DFN: heel/toe half as big as recommended?

[hephaisto]

librepcb-parts-generator/dfn_configs.py

Lines 33 to 44 in 76759e0

	# Toe and heel length as a function of pitch
	# According to IPC-7351C, see slide 26 of
	# http://ocipcdc.org/archive/What_is_New_in_IPC-7351C_03_11_2015.pdf
	LEAD_TOE_HEEL = {
	1.00: 0.35,
	0.95: 0.35, # not specified in standard
	0.8: 0.33,
	0.65: 0.31,
	0.50: 0.29,
	0.40: 0.27,
	0.35: 0.25,
	}

The linked document (only available via wayback machine or similar) shows the values given here as the value for both toe or heel (separately).
The usage

librepcb-parts-generator/generate_dfn.py

Line 219 in 76759e0

pad_length = config.lead_length + config.toe_heel + pad_extension

uses the value as total "oversize".

However, the length would be toe + original + heel = original + 2*toe_heel. I think the code generates only half of the "oversize" recommended in the presentation.

[ubruhin]

Hmm good point. I didn't check it in details, but just by opening an existing DFN package in LibrePCB, it is clear that this generator is quite outdated anyway since there are no footprints for the 3 density levels :-/ Also it looks strange to me that the pads don't have any heel at all, they only seem to have a toe.

Also note that there now exist footprint guidelines from Library Expert, which luckily are now public (in contrast to IPC7351 which was a proprietary, paid standard). For the newest generators we are now following that standard (which can be considered as the successor of IPC7351 IMO). For verification, I used the Footprint Expert Calculator, to be sure the dimensions of our generated footprints are identical to what their tool generates.

So in the end, I think we will have to rework the DFN generator to follow those guidelines and generating the different density levels accordingly. I also see that the silkscreen drawings need some rework too since they don't follow our own guidelines.