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.

TPS54060A: Bad regulation with some units

Part Number: TPS54060A
Other Parts Discussed in Thread: TPS54060EVM-590, , TPS54060, LM5161, LM5017

Hi!

We use the TPS54060A to generate +-17V from +24V (18..30V). As reference, we used the schematics of the evaluation board TPS54060EVM-590. At about 80% of all devices, the output voltage is stable. I mean if I varry the +24V to +19V, the output does not vary much, and for example only goes down 0.1V from 16.8V to 16.7V. The "good" devices also show not much impact when the current is increased from for example 80mA to 100mA. In contrast, the remaining 20% are very sensitive to the input voltage and also to the current drawn. Some exemplars show only +16.4V at +24V and less than +16V at +19V, so it varies with the input voltage. Therefore the negative voltage goes even more negative to for example -18V and less, I assume the sum stays approx. 34V. For regulation, we first used two resistors at input VSENSE (see the schematics below).

A hint by someone of this forum was to add a third resistor to GND (R99), to balance it out. But also this third resistor did not really help.

We then increased the resistor to the positive side (R40) to bring the voltage up. For example, we incresed the resistor so that a device that shows only +16.4V outputs nearly +17V. The drawback is that the negative voltage goes more negative, but this we think is no problem for us.

Unfortunately, this does not really help, as the devices showing this problem are also very sensitive to the current drawn. If we increase from for example 80mA to 100mA, some examples show a voltage drop of appprox. 0.5V.

Has someone an idea? - Can it be a capacitor that causes this? Or the inductor?

Thanks!

  • Can you share the waveforms for this design?

    Switch node, input, and output waveforms.

    Thank you.

  • Hi Marshall

    Wafeforms at +24V (pin VIN) and +17V in my design?

    What do you mean by "switch node waveform"?

    Many thanks!

  • Hello Stephan,

    Marshall is out for a couple of days so I will fill in. By SW node, Marshall is referring to the PH pin of the TPS54060 and also the anode of D2. Since diodes rectify the waveforms in your circuit, attention to ringing is important. This ringing is due to the leakage inductance for each winding of your transformer. Since this ringing energy is limited, often preloading the supply helps. The other condition that should be checked is if the part will go into skip mode. The power available to the secondary will be limited under this condition.

    Regards, Robert

  • Hello Robert

    Thank you very much for your informative answer! I really apppreciate it.

    I appended some screensshots, probably not exactly what you need (the VIN, +17V_BF, and -17V_BF signals only show some crosstalk), so please let me know.

    I have a two channel device, so I combined PH with VIN,, +17V_BF, and -17V_BF for two devices, a "good" one regulating fine, and a very bad one, reaching only approx. 15.2V at +17V_BF.

    Best regards,

    Stephan

    Bad_Device_PH_+17V_BF.PNG:

    Bad_Device_PH_-17V_BF.PNG:

      

    Bad_Device_PH_VIN.PNG:

    Good_Device_PH_+17V_BF.PNG:

    Good_Device_PH_-17V_BF.PNG:

    Good_Device_PH_VIN.PNG:

  • Hello Robert!

    Have you seen my sreenshots? - Probably not want you wanted to see from me? - Please let me know.

    Thanks & best regards

    Stephan

  • Stephan, Robert or I will get back to you shortly.

  • Thank you for the feedback!

  • Hi Stephan,

    I notice from the SW node voltage waveforms above that the converter is operating at the CCM-DCM boundary. Note that the isolated output may not regulate accurately in DCM, hence why you may be seeing some output voltage shifts. My recommendation is to increase the mag inductance of the coupled inductor, thus pushing the CCM-DCM boundary to lower current (worst case is max Vin). Also, you can check that regulation is okay at higher current levels to verify this is indeed the issue.

    As an aside here, note that it's important to minimize noise on the VSENSE line. This trace should be as short as possible and kept away from noisy nodes. Also ensure that the two feedback resistors are located close to the VSENSE pin.

    Regards,

    Tim

  • Hello Tim!

    Thanks for the ideas. i have already ordererd some parts for testing, it will take some days.

    Best regards,

    Stephan

  • Hello David!

    > Are the wave forms above taken from a bad unit?  

    from a good and from a bad. Please see the file names, the files are named with "good" and "bad".

    Best regards,

    Stephan

  • Thanks, Stephan.

    We need to keep this non-synchronous buck converter topology running in CCM to regulate correctly. If you need operation down to light load, please consider the LM5161 (100V, 1A) and LM5017 (100V, 0.6A) converters that are Fly-Buck compatible.

    Regards,

    Tim

  • Hello Tim!

    We first changed the resistors R40 and R36 from 499K and 12K to 240K and 5K9, but this brought no real improvement.

    We then changed the coupled inductor from 470uH to 680uH and this improoved it very much. With R40=240K and R36=5K9, the unit that even not reached +16V settled above 17V, even with VIN=18V, and the regulation is also acceptable. Just for a test, we increased from 680uH to 820uH, but then it again get's very worse with lover voltage reached and also bad regulation. We will change resistors and inductor on  set of "bad" devices and see if it is also fine with them.

    > If you need operation down to light load, please consider the LM5161 (100V, 1A) and LM5017 (100V,

    Strangely, it behaves not like this with my devices. The TPS54060A mounted on my test devices regulates fine and also reached a very balanced +-17V at low load (60mA @ 24V). Things get bad when I for example have a 100mA load.

    Thanks & best regards,

    Stephan

  • Hi Stephan,

    Just ensure that CCM operation is always present (check the SW node waveform for this - it should be rectangular). Also, make sure that the magnetizing inductance of the coupled inductor is not coming in low (leading to higher pk-pk ripple current) - this increases the output current level that sets the DCM-to-CCM transition. Another option to reduce pk-pk ripple current is to increase Fsw.

    Regards,

    Tim

  • Hello Tim

    We tested with various devices switching the coupled inductor from 470uH to 680uH, and the resistors to R40=240K and R36=5K9. The result shows that this brings good result for most devices. However, some stll not reached an acceptable voltage level. On these devices, we switched the 680uH inductor to another one of same dimension, and then it works fine and 17V are also reached. As these inductors we use have 20% tolerance (Bourns SRF1280-681M), we expected that the bad parts show a low inductance, but we measured and strangely, it is not the case. All measured (bad) coupled inductors show rather good combined values near the set point. So maybe the single coils inside are unbalanced?  (Btw. do you know how we can measure only a single coil of the cooupled inductor?).

    I know that you told me to switch to another converter, because of the low load we have, but what we see is that at for example, only 60mA load, the devices all work fine and that the problem starts with higher load (for example 160mA).

    Thanks & best regards,

    Stephan

  • Hi Stephan,

    The SRF1280-681 is 680uH with the windings tied in parallel. This means the nominal magnetizing inductance is 1360uH (measured across one winding with the other winding open circuit). Bourns is somewhat non-standard by quoting the paralleled inductance in their part numbers.

    Also, what is the leakage inductance of this part? It's measured across one winding with the other winding short circuited. Leakage inductance affects the current waveforms, making them look more resonant in nature.

    Maybe you can send the switching waveforms just before and just after the issue happens so we can investigate further.

    Regards,

    Tim