1
00:00:00,160 --> 00:00:06,799
In this bag is a piece of testing equipment so sophisticated and so

2
00:00:04,400 --> 00:00:13,040
expensive that for its cost I could literally hire a full-time butler. It

3
00:00:09,280 --> 00:00:15,759
has an arbitrary waveform generator,

4
00:00:13,040 --> 00:00:20,640
power analysis options, automotive protocol. Okay, to explain what the crap

5
00:00:18,320 --> 00:00:25,119
it is, we've got Lucas from the lab who's using it for our power supply

6
00:00:22,560 --> 00:00:30,160
testing processes and who can hopefully explain a little bit of what we're about

7
00:00:27,359 --> 00:00:36,520
to look at. All I know is that this is the Roden Schwarz MX058. 2 GHz, uh,

8
00:00:33,920 --> 00:00:40,320
eight analog channels, uh, 16 digital channels, good oscilloscope. The funny

9
00:00:38,960 --> 00:00:44,559
part is, if you break it down, an oscilloscope doesn't really do that

10
00:00:42,160 --> 00:00:49,280
much. Just like your multimeter at home, it measures voltage using probes. But

11
00:00:47,360 --> 00:00:54,239
what makes it special is the fact that it can show that voltage and changes in

12
00:00:51,760 --> 00:00:58,719
voltage over time. With this particular one being able to handle, what is it?

13
00:00:55,840 --> 00:01:02,239
4.5 million signals per second. That's a lot, right? That's a lot. Yeah.

14
00:01:00,480 --> 00:01:06,240
Waveforms. First, I want to have a look at the accessories. Sure. Yeah. And I'll

15
00:01:04,559 --> 00:01:10,240
try not to break anything. Okay. That's why he's here. Why don't you explain

16
00:01:08,080 --> 00:01:13,439
some of these probes to us? Sure. Yeah. So, all it comes with eight analog

17
00:01:11,760 --> 00:01:18,320
probes. These ones here like I believe they're 10 to one probes. Would you

18
00:01:15,200 --> 00:01:20,479
abbreviate that like an AL for the

19
00:01:18,320 --> 00:01:24,040
analog ones as opposed to like digi? You might say digi probes or like anal

20
00:01:22,560 --> 00:01:28,400
probes? No.

21
00:01:25,920 --> 00:01:32,400
So what do we use these ones for? So this will just be your like general

22
00:01:30,000 --> 00:01:35,680
probes for viewing any any waveform really. Um right but it gets the full

23
00:01:34,240 --> 00:01:41,040
you know bandwidth and we have digital probes as well but those will just be one or zero. This gets you know whatever

24
00:01:38,880 --> 00:01:44,720
voltages. Now this is pretty overkill for what we use it for. Is that correct?

25
00:01:42,880 --> 00:01:49,200
Yeah. Yeah. This is yeah is a lot for what we do. We just measure um the four

26
00:01:46,880 --> 00:01:53,200
or five rails on a power supply um and some inputs to it. Conceivably. Could we

27
00:01:51,280 --> 00:01:56,399
use this to look at something like memory signaling or is that too fast?

28
00:01:55,280 --> 00:02:02,399
That's probably too fast. There's fancier oscilloscopes for that. But we have used this for some LT projects like

29
00:01:59,439 --> 00:02:05,439
the Soviet keyboard. Um Oh, when we were trying to reverse engineer it. Yeah.

30
00:02:03,840 --> 00:02:08,959
Yeah. Yeah. RL and I worked on that pretty uh closely and like worked to

31
00:02:07,439 --> 00:02:13,360
Yeah. see what all the signals were saying and decode it. So that was pretty cool. That's super cool. There's a

32
00:02:11,920 --> 00:02:17,760
little more to these than just the probes themselves. I noticed these little color coding rings and whatever

33
00:02:16,480 --> 00:02:21,360
it is that you're putting back in that baggie. Yeah. Yeah. So, the color coding

34
00:02:19,840 --> 00:02:25,760
is just, you know, helpful for color coding.

35
00:02:23,080 --> 00:02:29,120
Um, this is a typical, you know, flying lead ground cable. So, obviously, you

36
00:02:27,440 --> 00:02:33,519
need to complete the circuit. Yeah. Um, it's got a little, um, wrench in here

37
00:02:31,280 --> 00:02:37,120
for controlling the capacitance of the cables. Why do I want to change the

38
00:02:34,959 --> 00:02:40,879
capacitance of my probe? To compensate for the inductance of the probe. So,

39
00:02:39,360 --> 00:02:44,000
because it's a wire, you know, all wires are have capacitance and inductance. Oh,

40
00:02:43,040 --> 00:02:48,480
yeah. Yeah, it's hard to explain, but you know, square waveform, it's uh you

41
00:02:46,720 --> 00:02:52,959
want to be exactly square, but any inductance or capacitance will skew

42
00:02:50,879 --> 00:02:57,840
that. So, uh cuz it takes time for voltage to change and rise. Um so, you

43
00:02:55,599 --> 00:03:01,840
can um compensate for that by adjusting the capacitance typically.

44
00:03:00,319 --> 00:03:06,319
Is that it for the accessories? Basically, it's just got lots of probes

45
00:03:03,760 --> 00:03:10,319
in it. It's got some vasa mounts, too. Oh, okay. So, you can mount it on a

46
00:03:09,040 --> 00:03:14,400
monitor. Yeah, you can like you can get an arm for your rack, too, to hang it

47
00:03:12,480 --> 00:03:18,480
off. Okay, cool. And then other than that, uh, manual, power cord. Let's have

48
00:03:17,040 --> 00:03:21,920
a look at the scope itself. And maybe you can show us some of the wizardry you

49
00:03:20,560 --> 00:03:29,200
do with it. Let's see how much of this I can figure out on my own. Oh, fuse,

50
00:03:25,159 --> 00:03:32,480
power, on, off switch. I got that far.

51
00:03:29,200 --> 00:03:37,360
Whoa. It has an SSD, I guess. I've never

52
00:03:32,480 --> 00:03:39,920
opened that. Remove. I don't know.

53
00:03:37,360 --> 00:03:43,680
What? What? What are you worried about? Never tried it. Well, it's first time

54
00:03:41,519 --> 00:03:51,560
for everything. Oh, cool. It's just an M.2 drive. I want to open

55
00:03:46,599 --> 00:03:54,159
it. What M.2 drive would Roden Schwarz

56
00:03:51,560 --> 00:03:58,319
trust for their oscilloscope. Oh, look at that. It's our many times sponsor,

57
00:03:56,080 --> 00:04:01,360
Kioxia. Nice. It's actually more important than you would think to choose

58
00:03:59,760 --> 00:04:06,400
a high quality drive for an application like this cuz I would think it would just be constantly overwriting or I

59
00:04:05,040 --> 00:04:10,080
guess a lot of it would actually go straight to RAM in memory probably.

60
00:04:08,080 --> 00:04:13,599
Yeah. and then only be output to the SSD under Yeah, this is only 256 gigs. Yeah,

61
00:04:12,480 --> 00:04:20,400
one of their advertising features is they have a ton of memory so they can store all those waveforms um for long

62
00:04:17,919 --> 00:04:26,160
history and long sample rate or fast sample rate and we have extra options on

63
00:04:23,840 --> 00:04:31,360
ours that expands the memory. Is that correct? Yeah. Yeah, I believe. Yeah. Uh

64
00:04:28,320 --> 00:04:34,800
all right. So Oh, I forget how this goes

65
00:04:31,360 --> 00:04:36,960
together now. H that may have been a

66
00:04:34,800 --> 00:04:40,160
tactical error. There are way faster oscilloscopes. I was talking to the the

67
00:04:38,960 --> 00:04:44,080
guy we know at Ro Sports. He's like, "Yeah, it's a good uh you know, standard

68
00:04:42,000 --> 00:04:48,479
lab oscilloscope, but there's definitely faster ones and more expensive ones. You

69
00:04:46,160 --> 00:04:52,320
know, they go into like like dozens of gigahertz and faster for communications

70
00:04:50,960 --> 00:04:58,240
and all that stuff." And do we have any idea what something like that would cost? Uh hundreds of thousands probably.

71
00:04:55,680 --> 00:05:02,080
Cool. Very uh very specialized. So this is like really fancy to us. And then

72
00:05:00,400 --> 00:05:06,800
we've got people who work at, you know, NASA or whatever that are like a Yeah.

73
00:05:04,639 --> 00:05:11,280
Yeah. Cute. Okay, let's try and put this SSD back in.

74
00:05:08,960 --> 00:05:16,000
Uh-oh. I know, Lucas. I'm trying to fix it. Okay, man. You seem stressed.

75
00:05:13,360 --> 00:05:19,520
Hopefully, it works. It Look, it just says do not remove during operation.

76
00:05:18,000 --> 00:05:23,600
Okay, it doesn't say don't. Did you think it just said do not remove? Did you not? I didn't read it. I didn't read

77
00:05:22,000 --> 00:05:27,680
it. All right, cool. What else are we looking at here? Well, obviously we've

78
00:05:25,520 --> 00:05:31,800
got rubber feet that allow it to be in I don't know, other orientations. They go

79
00:05:29,440 --> 00:05:37,039
all the way back. We've got a couple of 120. Jeez, this thing needs a lot of

80
00:05:34,880 --> 00:05:41,600
cooling. Is that just the processing needed for this level of sampling or

81
00:05:39,520 --> 00:05:45,199
what? It's doing a lot of compute. Um, it does dissipate a little bit of power

82
00:05:43,039 --> 00:05:49,360
inside as well, depending on what you're uh like probing. Oh, I see. Like not on

83
00:05:48,240 --> 00:05:54,639
purpose. Obviously, they would love to not take any power, but you know, in measuring, you have to take in a little

84
00:05:52,880 --> 00:05:58,639
bit of current or whatever. So, that makes sense. And then, oh, here we go.

85
00:05:57,120 --> 00:06:01,919
So trigger in that'll be if you want to have an external trigger to capture

86
00:06:00,240 --> 00:06:06,160
something specific like if you can't manage it in the software right and you

87
00:06:04,720 --> 00:06:10,800
want to capture something or if it's like something that happens unpredictably or it's too fast for you

88
00:06:09,199 --> 00:06:15,199
to like manually go oh yeah I want to capture now it's all too fast to manually capture now but you know it'll

89
00:06:13,840 --> 00:06:20,400
happen automatically and then gen one and gen two never use those but I believe those are just from the function

90
00:06:17,840 --> 00:06:24,880
generator the arbitrary waveform ref out 10 MHz that's also the function

91
00:06:22,639 --> 00:06:28,560
generator this one probably calibrate off of I'm not sure out is for when you

92
00:06:27,120 --> 00:06:34,759
trigger on something like one of the waveforms, then you can set a pulse up there. Got it. And a USB device. I'm not

93
00:06:32,720 --> 00:06:39,440
sure what that one's for. Well, it's a target port, so I guess if we wanted to

94
00:06:37,199 --> 00:06:43,840
hook it up to the computer via USB, that's possible. Meanwhile, on the other

95
00:06:41,280 --> 00:06:47,280
side, we can connect USB devices to it. Do we connect it to the network? I guess

96
00:06:45,360 --> 00:06:51,199
for Yeah, we do. Yeah. For all of our control and um for viewing. And then do

97
00:06:49,600 --> 00:06:55,520
you use the HDMI and DisplayPort out or do you just use the built-in display? Uh

98
00:06:52,960 --> 00:06:59,440
no, you can do it all over LAN uh like for the you can display from the

99
00:06:57,199 --> 00:07:03,599
computer. So you use LAN both for remote control of the unit and to output to a

100
00:07:01,759 --> 00:07:06,080
display. Uh yeah. Yeah, it's just for the same thing. But uh actually I was

101
00:07:04,960 --> 00:07:09,280
looking on the website today and they have headless units of this as well. So

102
00:07:08,080 --> 00:07:12,880
it's basically the exact same thing but without the screen. You can just use the

103
00:07:11,199 --> 00:07:16,319
display out or the land. Okay. But this one has a screen. This one does have a

104
00:07:14,400 --> 00:07:19,840
screen. Okay. Do you use the screen at all or is it just kind of superfluous?

105
00:07:18,000 --> 00:07:24,319
Yeah, I mostly use the screen. Uh, use the LAN just for monitoring if I'm, you

106
00:07:21,919 --> 00:07:28,400
know, at my other desk. Got it. So, there's your eight analogs and then I

107
00:07:26,560 --> 00:07:32,000
guess each of these does eight digital channels. Yes, that's correct. Yeah. Oh

108
00:07:30,240 --> 00:07:36,720
my god. Why does it have so many USB ports on it? We use it to save waveforms

109
00:07:34,160 --> 00:07:39,759
uh to to USB if you want. Um, or you can use it for some power. Can you run us

110
00:07:38,319 --> 00:07:42,639
through some of these controls here? Yeah, so a lot of them are multi-use and

111
00:07:41,680 --> 00:07:47,360
they'll do different things depending on the screen. And it's a big touch screen, so you can do a lot with that. Um, but

112
00:07:45,840 --> 00:07:52,160
you can control each of the eight channels plus any logic ones and math

113
00:07:49,919 --> 00:07:55,599
ones. Control the vertical scale and position to move waveforms up and down.

114
00:07:54,080 --> 00:08:00,479
Makes it a little easier to see what you're looking at, I guess. Yeah, I guess if you want to change view or see

115
00:07:58,240 --> 00:08:03,759
a certain section of something. Um, you have horizontal controls for horizontal

116
00:08:02,240 --> 00:08:07,120
scale or position. Why don't we plug it into something and then maybe you can

117
00:08:05,360 --> 00:08:10,639
show us this in real time then? Yeah. Okay. Yeah, let's do that. When it comes

118
00:08:08,879 --> 00:08:15,520
to this kind of stuff, they just kind of say, "Here's the one I want." And I go,

119
00:08:12,240 --> 00:08:16,879
"Is it a good one?" and they say, "Yep.

120
00:08:15,520 --> 00:08:21,840
Do you have like a power supply? Is that what we're going to look at or like what are we going to look at?" Oh, yeah.

121
00:08:19,280 --> 00:08:26,479
Well, do you want to start the um screen recording? Oh, yeah. Um and tell you

122
00:08:24,800 --> 00:08:29,759
about our sponsor. Thanks to Motion Gray for sponsoring this video. Their Erggo 2

123
00:08:28,479 --> 00:08:33,279
is one of the most affordable options for sit-tostand desks on the market,

124
00:08:31,520 --> 00:08:37,680
making your next work or gaming station upgrade even cheaper. It has a weight

125
00:08:35,360 --> 00:08:42,640
capacity of 176 lbs and a height adjustment range of 28 to 46 in. Powered

126
00:08:40,320 --> 00:08:45,440
by a silent but powerful electric motor. The whole setup even ships out in one

127
00:08:44,159 --> 00:08:51,120
box with all the tools you need for assembly included. Pick one up with our link for an exclusive 15% off on top of

128
00:08:49,200 --> 00:08:55,560
any discount they may already have. Hey, it's booted up. Now what? I had a little

129
00:08:53,360 --> 00:09:02,560
demo prepared. Ooh, a demo. Demolition. Uh, no. Demonstration.

130
00:09:00,480 --> 00:09:07,040
Okay, so we got a little SBC of some sort. What's that? Like an ESP32 I set

131
00:09:05,120 --> 00:09:11,279
up here and I've just programmed it to uh send a little command here. Okay. So,

132
00:09:10,080 --> 00:09:16,240
one of the neat things about this oscilloscope is that um it can do a lot

133
00:09:14,080 --> 00:09:21,200
of decoding of digital signals. Oh, so we tell it which pins to monitor out of

134
00:09:18,880 --> 00:09:24,800
our eight channels here. Uh yes. Yeah. Sorry. So, I plugged in the digital.

135
00:09:23,120 --> 00:09:30,560
This is again this is the thing about engineering people. They just start doing stuff assuming that everyone

136
00:09:27,600 --> 00:09:34,480
around them is like following along. Yes. Carry on. So, I've plugged in the

137
00:09:32,480 --> 00:09:39,920
digital probe here. Uh channel zero and one. Um, and I can go in here and go

138
00:09:36,720 --> 00:09:42,000
into the apps and protocol and to SPI.

139
00:09:39,920 --> 00:09:46,399
So, I'm comm communicating over SPI. Okay. And set this S clock on channel

140
00:09:44,880 --> 00:09:50,480
one, I believe. So, this is the kind of thing where if you didn't know how this

141
00:09:49,040 --> 00:09:54,800
was outputting, it would be a bunch of trial and error to figure out how to

142
00:09:52,480 --> 00:09:58,240
even monitor it like if you trying to reverse engineer something. Yeah. So, a

143
00:09:56,480 --> 00:10:01,680
lot of this like a lot of the setup it's done while you're testing. You know,

144
00:10:00,320 --> 00:10:05,920
some of these parameters, but you also have to just kind of find some of this.

145
00:10:04,240 --> 00:10:10,000
I have to also change the trigger so that it triggers onto that one. So,

146
00:10:08,480 --> 00:10:15,680
there's plenty of nice triggers in here, but we're just triggering off of a digital signal. Okay. And the edge. So,

147
00:10:13,600 --> 00:10:20,800
whenever it rises, then it will capture a waveform. And that's how we can see

148
00:10:18,079 --> 00:10:25,279
this here. Okay. And what are we looking at here? So, here it's got automatic

149
00:10:22,959 --> 00:10:28,720
decoding of this. So, we just capture a single waveform. You can just capture a

150
00:10:26,640 --> 00:10:32,240
single one instead of run stop as it was before capturing all the latest ones.

151
00:10:30,320 --> 00:10:36,160
Mhm. Um, you can see and it's just automatically decoding the message I was

152
00:10:34,079 --> 00:10:40,640
sending. Oh. Oh, Leila. Oh, look at that. It says PSU circuit. Yes. Which is

153
00:10:38,959 --> 00:10:44,480
the channel where you guys can see all of the power supply testing that Lucas

154
00:10:42,640 --> 00:10:49,040
does. What are the chances? A more practical case would be you have your

155
00:10:45,839 --> 00:10:51,360
PCB or some other device. Um, did you

156
00:10:49,040 --> 00:10:55,040
just make this be a keyboard essentially? I guess in a really slow

157
00:10:53,120 --> 00:10:58,399
way, yes, we can make this a pre keyboard. Yeah. Could Could you just

158
00:10:56,240 --> 00:11:02,240
type into Could it could it detect that as well if you set up your triggers so

159
00:11:00,399 --> 00:11:05,920
that it Yeah. And like decoding of whatever um communication standard

160
00:11:04,800 --> 00:11:11,120
you're using. Yeah. It's got a lot of different ones you can do like canvas and other things for more like

161
00:11:09,200 --> 00:11:14,000
automotive. Yeah. Exactly. Okay. That's and more complicated stuff. So this is

162
00:11:12,320 --> 00:11:19,200
like a really simple example, but if you want to see if your sensor was actually giving back the proper values, there's

163
00:11:17,360 --> 00:11:22,880
maybe a breakdown in your lines somewhere, you could monitor that and

164
00:11:21,120 --> 00:11:26,240
see, oh yeah, it is sending everything back correctly, just we're losing it or

165
00:11:24,560 --> 00:11:30,480
we're decoding it incorrectly on our own. Oh, that's super cool for

166
00:11:28,640 --> 00:11:34,320
diagnostics then. Yeah, exactly. And this is only using two of the 16

167
00:11:32,560 --> 00:11:38,800
channels. So if you had like a hugely parallel bus, then you can decode them

168
00:11:36,880 --> 00:11:43,279
all. So like this would be awful to, you know, count on your own and see one 0 01

169
00:11:41,680 --> 00:11:48,959
kind of thing. Right. Right. Right. Right. Cuz this is each of our characters here. Yeah. Yeah. It's also a

170
00:11:46,880 --> 00:11:53,600
touch screen. Right. Yep. You mentioned that. So we're back to our first signal

171
00:11:51,040 --> 00:11:59,600
here. This is a capital P. On the bottom, we have just our our clock that

172
00:11:57,120 --> 00:12:02,240
tells us when to sample a zero or one. And then on the top, we've got our

173
00:12:00,800 --> 00:12:08,240
signal. So, what we're looking at here then is our first clock telling us to

174
00:12:05,680 --> 00:12:13,360
sample. That's a zero. Then on our second sampling point, that's a one. On

175
00:12:10,959 --> 00:12:19,680
our third, it's a zero. On our fourth, it's a one. And for the last four, it's

176
00:12:15,920 --> 00:12:22,560
all zeros. So, this one would be 0 1 0 1

177
00:12:19,680 --> 00:12:25,360
0 0 0. And that's a capital P and ASKI then. Yep. So, then this would be how

178
00:12:24,160 --> 00:12:31,040
you guys reverse engineered that keyboard then. Yeah. A lot of that you have hook it up and see like when things

179
00:12:28,560 --> 00:12:33,680
are high and low. um and trying to figure out what all those random pins

180
00:12:32,720 --> 00:12:40,399
were cuz there were a couple we weren't even sure what they were. And I think that's how we found out that one of the

181
00:12:36,320 --> 00:12:42,399
decoding um like uh chips was broken and

182
00:12:40,399 --> 00:12:45,920
it was just sending all zeros like so it's sending the same kind of bits but

183
00:12:44,160 --> 00:12:49,839
then the second four bits was always zero for all of them right that I

184
00:12:48,160 --> 00:12:54,800
remember that the first half of each bite was working and then the second

185
00:12:52,480 --> 00:12:57,760
half of each bite was broken all zero because the chip was dead and so we

186
00:12:56,480 --> 00:13:02,160
could see that. All right, enough of the high school electronics class demo. Let's get this hooked up to a power

187
00:13:00,720 --> 00:13:06,560
supply and show you guys how we actually use it. So, these probes are all just

188
00:13:04,320 --> 00:13:09,920
like wired into our testing equipment. Yeah. Sorry, it's not uh you can't see

189
00:13:07,920 --> 00:13:13,920
it, but these run into the test chamber. Okay. Um connected to connection board,

190
00:13:12,240 --> 00:13:18,600
which is connected to the power supply here. I can show you guys what he's got

191
00:13:15,440 --> 00:13:21,680
in there. Here's our second

192
00:13:18,600 --> 00:13:25,519
one. So, all his probes are going into

193
00:13:21,680 --> 00:13:27,440
this and monitoring our various voltages

194
00:13:25,519 --> 00:13:32,000
from the power supply. And what's really cool is Lucas painstakingly soldered our

195
00:13:29,839 --> 00:13:35,519
measuring points to the bottom of our PCB so that we don't have to worry about

196
00:13:34,000 --> 00:13:38,800
this cable length affecting our measurements. What are we looking at?

197
00:13:37,200 --> 00:13:43,120
So, another great part about this oscilloscope is the um the

198
00:13:41,040 --> 00:13:47,440
automatability of it. So, we're able to do a lot of our tests um without, you

199
00:13:45,920 --> 00:13:51,120
know, actually touching a lot of it. We just have to set up all the programming

200
00:13:49,040 --> 00:13:55,920
beforehand and then as it's testing it'll send all the commands to properly

201
00:13:52,720 --> 00:13:57,519
configure it um and test without us

202
00:13:55,920 --> 00:14:02,720
having to stand here. But we really like standing here, don't we? We love it.

203
00:13:59,399 --> 00:14:06,440
Yeah. And hopefully this will work. I

204
00:14:02,720 --> 00:14:09,519
don't know. Somebody removed the SSD. So

205
00:14:06,440 --> 00:14:10,800
what what it's probably fine. Oh, should

206
00:14:09,519 --> 00:14:15,959
we talk about some of the things that ours is optioned with? Oh, sure. Yeah.

207
00:14:13,279 --> 00:14:22,000
It has a one gigabit GPTS memory extension. Uh yeah.

208
00:14:20,160 --> 00:14:25,760
So that's one gigap points memory extension. So that's another one that's

209
00:14:23,360 --> 00:14:29,680
lets you capture a lot more waveforms and it's a lot of memory like we said

210
00:14:27,120 --> 00:14:34,079
before. Oh, for longer waveforms. Yeah. Okay. And more of the same. So if you're

211
00:14:32,079 --> 00:14:37,839
measuring a really like we're measuring that thing of saying PSU circuit that

212
00:14:36,079 --> 00:14:42,320
was repeating like once a second or whatever, right? And so that allows to

213
00:14:39,519 --> 00:14:46,639
capture like multiple samples of it and you could average them or see if there's

214
00:14:43,839 --> 00:14:51,920
any like aberrations, right? So for diagnosis of say for example like like

215
00:14:49,120 --> 00:14:54,720
an an erratic error that only shows up once in a while that could be really

216
00:14:53,519 --> 00:15:00,720
useful then. Yeah, exactly. This is great for that and a lot of oscilloscope is used for that where um like 99% of

217
00:14:59,279 --> 00:15:03,839
your signals will be exact right waveform but you want to capture the one

218
00:15:02,320 --> 00:15:06,880
or two that is that isn't the intermittent problem. They call them

219
00:15:05,360 --> 00:15:10,480
like runs or whatever. Yeah. or it's like maybe it stayed high for too long

220
00:15:08,560 --> 00:15:16,519
or there's some kind of glitch. You know, there's a lot of people that find

221
00:15:12,160 --> 00:15:16,519
that word pretty offensive, right?

222
00:15:16,639 --> 00:15:23,760
This is our brown out test that we do for PSU circuits. Um, we basically just

223
00:15:22,160 --> 00:15:28,079
turn off the input power to the power supply for a very short period and see

224
00:15:26,160 --> 00:15:31,199
if the power supply survives. So, we can see here it's doing a lot of stuff on

225
00:15:29,519 --> 00:15:35,199
its own. It basically just configured itself so that it's got all the right

226
00:15:33,120 --> 00:15:38,959
measurements set up and it'll capture this waveform afterwards. So, okay,

227
00:15:37,760 --> 00:15:44,320
there we see here. So, what we're looking at here then is AC power going

228
00:15:41,839 --> 00:15:49,880
doing its thing, right? And then we're looking at Whoopsy Doodles. No AC power,

229
00:15:46,720 --> 00:15:52,000
but our 12vt on the power supply was

230
00:15:49,880 --> 00:15:56,880
uninterrupted. And what's our purple one here? Uh, yes, 3.3 volt. 3.3 volt.

231
00:15:55,040 --> 00:15:59,839
There's 5 volt and the power good signal. So that's the one from the power

232
00:15:58,560 --> 00:16:03,519
supply that's telling the motherboard and the computer like power is still

233
00:16:01,600 --> 00:16:06,800
good. I can still guarantee that we have the right output voltages. What if we

234
00:16:05,360 --> 00:16:10,480
gave it a longer brown out then? Exactly. We'll proceed through this.

235
00:16:08,720 --> 00:16:14,560
It'll go through and it saves the waveforms. This is how we save them back

236
00:16:12,560 --> 00:16:18,399
to a file and we can later graph those and create large graphics from them. And

237
00:16:16,720 --> 00:16:21,680
eventually that power good signal is going to be like no. Yes. Yeah.

238
00:16:20,480 --> 00:16:27,440
Hopefully we'll capture this. You know, it's always a lottery. We're doing it live. So we can see here it's in blue so

239
00:16:25,199 --> 00:16:30,959
it's hard but you see it cut it cut out here and went low for a while and came

240
00:16:29,279 --> 00:16:34,959
back up. So for that short period the power supply was saying you know I can't

241
00:16:32,800 --> 00:16:39,519
guarantee 12 volts output and if we have a good motherboard that behaves the way

242
00:16:36,880 --> 00:16:43,519
that it should it should say hey I don't have good power that'll shut down the

243
00:16:41,600 --> 00:16:47,040
computer. Yeah. So at that point it's like it's nice that it's kept the other

244
00:16:45,199 --> 00:16:50,320
voltages high but it's already turning off your computer because presumably

245
00:16:48,720 --> 00:16:53,839
that's enough for it to and this is exactly the behavior we'd want. We would

246
00:16:52,320 --> 00:16:57,199
want it to say, "Hey, the power's not good." before there's a major

247
00:16:55,600 --> 00:17:01,279
interruption to the voltage it's delivering to our components because

248
00:16:58,759 --> 00:17:04,799
otherwise, I mean, this is a very very small surge, but if we had a bigger

249
00:17:02,720 --> 00:17:09,199
surge, that could be an issue, right? Surge and just reliable things. So, it

250
00:17:06,640 --> 00:17:13,039
may be trying to save memory to uh like to, you know, storage or something and

251
00:17:11,199 --> 00:17:16,079
it might corrupt that or have some unpredictable effects. We'll see if

252
00:17:14,559 --> 00:17:21,360
we'll get one where it just fully drops out. We can see there's already little

253
00:17:18,160 --> 00:17:23,839
dips in the voltage, right? Oh, okay.

254
00:17:21,360 --> 00:17:28,319
So, it really didn't like 25 milliseconds of power being gone. About

255
00:17:26,000 --> 00:17:32,320
22 milliseconds or 24. I'm not sure exactly which one, but yeah, you can see

256
00:17:29,679 --> 00:17:36,240
it blips in the 12 volts and it just completely drops out, but the power good

257
00:17:34,559 --> 00:17:39,919
line dropped before that. Now, a lot of our testing is focused on the ATX

258
00:17:38,160 --> 00:17:44,000
specification and making sure that power supplies adhere to it. Have you

259
00:17:41,600 --> 00:17:48,160
encountered any that do not meet the ATX specification? Yeah, a lot of the power

260
00:17:46,000 --> 00:17:52,000
supplies will not quite meet in some places like oh maybe it's in the dynamic

261
00:17:50,240 --> 00:17:56,559
tests it doesn't quite meet the voltage regulations or one of the timings off.

262
00:17:54,640 --> 00:18:00,240
Um but a lot of them it's not a huge deal and it could be a part of our how

263
00:17:58,799 --> 00:18:04,160
our setup test setup is different from the ATX. Got it. So is that why we don't

264
00:18:02,240 --> 00:18:07,840
necessarily publish every result that we record right now cuz we're kind of

265
00:18:05,520 --> 00:18:11,120
waiting to make sure that it's on the up and up that we're in we're confident in

266
00:18:09,600 --> 00:18:17,280
it and we don't make any claims of like full ATX certification with that. So um

267
00:18:14,799 --> 00:18:20,320
yeah it's just you know indicative and one of the things that we really care

268
00:18:18,960 --> 00:18:24,480
about a lot though and that we are publishing now is any failures in safety

269
00:18:23,039 --> 00:18:29,200
mechanisms that are built into power supplies because you know I don't think

270
00:18:26,960 --> 00:18:34,240
that most people are going to look at a slight blip in the 12vt rail when they

271
00:18:31,360 --> 00:18:38,000
have a 15 millisecond power outage and go I'm not going to buy that power

272
00:18:36,160 --> 00:18:43,200
supply but we've had some that will just outright fail under an overcurren

273
00:18:40,480 --> 00:18:47,520
scenario where we are drawing more power than the power supply expects. Yeah.

274
00:18:45,200 --> 00:18:50,520
Which we don't expect it to be able to provide more power than it's rated for,

275
00:18:49,200 --> 00:18:54,480
but we do expect it not to catastrophically die. And we have had a

276
00:18:52,960 --> 00:18:58,000
couple die during this test where we're just shutting it off for brief periods

277
00:18:56,000 --> 00:19:01,760
of time. Can't explain those ones, but cuz it's not super ownorous, but uh

278
00:19:00,400 --> 00:19:04,880
yeah, like this is something that a power supply should be able to handle.

279
00:19:03,440 --> 00:19:10,080
Yeah, we do it a lot of times and we do it at 0 degrees, 20°, and 40°, but it

280
00:19:08,000 --> 00:19:13,120
should be, you know, fairly fairly survivable, I think. Do you have any

281
00:19:11,520 --> 00:19:17,280
more tests for us to look at? Yeah, we also have the timing test that we use the oscilloscope for. And I think

282
00:19:15,919 --> 00:19:21,039
that's, you know, more interesting than the brown one. When you say timing, what

283
00:19:19,600 --> 00:19:25,520
what do we want to know? What how does the timing matter of a power supply? I

284
00:19:22,799 --> 00:19:28,559
mean, it's DC. There's no signaling. There's no waveforms even. There is

285
00:19:27,039 --> 00:19:32,799
signaling. So, there's a couple signal channels on that like the power good and

286
00:19:30,559 --> 00:19:36,640
the power on, but there's specific time they have for how long it takes for the

287
00:19:34,400 --> 00:19:41,280
voltage to rise to a certain level or to drop or delays between those. So this

288
00:19:39,440 --> 00:19:46,559
configures the oscilloscope for a turn on waveform which we get here. So do we

289
00:19:43,760 --> 00:19:49,919
want to go from 0 to 12 volts really quickly or do we want it to be slow?

290
00:19:48,480 --> 00:19:56,080
What do we want? I believe there is a maximum time I can I don't know the exact timing now but the maximum time

291
00:19:52,720 --> 00:19:58,480
for 0 to 90% of the voltage spec. So of

292
00:19:56,080 --> 00:20:03,039
12 volts 5 volts and 3.3 volts. Got it. And then there is also a spec for the

293
00:20:00,080 --> 00:20:08,000
delay between them going high and the um power good signal going high. it'll save

294
00:20:05,520 --> 00:20:13,360
this waveform and then reconfigure so that it can um save the turn off

295
00:20:10,600 --> 00:20:16,880
waveform which is also important in for like brownout and stuff. So this one the

296
00:20:15,280 --> 00:20:20,480
first one will be basically just flat lines once it's saved because there's

297
00:20:18,799 --> 00:20:25,280
very little load on the power supply. Yeah. Um but future ones um do actually

298
00:20:23,520 --> 00:20:30,960
drop and you can see them there. It takes about two days to fully test a

299
00:20:28,400 --> 00:20:35,679
power supply. And if we have a failure, we always obtain another unit and fully

300
00:20:34,000 --> 00:20:40,320
test that additional unit to see if it was just a one-off bad unit. What's

301
00:20:38,080 --> 00:20:45,120
interesting to me though is so far we haven't had a lot of bad units that

302
00:20:42,400 --> 00:20:49,280
weren't just a bad power supply. All but I believe one of the ones that have

303
00:20:47,120 --> 00:20:52,880
failed, the second unit has failed as well. Yeah, there's one or two that the

304
00:20:51,679 --> 00:20:57,520
first one failed and the second one survived, but so far it's been Yeah, they both die and often run on similar

305
00:20:56,080 --> 00:21:01,520
tests. So, here's another one. So, this is trying to turn on with a load and

306
00:20:59,600 --> 00:21:04,720
obviously it failed that. Um, a lot of power supplies do even though I think

307
00:21:03,200 --> 00:21:08,880
it's in the spec, the ATX spec that have to be able to turn on with a load. Um,

308
00:21:06,640 --> 00:21:13,200
they just can't because you know it's not typical. I guess your GPU isn't

309
00:21:11,120 --> 00:21:16,640
pulling power before it has, you know, a voltage. We can also use this for ripple

310
00:21:15,200 --> 00:21:21,120
testing of a power supply. Is that correct? Yes. Yeah, we do that same. Can

311
00:21:18,480 --> 00:21:26,240
you explain like on five why ripple matters? Because a perfect DC voltage

312
00:21:23,760 --> 00:21:30,080
output like 12 volts, it'll be perfectly flat, but because we're in the real

313
00:21:28,159 --> 00:21:34,240
world, there's capacitance, inductance, lows, everything else, it's not going to

314
00:21:32,400 --> 00:21:37,440
be flat. Um, there's some ripple to that. We can see that. We can see it.

315
00:21:36,400 --> 00:21:43,360
Little little bit of noise in there in the line there. Why does excessive ripple matter? because the components

316
00:21:41,600 --> 00:21:47,760
later in your like in your computer are expecting a really constant um supply

317
00:21:46,159 --> 00:21:52,640
and something they can predict and you know operate um consistently on and they

318
00:21:50,799 --> 00:21:56,559
further regulated down to other voltages like CPU like 1.2 volts and stuff but

319
00:21:55,039 --> 00:22:00,480
even those regulators want a really consistent input voltage. Got it. So it

320
00:21:58,799 --> 00:22:04,960
just helps with general stability and reliability. Could it improve longevity

321
00:22:02,720 --> 00:22:08,799
to have low ripple? Uh yeah it'll put less load on the capacitors over time.

322
00:22:07,280 --> 00:22:12,799
Put less energy in and out of them. Here I have another way from where this one actually went down. So this is another

323
00:22:11,360 --> 00:22:17,360
one. So during the timing test, we can measure how long it is between the power

324
00:22:15,200 --> 00:22:20,559
like it turning off and the um power supply shutting down and how long it

325
00:22:19,039 --> 00:22:24,720
takes for each of the voltages to drop. This one's at like a 10% or 20% load. So

326
00:22:23,440 --> 00:22:30,559
this is the whole thing where you hit the power switch on something and it takes a minute for it to like for the

327
00:22:28,559 --> 00:22:33,840
LEDs to you know Yeah, exactly. So that' be like for a small load and you'll see

328
00:22:32,240 --> 00:22:38,159
with like even larger loads, it's almost instantaneous the the voltage drops. Oh,

329
00:22:36,320 --> 00:22:44,640
one more thing. Actually, we've alluded to it, but I don't think we've explicitly said what one of these

330
00:22:41,960 --> 00:22:47,919
run-of-the-mill benchtop oscilloscopes is worth. It's a bit of a range

331
00:22:46,159 --> 00:22:54,960
depending on options, but it's about 20K to 50K US,

332
00:22:50,559 --> 00:22:58,559
right? Yeah. Go and buy one now. Yeah.

333
00:22:54,960 --> 00:23:02,600
Use our affiliate code. Lucas from the

334
00:22:58,559 --> 00:23:05,679
lab, everyone. And the Roden Sports MX05

335
00:23:02,600 --> 00:23:08,960
series. Subscribe to ShortCircuit.

336
00:23:05,679 --> 00:23:08,960
and power supply circuit.