Disclaimer: I received this a review sample for free as part of the Overclock Labs program. No form of compensation is given for this other than access to the product. From my current understanding, I do not own the product and I am asked to hold onto it until it is needed for future for giveaways or for future use in the lab that OCn will setup.
I will focus on the technical details because
1. Ino and others will do a much better job at giving a general overview and comparing against similarly shaped mice.
2. The shape/size isn't one that I'm instantly comfortable with in-game, so I have not used the mouse extensively.
Initial impressions
Packaging: Fairly standard. Comes with the mouse, a quick start guide card (with no useful information), and a warranty card.
Cable: The Revel has an unbraided cable. Initially I was concerned about its thickness, as it instantly reminded me of the WMO cable. But in actual use, I found it to be quite unnoticeable. The cable feels rubbery and is very flexible.
Shape: Copy of Sensei? Not much more to say. Unfortunately last year my dog chewed up the only Sensei I had so I can't compare details. I know many people love this shape, but personally I greatly prefer smaller shapes (kinzu, g100s).
Sensor position: It's a bit higher than average. About 5mm higher than what I'm comfortable with.
Feet: Three large black feet. I have no complaints about them but I'm not particularly picky about mouse feet. Although there is a circular groove around the sensor opening, there is no sensor feet.
Liftoff distance: Same as that of any logitech 3366 mouse on the default surface calibration setting.
Weight and balance: 82-83g. Center of mass is directly over the sensor, at least from what I can tell with just my fingers.
Build quality
These comments are for my particular unit, which is not representative of all Revels. But from the little I've read from the forums, some of these issues are common to others' Revels as well.
Wheel: Definitely a mechanical encoder, but the housing for the wheel is too loose, resulting in a rattling sound when scrolling, especially when scrolling upwards. The encoder itself has a bit more friction than I like, but that could change with more use.
Buttons: L + R have significant pre-travel. The clicks do not feel well-defined or crisp. They are not unusable but it's disappointing. Wheel click is slightly stiff but reasonable. Side buttons are placed a few mm too far back. The lever system inside gives the side buttons too much travel and too little actuation force, leading to a mushy feel.
Coating: Mine has a rubberized black coating on the top shell. Who knows how durable it will be. I believe the sides are raw ABS, so those will probably become smooth after a few months.
Overall: Other than the moving parts, the build quality is good. The shell has no sharp edges from poorly fitted parts
Tracking, subjective impressions
Given that I've been using a Kinzu v1 for the past few weeks at half my normal sensitivity, which I've used since 2013, well everything feels a bit off to me right now. So these impressions are not worth much. But from a total of an hour or so of dm-ing in csgo, there are no glaring issues with the Revel's tracking. I don't think I can answer questions like "does it feel as good as the 3366" because for me, any details of its tracking are overshadowed by my discomfort with the sensor position and the shape. Again, it's not that the shape is bad; I'm just too accustomed to smaller shapes.
Mousetester, mspaint, etc... results
note: the new firmware:
http://www.overclock.net/t/1608034/nixeus-revel-review/100_100#post_25517281
disables mcu smoothing. all the tests and data below were from the release firmware.
DPI
Does it drop or skip usb reports?
Malfunction speed
Motion latency
MCU smoothing
Ripple at 12000dpi
Speed-related accuracy variance
Lens rattle
Internal pictures
The Revel is fairly easy to open. Just poke holes through the bottom feet, unscrew, and push the cable downwards and towards the rear of the mouse to separate the shell.
The internal layout is quite simple and has plenty of room for modding.
The single PCB is held by one screw and 4 plastic alignment posts in the corners. DPI and wheel buttons are "YSA"
MCU is a Holtek HB68FB560. Sensor is a PMW3360 as advertised. Note that the lens is not heat-staked.
L,R buttons are standard 20M omrons
I don't know what brand the wheel encoder is.
Quote:
USB wire order is: red, black, green, white, shield. Plug is the standard/most common size. See CeeSA's thread for more info.
Flipping it around now...
Apparently my 3360 is slightly scratched...
Right button has quite a bit of residual flux from soldering. And apparently the bottom pins of the buttons are not soldered... which doesn't really matter
PCB images:
Button latency
Wired test vs my teensy-modded g100s, which has less than 100us latency vs. the theoretical minimum for 1000hz usb polling.
7.7ms slower on average
revelteensy.txt 3k .txt file
For reference, g303 is ~5ms slower than my teensy-modded g100s.
uaok's tapping test:
I will focus on the technical details because
1. Ino and others will do a much better job at giving a general overview and comparing against similarly shaped mice.
2. The shape/size isn't one that I'm instantly comfortable with in-game, so I have not used the mouse extensively.
Initial impressions
Packaging: Fairly standard. Comes with the mouse, a quick start guide card (with no useful information), and a warranty card.
Cable: The Revel has an unbraided cable. Initially I was concerned about its thickness, as it instantly reminded me of the WMO cable. But in actual use, I found it to be quite unnoticeable. The cable feels rubbery and is very flexible.
Shape: Copy of Sensei? Not much more to say. Unfortunately last year my dog chewed up the only Sensei I had so I can't compare details. I know many people love this shape, but personally I greatly prefer smaller shapes (kinzu, g100s).
Sensor position: It's a bit higher than average. About 5mm higher than what I'm comfortable with.
Feet: Three large black feet. I have no complaints about them but I'm not particularly picky about mouse feet. Although there is a circular groove around the sensor opening, there is no sensor feet.
Liftoff distance: Same as that of any logitech 3366 mouse on the default surface calibration setting.
Weight and balance: 82-83g. Center of mass is directly over the sensor, at least from what I can tell with just my fingers.
Build quality
These comments are for my particular unit, which is not representative of all Revels. But from the little I've read from the forums, some of these issues are common to others' Revels as well.
Wheel: Definitely a mechanical encoder, but the housing for the wheel is too loose, resulting in a rattling sound when scrolling, especially when scrolling upwards. The encoder itself has a bit more friction than I like, but that could change with more use.
Buttons: L + R have significant pre-travel. The clicks do not feel well-defined or crisp. They are not unusable but it's disappointing. Wheel click is slightly stiff but reasonable. Side buttons are placed a few mm too far back. The lever system inside gives the side buttons too much travel and too little actuation force, leading to a mushy feel.
Coating: Mine has a rubberized black coating on the top shell. Who knows how durable it will be. I believe the sides are raw ABS, so those will probably become smooth after a few months.
Overall: Other than the moving parts, the build quality is good. The shell has no sharp edges from poorly fitted parts
Tracking, subjective impressions
Given that I've been using a Kinzu v1 for the past few weeks at half my normal sensitivity, which I've used since 2013, well everything feels a bit off to me right now. So these impressions are not worth much. But from a total of an hour or so of dm-ing in csgo, there are no glaring issues with the Revel's tracking. I don't think I can answer questions like "does it feel as good as the 3366" because for me, any details of its tracking are overshadowed by my discomfort with the sensor position and the shape. Again, it's not that the shape is bad; I'm just too accustomed to smaller shapes.
Mousetester, mspaint, etc... results
note: the new firmware:
http://www.overclock.net/t/1608034/nixeus-revel-review/100_100#post_25517281
disables mcu smoothing. all the tests and data below were from the release firmware.
DPI
note: this is only for my particular unit.
5 measurements of the 400dpi step against 10in of a ruler:
409.2, 408.9, 410.1, 408.6, 409.9
So the dpi steps are about 2% too high. Fairly good.
5 measurements of the 400dpi step against 10in of a ruler:
409.2, 408.9, 410.1, 408.6, 409.9
So the dpi steps are about 2% too high. Fairly good.
Does it drop or skip usb reports?
nope
![thumb.gif]( "thumb.gif")
Malfunction speed
As with the 3366, it's too high to measure. In this picture it's tracking at nearly 8m/s before reaching the end of my mousepad.
Motion latency
the new firmware:
http://www.overclock.net/t/1608034/nixeus-revel-review/100_100#post_25517281
should reduce motion latency by 1ms by disabling mcu smoothing. the data below is for the release firmware
note: I used this method, which is only precise to around 1ms.
At <= 2000dpi, the Revel was 1-2ms slower than a Logitech 3366 mouse (which has less than 1ms motion latency than what's possible given the sensor).
At >=2000dpi, the 3360 sensor has 32 frames of smoothing, which corresponds to an additional ~4ms delay of motion in the lowest framerate mode, (which is what is used when motion begins, until up to aorund 0.4m/s). See this and this for more info. I determined the number of 32 by interfacing a teensy with a 3360 board to act as a 8000Hz mouse (thanks to SweetLow's driver) and counting the number of points framerate transition were blurred by in plots like this.
Here's one image for 3200dpi showing a ~5ms delay between the Revel and a Logitech 3366 mouse.
http://www.overclock.net/t/1608034/nixeus-revel-review/100_100#post_25517281
should reduce motion latency by 1ms by disabling mcu smoothing. the data below is for the release firmware
note: I used this method, which is only precise to around 1ms.
At <= 2000dpi, the Revel was 1-2ms slower than a Logitech 3366 mouse (which has less than 1ms motion latency than what's possible given the sensor).
At >=2000dpi, the 3360 sensor has 32 frames of smoothing, which corresponds to an additional ~4ms delay of motion in the lowest framerate mode, (which is what is used when motion begins, until up to aorund 0.4m/s). See this and this for more info. I determined the number of 32 by interfacing a teensy with a 3360 board to act as a 8000Hz mouse (thanks to SweetLow's driver) and counting the number of points framerate transition were blurred by in plots like this.
Here's one image for 3200dpi showing a ~5ms delay between the Revel and a Logitech 3366 mouse.
MCU smoothing
the new firmware:
http://www.overclock.net/t/1608034/nixeus-revel-review/100_100#post_25517281
disables mcu smoothing. the info below is for the release firmware
The Revel with its release firmware does indeed have MCU smoothing (which appears to be extremely common to the firmware of mice with Holtek MCUs). The algorithm is something like: the current thing counts over usb is the average of the previous counts and the latest counts from the sensor. In other words, an exponentially weighted moving average.
If there is no mcu smoothing, one expects to see well-defined periodic patterns like this, due to the lack of synchronization between the sensor framerate and the usb polling.
This form of MCU smoothing is recognizable in mousetester as patterns like *-..__*-..__*-..__ where the peaks take a sawtooth-like form instead of something like *___*___*___.
http://www.overclock.net/t/1608034/nixeus-revel-review/100_100#post_25517281
disables mcu smoothing. the info below is for the release firmware
The Revel with its release firmware does indeed have MCU smoothing (which appears to be extremely common to the firmware of mice with Holtek MCUs). The algorithm is something like: the current thing counts over usb is the average of the previous counts and the latest counts from the sensor. In other words, an exponentially weighted moving average.
If there is no mcu smoothing, one expects to see well-defined periodic patterns like this, due to the lack of synchronization between the sensor framerate and the usb polling.
This form of MCU smoothing is recognizable in mousetester as patterns like *-..__*-..__*-..__ where the peaks take a sawtooth-like form instead of something like *___*___*___.
Ripple at 12000dpi
As expected due to smoothing, the 3360 has noticeably less ripple than the 3366.
![IZHzyi2.png]( "IZHzyi2.png")
Speed-related accuracy variance
Also known as "acceleration".
This can be tested via a fast (~2m/s) turn to the right and a slow return, with the initial and final position lined up against a keyboard or some other solid edge, and reading out the angular deviation. For instance watch
.
Due to my sensitivity, my turns covered ~360 degrees instead of 180 as Ino did. The angular deviation from 10 tests are:
0.27, -1.36, -0.55, -0.55, 0.07, -0.89, -1.77, -0.68, 0.14, -3.27
Except the the final outlier, this is about as good as it gets and meets what I've seen with Logitech's 3366 mice.
This can be tested via a fast (~2m/s) turn to the right and a slow return, with the initial and final position lined up against a keyboard or some other solid edge, and reading out the angular deviation. For instance watch
Due to my sensitivity, my turns covered ~360 degrees instead of 180 as Ino did. The angular deviation from 10 tests are:
0.27, -1.36, -0.55, -0.55, 0.07, -0.89, -1.77, -0.68, 0.14, -3.27
Except the the final outlier, this is about as good as it gets and meets what I've seen with Logitech's 3366 mice.
Lens rattle
IMO it's quite silly to check for lens rattle by listening or by poking the lens.
A good way to check for lens rattle that I recall from the g303 thread is to:
1. open paint, use the pencil tool
2. set the dpi to 12000
3. put a piece of tape over the sensor hole. Make sure the tape's surface is trackable by tapping it lightly. At 12000dpi, the cursor should jump quite a bit.
4. lift up the mouse, click to use the pencil tool of paint, and shake the mouse midair.
My particular Revel has no visible lens rattle at all.
Here's a comparison to some of the 3366 mice I've used (uh, don't ask me about A,B,C,D,E):
Note that the 32 frames of smoothing present at 12000dpi means that the test isn't quite as sensitive as it is for 3366.
For my g900, the poor result is likely related to how I've opened it multiple times and probably didn't tighten everything properly.
A good way to check for lens rattle that I recall from the g303 thread is to:
1. open paint, use the pencil tool
2. set the dpi to 12000
3. put a piece of tape over the sensor hole. Make sure the tape's surface is trackable by tapping it lightly. At 12000dpi, the cursor should jump quite a bit.
4. lift up the mouse, click to use the pencil tool of paint, and shake the mouse midair.
My particular Revel has no visible lens rattle at all.
Here's a comparison to some of the 3366 mice I've used (uh, don't ask me about A,B,C,D,E):
Note that the 32 frames of smoothing present at 12000dpi means that the test isn't quite as sensitive as it is for 3366.
For my g900, the poor result is likely related to how I've opened it multiple times and probably didn't tighten everything properly.
Internal pictures
The Revel is fairly easy to open. Just poke holes through the bottom feet, unscrew, and push the cable downwards and towards the rear of the mouse to separate the shell.
The internal layout is quite simple and has plenty of room for modding.
The single PCB is held by one screw and 4 plastic alignment posts in the corners. DPI and wheel buttons are "YSA"
MCU is a Holtek HB68FB560. Sensor is a PMW3360 as advertised. Note that the lens is not heat-staked.
L,R buttons are standard 20M omrons
I don't know what brand the wheel encoder is.
Quote:
Originally Posted by Dahvi
its TTC scroll, that 11 in circle is new logo of TTC for scrolls
USB wire order is: red, black, green, white, shield. Plug is the standard/most common size. See CeeSA's thread for more info.
Flipping it around now...
Apparently my 3360 is slightly scratched...
Right button has quite a bit of residual flux from soldering. And apparently the bottom pins of the buttons are not soldered... which doesn't really matter
PCB images:
Button latency
Wired test vs my teensy-modded g100s, which has less than 100us latency vs. the theoretical minimum for 1000hz usb polling.
7.7ms slower on average
revelteensy.txt 3k .txt file
For reference, g303 is ~5ms slower than my teensy-modded g100s.
uaok's tapping test:
