> Can I use a pot as a voltage divider
Insufficient data to be sure. Even AFTER I found and studied Primotec's data-sheet.
I would "assume" that the speed control input is HIGH impedance. I would like to know how high. I might want to wiggle it with something weak. Nano-power CPU, sun sensor, who knows?
I'd go with a 10K pot and see what happens.
The total 5V load is teeny. Perhaps 0.5mA for the pot, up-to 0.5mA for the opto-coupler (which seems to do nothing but invert tach pulses?). For constant speed you need constant voltage, true. But at the 1mA level, you could take 3K 0.5W from the 24V supply to a 5V Zener, save a transformer.
What's with this 24V supply? 50V raw, 50 ohm 50W? OK, 50V through 50 ohms means you won't get over 1 Amp. Then a regulator, which will shut-down near 1A. And the motor (which could want 2A) is self-protected, doesn't nominally need protection. BTW, 50V raw is well above the rated input to 7824.
470uFd is light. In audio, for 24V 1A I would think 1,000uFd min, maybe 4,700uFd. If you go much lower, the cap is not physically large enough for the large ripple current.
The motor will eat 14V to 30V. The obvious supply is one 17V 3A AC winding to FWB to say 2,200uFd 35V. Or 17-0-17V 1.5A to 2-diode rectifier, and on with 2,200uFd. This gives 26V-20V 1A DC for the motor. Tap 3K 0.5W to 5V Zener for your teeny 5V needs.
Direction wire not shown.
The FG wire is rated 4K output. It isn't clear that it will flash an LED. Ass-uming the 4K is from the 24V, that's 5mA or 6mA, OK, LED will blink. But it is 36(??) pulses per revolution, which is perhaps awkward for a "winder" where you count full turns. I also can not see why you would invert (LED+opto), since high and low have no specific meaning, only the rate/count. Bring the FG wire out through another 5K to a 5V Zener (assuming your counter will take 5V pulses). _OR_ ferget the FG pulse, put LED and photoresistor (or two strips of copper clapping) ON your winding spindle, count actual turns not motor*36.