Overclock.net banner

g102 mercury sensor info

195K views 262 replies 69 participants last post by  qsxcv  
#1 ·
disclaimer: i received this mouse for free. not a review. just sharing info and pictures for those who are curious.

the logitech g102 has a new optical sensor (i.e. led illuminated, image correlation) that is not made by avago/pixart



the sensor is connected to the main pcb with some weird surface-mount soldering. not sure what it's called.

illumination is external as you can see.

the lens is sort of clipped onto the sensor block. there is a little bit of wiggle possible. i doubt anyone will complain about "lens rattle" like for some of the 3366 mice, but we'll see.


some quick probing around with a multimeter and an oscilloscope reveals most of the pins' functionalities:
(1 is bottom right, closest to the mcu. numbers go counter clockwise)

Code:

Code:
1    ss
2       mosi
3       miso
4       sclk
5       vdd?
6       gnd
7       ??

8       ??
9       gnd
10      ??
11      gnd
12      ??
13      gnd

14      ??
15      gnd
16      led-
17      gnd
18      gnd
19      led+
20      gnd
anyway...
mousetester looks unlike anything i've ever seen.
Quote:
Originally Posted by qsxcv View Post


at low speeds there are discrete steps. at high speeds eventually it gets to something above 10000fps. for now, i'm too lazy to figure out exactly

i was quite confused by the discrete steps, but here's a plausible explanation
it appears what is going on is that for moderate speeds, the framerate is continuously adjusted such that the amount of motion in a frame is nearly constant.

i suspect this is a very low power sensor which will eventually be used in a wireless mouse
well my hunch about the continuous framerate was sort of right. hooking up a wire to the led cathode gives a direct view of the timing of the frames:

when not in motion, the sensor runs at about 1000Hz.


the sensor frames are not synchronized to the usb frames, which can be easily shown by taking a few snapshots:
yellow trace is the led cathode. blue trace is USB D+.


if you don't touch the sensor for about 100 seconds, it goes into a 50Hz rest mode. I'm not sure whether there are deeper rest modes though.

k so what happens when you move it?

the increasing intensity reflects increased density of the led's toggling. i.e. higher framerate

zooming in:



quite surprising how sporadic it is. not sure if this is a good or a bad thing. i would prefer to see either a constant framerate or at least one that always has periodic timings, but it's possible that this is intentional. i don't know.
but anyway, unlike the 3310, 3360/3366, etc... the framerate isn't adjusted by entering different modes with predefined framerates. or so it appears at least

the maximum framerate is something around 12kHz


anyway...
some initial info about the spi:
6 bytes of communication every 1ms.

spi runs at 8MHz
 
#6 ·
Quote:
Originally Posted by dobragab View Post

Great, thanks for the info!

Have you tested its perfect control speed or smoothing?
biggrin.gif
woll3 said it has no smoothing at lower CPI and said that if it has smoothing at higher CPI it's not enough to counteract the jitter of higher CPI from what I understand, so I guess it has no or very little smoothing.

so how's the SRAV? and the sensor being PCB mounted is...interesting, to say the least, weren't the MLT04 chips like that as well?
 
#10 ·
Quote:
Originally Posted by Alya View Post

so how's the SRAV?
I don't think qsxcv has tools to test that... if he does that would be cool though.

My theory is that the weird granular dynamic framerate is to reduce one source of SRAV.
Quote:
Originally Posted by chr1spe View Post

So the frames aren't separated by some multiple of the minimum interval between frames at all points or anything? Seeing those patterns I would think the gaps would all be 0.1ms, 0.2ms, 0.3ms or something, but I guess that doesn't look to be the case.
If I'm right and the sensor is trying to get an integer amount of pixel motion in between each frame, it might be elongating every few frames because of its internal subpixel motion tracking falling ahead or behind. Also, if they wanted to get an integer-pixel-offset for each axis, they would have to capture two correlation frames for each reference frame (if possible for the given movement), which would be silly and cause a lot of DSP programming difficulties... would it?
 
#11 ·
Quote:
Originally Posted by wareya View Post

I don't think qsxcv has tools to test that... if he does that would be cool though.

My theory is that the weird granular dynamic framerate is to reduce one source of SRAV.
If I'm right and the sensor is trying to get an integer amount of pixel motion in between each frame, it might be elongating every few frames because of its internal subpixel motion tracking falling ahead or behind. Also, if they wanted to get an integer-pixel-offset for each axis, they would have to capture two correlation frames for each reference frame (if possible for the given movement), which would be silly and cause a lot of DSP programming difficulties... would it?
Well, I'm just trying to figure out if that is the case. I'm having trouble telling with the first image, but the second one all of the gaps are actually 0.1ms or 0.2ms which would make me think that the others are too. Its difficult to read the other one for me though.
 
#13 ·
Quote:
Originally Posted by wareya View Post

Quote:
Originally Posted by Alya View Post

so how's the SRAV?
I don't think qsxcv has tools to test that... if he does that would be cool though
i can only test with my hand. so not very accurate of course, but for the prototype i used it was as good as the other logitech mice. i assume this one is the same
 
#19 ·
Quote:
Originally Posted by v0rtex-SI View Post

Its on german ebay.. Just search for logitech g102 and buy it
biggrin.gif
Yeah but if the pricing is going to be similar in retail shops in our own countries it's better to buy it there considering the return rate on mice, monitors, etc. is staggering due to poor QC or bad design.
eBay is sort of buy and keep, not buy and return if you don't like it or the one you got is faulty (such as inconsistent buttons or other stuff that doesn't make it DOA).

At least that is why I wait, to have the option to return it if prices are very similar.
 
#20 ·
Quote:
Originally Posted by JackCY View Post

Yeah but if the pricing is going to be similar in retail shops in our own countries it's better to buy it there considering the return rate on mice, monitors, etc. is staggering due to poor QC or bad design.
eBay is sort of buy and keep, not buy and return if you don't like it or the one you got is faulty (such as inconsistent buttons or other stuff that doesn't make it DOA).

At least that is why I wait, to have the option to return it if prices are very similar.
You can always sell the mouse? Sure you lose an euro or two but theres no need to keep it if you dont like it.
 
#26 ·
You mean patented common sense by pixart... :/
I bet it's some Chinese sensor maker and IMHO they shouldn't give a damn about patents and IP and what not as long as they develop it themselves not just copy paste steal it.
The patents and IP nonsense really needs to die.