• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 4 answers
  • Subscribers 63 subscribers
  • Views 167 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.

LM5143: Only works in the default 5V/ 3.3V mode. Does not work using feedback

john pittenger
Prodigy 40 points
Part Number: LM5143

Tool/software:

I am completely dumbfounded. Using the VDDA & Ground on the feedback pins generates 3.3V and 5V as advertised. When changing the feedback resistors to 110K over 12K for a roughly 6V volt out, output jumps to about 14 volts at power up for roughly 100 mSec,  then drops to zero. I have grounded the VCCX pin, Vin is 32 volts.

It acts like it does not recognize the resistor divider configuration and does not go into that mode of regulation. 

This seems like a simple thing to do, yet I can't get it to respond across multiple boards. 

What am I missing?

  • 0
    TI__Guru 73952 points

    Hi John,

     Please send a completed schematic and quickstart file for this design. I assume this is your board, not the TI EVM.

    110k and 10k feedback resistors should be fine (the parallel is greater than 5kΩ, so it does not detect FB = low at startup). Are you sure there is no open circuit in the feedback divider? Even if there was, Vout should operate close to Vin and not drop to zero.

    Regards,

    Tim

  • 0 in reply to Timothy Hegarty
    Prodigy 40 points

    Hi TIm. Things are evolving. The design is your EVM, but we have swapped out the relevant caps and FETs to be able to run this at 60V in and 28V out.  The issue I wrote about turned out to be a dead board.  Subsequent copies came up working. ( that's frustrating). I am seeing a different issue, that appears to be thermally related. If we allow the internal LDO to supply the 5V, the part gets hot quickly and shuts down. Supplying the VCCX pin with an externally generated 5V (thereby bypassing the internal LDO) the part appears to run maybe 20 degC cooler. The power supplied to this VCCX  reports to be about 0.8 watts.  After 5 minutes, the part shuts down again. Now, with the part shut down, the VCCx supply is reporting about 5 to 6 watts. That surprised me. Our target application will be in the vicinity of 300 watts and I am having concerns if this part is up to the job.  The part specifies a Vin max of 65 volts, and a dual path combined current output of 14 to 15 amps. But, maybe the combination of the two is asking too much?

  • 0 in reply to john pittenger
    TI__Guru 73952 points

    Hi John,

    0.8W/5V = 160mA VCC current. This seems extremely high. The DC current from VCC to drive each FET is Qg*Fsw, where Qg is taken at Vgs = 5V. Note that the EVM is setup for 2.1MHz, so you may want to run at 400kHz if Vin = 60V.

    Please send a completed quickstart calculator so I can review power losses.

    Regards,

    Tim

  • 0 in reply to Timothy Hegarty
    Prodigy 40 points

    Well, filling out that calculator was a disaster. My choice of FET was abysmal. I thought it was a close approximation to the EVM FET, but the efficiency plot dropped off the bottom of the grid.  So, is there a tool for recommending FETs given my design constraints?

  • 0 in reply to john pittenger
    Prodigy 40 points

    I got ahead of myself and analyzed the wrong FET. The one I'm using is not too bad. I am sending the quickstart filled out  with the FET values per the one I am actually using.(iAUC50N08). LM(2)5143-Q1 quickstart design tool - revB2_filled out.zip

  • 0 in reply to john pittenger
    TI__Guru 73952 points

    Hi John, can you check if this is actually a logic-level FET (Rdson rated at Vgs = 4.5V).

    Comments on the quickstart:

    1. Fsw = 2.1MHz is really too high for Vin = 59.5V. Try 400kHz.
    2. The shunt value seems too high (current limit is lower than the full-load rating - it's typically 130-150% of Iout-max).
    3. Inductor values are way too low - target 30-40% ripple current.
    4. Make sure ceramic Cout is derated for voltage.

    Regards,

    Tim

  • 0 in reply to Timothy Hegarty
    Prodigy 40 points

    Thanks Tim.  I was hoping to wiggle as few things as possible from the stock EVM design. It looks like that is necessary.  As an update, I did change the frequency to 500 KHz, per your earlier direction and saw a 20 degC drop in the package temperature. I ran the design at 35V in, 28Vout 4 amp load for 100 minutes yesterday.  In that time frame, it shut down 4 times about a two minutes apart roughly around the 30 minute mark, each time restoring when I cycled the load. It then ran steady for the next 60 minutes uninterrupted.  

    As an additional datapoint, this morning I dug out our one remaining untouched EVM and configured it to run in combined mode, with Vin at 12V, Vout of course defaults to 3.3V, and I loaded the output at 10 amps. It ran for 20 minutes before dying.  I had glued a thermocouple on the package and it rose to 49 degC +/- 1 degC and stabilized there. Thus was the temperature when it died.   I'm interested in your thoughts here. 

    ~John

  • 0 in reply to Timothy Hegarty
    Prodigy 40 points

    The FET claims to be logic level with Vgs threshold @ 1.6V and Vgs = 4.5 V with Rds @ 10.3 mohm.nominal

  • 0 in reply to john pittenger
    TI__Guru 73952 points

    Hi John,

    3.3V/10A on an EVM ...what was the Fsw here? 400-500kHz should be no problem. What do you mean by combined mode?

    Can you measure the current into the VIN pin of the IC (normally there is a resistor in series that is useful for this). The IC dissipation is then Vin *Ivin.

    Regards,

    Tim