Disclaimer: Although the review sample was provided by Endgame Gear, the review itself is 100% unbiased. All opinions are my own.
The Endgame Gear XM1 is an ambidextrous gaming mouse with side buttons on the left side only (six buttons in total) that comes with Pixart’s latest top optical sensor, the PMW3389, which is capable of up to 16,000 CPI. Furthermore the XM1 features a lightweight construction of just 70g, 100% PTFE mouse feet, pre-selected Omron switches for the main buttons (rated for 50M clicks) and lastly a patented analogue switch contact technology, which promises a click delay of <1 ms without any unwanted double click actuation. A button for on-the-fly CPI and polling rate adjustment on the bottom along with software functionality are present too.
The XM1 comes in a simple yet solid box, with the mouse sitting nicely cushioned in a foam package. The cable is somewhat tightly folded, so it may take a while until the kinks disappear. There are no replacement mouse feet in the box, but the provided ones should already last a while. Endgame Gear will also sell replacement feet.
Base length: 11.3 cm
Length including overhang: 12.2 cm
Height (highest point): 3.6 cm
Front height (lowest point): 1.3 cm
Width (widest point): 6.6 cm
Grip width (narrowest point): 5.5 cm
My scale returns a weight of 69g, which is 1g less than the proclaimed 70g weight.
Weight balance is very good as well, which makes the XM1 feel even lighter than it already is.
Comparison: Steelseries Kana at the top, Nixeus Revel at the bottom
The shape of the XM1 has been co-developed by CS 1.3/1.5/1.6 AWP legend Johnny R, who also had a hand in designing the Steelseries Sensei, Kana and Kinzu shapes. Hence it comes as no surprise that the shape of the XM1 is rather close to that of those three mice. As you can tell from the comparison pictures, the XM1 is actually most similar to the Kana, with the greatest differences being a slightly more tapered butt, the hump being more towards the back and the XM1 being slightly shorter and wider. The uniquely curved sides along with the widened butt are strongly reminiscent of the Sensei and Kana. Since I’m a big fan of the Kana the XM1 is right up my alley shape wise. The feel in hand isn’t exactly the same as the Kana though, but quite close.
As for grip styles, I’d say the XM1 should be suitable for all three grip styles as well as most hand sizes (it’s a medium size mouse). However, I reckon that it’s most suitable for claw and fingertip grip and least suited for palm grip. Although there is some palm support due to the hump being positioned towards the back, it’s too flat to provide full support.
Comparison: Microsoft WMO at the top, Zowie FK2 in the middle, Dream Machines DM1 Pro S at the bottom
Mouse feet and cable
The feet are made of 100% PTFE, which is why they’re white instead of the usual dyed black. 100% PTFE feet should result in greater longevity compared to non-100% PTFE feet. Glide is excellent on the cloth pads I tested it on. There are two medium (at the back) and two small (at the front) feet, the latter of which could be difficult to replace with generic feet. Since Endgame Gear will provide fitting replacement feet this is a rather minor complaint, however.
The cable is a standard rubber cable, which is not as flexible as the Model O or DM1 FPS cable, but still above average. I’d say it’s roughly on par with Zowie cables. Thickness is standard, so it should fit into any mouse bungee (tested with a Zowie Camade). Cable length is 1.80 m (measured). There is a ferrite bead attached to the cable which helps lessen noise.
Despite being a very lightweight construction the XM1 has a sturdy and solid build. There is no rattle, no loose buttons, no creaking and no flexing. The main buttons have very low pre- and post-travel, giving them a nice and snappy, fairly light click. The right click has a hollow and subdued sound on my copy, however, which isn’t present on the left one. This might be fixable by loosening the screw directly below the affected click, but I haven’t tried it so far. Side buttons have low pre- and moderate post-travel, with the forward button being the better one. The scroll wheel has very distinct and tactile steps while being very quiet to scroll in both directions. Actuating the scroll wheel requires an above average amount of force.
The XM1 comes with a black matte coating, which provides decent grip while not attracting dirt too much. I’d say it’s roughly comparable to current Zowie coatings (slightly grippier and not as much of a dirt magnet though).
On my copy there is a faint high-pitch noise (‘coil whine’) audible when the sensor is having contact with a surface. It can only be heard at certain angles and it’s not as loud as the high-pitch noise on the early batches of the EC-B series, but noticeable still. The noise may be inaudible entirely for those who are not susceptible to hearing high frequency noises in the first place. The factory has already been informed of my findings, so we may see this addressed in the future, provided that it’s not a one-off occurence.
For this test I’m checking whether the actual CPI values match the nominally stated CPI values. The default (out of the box) nominal CPI values are: 400/800/1600/3200. The actual (tested) values, however, are: 410/816/1638/3270. In short, actual CPI is consistently higher than nominal CPI, but only by a small margin. Note that this variance will itself vary from unit to unit, although the general trend should persist.
Sensor smoothing (motion delay):
In this test I’m gauging whether the sensor performs how it’s supposed to. ‘Sensor smoothing’ describes an averaging of motion data across several capture frames in order to reduce jitter at higher CPI values, which increases motion delay. The goal here is to have as little smoothing as possible. According to specification the 3389 has no visible smoothing up to 1800 CPI and 32 frames of smoothing at and above 1900 CPI, which doubles twice at 6000 CPI and at 11300 CPI. Let’s see how the 3389 in the XM1 fares in this test.
First I’m doing a basic test in which I’m looking at a plot of the raw motion data, aggregated as xCount. In an xCount plot smoothing can be recognised by any ‘kinks’ being visible, which show up at framerate transition points. As you can see no kinks can be observed, which means that no smoothing is present.
In order to ensure that my result from the xCount test is correct I’m taking a look at a plot that compares the XM1 to a G403 (3366, no smoothing across the whole CPI range) at 3200 CPI, aggregated as xSum. Any CPI deviations between the two mice have been normalised. As you can see there is about 0.5 ms between the mice, which is within margin of error (G403 was also physically moved before the XM1) and confirms the results from above. The XM1 does not have any visible smoothing at the tested CPI steps, which is a first for a 3389 mouse.
This test is used to indicate any potential issues with angle snapping (non-native straightening of linear motion) and jitter. There may be some added jitter at 3200 CPI (would be expected due to the lack of smoothing), but it’s very minor. Other than that no issues can be observed.
‘PCS’ is short for ‘perfect control speed’. The PCS indicates the maximum speed up to which the sensor functions flawlessly (i.e. without malfunctioning in some way). The nominal PCS of the 3389 is 400 IPS, which translates to 10.16 m/s. I only managed to hit a measly 4 m/s. As you can see there are no signs of the sensor malfunctioning at that speed.
One of the main selling points of the XM1 is its unique (and patented) analogue switch contact technology. Since this is something that hasn’t been done before, I figured some technical details would be interesting to some, which is why I’ve asked one of the developers behind the XM1 to provide a rough explanation. I’ll provide his quote in full:
“Analog Key Sensing Technology does not debounce at all. Debounce is something from the digital world. The switch itself however is still the same switch, causing the same bouncing as before, but the Analog Key Sensing takes sample snapshots at an incredibly high speed (very much like a scope) and the bouncing of the switch is summed into this analog sampling signal and therefore the bouncing is transformed into a mathematical function rather than a digital bounce (0/1), thus eliminating the need to debounce altogether. This works similar to how different brightness levels are achieved for LEDs traditionally. LEDs are pulsed to achieve that, being always driven at their max brightness for small periods of time. The more times that happens the brighter the LED is perceived. Analog Key Sensing is the reverse of this and using true analog to actually see the ‘brightness’, i.e. the key level.”
Onto the testing. I’m only able to test the click delay with a program called Bloody KeyResponse, which not only accounts for any differences in terms of hardware debounce delay but also in terms of click actuation weight, which means that a mouse with lighter clicks may return lower delay values, despite not being actually faster. In order to minimise any inconsistencies I’ve tested the XM1 against several different mice with varying click actuation weight (with the G403 having the lowest). All lenses were covered during the test, as some mice have higher debounce during lift-off. Here are the results (values in brackets indicate the delay relative to the Steelseries Ikari, which acts as a baseline):
Logitech G403 (4.2 ms): XM1 was slower by 2-4 ms
Nixeus Revel (6.2 ms): XM1 was faster by 5-7 ms
Microsoft WMO (12 ms): XM1 was faster by 14-16 ms
Based on these values (the G403 appears to be an outlier, due to the very low click actuation weight) I’d say the XM1 should be around +0-2 ms relative to the Ikari, which is an excellent result. Of course, I’m unable to test whether the ‘no double clicks’ claim holds, only time will tell that.
Since the XM1 already allows for CPI level and polling rate adjustment without any software, the software is kept very basic. It allows for button remapping (in theory, appears to be nonfunctional currently), CPI adjustment (four levels, in increments of 50), LOD adjustment (2mm or 3mm, the former being the default setting) and enabling or disabling ‘ripple control’.
The most interesting option is actually ‘ripple control’. As seen in the sensor testing section, the XM1 does not have any smoothing at the default CPI levels. In fact, I went ahead and tested 16000 CPI and still could not detect any smoothing (resulting in jitter and ripple being out of control at that setting). Upon enabling ‘ripple control’, however, the usual behaviour from a 3389 can be seen: 32 frames of smoothing at and above 1900 CPI, which doubles twice at 6000 CPI and at 11300 CPI. Personally I’m a big fan of giving users control over options like this one, as both people who want high CPI but don’t like smoothing as well as people who want high CPI but don’t want ripple and jitter can be satisfied.
Once again I’m getting to review the first mouse release of a new company, Endgame Gear, and once again I’m impressed. The XM1 combines a very good (and unique) shape, a flawlessly implemented top tier sensor, innovative technology (analogue debouncing) and very low weight (without having to resort to holes), without having any of the rookie issues you might expect from a new competitor. My copy had some build quality concerns (button uniformity, high-pitch noise), but they weren’t major enough to negatively impact my experience with the XM1. Recommended.