• State TI Thinks Resolved
  • Locked Locked
  • Replies 15 replies
  • Answers 2 answers
  • Subscribers 63 subscribers
  • Views 914 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-POWER-DESIGNER: How does webench calculate RCD snubber power loss?

Thomas Nilsson
Prodigy 195 points
Part Number: WEBENCH-POWER-DESIGNER
Other Parts Discussed in Thread: UCC28C54

Hi

In the attached Webench design of a flyback AC/DC converter, Webench has calculated the RCD snubber power loss to 1,202 W.

According to the article on Snubber design guide lines from the TI website, linked below, the following formula is presented for calculating snubber power loss.



The webench design operating values gives the following parameters

Vsn = 214,2V

Ipeak = 1,847 A

n = 119:12 = 9,92

Vo = 12 V

fs = 41951 Hz

Webench doesn't specify any leakage inductance Llk.

So, my first question is - How does Webench calculate the snubber power loss?

Using the given values and solving for the leakage gives Llk = 7,5uH

Webench has specified a 1,552mH ETD34 Wurth transformer, looking at other ETD34 transformers, leakage current is typically about 10 times this value.

So my second question is - Is this really a realistic leakage current and power loss, and if so, what examples of a transformers with similar specs are there?

/cfs-file/__key/communityserver-discussions-components-files/196/Design-Guidelines-for-RCD-Snubber-of-Flyback-Converters_2D00_Fairchild-AN4147.pdf


2352.WBDesign5.pdf

  • 0
    TI__Guru 73575 points

    Hi,

    It looks like the Llk should be around 17uH, or 1.1% of Lm.

    Vsn is about 2 to 2.5 x nVo, nVo = 9.9 x (12V +1.05V) = 129.2V, so if assuming 2.5x

    Vsn = 2.5 x nVo = 2.5x 129.2V = 323V - this is different from your 214.2V.

    Ipk = 1.735A, for Pin = Po +Pd = 84W + 14W = 98W, not 1.847A needed

    Besides, Psn is more than snubber Pd because not all Psn dissipatate to snubber resistors, some of Psn also dissipated to MOSFET M1

    Based on M1 part IPP65R190CFD, its Coss = 86pF, which dissipate about Pm1 = 0.56W through M1 channel when it turns on.

    So Psn = Pd + Pm1 =  1.2W + 0.56W = 1.76W

    with the above adjustment, and the formula from your message, the leakage inductance is about 17uH, 1.1% of Lm.

    This is typical from our design experience but towards lower end while still in the reasonable range. 

    Note, the leakage inductance usually not very well specified, and also usually with a much larger than real specs from what we know working with various vendors for long time, the main reason is the leakage is not a production test in general so they would loosely specified to cover their reliability. From our experience, the leakage is typically with in 1% to 2% of Lm, this webench estimation still in the effective range.

  • 0 in reply to Hong Huang
    Prodigy 195 points

    Could you please take part of the information and values in the post and the attached data and base an answer on the information given, instead of your own made up stuff that is not relevant to the question asked.

    Vsn is given in the webench document just like I wrote in my first post above, it is stated Vsn = 214,2 V

    Ipeak is also given 1,847 A. NOT 1,735 A like you say. 

    If the power loss given by webench also includes losses in the MOSFET, then the losses in the snubber circuit is even lower than 1,202 w , which just makes my question even more valid.


  • 0 in reply to Thomas Nilsson
    TI__Guru 73575 points

    Hi,

    Ok, I will see if I can get the webench team to answer how Pd = 1.2W is determined. This may take some time as I am not sure the designer for this model still available.

  • 0 in reply to Hong Huang
    TI__Guru 73575 points

    Hi,

    The Webench team received my request about your question how snubber Pd is estimated. The team is to find the answer.

  • 0
    Prodigy 195 points

    Ok, I am waiting for the answer.

  • 0 in reply to Thomas Nilsson
    TI__Intellectual 2330 points

    Hi Thomas,

    We are working on this post. Kindly allow us sometime for us to get back on your query. 

    Appreciate your patient support.

    Thank you

    Best Regards,
    Sateesh A G 

  • 0 in reply to Sateesh Goudappanavar
    Prodigy 195 points

    Ok, thanks for the update.

  • 0 in reply to Thomas Nilsson
    TI__Intellectual 2330 points

    Hi Thomas,

    Kindly allow us some more time to respond to this query. 

    Thanks for your patient support.

    Best Regards,
    Sateesh A G

  • 0 in reply to Sateesh Goudappanavar
    Prodigy 195 points

    How are you doing with this?
    It really doesn't seem like a hard to answer question.

  • 0 in reply to Thomas Nilsson
    TI__Intellectual 2330 points

    Hi Thomas,

    Greeting of the day!

    UCC28C54 device had a small bug in its WEBENCH model and we had to fix it first.

    We also noticed that there is difference in Snubber Pd calculations in WEBENCH model with respect to Snubber Pd formula shared by you. On WEBENCH, we are considering Rsnubber loss and Diode losses as a part of snubber loss, this value is not matching with the above equation shared by you, we are confirming this with product experts on this.

    Kindly allow us some more time to get confirmation.

    Appreciate your continued support. 

    Thank you

    Best Regards,
    Sateesh A G

  • 0 in reply to Sateesh Goudappanavar
    Prodigy 195 points

    There are obviously several very big bugs in the WEBENCH model.
    But you really don't need to fix them in order to answer my question.

    The equation is not shared by me, it is shared by you, I just referenced your documents.

    There seems to be a lot of things that doesn't match with WEBENCH and the data sheet, but totally impossible to get any answers.

  • 0 in reply to Thomas Nilsson
    TI__Intellectual 2330 points

    Hi Thomos,
    We are still working with product experts on two queries you have asked.

    Webench calculates snubber power dissipation using different equation than shared formula above. So you might not able to back calculate the leakage inductance using the power dissipation from Webench.

    Typically values of leakage inductance is within 1 to 2% of Lm.

    Meanwhile we will check if the Webench formula used in snubber calculation is giving correct estimations. We will share our observations with you once I have an update from product expert.

    Thank you

    Best Regards,
    Sateesh A G

  • 0 in reply to Sateesh Goudappanavar
    TI__Intellectual 2330 points

    Hi Thomas,

    The WEBENCH team has reviewed the formula used to calculate snubber power dissipation on WEBENCH. Please see our observations below:

    1. The power dissipation formula used on WEBENCH is incorrect and does not match the Fairchild App note. We are currently in the process of updating the power dissipation formula on WEBENCH. Product experts are discussing the accuracy of the Fairchild App note formula. Once a conclusion is reached, the WEBENCH team will update the model.

    2. On WEBENCH, leakage inductance is assumed to be 2% of primary inductance for all calculations, including power dissipation. Since the power dissipation formula was incorrect, the back-calculated value of leakage inductance turned out to be incorrect as well.

    3. For now, please consider that leakage inductance is assumed to be 2% of primary inductance for your calculations as well (as shared by @Hong Huang in the above reply). Meanwhile, we will update the WEBENCH model

    Thank you

    Best Regards,
    Sateesh A G

  • 0 in reply to Sateesh Goudappanavar
    Prodigy 195 points

    Hi Sateesh

    Thanks for the info. I am pleased to see that you have fiiiiiiiiiinally realized that Webench is incorrect. Please have a look at my other posts on the topic of the UCC28C54 and Webench design, as there are many more parts that are totally wrong in the designs presented by Webench.

    Best regards, Thomas

  • 0 in reply to Thomas Nilsson
    TI__Intellectual 2330 points

    Hi Thomas, 

    Appreciate your patient support. We are closing this thread. 

    Kindly request you to select "Responsible Organization" as "--WEBENCH Tools" to post further queries so that it'll reach us.

    Thank you

    Best Regards,
    Sateesh A G