I recently obtained the STEVAL-TSP004V2 eval board, and tested it with Tripp-Lite NPOE-30W-1G injector. As supplied the eval board worked fine.
I then modified the eval board as I require a 24V output, so :
- removed T5 and fitted a Coilcraft POE300F-24L on the T6 footprint
- uprated all the output caps
- R93 changed from 1K to 4K7
- R94 changed to 330K
- R111 to 18K, R112 removed
- LED bias resistors swapped from 3K3 to 22K.
- R104 swapped to 150K.
The above changes excluding the transformer swap result in an output of 23V, but the transformer does warm up a tad with no load - no unexpected as it is designed for 5V. Swapping to the POE300F though results in a POE detection error.
The eval board datasheet, nor PM8803 datasheet, really cover these bias requirements but the above have been worked out by assessing the current circuit for 5V against 24V, comparing with the 3V3 and 12V eval boards (the latter using a PM8800). However I'm not getting any joy.
I don't think the Q12 drive circuit, not transformer auxillary coil circuit should need to change - but are there any ideas or pointers as to where I'm going wrong please?
Edited to add that I have found schematics for an EVALPM8003-FLY which is in flyback configuration and has allowances for the POE-300F transformer. This has a diode feed to the transformer rather than a pair of resistors, and a simpler (ie direct resistor) for the Rcls calibration. I have changed these across, with no difference in behaviour (and rechecked with original transformer, which it still works with.
It does try, and very briefly flicks the output on, but the PoE injector flashes blue (error).
