Wireless Mouse Technology Testing at the Logitech Daniel Borel Innovation Center

Linus Tech Tips ·Linus Tech Tips ·2016-05-06 · 3,829 words · ~19 min read
Floatplane YouTube

Transcript

JSON SRT VTT 260
0:00 Logitech brought us here to Switzerland to check out the G900 Chaos Spectrum. An
0:04 interesting looking mouse, but with some really bold claims, especially when
0:07 comparing it to a wired mouse, which traditionally would have been a lot
0:11 better in like almost every possible metric. They claim, however, that it has
0:16 now reversed. They're claiming that the G900 is better than a wired mouse. Let's
0:21 do some science and figure it out.
0:31 So, we're starting off in the RF Labs. This is probably one of the most important things about this mouse is its
0:36 wireless performance. That's what they've been trumping really, really hard this whole time. So, we need to
0:41 benchmark it. We're going to test it with huge amounts of interference,
0:44 without interference, all this different kind of stuff, and see how it compares
0:47 to other mice. We're going to do that by coming inside of an anooic chamber, but
0:51 this time it's not an anooic chamber in terms of sound. It's an anooic chamber
0:55 in terms of signals. So RF, Wi-Fi, Bluetooth, all that kind of stuff isn't
0:59 going to be able to leave the room once the door is closed and isn't going to be
1:02 able to come on into the inside. But to
1:05 cause that interference, we have this antenna right here, which is going to be
1:08 shooting out massive amounts of amplified RF, Wi-Fi, and Bluetooth
1:12 signals. What they did was they went to a LAN of about 300 people, measured the
1:16 signal noise, and then amplified that by a little bit more to make a crazy worst
1:21 case scenario. Over on the other side of the room, there's all this foam kind of
1:25 stuff everywhere. That's so that signals aren't bouncing around throughout the
1:28 room, and you just get a solid trajectory, so it's better and more
1:31 consistent for measurement. Here we have the mouse rig. So, this is a jogger that
1:35 keeps a pad moving under the mouse so that the uh actual sensor is constantly
1:40 firing and sending information, which is important because you need the mouse to be actually working. And then the mouse
1:44 is held in place above that. Across the
1:48 room a little bit right here is the receiver. So, this guy is actually
1:53 talking to the mouse. That wire goes through a tube to the outside and then
1:56 we're measuring actually the tracking of the mouse out there where the actual
2:01 interesting stuff is. So, let's check that out. So, as you can see back here
2:04 on the spectrum analyzer, they have a crazy amount of noise being fired in
2:08 that room right now. But over here, you
2:11 can see a tracking tester where it's just tracing the cursor movement of the
2:15 mouse as that jogger is going around. And it is in a very I'm going to reset
2:19 this. You can see it drawing. It isn't a very consistent kind of pattern, which
2:23 is good. There's going to be some drift from the sensor, but that's actually fine. All sensors have some amount of
2:27 drift. Um, but the good thing about it is that it's consistent. It's not going
2:31 all over the place. And if you're not making robotic exact circular movements,
2:36 it's going to be extremely predictable where everything's going to be anyway. So, this is a good sign. And this is for
2:41 the G900. So, now we're testing the Razer Mamba. Exact same scenario. This
2:45 time we're going to try it with the signal noise on and with the signal
2:48 noise off so you can see what it is in a more perfect scenario. The Mamba is
2:53 currently set up and good to go. Powered on. Jogger is running. Antenna is over
2:57 here. Let's go check the results. All right, so the Mamba 2015 is hooked up
3:01 inside. That's what's running right now on the tracking tester. As you can see
3:05 on the spectrum analyzer in the back, there's nothing really going on. That's
3:08 because we have the antenna off. So, what I'm going to do is come back here,
3:13 crank the antenna on. So, I'm going to turn this dial a little bit there.
3:18 That's all good to go. The spectrum is now crazy. You'll see it actually moves
3:23 where it's where it's kind of firing huge amounts of spectrum. So, it's not
3:26 going to be a consistent problem. You can already see on here. I'm going to
3:29 reset it so that we have some more room to actually see it get messed up. But,
3:34 it's jumped all over the place. These lines that you see is where it stopped
3:38 tracking completely for a second and then engaged back in somewhere else. You
3:41 can see that instead of just simple sensor drift, it's actually jumping all
3:45 over the place. I'm going to reset it here. It's going okay at the start, but
3:49 then you see one big jump there. I'll wait for it to do a few more.
3:54 There's one jump right there. So, there's going to be bad spots. Sometimes
3:58 it'll be okay because maybe it's away from where the antenna is firing huge
4:02 amounts of information. But then other times, as you saw before, it might
4:05 actually start jumping all over the place because it's kind of getting overpowered, which is not very good. So,
4:10 there's a line there. There's another line there. It's getting a little bit worse now. Yeah. So, that's the Mamba
4:15 2015. Let's move on. All right. Next up on the chopping block is the Oraoros. A
4:20 little bit of an older mouse. It's currently running on the jogger. Its
4:23 sensor is currently over here. We're good to go. One other little bit of
4:26 information that I want to give you guys is this is the antenna that's receiving
4:30 the signal. So when we show you guys the spectrum analyzer, this is the thing
4:34 that's outputting that information just in case you're interested. So next up,
4:37 Razer Orbus. That's on the inside right now. You see it going with nothing
4:42 really happening on the spectrum analyzer. Seems fine. A little bit of
4:45 sensor drift, which is completely normal. Now let's hurt it a little bit.
4:49 So going to turn things up.
4:55 There we go. Spectrum analyzer is going like crazy. Stuff sweeping across like
4:59 we saw before. Moving back. Now that
5:03 happy little bit of sensor drift is no longer really very happy. As you can
5:07 see, there's already two major jumps and it's completely changed direction. Three
5:11 major jumps now. Yeah. So, not the clean nice patterned
5:17 thing that we saw with the G900. It's a little bit more of a mess. Now looking
5:20 back, there's 1 2 3 4 5 six major jumps,
5:26 I think. And instead of just drifting nicely upwards like it was before,
5:29 you're going up and then cranking way over to the right. And it's probably
5:33 going to screw up again later on. Not the end of the world, but not perfect.
5:37 Last, but potentially not least, is the Sensei. So, we have the Sensei rigged up
5:41 on the jogger, and we have its receiver on the other side. Something to notice
5:45 about the configuration of things as well. The receivers are always placed in
5:49 line with the very front of the mouse. And the way that the jogger works,
5:53 there's little arms going all the way around so that it can hold it in place
5:56 perfectly despite the movement of the actual jogger. And it doesn't have to
6:01 press any buttons. So, it's not pressing left or right buttons. It's not pressing the back button. So, it's not doing any
6:05 of that kind of stuff. And it's holding it in place, which is pretty cool. So,
6:09 the Sensei isn't doing so great. You can
6:12 see on the spectrum analyzer there's nothing really going on. We basically
6:16 have it off. Um, and that's not just
6:20 simple normal nice unified sensor drift.
6:23 That's quite messy. Let's start over again somewhere clean. Give it a nice
6:27 clean slate and then hit it with the antenna as hard as we can go
6:35 just to see what happen. Wow. Okay. So, this one
6:40 that is just that is horrible. So, one of the interesting things that this one
6:44 does that none of the other ones actually had a problem with is you can
6:47 see the lines get kind of squiggly and they start doing weird spirals and
6:53 actually making things that aren't at all the shape of what it's doing, which
6:57 is a huge unusable problem. Again,
7:01 something to remember, this is an insane worst case scenario. It's beyond like a
7:06 normal kind of worst case scenario because they went from a like 300 person
7:10 computer LAN, which is an insane scenario for any wireless device, then
7:14 made it a little bit harder. But this thing is completely failing. The other
7:19 ones were failing pretty hard, and the G900 had zero issues whatsoever. So, I
7:24 think that's pretty cool. So, to create the device, the G900 that beats all of
7:29 the other devices in this crazy worst case scenario test, it takes a lot of
7:33 things. skills, science, testing, all that kind of stuff. It's not just buy
7:37 the better thing, put the better thing in, and then you have a better object.
7:40 It's not that simple. So, what they do is they use simulators in order to test
7:45 what the actual end result is going to be. That's not always perfect, and just
7:50 putting something in there and running a simulator won't magically solve all of your problems. That's where the skill
7:54 component comes in. They're able to get their RF simulator results to be
7:58 extremely similar to their actual measurement through skill. That's
8:03 extremely important because it helps with your iteration process. That way,
8:07 you don't have to build a new full device every single time you want to
8:11 test it, which is going to help a lot because they have to release a product
8:14 sometime. Then they also have to pay attention to where different components
8:19 within the mouse go, like the battery. battery is a huge deal. With some older
8:24 mice that even Logitech had, the positioning of the battery could cause
8:28 weak spots in your signal as it went around. But the sizing and the
8:32 positioning of this battery is a little bit better. They they have it on a different Z-axis within the mouse so
8:37 that the emitter should be able to have a strong signal going all the way
8:40 around. Super cool. Next up, we've got the sensor lab. So, what we're going to
8:44 be doing here is testing motion latency. So, how long it takes for you a movement
8:50 of the mouse to actually go to the computer. We're not worried about the less rest of the latency in the system.
8:54 This is just what Logitech would have control over. So, we're going to do that
8:58 by coming in here and using this turntable. The service on here is just
9:02 like a Steel Series QCK type surface. So, probably something that's pretty
9:06 common in someone's home. We're going to start with the Logitech G900, which is
9:10 already on here. It's not torqued down. It's not too light. It's just kind of
9:15 exactly how much someone would be pushing down on the surface cuz you
9:18 don't want to grind against the mouse pad. You don't want to do anything like that. That would mess up with the
9:22 results. Next up, we'll have a Razer Chroma, which is actually a wired mouse.
9:26 So, we'll compare it against that. And then we have a Mamba 2015, which is
9:30 another competitor wireless mouse. There's a little bit of plus minus here.
9:34 that it could be plus 0 to 1 millisecond
9:38 or anywhere in between due to polling rate. But spoiler alert, it's going to
9:41 win by more than that. So, it doesn't really matter. Another thing I forgot to
9:45 mention is that the mouse is placed very specifically on the turntable mouse pad.
9:50 It's placed at a 45° angle so that X and Y axis are contributing the exact same
9:55 amount. They have to be very precise about these things cuz extra variables
9:59 like that could screw with something in a way that would mess up your results.
10:02 So they have the right amount of tension down, they have the right angle, and
10:07 they have a huge amount of samples, all of which are good. All right. So we're
10:10 going to run our first test here. You can go right now. And what you'll see on
10:14 the screen in a moment is it's going to accumulate data over time. So what's
10:18 actually going to be displayed graphically is rounded a little bit just
10:23 more for presentation so that you can kind of see whereabout things are. But
10:27 below that is more of the hardcore math. No rounding and you can see the mean.
10:32 So, we have to wait a little bit for it to gather some data and then we'll give
10:35 you a more accurate mean. So, after taking 33 samples of movement, the mean
10:40 latency was 4.79
10:44 milliseconds. So, that might not sound impressive yet, but just wait till we
10:47 try the other mice. So, we're about to run test two. This is against the Razer
10:51 Chroma, which is actually a wired mouse. You can see here in the device info, it
10:55 is in fact the Razer Chroma. We're going to be comparing it against our previous
10:59 file, which was running the G900. So, we're going to have blue data showing up
11:03 on the chart. That's for the previous mouse. And the green data you, as you
11:08 can see in here, color green, will be actually for the Chroma. So, let's run
11:12 the test. So, we have our first result, the G900 versus the Razer Chroma. As you
11:17 can see here, the green is much higher than the blue, which means that the
11:21 green lost. This is motion latency. So, you want a low number. As you can see,
11:26 the median for the G900 was 4.8, as we
11:29 already discussed. The blue is the same numbers that we had running on our
11:32 previous test, but the mean for the chroma is about 6.88,
11:38 which is 2 milliseconds more, which is a lot because you have to pay attention to
11:43 each segment of your latency throughout the entire entire uh equation. So, your
11:48 graphics card, your monitor, your mouse, all that kind of stuff. And reducing
11:52 each one by relatively small amounts is
11:55 a huge deal. So, you can get the total number lower. So having Logitech's
11:59 number of the input from their mouse due to motion lower is good. And that means
12:03 that it's lower than a wired mouse, which is pretty damn impressive. Another
12:06 kind of cool thing to look at is that the lowest, so the fastest rate at which
12:11 the motion was able to go from the mouse to the computer for the Razer mouse was
12:16 about 5.5, which tied the highest or
12:20 worst result from the G900. Just kind of a fun little tidbit. So this time the
12:24 mouse we are testing is the Razer Mamba. As you can see in the top, we're
12:28 comparing it yet again against the G900. So the compare with file is the original
12:32 G900 test that we ran again. We're doing 33 33 trials, the exact same that we're
12:37 doing every time. This time it's going to show up as pink. So yeah, let's get
12:41 started. Okay, so we're done testing the Mamba 2015. And this is kind of why I
12:47 haven't liked wireless mice in the past and probably why a lot of people haven't
12:51 either. Look at the data. It's a mess. You've got results from 0.5, which are
12:56 junk data. We'll explain why that's junk data in a moment, all the way up to
13:00 18.5, which is probably caused due to
13:04 lag in the transmission of data over the wireless signal. If it if it misses a
13:08 report and then has to do it later, it's going to be massively delayed. your mean
13:12 is at about 9.3 which is much higher than the 4.79
13:19 of the G900 but probably more within the
13:22 actual capabilities of the sensor. The problems with the junk data and the
13:27 really really high late reported data is
13:30 more due to the implementation of the sensor instead of the standard
13:34 capabilities. Last but not least, we have click latency. So realistically,
13:38 what all of these different stations have been explaining is that we don't
13:41 necessarily need to be afraid of wireless anymore. I was afraid of
13:45 wireless before, but I think that was justified before this mouse came out
13:49 because, as we've seen, there were some problems with wireless mice. Now, click
13:53 latency would be a huge one if it was worse than say a wired mouse. So, what
13:57 we're going to do here is actually measure that. So, I'll explain how we'll do that. Now, they have a wire going
14:02 into here, but you can see that it's not the actual normal one. This is wired
14:06 directly into the switch instead of actually um the mouse itself so that you
14:11 have your wireless adapter still running. That has been done on the G900,
14:16 a Rival 300, and a Razer Chroma. So, you
14:19 can see there's two wires going into each one of these cuz these are more naturally wired mice. Now, I'm going to
14:24 move these guys out of the way. This uh
14:28 wire from the switch is going into a signal adapter. This then sends it to a
14:34 latency measurer. So what this one is recording off the bat is exactly when
14:39 the uh switch is pressed. Um and then it
14:43 starts a timer from that point in time to when this adapter is able to send the
14:48 signal to the USB analyzer. So that's when it actually receives the signal
14:53 that the switch was pressed because that one isn't wired in. That's a wireless
14:57 signal. So you get to see the total latency of the mouse within your system.
15:01 Similar to how we were testing total latency of the mouse in previous
15:05 sections. This isn't the total latency until something would appear on your
15:09 screen because there's other things within that equation like your graphics card and your monitor and whatnot, but
15:14 this is what Logitech can control. So in order to actually run the test, they set
15:18 up this kind of game thing where you go click on the lit up circle. This is just
15:23 causing you to click a number of times at different intervals so it can be
15:27 recorded over here. Much like the motion testing, motion latency testing that we
15:31 did earlier. It'll show up a graph which will have some slightly rounded numbers,
15:34 but we'll give a good visual representation of the results. So, we're going to run this about 30 clicks and
15:39 then get a result after that.
15:43 So, as you can see graphically, our average is at about 4.5. It kind of all
15:48 huddles in a nice bell curve right around that area. And our mathematical
15:52 not really rounded as much average is at about 4.3. Surprisingly similar to the
15:58 latency in our motion input latency test. So yeah, let's move on and try the
16:03 other mice. The first contender is the Steeler Rival 300, which has its
16:07 firmware updated to the most recent one since last Friday. It's a wired mouse,
16:11 so theoretically up until now, it should be faster. So we'll see if it reigns and
16:16 holds that title. Let's begin the results and then compare it against the
16:19 G900. Okay, so looking up here, we can see the
16:24 blue data is from the Logitech G900. The
16:27 green data is from the Steel Series Rival 300. Um, you can see again the
16:33 mean or general average from the Logitech G900 was about 4.5 and now with
16:39 the Rival 300, it's kind of less of a
16:42 more perfect bell curve. The mean looks somewhere around 7.5, though. Moving
16:46 over to the actual math of things, it looks like the average is, yeah, about
16:50 7.6, which is a pretty significant increase, which is not good. You want a
16:55 lower number again. It's more like golf. So, it's not over yet. Let's try one
17:00 more. So, the program can only store two sets of data at a time. So, for round
17:04 three, I'm actually going to have to redo the G900. So, here we go. G900 test
17:09 two, and then the G900 test two will be
17:12 compared against the Razer Chroma.
17:16 Okay, so that's not really very good for
17:20 the other team. Uh, looking at the graphical average now, it's probably at
17:24 like a little bit below 11. And uh,
17:28 yeah, that's about right. 10.9 milliseconds compared to 4.29.
17:34 Sounds like wireless wins. Okay, that was actually kind of awesome. Uh, I'm
17:39 pretty convinced. Let me know what you guys think in the comments down below or on Twitter at Luke Lafr. Thank you to
17:45 Logitech for sending us here. If you guys want to see more stuff, click up
17:49 here. We've got videos that we show. Usually, this goes at the end of the
17:53 outro. This I'm switching the game up. It's going to the beginning this time.
17:56 Like the video if you liked it. Dislike the video if you disliked it. Be sure to stay subscribed so you can see all of
18:01 our Linus Tech Tips content. If you want to use our Amazon affiliate code, that's
18:04 cool. You can buy stuff. We get a fun kickback. It's all nice. If you want to check out the forum, become a
18:08 contributor there. That's super helpful. And if you want a shirt, check out the description down below. See you guys
18:12 next time.