I've found the light shape from reds to be slower and curved, and super bright green and blue the fastest and squarest.
I won't be trying to get the fastest speeds out of the green and blues because I've got a 100W white COB waiting for >>100W testing!
I've tried also with some green led and found them slower. I used an IR led and it was the fastest.
I'm trying to build this little probe because i need it to measure a UV source (20W 365nm) that i've already built and rise/fall time are really important.
I suppose white LED are really bad (microseconds?) at turning on/off because of phosphorescence (turn off especially).
I've tested a variety of LEDs and had mixed results.
White is interesting. The initial turn off is very fast. There is some afterglow, but it's very weak, several orders of magnitude dimmer than when driven at the rated current, so it's not a problem, Even the warm white Cree CMA3090 I tested was fast enough for use in a 1µs strobe application.
Some phosphor converted LEDs are slow. The amber one I tested was extremely slow. I can't remember the exact part number. It was a Lumiled/LUXEON brand.
https://www.lumileds.com/uploads/161/DS58-pdfhttps://www.luxeonstar.com/assets/downloads/ds62.pdfHow bright is the source shining on the photodiode? If it's bright, you need a large revere bias voltage to prevent saturation, as well as speed it up. At very high intensities, no amplification is required. As you can see from the above schematic, I connected the photodiode directly to the 50Ohm input of an oscilloscope and applied a bias voltage to it.