• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 354 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

WEBENCH® Tools/UCC28634: Webench design violates datasheet constrains

Piotr Wyderski
Prodigy 150 points
Part Number: UCC28634
Other Parts Discussed in Thread: UCC28711, UCC28701, UCC28631, UCC27531

Tool/software: WEBENCH® Design Tools

I have created a flyback design based on UCC28634. The parameters are: VIN=180..470VAC, 50Hz, VOUT=12V, IOUT=4A, Temp=50 degrees. Webench told me to:

1. Select RHV1 (the resistor connected to the HV pin) of 100kOhm. This violates the requirements from table 7.3, which say 180..220kOhm.

2. Select R_CS=330mOhm, transformer's L_primary=552uH and in the design report it says that T1 Iprim pk=1.794A. But then:

a) section 9.2.4.7, equation 29 says that the saturation current of the primary should be 2.5041A, assuming V_cs_max=800mV. This is a huge difference, affecting core selection. Webench designs a transformer for I_pri_pk=1.94A, which is reasonable for 1.794A, but would explode at 2.5A.

b) section 8.3.7, equation 16: R_cs/L_pri=597.83, which should be less or equal to 431.285 (V_IN_PK_MAX=470*sqrt(2), v_cs_min=172mV, t_on_min=600ns).

Webench and the datasheet cannot be both right at the same time, so could you please resolve the three cases described above?

  • 0
    TI__Guru* 86266 points

    Hello,

    1. The RHV1 resistor is for controlling the XCAP discharge.  The data sheet does recommend between 180K and 220K.  So you could increase the resistor value to 180 K in your design.

    2. The data sheet example is for 19.5V 65 W design, your design is for 48W.  I would expect the Webench values would be different compared to the design example.

    3. Depending on what assumptions the we bench designer made the calculations and values could be different.

    4. There is an excel and Mathcad design tool that you can use as well to check your design. You can find them in the following link. https://www.ti.com/product/UCC28634?keyMatch=UCC28634&tisearch=Search-EN-everything&usecase=part-number#design-development

    Regards,

  • 0 in reply to Mike O'
    Prodigy 150 points

    Hello Mike,

    Mike O' said:
    1. The RHV1 resistor is for controlling the XCAP discharge.  The data sheet does recommend between 180K and 220K.  So you could increase the resistor value to 180 K in your design.

    There are no XCAPs in this design, this is a pure DC-powered HV flyback operating directly from the PFC output capacitor and driving a SiC switch. As you can see in the tags, the controller is UCC28634, which has no XCAP discharge support. I do not want to have any resistor there, as the UCC28634 startup cell is rated only to 700V, which is not sufficient here. Hence the question about the active high voltage steerable current source based on a depletion mode MOSFET in a cascode configuration with the HV pin of the controller. Is this allowed or not? If not, why not? Do you happen to have the detailed knowledge of the internals of this built-in startup cell? It seems to be necessary to answer this question.

    Mike O' said:
    2. The data sheet example is for 19.5V 65 W design, your design is for 48W.  I would expect the Webench values would be different compared to the design example.

    Of course the values are different and the output power level is even not the most important driving factor here. I am not referring to the values, but to the very strict design procedure described in the datasheet. If you follow this procedure with the values calculated by Webench, you will quickly discover violation of several constraint mandated by the datasheet. This is exactly what I have done. Please refer to my previous post to find the details; I can also support you with the project, if needed. Something is clearly very wrong here and the TI support is supposed to sort out what.

    Mike O' said:
    3. Depending on what assumptions the we bench designer made the calculations and values could be different.

    Very well, but whatever the assumptions Webench has made, they must not violate the constraints defined in the datasheet or the datasheet is wrong and urgently needs to be updated. There is no third option. All I have is the datasheet, so I cannot resolve this issue on my own. You can ask the designers if neecesary, I cannot.

    Thank you, I will have a look at the spreadsheet.

       Kind regards, Piotr

  • 0 in reply to Piotr Wyderski
    TI__Guru* 86266 points

    Hello,

    These devices have an internal depletion mode mosfet.  The data sheet explains that this resistor is used for Xcap discharge timing.  It does not show any examples without the resistor. 

    We have a UCC28711 device that has high voltage startup that I know that does not require this resistor. 

    I will let the Webench team know that they their design tool violates the data sheet recommendations. 

    Regards,

    Mike

  • 0 in reply to Mike O'
    Prodigy 150 points

    Great, thank you! Could you please suggest your technical authors adding an example without the resistor or at least emphasising the fact that it is solely for discharge timing and has no purpose during the startup phase, except for obviously slowing it down?


    Additionally, a prototype implemented with UCC28634 according to the Webench design does not lift off at low voltage (but still 40V above the minimum assumed by Webench). It goes into an infinite startup-reset cycle. The waveforms at the drain, the current sense resistor of the MOSFET and the low-voltage windings look basically right, but the VDD voltage is steadily decreasing, down to the restart level. The controller leaves the 3 exploratory pulses phase, there are maybe 100 follow-up pulses, and then the restart cycle happens. Shorting the upper 33k VSENSE resistor with a 150pF capacitor makes it start, and so does replacing it with a potentiometer in series with a 10k resistor and dialling some low range setting. The controller is then PSR regulating something around 14V, not sure why. It does so with the minimal 3k3 load resistor at the secondary side, attaching a 24-ohm load (0.5A out of assumed 4A current budget) makes the unit stop. Both the secondary and the auxiliary windings have the same number of turns and are aimed at 12V. I am writing this to complete the case. On the practical side: first, I am tired of debugging the Webench design -- winding a sufficiently small planar transformer with the required ratings was enough pain on its own. Secondly, repeating the calculations in Matlab strictly following the datasheet procedure indicates that it is just impossible not to violate the previously mentioned constraints with the selected core. Your Excel spreadsheet confirms that. The L_pri/r_cs ratio and I_sat call for a huge core (E38) due to the 600ns t_on_min limit of the controller. UCC28701 has 300ns and is a strictly DCM part, so I am going to give it a try with the already wound EQ30 transformer.


    An earlier prototype based on UCC28631 with a much bigger PQ26/20 transformer and 19.5V@3A V_OUT (1280uH of L_pri instead of the 552uH of the transformer I have now) basically worked correctly in a similar setup. Unfortunately, I no longer have it and cannot just replace the transformer with the (wrongly designed) planar one to see what happens.

    The UCC28711 appears to be UCC28701 with a startup cell, so it can simplify the startup circuit design. Unfortunately, it still has that 25mA gate drive limit and the 14V gate voltage clamp, which does not allow seamless integration with a SiC part like the SCT2750.  A UCC27531 will be needed to supplement that. But at least the controller can operate in the 21+V VDD range, as opposed to the UCC28634, so a single VDD for both chips should suffice. BTW, are you considering adding a SiC-tailored PSR controller to your portfolio?

    OK, I will start prototyping with the UCC28701, because I already have one, and if satisfied with the results, consider the UCC28711 to simplify the external D-mode FET startup cell. Thank you, this might work.