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.

LM5085: Current runaway concern

Part Number: LM5085
Other Parts Discussed in Thread: LM5141

I am looking to use the LM5085 in an application where the current limit runaway is a real concern.  I have a relatively large bulk capacitor on the LM5085 output (which will look like a short circuit until it is charged) and also the risk of an actual short circuit on the output.  I have gone through datasheet section 7.3.5 - Current Limit Off-Time.  For my 30V input I get a required Vfd + Vesd of about 300mV.  This was using the nominal values for Ton_min and Toff_min.  

Does this result seem correct?  This seems easy enough to meet, as the diode Vf is at least 400mV.  I do see the note that mentions that variations in t_on and t_off tend to move together.  I assume this means that I do not need to worry about a minimum t_on and maximum t_off, which would give me a much higher requred Vfd + Vesd of about 800mV.

I guess the question comes down to - how common is it to have trouble with current runaway for this part?  And if it is a real concern, what steps might you recommend to make sure it cannot happen in actual usage?

  • Hi David,

    Ton-max is 237ns plus PFET turn-off delay 50ns

    Toff-max is 8670ns plus PFET turn-on delay 50ns

    Vin=30V

    So VF+Vdcr=0.987V

    VF=0.4V, 

    so Vdcr should >=0.587V at peak current

  • Hi Andy,

    Can you help me understand where those numbers came from?

    For Ton-max, I see a worst-case value of 235nS for Ton-4 in Electrical Characteristics.  Is that where your number is from?

    For Toff-max, I used equation 10, but this gave me 16uS.  How did you calculate 8670ns?  Is that my 16us with a tolerance applied?  Or something else?

    Rather than adding the PFET turn-on and turn-off delays, the last paragraph on page 13 seems to state that you add the difference of the delays to Ton-max, rather than the absolute values.  In this case the difference is 0, which would make a significant difference.  Is that correct?

    Lastly, are your numbers the largest Ton-max and smallest Toff-max, to get the absolute worst case result?  I typically would design for that, but I want to make sure I understand the note in the datasheet that seems to mention that this combination will not happen - "A device which has an on-time at the high end of the range will have an off-time that is at the high end of its range" - does this mean that a device with a worst-case Ton-max of 235ns would have a Toff-max higher than the typical value, rather than smaller?

    However the numbers end up, other than selecting a higher Vf and DCR, is there anything you would recommend to insure that the current cannot run away?  It seems a shame to intentionally make the design less efficient to deal with a potentially fringe case.

    Thanks,

    Dave

  • Hi Dave,

    Yes, the Ton-min 235ns is from the data sheet Spec. table. 

    you can calculate Toff with Vfb=0V and Vin=30V by equation 10, and add +/-50% tolerance for T-off min/max value. 

    Use 235ns as the max Ton-min value and Toff-max value at 30V to calculate Vf+Vdcr, also verify it with 55ns as Ton-min and T-off min at 30V, pick the larger result.

    Do you consider LM5141 for your application?

    B R

    Andy

     

  • Hi Andy,

    If I use 235ns as the max Ton-min and the maximum Toff-max (calculated + 50% = 24uS) I get Vf+Vdcr = 322mV.

    If I use 55ns as the max Ton-min and the minimum Toff-max (calculated - 50% = 8uS) I get Vf + Vdcr = 304mV.

    Is this what you mean?  To calculate using both maximums (235nS, 24uS) and both minimums (55nS, 8uS) and select the worst case?  

    If this is correct, is approx. 322mV the number I should use?  Or is the answer 987mV but I am misunderstanding something?

    I have some application-specific reasons why I cannot use the LM5141.  Would it be possible to contact you off of the forum about this?

    Thanks,

    Dave

  • Hi Dave,

    Ok, pls send me email at andy.chen@ti.com, 

    I will close this thread first.

    B R

    Andy