reference design or starting from scratch

[rwDDD]

hi all
I’ve been thinking about how different everyone’s workflow is. When you start a new project (work or DIY), do you actually build the schematic from the ground up, or just 'Frankenstein' a bunch of reference designs together?
I used to be a hardware dev at a drone countermeasure startup. Honestly? From the day I joined until I left, I barely touched a blank schematic. The original design was total garbage .super inefficient and buggy,but nobody had the guts to start over.
We just kept patching the same mess because 'it works (barely) and we have a deadline.'
Is this just the industry standard now?
I’m curious—how many of you guys actually design from first principles anymore

[Nominal Animal]

When you start a new project (work or DIY), do you actually build the schematic from the ground up, or just 'Frankenstein' a bunch of reference designs together?

I only do hobbyist schematics, but the way I solved this during my 30+ years of paid software development in a dozen different programming languages might be useful here too (because I find myself doing the same with my hobbyist schematics also).

I keep a library of commented, executable pluripotent algorithm implementations, that I adapt instead of copy-paste.  I trim off all the unneeded features, and optimize the rest, and only then integrate to the overall design.  The schematic/PCB equivalent would be a subsystem, perhaps a DC-DC power supply, a transceiver for some high-speed bus, the verified-working way to connect some specific MCU/SoC to DRAM/SDRAM, and so on.  I always have a README file describing the design choices and limitations, and since many of my algorithms involve math, a PDF with its source deriving the math expressions used in the code.  I believe that for schematics, you'd very likely need PCB layout snippets, perhaps for different numbers of layers too.

This is neither "build the schematic from the ground up", nor "Frankenstein a bunch of reference designs together".  It is something very clearly in between; more like proper integration of well-understood parts than either one.  But note that the adaptation part is still quite a bit of work, and is absolutely crucial so that you don't fall into the Frankenstein copy-paste hell.  Fortunately, my own experience shows that it is possible to learn to do; you don't need any specific innate ability to be able to do that.

What I do not know is whether this can be done in the short timespans allowed nowadays for such work.

[ftg]

At work we have a catalog of existing, proven design blocks, so for some DC-DC I can sometimes find something that is known to work and the parts can be sourced.
For hobby use I have my own library of building blocks and parts.
But unless I'm making a derivate of some existing project of mine, it always starts on a blank schematic sheet.

[nctnico]

For a more complex designs (think SoC + DDR + power management) I like to use the schematics for a reference design as a starter. Sometimes just to copy the symbols and footprints if they are any good. A demo board rarely suits my needs though so I quickly move to doing a custom design. I'm currently in the process of evaluating a new-to-me FPGA and I'm creating a custom designed dev / test board. Typically doing this reveals some areas where more attention is needed so in the long run it prevents doing a respin of the final product.