I think the S21B uses a ∅20mm driver, but since the S21B host advertisement lacks this information the only thing worth saying is the following: Simon, pay attention and good care of the information provided with your products. Your product advertisements must provide all customer required information.
Now that I'm here, I will also seize the opportunity to say that as a rule anti-reflective coated lenses damage the tint and CRI of flashlights. This of course is a way to tell Simon that some customers would prefer to order flashlights with good quality non-coated lenses, namely those who care for tint and CRI and less about throw.
Single layer AR-coated lenses increase output very little and the price is the obvious “beam greenery”. The solution is multi-layer AR coatings (5+ layers), or no coating.
The information on the driver diameter has been kind of inconsistent, as you said the host listing is sparse, and the only information I’ve been able to find was this comment saying it was 22mm and this comment asking about using a 20mm to 17mm adapter to put in a different driver.
Now with a bit of detective work I’m sure the S21B uses 22mm drivers not 20mm, check these images I pulled from AliExpress:
Customer picture of S21B
22mm SST40 driver
20mm SST40
I still want Simon to confirm compatibility with S21B and the new 22mm 8A driver, though, since for one thing the new driver uses a pin contact instead of a spring and there might be other things I haven’t considered.
Reason for the pin contact in the 8A “buck” (hope it were boost-buck) driver is conductivity, as even high conductivity springs could have problems at 8A or close (with just 10mΩ of resistance a spring dissipates 0.64W as heat when 8A crosses it). Simon and/or the driver manufacturer played it safe this way, but it has drawbacks, namely for people who really need damping or shock absorbing contacts (i.e. hunters for example).
The solution, of course, are really high conductivity springs. A bypassed spring can do, but a durable spring bypass requires careful work, giving the wire a coiled shape. Problem is the contact surface under the pin of the driver, it just has the size of the pin contact and is too small, unnecessarily forcing spring modders to remove the solder mask around it.
Let me also say that, in all honesty, I think 8A is probably a bit over the top. We know the CSLPM1.TG peaked at 8A or just a hair above, and imho it's over the top for it. The wrong thing about this is we don't have any CULPM1.TG tests. So actually we still don't know how much higher is the 4040 CULPM1.TG going to peak over the 3030 CSLPM1.TG. And honestly, it's the same led, only the footprint changes. Nothing else to say, except that being able to configure a driver's output current in firmware would be a godsend.
P.S.: low voltage cut-off is overall set too high in drivers. Bear in mind that drivers don't see the actual battery voltage, they see it after it goes through springs, contacts and switch. All of this causes voltage drops, the higher the current the larger the drops; therefore, the actual cut-off ends up being at a higher voltage, namely if running in high current modes. In essence, I think driver cut-off voltages should at least drop down to 2.7 or 2.5V. After all, I've yet to see a battery complain about a low cut-off, namely when datasheets specify 2.5V (and I've seen lower).
Okay, my new finding is the M21C-U is doing fine, in fact very well, after the light has warmed up. It step down from high(100) to moon about 3.3V. This time I’m using a Vapcell T50, previously on P42A and 40T. Taking into account all factors and voltage sag. It’s about right that it step down about 3.3V on high(100) to moon.
I will test again with the P42A and 40T. Will also test with Vapcell G50 and Vapcell Red 4500mah(20A) and update here again.
Even before this finding. I love the M21C-U. It’s my most carried light for all my night walks.