choose nmos width of * Latches are often thought of a "latch" a signal, makes sense! In this case it's often playing the role of a comparator, or quantizer, or sampler. Another kind of whole diffent view, and often for a different purpose, is for "state". In analog, capacitors and inductors hold state. The state can change up and down, usually in a continuous fashion, and the state often controls something - a frequency, a phase, a duty cycle, an amplitude, an offset, an equalizer coefficient. We want an element that hold state, it doesn't really sample something. It can be filled up or down, like a cap fill with charge, or an inductor with current, or a water tank filling with water, or a fly wheel momentum. We need the state to be able to be moved up and down by some input, and then we need to be able to use the state to control something. For digital, we have non-ideal issues of glitches, linearity, decodeing, and most of all metastability and illegal states!! * I wish I had invented it, but I can take some credit for spuring it on. I was working on VCO's and phase selection circuits (and the issues of glitches, and boundaries, ...). I came up with the ultimate state code that would be the most optimal, but now idea of a good way to make it. Tim, just needs a challenge and to be motivated. * the holy grail is to design state machines with no resets, no asynchronous, no initialization, just plain bulletproof * The basic mobius structure will be explained with a series of highlighted pictures, k1_k0 four possible states, for each of these two muxes are latched, so there are two possible output states for each input state clock forward 8 times, with 8 pictures, each time adding the lastest state Do the same thing with a divide by 2, divide by 4 * Use inkscape starting point in Dropbox/howto/inkscape startInk2018.svg * To do animations, let's work locally in Desktop MWork mobAna is 8" x 6" (100 units / in.) Export using * convert -delay 10 -loop 0 mobAna*.png junk.gif