It's still not worth making a buck switcher for 12v to 12v system.
If your 12v panel makes an open circuit voltage of 21v it's probably a crystalline type that produces peak power at about 17v.
On a 12v to 12v system you need to protect the panel with a reverse protect diode, even a decent schottky will drop about 0.4v and you will lose about 0.3v in wiring and terminations etc.
So with your panel at peak power output your SMPS will receive 16.3v input. Now assume your battery is at a nominal 13.8v operating voltage. You have a SMPS with 16.3v input and 13.8v output. Now if you can make a SMPS as good as the commercial one I designed you might get 90% efficiency at 16.3v->13.8v.
So that's (16.3 * 0.90) / 13.8 = current gain of 6.3%.
So you think "that's ok, it's still worth making the MPPT to get 6.3% more current into the battery".
But it's rarely worth it, because if the battery is close to full charge (over 14v) it won't produce much gain at all. Also, the max panel power is rated at say 20W (17v 1.17A) BUT if you connect to a lower load voltage the panel power is on a curve and it will make more current into a 13.8v load than it will into a 17v load.
See this curve, it's pretty typical;
**broken link removed**
So if you directly connect the panel to the 12v battery you will get about 6% more current into the battery because it is driving a lower load impedance, which is the same power output as you got from the MPPT.
I've already said it above, it's not worth doing on a 12v->12v system which is why there are practically no 12v MPPT devices on the market. And you completely ignored my post, you are asking for people who have experience to help and then ignoring them...