To my knowledge, there's still some "hangers on" applications on the hardware-side of things, but not in satellites. The only place I know these to be used is IC-card billing systems, and this is because of all of the legacy hardware. My IC-card that I used in the Tokyo subway in 1999 still works.
Satellites often have a strongly encrypted control channel, AES for example, but the data channel has something else that's less secure. This is a function of transfer windows and memory. Due to time delay, we have 64-Mbit frames for data, so you end up with weak stream cipher and you just change the key often via the control channel. It's been 10 years since I've looked at this, but it hadn't changed much in the 10 before it.
I believe that 3DES could still used in some power constrained, silicon implementations. The cost of make a 3DES IC is on the order of 260k~ transistors (pulling from memory here, so corrections are welcome), which is about 65k~ logic gates. It's easy to get enough power to run this off NFC power from magnetic sources. My passively-powered RFID tags only have 260k~ transistors so generally cryptographic cores are really expensive from a transistor point of view. This is particularly true with a Flip-Flop that stores a bit, which takes 78-transistors (not counting clock drivers), which gives you 20k~ transistors for AES128 just to hold state, not event to move the data into and out of the key and data registers.
Another note why an engineer (not a cryptographer) would want to use DES/SIMON or something else that is shorter for the keys is that when you are "loading" data, you aren't collecting power in passively powered systems. The area cost of an analog front-end is a lower for fewer bits. I always thought that SIMON48/96 would be a better choice for passively-powered near field (NFC) or far field (RFID) than the other ciphers for this reason.