as a function of Visense (pin3) or as a function of gate drive duty cycle?

    Jeff.

  • John,

    Thank you for confirming the linear operating range for the current amplifier section.

    Is there any way to calculate Vicomp (pin2) as a function of Visense (pin3) or as a function of gate drive duty cycle?

    Jeff.

  • Hi Jeff,

    You can use equations 76 to 94 of the data sheet for the UCC28019A to calculate the current amplifier gain.

    With these equations you have a relationship between the input (pin 3 ISENSE) and the output (pin 2 ICOMP)

    The equations are valid up to V( ICOMP) = 5.6V

    This will give you a max value for V(ISENSE)  = 5.6/ G where G is the gain of the amplifier.

    Regards

    John

  • John,

    I have been working through formulas 76-94 however I have not been able to figure out how to determine the gain (G) of the current amplifier.

    Formula 76 yields an M1M2 value of  0.239V/uS for my design.

    Using equations 81-89 I created a high resolution version of Figure27 with tabular data and found that if M1M2 = 0.239 then Vcomp = 3.71V

    Consequently M1 = 0.403  and M2 = 0.597

    Based on the Functional Block Diagram on datasheet page 11, it looks like parameters K1( K1 = 7) and M1 set the current amplifier gain but I could not find an equation that defines how to use K1 and M1 to calculate G.

    Can you please provide me with the appropriate equation?

    Also, datasheet equations 92,93,94 reference "non-linear gain variable M3" but I cannot find M3 anywhere else in the datasheet.  What is the significance of M3?

    Jeff.

  • John,

    I have been working through formulas 76-94 however I have not been able to figure out how to determine the gain (G) of the current amplifier.

    Formula 76 yields an M1M2 value of  0.239V/uS for my design.

    Using equations 81-89 I created a high resolution version of Figure27 with tabular data and found that if M1M2 = 0.239 then Vcomp = 3.71V

    Consequently M1 = 0.403  and M2 = 0.597

    Based on the Functional Block Diagram on datasheet page 11, it looks like parameters K1( K1 = 7) and M1 set the current amplifier gain but I could not find an equation that defines how to use K1 and M1 to calculate G.

    Can you please provide me with the appropriate equation?

    Also, datasheet equations 92,93,94 reference "non-linear gain variable M3" but I cannot find M3 anywhere else in the datasheet.  What is the significance of M3?

    Jeff.