Stress tests are designed to push your cpu as hard as it can be pushed, and are generally used by people wanting to check CPU reliability when overclocking to GHz/MHz speeds they've decided to use. If you use your cpu at stock clocks and voltages there is no sense in running a stress test--as it will only tell you what you already know--that your cpu will run when stressed without errors and get hotter as a result. (Most games, for instance, never stress a cpu like Prime '95 or similar programs will.) That's true for every CPU made. Again, if you are not overclocking or overvolting there's no point in running a maximum stress test--at least, I wouldn't bother...;)
I'm on an x570 motherboard, and running CPU-Z, I see the temps fluctuate constantly, dropping down to a low of .2v--if not lower. So, that is what you should be seeing. Try this: go to your desktop and run CPU-Z and then let the system sit there doing absolutely nothing--take a 2-3 minute break away from your chair just to make sure--then come back and watch the CPU-Z voltage readings carefully. The cpu is now idling, or rather should be idling, so you should see significant drops in voltage below 1v that fluctuate between those lows and the highs that you have already seen. This constant flux is normal and is what you should be seeing. Also, for some reason I cannot fathom, the latest Ryzen Master version reports *average* voltages across the cpu sensor net and does *not* support showing these dips in voltage that I see, and you should see, when the CPU is idling per CPU-Z. Robert has officially endorsed CPU-Z, and I think it would behoove AMD to raise the Ryzen Master voltage sensitivity to at least match that of CPU-Z.
OK, what if you are not seeing these dips in voltage (to as low as .2v or lower) in CPU-Z when the CPU is idling? (Assuming you are running Win10 and the latest chipset-driver version of the Balanced plan, with the CPU set to 99% minimum and 100% maximum, active cooling--which you have said you are using.) If you still cannot see voltage dips significantly below 1.4v in CPU-Z, then the only explanation I can think of to explain that is that the bios supplied by the manufacturer of your motherboard--evidently MSI--is not handling voltage to the Ryzen 5 3600 correctly! If so, I would be on the case of MSI to get out a bios that does, like one built on AGESA 1003ABB, for instance. Remember that the B450 motherboards were not designed with the physical component environment that Ryzen 3k cpus require. Indeed, the B4xx class of motherboards were designed a level below both x370 and x470 motherboards in terms of physical features & component quality--as such, they cost less than x3/470 boards, which also were not designed with Ryzen 3k cpu support as a design target.
In your shoes, so long as you elect to run a Ryzen3k series cpu in a B4xx-class motherboard, I think buying a new cooler is a wise move. It certainly can't hurt. MSI and every motherboard vendor are charging significantly more for x570 motherboards than for B4xx-motherboards, which were "budget" class when they were new, and strictly designed for Ryzen, and Ryzen Plus CPUs (Ryzen Plus CPUs were carbon copies of Ryzen 1, just manufactured at 12nm instead of 14nm.) If they maintain that apart from PCIe4 support there's "no difference" between a B4xx mboard and an x570 mboard, then we might ask them why in the devil they are charging so much for x570 mboards, eh?
Good luck--try the CPU idle test with CPU-Z--maybe your B4xx mboard is supporting the voltages correctly, after all!
