Other Parts Discussed in Thread: TPS54531
Hello,
I have the task to design a non isolated power supply that converts 12-48V DC to 6-48V DC, because of my good experience with the TPS54531 I have chosen the TPS54560 to do so.
Requirements:
Input: from 12 to 48V DC
Output: adjustable with potentiometer from 6 to as close as possible to the input Voltage.
Output current, fluctuating from 0-4,8A (the output is switched into a load with around 50ns rise time and the same fall time, worst case 0 to 4,8A full current is switched with this rise time)
Output voltage should fluctuating as little as possible, but the design needs to be very small, so big output caps are not possible, I can take 5% voltage fluctuations.
Issues:
I can´t get it to work properly, even if the load is only resistive and not switched in any way or form.
The TPS does work at no load, the output voltage is stable and variable with the potentiometer. If I connect a resistive load e.g 10R the TPS works if the output voltage is set to below approximately 12V (varies with varied input voltages) above that the TPS enters what looks to be over voltage/current protection, the waveform on the SW pin does not switch with a "constant" duty cycle and skips pulses, therefore the Output voltage varies a lot (around 3v pp). If if I try to raise the output voltage further the dutycyle at the SW pin gets smaller (because of more skipped pulses) and therefor the output voltage drops and varies more.
Additional I have problems with the TPS frying if I connect 48V directly/hot onto the board, I know this happens do to transients and I think I can solve that with a additional tvs at the power input.
Assumption:
I have made a huge mistake at the layout but I don't know where, except for the bad placement of the FB line.
I have problems with noise pickup at the FB line, however I tried to probe that and the noise at the FB line is lower than my RIGOL DS1054 can measure.
Things tried to solve the problems:
Disconnecting the FB line on the PCB and soldering a wire to it thats attached to a potentiometer thats directly soldered on top of the ceramic capacitor -> no real change (thats baffling me)
Increasing the input capacitance -> no change
Increasing the output capacitance -> no change
Changing the switching frequency -> no change/ above around 600khz the TPS/diode gets hot because of the increased switching losses (I assume?)
Soldering a capacitor from FB to GND directly at the TPS -> worse
Lowering the FB resistive divider in value to reduce the picked up noise -> slightly better but I can only get so far with it because I need SMD potentiometer and these have a not so high power rating
Different compensation components -> changes behavior slightly and does improve it however, I do not know what components should be used because of the varing input/output Voltages. I have tried various suggested components from the Webbench simulation software, however all works worse than a single 50nf to ground.
Components used/chosen:
Input Caps are: 2x 10uf/50V Ceramic directly at the TPS and 6x 10u/50V at other places on the board
Output Caps are: 1x 100uf/50V polymer ESR@100khz: 25mOhm rated Ripple: 2A (HHXB500ARA101MJA0G) in parallel with 2x 10u50V Ceramics
Inductor: 22uH Ir: 5.7A Isat: 10.2A Rdc:37mOhm (Würth: 744373965220)
Diode: STPS5L60
Timing Resistor: 220k -> around 450khz switching frequency
Schematic:
NOTE: The same Board is shall be used with the TPS54531 therefore the timing resistor is drawn as a capacitor, on the actual board a resistor is populated. I know I screwed up the compensation network, however I soldered the extra components ontop of the layouted capacitor footprint. I know its not optimal to lay the FB line directly under the SW node.
Layout:
I have made a test Layout you can look at it below. Layer order as following:
Top: Red
2. Layer: Purple
3. Layer: Orange
4. Layer: Green
5. Layer: Yellow
Bottom: Blue
All Layers are filled with ground.
I have tried to minimize loop area I don't know how well I managed to do so..
Needed:
Help to point me in the direction I should be looking, what im missing?
Maybe the only problem is that I don't know/understand how to design the compensation components for different Input/output Voltages.
Is my Layout really this bad? (FB line under SW node / diode / inductor placement) what can I do to improve it further?
Thank you for your time to read it.I really appreciate it. If some additional information is needed don't hesitate to ask for it, I can provide scope screenshots as well.
Please excuse the crude englisch, im not a native speaker.