CABD
Most force (inversely proportional to speed) for a given motor torque and step rate: CADB (the last two are a toss-up, effective diameter looks to be the same).
Best grip depends on the idler tension (B will grip well if you smash it hard enough to cause other problems, especially with retraction), but I’d say CDAB. D should beat C due to greater contact area, but the teeth on D look pretty crappy.
D has more grip/less torque . C has more torque / less grip than D. B reminds me Qu-bd design - no grip when used with original designe with set screw .
CADB
From best to worst I would guess C, A, D, B.
I’ve only run D but got a C and B on my desk.
By the looks of them I would not expect B to perform at the top of the pile under any conditions except maybe very soft filament.
Hmmm. If diameter is the dominant factor, then C is the clear winner. If any of them slip before the stepper stalls, then diameter may not be dominant. I’m not sure which way to guess. I’m very curious to see the results.
B is the winner. Large sharp grip surface with a high diameter.
Built my own spincaster, and really, something like this would be super easy to cast in Zamak, if a good design were worked out.
I know people are custom cutting hobbed gears all the time. But really, could churn these out easily if a good design were worked out. I’d just need one for molding. Have to tap the hole for the set screw later, but Zamak should work fine.
So, here is the result from the benchmarking test.
Thank you all for participating.
Ranking order from best to worse for pushing power:
C,A,D,B
Well done to +Jesper Lindeberg, +Andreas Thorn.
C, MK8 Drive Gear 4.2kg +
A, Plain Insert Gear 3.6kg +
D, MK7 Drive Gear 2.7kg +
B, Raptor Drive Gear 1.8kg +
C and D is the best choice for normal use and ease of set up, A and B are most difficult to set up for normal use and should be avoided. The performance of D will depend on how well tuned your stepper drivers are and could do better than 2.7kg for direct drive. Personally, I think you need at least 3.5kg of force for good reliability for most extrusion conditions.
Without a force sensor it would be difficult to get the best out of C,A and B because of the narrow margin between filament slipping and stepper motor stalling. Filament slipping is silent and difficult to detect but I believe C can be set up reliably to out perform D without a force sensor.
Full report will be published on the blog soon.
Thanks for the test. I have C and D myself. Really like C. I use it with direct drive with my own extruder design and a j-head. 1.75mm filament. Runs great 
Ultimaker was (at the time) running 3mm with gearing (probably in the neighborhood of 3:1) with a diameter comparable to gear C, so 10Kg sounds low by comparison.
I think Ultimaker is making a mistake with the knurling. It probably beat-out a poorly-hobbed bolt that they knew how to make at the time, and they never thought about going back to a straight-toothed design once they had the information and resources to make a better one because they got it in their heads that knurling was better. Purpose-made straight-toothed drive gears (QU-BD’s attempt notwithstanding) keep getting better, while knurling is and old technique that’s as good as it’s ever going to be.
@Whosa_whatsis , what about straight knurling? Has it been determined what the ideal tooth profile would be? Has anyone published grip figures for different designs other than this and @nop_head 's blog from years back?
The MK8 Drive Gear was extruding 110mm/s in air during the test and that was through a hot end that is long past it’s best.
However, flow rate can’t be compared to extrusion force since temperature also effects extrusion force. There was not enough consistency between flow rate and force during the test to use flow rate as a performance gauge alongside force measurements.
I’ve graph the effects of over tightening the idler bearing on to the filament and found stepper motor performance was reduced by a large margin; in particular with the Raptor gear and the plain insert.
Do you try this one?? Is near of MK7, but the pattern is little different
http://blog.arcol.hu/chrome-beauty