I want to implement a 12v PoE+ PD power supply, and have settled on the LTC4269 for use in this application. Unfortunately I'm having problems getting my head around everything that is going on.
A demonstration circuit is available which provides 5v output (DC1351B) and I have one of these boards. I took some measurements of the startup and ripple at Vout in it's off-the-shelf state at various resistive loads (no load, 15Ω, 7.5Ω, 5Ω, 2.5Ω and 1.25Ω) - needless to say, they all look great.
Startup and Ripple at 1.25Ω / ~4A / ~20W.
I then modified it for 12v output (more details at the end), after which it failed to startup, or would fall over under both no load and 30Ω (~400mA / ~4.8W). It also made nasty crackling / clicking sounds...
This screenshot shows a failed start followed by a successful start. Note that this does not show the switching frequency, but a much lower frequency component.
I have a feeling that this is because of flux buildup in the transformer, but I'm really not sure, and this is what I'd like to understand with your help.
Unfortunately I didn't get any measurements of the transformer coil's switched side before my modifications, so I don't have a known "good" trace of my own, but a series from afterwards are below. I have ordered another board to keep as a control so that I can compare properly.
- Yellow is Vout
- 3v/div, aiming for 12v
- Blue is the switched side of the transformer's primary coil (pin 2/3 of T2)
- 50v/div, the PoE supply is nominally ~54v.
As time progresses, the waveform becomes increasingly distorted. Afterwards (in the last frame, when the switching stops) there is significant oscillation on the transformer.
A full video of this trace is available here (YouTube), and a wide screenshot of everything is here - there are significant transients later in the trace...
I believe that the yellow trace below (from the DC1351B quick start guide) is equivalent to my blue above. This implies that ~80v is not unreasonable, and more when the output current step occurs - unfortunately it doesn't show the detail I'd like.
Am I correct in thinking that this is flux buildup? If so, what should be done to avoid it? The duty cycle already appears to be quite low.
I'm still trying to fully understand the Forward Converter topology and the constraints that are in place. I've watched videos from The University of Colorado (e.g: here).
My modifications to the DC1351B demo circuit are below:
- Replace T2 with PA2649NL
- originally PA2431NL
- from the EP13 series - datasheet
- note reduced primary inductance from ~429.3µH to ~100.5µH (I suspect this is very relevant)
- Replace R29 with 180kΩ
- target switching frequency of 200kHz
- Replace R20 / R33 potential divider with 13.3kΩ / 1.54kΩ
- D21 has a 1.24v reference
- ~11.94v Vout
- Replace R19 with 5.6kΩ - for 12v operation
- Replace R8 with 2kΩ - for 12v operation
- Replace C9 with 16v rated capacitor - for 12v operation
- 220µF low-ESR (0.1Ω) Tantalum
The full schematic (with annotated modifications and probe points marked) is below:
The transformer windings have gone from PA2431NL: (switched input on pins 1 & 2)
to PA2649NL:
Please note, this is largely a learning exercise, and I've not yet been in contact with LT.