Eldritch Tempest says it does +5[W], +10 Critical Threat Range, +3 Critical Multiplier.
It is currently doing +10[W], +20 Critical Threat Range, +6 Critical Multiplier.
.
I had been struck for awhile about how well concentrated the values for Eldritch Tempest were, which suggested its multiplier was higher than documented - since critical hits aren't actually "multiplied" but rolled again, rolling any given die many more times will end up bunching observed values together. So I took a character with Eldritch Tempest, IC: Slashing, Improved Knight's Transformation, and no other modifiers to critical threat or multiplier and tested in the Waterworks with five weapons:
a Barovian falchion - 1.5[2d4] + 36 and 15-20 x3,
a +1 quarterstaff of everbright - [1d6] + 32 and 20 x3,
a +1 heavy pick - [1d6] + 21 and 20 x5,
a greatsword - [2d6] + 31 and 17-20 x3, and
a +1 club - [1d6] + 21 and 20 x3.
The character also has 2 melee power with two handed weapons.
With at least 11 hits each, I observed the following averages on critical hits compared to the predicted:
Code:
actual pred weapon
861 419 barovian fal
644 324 +1 qstaff of everbright
657 336 +1 heavy pick
988 447 greatsword
539 252 +1 club
I also observed only critical hits before confirmation. Even the quarterstaff with a cool range of 20 managed to threaten on a roll of 3(!!!). I didn't happen to see any rolls of 2 but it beggars credulity they wouldn't crit too. Notably, I did observe six cases where rolls of 1 on the critical confirmation roll with various weapons resulted in ordinary hits. I did not include those in this analysis.
O.K., so the bonus threat is definitely not functioning as documented since (before confirmation which can auto-fail on 1) it's effectively auto-crit, but things are off in damage too. No matter what the +[W] is, the ratio of heavy pick to club will be governed entirely by multiplier because they have exactly the same base damage and bonus. If we assume that it is an additive bonus to multiplier of the form +X, we only need to find the closest approximation for...
(5 + X) / (3 + X) = 657 / 539 = about 1.22
...and that X happens to be 6, because 11 / 9 = 1.222 = about 1.22
Great! We can also look at the ratio of greatsword to quarterstaff since those have the same multiplier and different [W] to see what kind of +[w} we're getting. This time...
(7 + 31 + 7 * Y) / (3.5 + 32 + 3.5 * Y) = 988 / 644 = about 1.53
...and that Y happens to be 10, because 108 / 70.5 = about 1.53
.
And just to make sure, let's plug in +10[W] +6 Critical Multiplier and see what values our new model expects:
Code:
actual pred weapon
861 858 barovian fal
644 635 +1 qstaff of everbright
657 655 +1 heavy pick
988 991 greatsword
539 536 +1 club
Not bad at all! 
.
I have no idea if this is intended. It's enormously suspicious that each value is exactly double the documented value, of course! For what it's worth I think it would be good to have it this way, that neutralizes the advantage of broader crit range since Improved Knight's Transformation so heavily favors weapons with it, and this way it's not a no-brainer to get the best crit range weapon as an eldritch knight. It's probably still the best choice because of how long the cooldown is, but still.
I've updated the wiki to reflect this research. I don't have a sorcerer EK so I can't test with them. It seems really unlikely it would be any different, but I have included that caveat in the wiki.
.
.
Appendix of the damage by hit:
Code:
875 869 835 881 860 851 865 891 843 840 865
628 644 637 654 632 632 657 643 651 681 650 656 640 631 639 631
673 652 638 646 669 664 663 675 667 645 634
1031 953 981 989 1026 988 956 972 978 976 984 1027 1011 966
535 526 520 544 558 544 547 554 529 554 513