Shelford Silk-Modes Don’t Stay

RND probably using a form of volatile memory for digital controlled analog switching scheme - power goes away and settings go back to default on next power-up. It saves some complexity, some money (and less parts to break usually means less time in repair shop) but I can see this being the drawback. 

Added later: a long while back I posted a plot with a transformer core B-H curve with hysteresis showing a possible way the "silk" might actually be implemented (basically biasing the gapped core more or less towards a certain non-linear portion of the curve) using a DC current resulting in "pleasing" harmonics.

The  point above might actually also be a reliability issue with the transformer in that the current levels are only controlled after proper power-up is over and not during. It's likely easier to just have the user reprogram the setting manually than to deal with it using a collection of extra devices in a dedicated circuit. Too much current could potentially alter the desired magnetic behavior of the transformer core. Of course, this is just speculation on my part.

Don't get me started, I had a Shelford and as much as liked it some things drove me nuts. Add power switches on the back to this list..

I think that’s how the silk circuits work in all their gear. Definitely in my 542’s.

You just need to get in the habit of noting your settings for recall. I always have a notebook handy for these sorts of things.