• State Resolved
  • Locked Locked
  • Replies 17 replies
  • Answers 1 answer
  • Subscribers 68 subscribers
  • Views 8076 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

UCC28070

Winston Chang
Prodigy 100 points
Other Parts Discussed in Thread: UCC28070, UCC28700, UCC28220, UCC28070A

Converter Requirements:

Vac:  80-265V

Vout:  390V

Pout: 1400W

PF:  >0.99 at rated load

Architecture:  2-phase Interleave

Controller:  TI UCC28070

Design Details:

Boost Inductor:  160uH-->260uH

Voltage Loop BW:  7.5Hz, PM=52 deg calculated

Current Loop BW:  Changed from 23KHz, PM = 52 deg to 10Khz, PM=40 deg

Cbulk: 492uF

Problem Description:

PFC converter becaomes unstable at full load when Vacin is slowly decreased across QVFF boundaries, particularly at low lines, <120Vac.  The disturbance is recoverable at light to mid load but is irrecoverable at full load.

Steps Required to Recreate Problem:

Set Pout=1400W, Decrease Vacin gradually from 135Vac to 75Vac.  Three QVFF boundaries will be encountered.  At each boundary the perturbation can be observed on the oscilloscope.  At some boundaries, particularly the lowest two, the converter will break into an irrecoverable unstable state.  The symptom is recoverable at Pout of 1000W and lower.

Discussion:

When converter entered the "unstable" state, it seems the controller is transitioning between QVFF states, ie insufficient input voltage hystereis.  During the symptom, all UCC28070 control signals are behaving "true to form".

Question:

Has anyone encountered a similar problem with this controller and how was it resolved?

  • 0
    TI__Guru 66135 points

    Hello Winston,

    The QVFF is supposed to have hystersis and should not be behaiving as you describe.  If there is exesive noise on the VINAC pin it could cause the QVFF circuitry to misbehaive.  A noisy ground poor layout could also cause the QVFF circuit to misbehaive.

    I have seen issues with input current distortion due to disconiuouse inductor current causing a false current sense signal at the CSA/B pins.  This casues the current loop to misbehaive.  This is covered in pages 8 through 11 in the UCC28700, 300W Interleaved PFC Pre-Regulator application note SLUA479b.  The following link will get you to the application note.  http://www.ti.com/litv/pdf/slua479b  Not sure if this issue but it is a place to look.

    I believe that I might have one of the 300W UCC28070 EVMs and will evalute it too see if I observe the same behaivor.

    Regards,

    Mike

     

  • 0 in reply to Mike O'
    Prodigy 100 points

    Hello Mike,

    Thank you for your response. 

    I have been using all available application information on the TI site including the document you suggested.

    Per the UCC28070 spec, the QVFF incorporates a 5% hysteresis to miminize chattering.  At high line, Vinpk=345V, this represents 17.25V.  At low line, Vinpk=130V, this is only 6.5V.  At low line the input current and input ripple current are higher.  The higher ripple current will induce a ripple voltage as measured at the Vinac input of the UCC28070.  The magnitude of the voltage will be dependent upon souce impedance and also the impedance of the EMI line filter.  I have confirmed that the symptom is exacerbated when a inexpensive Variac is used.  Six volts of hysteresis is simply not sufficient.

    When one transitions across a QVFF boundary, irrespective of loading, one can observe a momentary disturbance on the input current waveform.  In my case, the disturbance is recoverable up to Pout=1000W, down to the lowest QVFF boundary.  Only above Pout=1000W is the symptom irrecoverable.  If the 300W EVM design has a decent layout, you should see a recoverable event.

    I am reposting the waveforms I have previously captured and provided to members of your internal team.  The VINAC waveform is actually quite clean when a "short probe" is used.  I did not capture that waveform because at the time I did not have enough hands and did not want to risk blowing the unit up in case my hand slipped.

    Can you provide your phone number for further discussion?

    Regards,

    Winston

    3240.110926 1U 51-0286 Debug.pdf

     

     

     

  • 0 in reply to Winston Chang
    TI__Guru 66135 points

    Hello Winston,

    It sounds like you understand why the multiplier output is jumping in  your design.  The ripple voltage at VINAC is greater than the hysteresis of the UCC28070 QVFF function. The interaction of the source impedance and EMI filter is causing this misbehavior to be unrecoverable.  The hysteresis of the UCC28070 QVFF steps are fixed and cannot change so we will need to find a work around for your application.

    There is one thing you can try that may resolve the issue.  Try taking the VINAC signal before the EMI filter.  This just requires adding two diodes as a full wave rectifier before the EMI filter. One anode to each input leg and tie the cathodes to the VINAC divider.  The return path will be through the bridge rectifier.

    Regards,

    Mike

     

     

  • 0 in reply to Mike O'
    Prodigy 20 points

    Hi Mike,

    Would the UCC28070 still work with DC input? We've actually used it on a interleaved Boost converter. Vin = 40-90V, Vout = 147V.

    I am trying to use the same circuit but scaled to Vin = 15-38V Vout = 74V. But it seems to go to instability when I apply a step load from 10-13A

    Lawrence

  • 0
    Prodigy 50 points

    Winston, I observed the exact same phenomenon with the UCC28070 and asked TI about it at a seminar and via e-mail have not recieved any reply.  I also posted it on E2E November 02 and also have not recieved any replies on it.  My only current solution is to increase the amount capacitance of the 400V raw supply which is not a desired solution from a cost or a volume point of view.  I believe that  the descrete voltage feedforward steps acts as a distrurbance in the voltage loop which with falling AC input voltage causes an transient increase of the output voltage.  With the OVP being very close to the regulated voltage (as low as 104%) and the voltage loop being necessarily slow the OVP trips the PFC off until the DC voltage falls sufficiently.  This is typically a 1/2 cycle of the AC input so the PFC can end up operating on only one polarity of the AC input voltage which is not very good indeed.

    I would like to know a better solution. 

    I think TI should increase the OVP level somewhat for this IC or go back to the tried and true analog voltage feedforward method.

         

  • 0 in reply to Donovan Davidson
    TI__Guru 66135 points

    Hello Donavan,

    Sorry that the TI field engineering team did not look into this for you. 

    I did look at the thread and it looked like you had resolved your issue by adding more output capacitance.  Even though that is not a desirable solution for you.  How does the size of your output capacitor compare to the recommended value from the application note and excel design tool?

    I do agree with you that the analog multiplier seemed to be much easier to use than the QVFF.  I have discussed this with systems engineering team hear at TI on many occasions and they awer of it.  Unfortunately there are no plans at this time to redesign the UCC28070. There is a two chip solution using the UCC38528 and the UCC28220 that uses a traditional multiplier that you can use.  This link will get you to the application note for it.  There is also an EVM for it. http://www.ti.com/litv/pdf/slua369b   

    Regards,

     

  • 0
    Prodigy 50 points

    Winston, you are correct that there are two problems with the quantized feed forward in the UCC28070, one I described earlier which can be 'fixed' by increasing the raw supply capacitance. The second, which you are currently experiencing is due to the quantized feed forward hysteresis being insufficient to account for reasonable AC line source impedance resulting in the QVFF transitioning between states. This has no solution except to tell you customer to use a lower AC line source impedance or for TI to go back to the tried and true analog multiplier divider approach.

  • 0
    Prodigy 140 points

    Hello Winston,

    I am facing a similar issue. Did you find a solution to overcome this instability?

    Looking forward for any suggestion.

    Regards,

    Amardeep

  • 0 in reply to Amardeep Shetty
    Prodigy 100 points

    Hello,

    Sorry, it appears folks are still experiencing the same problem.  The issue is multi-factored but boils down to the Qff states at low ACinput line introducing sufficient output bus disturbance that exceeds the OVP limits of the UCC28070 (if I recall correctly OVP is 104-108% nominal).  TI offers a minimum output capacitance to output power ratio to circumvent the problem but if your power density needs to be high, the proposal is basically not usable.  What I did to overcome the issue was to introduce an active clamp with precision resistors to ensure the sensed output voltage cannot reach 104% nominal.  You are better off using an external independent OVP network anyway.

    I hope this helps!

    Winston

  • 0 in reply to Winston Chang
    Prodigy 50 points

    Winston.  It is sad that TI has not resolved this issue.  The UCC28070 is otherwise a good IC and is used by several of my customers regardless of my suggested alternate approaches and I am sure by many others.  They should come out with a UCC28070A with a higher OVP voltage and much less hysteresis in the OVP levels and an analog voltage feed-forward divider circuit.  As it is there are many compromised products out there. 

  • 0 in reply to Winston Chang
    Prodigy 140 points
    Thanks for the inputs Winston, really appreciate your prompt response. The issue was bugging me since quite few days, will incorporate the same approach on my board and check if it addresses my issue.

    Regards,
    Amardeep
  • 0 in reply to Amardeep Shetty
    Prodigy 140 points
    Winston,

    I have a query, hope I am not bothering you too much with my questions!
    Well I have added an active clamp to the Vsense signal so that it never reaches 104% of nominal. Also, added an external OVP signal which goes low at OVP voltage(415V) and have fed this to the SS pin of the UCC28070; so the PFC controller is disabled at OVP.
    Now, when I test, the PFC works till around 50-54% load (i.e almost 2.5kW) above that the current rectification seems to fail momentarily and then it recovers.
    I didn't test it for a longer duration as I don't want to blow up any more power devices at this stage.
    Did you encounter similar behavior in your design ?

    Regards,
    Amardeep
  • 0 in reply to Amardeep Shetty
    Prodigy 100 points
    Amardeep,

    The independent OVP circuit should not be connected to the SS pin of the UCC28070. Should you do so, once you exceed OVP threshold (which you will at heavy load and decrease the Vacinput) the controller will discharge the SS capacitor and shutdown requiring substantial time (depending on the value of SS capacitor) to return to the original operating state. This will cause the bus voltage to dip more than necessary and in the recovery stage and cause the converter into a perpetual off/on condition.

    I would prefer you consider connecting the OVP comparator to VAO (voltage amp output), so that the bus voltage is clamped to the OVP threshold during the QFF disturbance.

    We are able to smoothly decrease Vacinput at full rated load w/o having to resort to increasing the bus capacitance.

    Good luck!

    Winston
  • 0 in reply to Winston Chang
    Prodigy 140 points
    Winston,

    I connected the OVP comparator to VAO, the problem still persists. Its recovering faster than earlier but it still looses control intermittently which is quite evident in the current waveforms & also not acceptable.
    I will have to analyse further to see if this QFF issue can be resolved in our circuit.
    Thanks Winston for your prompt replies and guidance.

    Regards,
    Amardeep
  • 0 in reply to Amardeep Shetty
    Prodigy 30 points

    HI Amardeep Shetty
    I have you problem whit ucc28070. i use more cap in IMO pin that reduce instability but not solve complet.are you can solve it?

  • 0 in reply to jafar faroughi
    Prodigy 140 points

    add a SMD cap on IMO Pin? what is the value of this cap? how to solve this QVFF issue?
    i had seen many projects which added a cap on IMO PIN.

  • 0 in reply to jafar faroughi
    Prodigy 140 points
    Jafar,

    Mine worked for 4.8KW with a resisitive load but failed to work above 3.5KW (approx.) with other load types. I cant afford such failure rate and I am looking at other PFC controllers.
    The cap on IMO solution I tried but more cap will lead to poor response and poor shaping of the input current and in turn poor power factor.