From customer.
What I'm trying to do is to design a power supply that takes a +19V input and creates a +48V @ 300mA max isolated output. This is for a PoE PSE application.
The TPS55340 part looks ideal for this. I bought the TPS55340EVM-148 eval board that uses this part in an isolated flyback configuration. After verifying the operation of the eval board, I modified it to change it's output voltage from +5V to +48V. This involved changing many parts, but primarily the transformer and output components to work at the higher voltage. The unit works but is unstable; i.e. it goes into sub-harmonic oscillation and pulse skipping at light, nominal, and heavy loads. I've tried chaning the compensation but have not had any luck. I'm doing all this based on the app notes and other information I could find as well as trying to measure the network response of the circuit. However since I'm not an analog or power engineer I'm not sure I'm doing this correctly or even how to interpret the results. So that's why I need the help of an experience analog/power designer.
Note: I pointed out that the TPS55340 has a max Vout of 38V. Customer responded with below.
Yes you are correct on the max of 40V. But in a flyback configuration the max output voltage experienced by the IC (or switching FET) is the input voltage plus the reflected output voltage; the latter being the output voltage divided by the transformer turns ratio. In my application I'm using a 1:3.43 transformer which yields a voltage of 33V (19V + 48V/3.43) which is confirmed by lab measurement. (I should note that there is some overshoot due to leakage inducatance,etc. that hasn't been fully negated via a properly tuned snubber. I'm waiting to get the supply stable before tackling that.) Additionally, simulations show that the max current in the transformer primary is around 2A with a nominal load, so in theory the TPS55340 should be well within its limits for both voltage and current.
The reason I gravitated to the TPS55340 was its small solution size. I didn't mention it in the first email but board space is a critical factor in the design and effeciency is probably equally important. Both for the same reason: the design must fit within an existing enclosure that was not originally intended to house a PSE supply. Therefore space and internal heat dissipation both need to be kept as small as possible.
So in conclusion I'm still wanting to see if you can help stabalize the control loop of the TPS55340 in this flyback configuration. Again it is the EVM-148 Eval design modified for 48V out by changing T1, D1, C8, R14, and the compensation caps C15, C13, and C17. I've attached the datasheet of the transformer I'm using, PA1138NL, since it probably is the single biggest change from the eval board. The output diode, D1, was changed to a MURS120 and the output cap, C8, is a 47uF, 100V Electrolytic with ESR of 0.32 ohm; C5 and C6 were removed. The feedback resistor R14 is now 800 ohms (R8 didn't change). With these mods the output was ~48V but unstable - the IC cycled between CCM, DCM, pulse skipping and sub-harmonic oscillation
