Tinypol

48-V Input Affordable 1.2-A Output Converter / SY8502

Overview

This design is based around the Silergy SY8502FCC converter.
IC Datasheet. Board Schematic.
For layout, see the .kicad_pcb file in the repository.

Use

Connect the input side to the power source, and the output side to the load. The converter powers on automatically above approx. 6.2V input. Output is programmed to 5.2 volt, can be changed by replacing R3 and/or R4. Reference voltage is 1.2 V. Supported input voltage is around 56-V, above which the D1 TVS diode will start to clamp. You can remove this to increase maximum input voltage to 85V abs max, but then you should avoid hot-plugging the board.

There are two ways to shut down the converter while input power is present.

• Either apply a logic high to input labled OFF (J4) don't exceed +/- 12V on this pin, or
• pull EN input low. The EN pin is pulled up on-board to VIN with 100kOhm, therefore the circuit you use to discharge EN must withstand the maximum VIN you plan to apply.

Alternatively, you can disconnect the input.

Properties

Planned Future Changes

• Add a capacitor at SYE pin to de-glitch.
• Add an output indicator LED.

Test report

The IC looks quite OK! Output has little noise (thanks to the plentiful output MLCCs) but high ripple because of low switching frequency and hysteretic operation. Rectifier MOSFET control looks good. Short circuit test looks good. Startup after short is good.

The SYE (enable) pin does not seem to be properly deglitched internally. I can make it behave erronously by feeding noise into EN. The on-board 100k pullup is strong enough to counteract spurious noise, but you can drive it externally to make it do weird things.

I can tickle the IC to make the output oscillate, but it settles rather quickly. Would not use it to power sensitive circuits directly, without a second regulator.

Driving it close to 2 A output makes it run very hot. The thermal pad below the IC seems to be well soldered and the board is sinking the heat. Inductor heats up quite a bit; a better one would probably help increase efficiency, but it is okay for its typical use case, and you would not get much more current out of this anyway. The board has nice thermal mass!

Tried adding 2200uF electrolytic cap to the output, this did not reduce ripple (because it is determined by the hysteresis of the comparator) but pushed the switching frequency down to the audible range with low load.

One weirdness I observed: when running it with a hard short at its output the output current is initially clamped at 2A, at which the converter heates up, which changes the overcurrent comparator's threshold. After thermal shutdown kicked in, I let the IC cool down, and repeated the test. Now the output current into a short was only 1A. This did not change after power cycling the IC, and seems to be permanent. I will repeat this test on different boards too, but I want to set up a test script first so I can compare chips before-after, and that needs more work on our lab instrumentation.

Either way, this IC looks very usable, while being much cheaper than any of its alternatives.

Planned Future Tests:

• Characterize:
  • Iq over Vin with EN=low and EN=high
  • Efficiency curve (needs work on lab equipment)
• Increasing Cff to optimize load step behavior
• Adding C to SYE to test glitch improvement