fatal1ty x399 blue screen after RAM upgrade

Hiya!

CPU Type   AMD Ryzen Threadripper 1950X 16-Core
CPU Clock 3750 MHz [Turbo: 3750.3 MHz]
GPU Asus Nvidia 1060 6GB
RAM (32GB) 8GB DDR4 XMP PC4-24000 X 4
           G Skill Intl / F4-3000C15-8GVRB
           15-16-16-35 / 3002 MHz / 1.350V
(installed based on 4-DIMM config on manual. Single Channel Memory set in BIOS)

Running for a year without issues doing mostly 3D rendering in Maya, no gaming.
Last December, I added two bands: (DDR4_D1 and DDR4_B1 slot to stay away from CPU, IIRC)
16GB DDR4 XMP PC4-24000
G Skill Intl / F4-3000C16-16GVRB
16-18-18-38 / 3002 MHz / 1.350V

Daily usage like browsing is fine and not much issue. But recently I got Maya crashes, fatal error during rendering, or even blue when rendering some scenes that used to finish without issues.

I ran a memTest86 and got memory errors like below.
Test Start Time     2019-02-02 13:47:59
Elapsed Time     16:36:49
Memory Range Tested     0x0 - 1040000000 (66560MB)
CPU Selection Mode     Parallel (All CPUs)
ECC Polling     Enabled
# Tests Passed     33/48 (68%)
Lowest Error Address     0x444339EC0 (17475MB)
Highest Error Address     0xC3EB34400 (50155MB)
Bits in Error Mask     0000000000080008
Bits in Error     2
Max Contiguous Errors     1
Test     # Tests Passed     Errors
Test 0 [Address test, walking ones, 1 CPU]     4/4 (100%)     0
Test 1 [Address test, own address, 1 CPU]     4/4 (100%)     0
Test 2 [Address test, own address]     4/4 (100%)     0
Test 3 [Moving inversions, ones & zeroes]     4/4 (100%)     0
Test 4 [Moving inversions, 8-bit pattern]     4/4 (100%)     0
Test 5 [Moving inversions, random pattern]     4/4 (100%)     0
Test 6 [Block move, 64-byte blocks]     1/4 (25%)     52
Test 7 [Moving inversions, 32-bit pattern]     0/4 (0%)     1054
Test 8 [Random number sequence]     0/4 (0%)     416
Test 9 [Modulo 20, ones & zeros]     0/4 (0%)     62
Test 10 [Bit fade test, 2 patterns, 1 CPU]     4/4 (100%)     0
Test 13 [Hammer test]     4/4 (100%)     0
Last 10 Errors
2019-02-03 04:03:52 - [Data Error] Test: 9, CPU: 5, Address: BD28F0A40, Expected: 2B21A79A, Actual: 2B29A79A
2019-02-03 04:02:48 - [Data Error] Test: 9, CPU: 25, Address: BA36F2800, Expected: 02D1A71E, Actual: 02D9A71E
2019-02-03 04:02:48 - [Data Error] Test: 9, CPU: 26, Address: BA706C6C0, Expected: 02D1A71E, Actual: 02D1A716
2019-02-03 04:01:09 - [Data Error] Test: 9, CPU: 7, Address: A58088E80, Expected: EA57A044, Actual: EA57A04C
2019-02-03 03:58:40 - [Data Error] Test: 9, CPU: 23, Address: 999A47400, Expected: 59A4DC8A, Actual: 59A4DC82
2019-02-03 03:58:32 - [Data Error] Test: 9, CPU: 1, Address: 9429ECEC0, Expected: E1E29A13, Actual: E1EA9A13
2019-02-03 03:58:09 - [Data Error] Test: 9, CPU: 11, Address: 8E8A4A840, Expected: F40E34EE, Actual: F40634EE
2019-02-03 03:57:54 - [Data Error] Test: 9, CPU: 19, Address: 90882E2C0, Expected: F40E34EE, Actual: F40E34E6
2019-02-03 03:57:47 - [Data Error] Test: 9, CPU: 24, Address: 91F413AC0, Expected: 0BF1CB11, Actual: 0BF9CB11
2019-02-03 03:55:49 - [Data Error] Test: 9, CPU: 17, Address: 800EC3400, Expected: 89B92237, Actual: 89B12237

Not an expert on reading the memTest86 result. Does this mean I have a memory compatibility issue and that causes blue screen?
Is it ok to install memory of slightly different timing? (15-16-16-35 vs. 16-18-18-38)
Or it has to do with me mixing memories of two different sizes (8GBx4 + 16GBx2)?
Finally, is there a safe way to best use the 6-DIMM I currently have?

I don't think I can return the two additional 16GB memory now after two months and I am trying to get the best out of the $250 I spent.

Thanks!

You have to slacken the timings and increase DRAM + SOC voltage to overclock memory from the defaults.