1
00:00:06,320 --> 00:00:11,499
[Music]

2
00:00:15,679 --> 00:00:18,480
hello and welcome back to the open

3
00:00:17,039 --> 00:00:20,000
hardware mini conf

4
00:00:18,480 --> 00:00:22,400
now according to the original schedule

5
00:00:20,000 --> 00:00:25,199
steph piper was going to be doing a talk

6
00:00:22,400 --> 00:00:27,199
now but the amazingly talented stuff is

7
00:00:25,199 --> 00:00:28,000
unwell at the moment so i'm filling in

8
00:00:27,199 --> 00:00:29,359
and

9
00:00:28,000 --> 00:00:31,679
what i'm going to be talking about is

10
00:00:29,359 --> 00:00:34,000
when things go wrong so

11
00:00:31,679 --> 00:00:36,880
this whole presentation

12
00:00:34,000 --> 00:00:39,200
could fail terribly and i will still

13
00:00:36,880 --> 00:00:41,760
have succeeded i hope that's my theory

14
00:00:39,200 --> 00:00:44,399
if i set out to do a talk about failure

15
00:00:41,760 --> 00:00:45,200
and it fails then that's a win

16
00:00:44,399 --> 00:00:47,280
so

17
00:00:45,200 --> 00:00:48,320
there were a few difficulties that we

18
00:00:47,280 --> 00:00:50,320
faced

19
00:00:48,320 --> 00:00:52,800
oh and also just a reminder if you have

20
00:00:50,320 --> 00:00:55,920
questions and things i'm not necessarily

21
00:00:52,800 --> 00:00:57,440
going to see them live but

22
00:00:55,920 --> 00:00:59,840
drop them in the chat

23
00:00:57,440 --> 00:01:01,039
and

24
00:00:59,840 --> 00:01:02,960
mark said

25
00:01:01,039 --> 00:01:04,400
did i steal his slide of

26
00:01:02,960 --> 00:01:06,080
if you didn't design a fuse in your

27
00:01:04,400 --> 00:01:07,600
circuit the circuit will find one for

28
00:01:06,080 --> 00:01:10,479
you

29
00:01:07,600 --> 00:01:11,840
yes that is actually failure number two

30
00:01:10,479 --> 00:01:13,840
we're going to get to that mark we have

31
00:01:11,840 --> 00:01:15,200
a good example of that

32
00:01:13,840 --> 00:01:16,000
so

33
00:01:15,200 --> 00:01:16,880
uh

34
00:01:16,000 --> 00:01:19,280
this

35
00:01:16,880 --> 00:01:21,200
project and this collection of projects

36
00:01:19,280 --> 00:01:23,040
i should say because there for the

37
00:01:21,200 --> 00:01:24,240
hardware project this year we really had

38
00:01:23,040 --> 00:01:26,479
three

39
00:01:24,240 --> 00:01:28,080
loosely associated boards there is the

40
00:01:26,479 --> 00:01:29,600
swag badge

41
00:01:28,080 --> 00:01:32,560
the rockling which is the sailor that

42
00:01:29,600 --> 00:01:34,880
goes on top of it with the fpga and the

43
00:01:32,560 --> 00:01:37,280
party button which steph designed

44
00:01:34,880 --> 00:01:39,200
and we've had issues with all of them so

45
00:01:37,280 --> 00:01:41,520
i'm just going to talk about a couple of

46
00:01:39,200 --> 00:01:42,240
the things that we faced

47
00:01:41,520 --> 00:01:44,240
and

48
00:01:42,240 --> 00:01:45,680
ways we worked around

49
00:01:44,240 --> 00:01:47,280
those issues

50
00:01:45,680 --> 00:01:48,720
so i'm going to try to show you some

51
00:01:47,280 --> 00:01:51,360
live stuff

52
00:01:48,720 --> 00:01:52,880
i have zero slides it's just going to be

53
00:01:51,360 --> 00:01:55,119
looking at hardware and messing around

54
00:01:52,880 --> 00:01:56,880
with it and if we're really lucky we'll

55
00:01:55,119 --> 00:01:57,759
let some smoke out

56
00:01:56,880 --> 00:02:00,560
so

57
00:01:57,759 --> 00:02:01,840
uh where to begin okay so

58
00:02:00,560 --> 00:02:05,200
the um

59
00:02:01,840 --> 00:02:08,640
this is the rockling which is the little

60
00:02:05,200 --> 00:02:11,039
fish-shaped fpga board with the

61
00:02:08,640 --> 00:02:12,239
uh the theremin front end on it which

62
00:02:11,039 --> 00:02:15,200
was the

63
00:02:12,239 --> 00:02:17,440
outline of this was designed by andrew

64
00:02:15,200 --> 00:02:19,440
and the pcb layout was done by bob who

65
00:02:17,440 --> 00:02:21,599
you can see over my shoulder on

66
00:02:19,440 --> 00:02:23,120
this side

67
00:02:21,599 --> 00:02:25,200
and so

68
00:02:23,120 --> 00:02:27,200
there are some very interesting aspects

69
00:02:25,200 --> 00:02:29,680
to the design here

70
00:02:27,200 --> 00:02:30,879
which caught me out when it came to

71
00:02:29,680 --> 00:02:32,800
assembly

72
00:02:30,879 --> 00:02:34,400
so the first thing is that because this

73
00:02:32,800 --> 00:02:36,800
is being designed

74
00:02:34,400 --> 00:02:39,440
primarily well not primarily but one of

75
00:02:36,800 --> 00:02:41,440
the major aspects to it is the visual

76
00:02:39,440 --> 00:02:43,280
the way the board was designed we wanted

77
00:02:41,440 --> 00:02:46,080
this to be the front that is presented

78
00:02:43,280 --> 00:02:48,239
to you when it's in use

79
00:02:46,080 --> 00:02:50,239
unlike most pcbs we didn't want all of

80
00:02:48,239 --> 00:02:52,239
the parts to be on the visible side we

81
00:02:50,239 --> 00:02:56,000
wanted them to be on the back

82
00:02:52,239 --> 00:02:58,319
so andrew designed the layout of the pcb

83
00:02:56,000 --> 00:02:59,920
and then bob did the layout and putting

84
00:02:58,319 --> 00:03:01,519
all the parts on the back and he's going

85
00:02:59,920 --> 00:03:03,040
to talk about that

86
00:03:01,519 --> 00:03:04,480
right at the end of the mini conf so

87
00:03:03,040 --> 00:03:05,680
that'll be really interesting to see as

88
00:03:04,480 --> 00:03:08,159
well

89
00:03:05,680 --> 00:03:10,400
but one of the implications of that

90
00:03:08,159 --> 00:03:12,640
is that when the production files are

91
00:03:10,400 --> 00:03:14,800
generated out of keycad

92
00:03:12,640 --> 00:03:17,599
it has all of the parts on the back of

93
00:03:14,800 --> 00:03:20,640
the pcb and not on the front

94
00:03:17,599 --> 00:03:22,959
now that's not really a big deal as long

95
00:03:20,640 --> 00:03:25,200
as you are paying attention which

96
00:03:22,959 --> 00:03:26,159
apparently i was not

97
00:03:25,200 --> 00:03:28,080
and

98
00:03:26,159 --> 00:03:29,920
it led to a bit of a problem

99
00:03:28,080 --> 00:03:32,000
so what happened was that the production

100
00:03:29,920 --> 00:03:34,400
files were generated

101
00:03:32,000 --> 00:03:37,440
and i uploaded the gerbers

102
00:03:34,400 --> 00:03:39,840
to jlc pcb who did the fabrication of

103
00:03:37,440 --> 00:03:43,599
them so it's a four layer pcb

104
00:03:39,840 --> 00:03:45,440
and on the order form for the pcbs there

105
00:03:43,599 --> 00:03:47,680
is a little tick box where you can

106
00:03:45,440 --> 00:03:50,959
select whether you want a stencil

107
00:03:47,680 --> 00:03:53,360
so the solder paste stencil is used to

108
00:03:50,959 --> 00:03:55,439
used as a mask to control where solder

109
00:03:53,360 --> 00:03:57,760
paste is applied to the board

110
00:03:55,439 --> 00:03:58,640
i ticked that box didn't think any more

111
00:03:57,760 --> 00:04:01,120
of it

112
00:03:58,640 --> 00:04:03,840
and selected the solder mask

113
00:04:01,120 --> 00:04:06,159
for the top of the pcb

114
00:04:03,840 --> 00:04:08,720
and when it got back when it arrived i

115
00:04:06,159 --> 00:04:10,879
opened the box and this is the stencil

116
00:04:08,720 --> 00:04:12,879
they shipped me

117
00:04:10,879 --> 00:04:15,120
you may notice there's something missing

118
00:04:12,879 --> 00:04:18,479
a little thing called apertures apart

119
00:04:15,120 --> 00:04:21,199
from a couple here which are apertures

120
00:04:18,479 --> 00:04:23,600
that are on the bottom of the pcb

121
00:04:21,199 --> 00:04:25,120
all of the important stuff all of the

122
00:04:23,600 --> 00:04:27,280
apertures that we need

123
00:04:25,120 --> 00:04:29,280
for applying the solder paste

124
00:04:27,280 --> 00:04:31,840
but just not there

125
00:04:29,280 --> 00:04:35,240
it is entirely my own fault

126
00:04:31,840 --> 00:04:35,240
oh and

127
00:04:42,560 --> 00:04:46,479
we're back from that little video

128
00:04:43,840 --> 00:04:47,199
failure it is entirely my own fault

129
00:04:46,479 --> 00:04:49,440
for

130
00:04:47,199 --> 00:04:50,960
not being aware when i was ordering it

131
00:04:49,440 --> 00:04:52,720
now the thing is that these production

132
00:04:50,960 --> 00:04:55,199
boards we were starting to run a little

133
00:04:52,720 --> 00:04:58,800
bit short on time so there wasn't

134
00:04:55,199 --> 00:05:00,960
time to simply order a new stencil

135
00:04:58,800 --> 00:05:02,400
luckily bob remembered that we still had

136
00:05:00,960 --> 00:05:03,919
the stencil from the first round of

137
00:05:02,400 --> 00:05:04,800
prototypes

138
00:05:03,919 --> 00:05:06,320
so

139
00:05:04,800 --> 00:05:08,880
right at the start of the process we had

140
00:05:06,320 --> 00:05:11,840
ordered a small number of prototype pcbs

141
00:05:08,880 --> 00:05:13,759
and on that order the stencil was on the

142
00:05:11,840 --> 00:05:16,160
correct side of the board

143
00:05:13,759 --> 00:05:19,039
and this is the stencil right here so

144
00:05:16,160 --> 00:05:20,960
this is the stencil for the prototype

145
00:05:19,039 --> 00:05:23,199
now the difficulty is that the stencil

146
00:05:20,960 --> 00:05:25,440
for the prototype was not quite the same

147
00:05:23,199 --> 00:05:27,600
as the stencil that we needed

148
00:05:25,440 --> 00:05:29,440
so it's not just a simple matter of

149
00:05:27,600 --> 00:05:31,360
putting it down putting on the paste

150
00:05:29,440 --> 00:05:34,240
because some of the parts had changed

151
00:05:31,360 --> 00:05:35,680
and a couple of the parts had moved

152
00:05:34,240 --> 00:05:37,680
so

153
00:05:35,680 --> 00:05:39,280
we then had to try to figure out how to

154
00:05:37,680 --> 00:05:40,960
get the stencil and how to get the

155
00:05:39,280 --> 00:05:42,479
solder paste onto all of the boards in

156
00:05:40,960 --> 00:05:43,600
an efficient way

157
00:05:42,479 --> 00:05:46,080
and

158
00:05:43,600 --> 00:05:47,759
one thing i had heard of a story i'd

159
00:05:46,080 --> 00:05:50,240
heard many years ago

160
00:05:47,759 --> 00:05:52,560
was about people modifying solar

161
00:05:50,240 --> 00:05:54,800
stencils by

162
00:05:52,560 --> 00:05:56,720
stencils by soldering closed apertures

163
00:05:54,800 --> 00:06:00,000
that they don't want

164
00:05:56,720 --> 00:06:02,240
and so i thought why not give it a try

165
00:06:00,000 --> 00:06:04,800
however i very quickly discovered that

166
00:06:02,240 --> 00:06:06,960
it is pretty much impossible to solder

167
00:06:04,800 --> 00:06:09,039
stainless steel

168
00:06:06,960 --> 00:06:10,560
and when i tried i tried different

169
00:06:09,039 --> 00:06:12,960
fluxes and things i just couldn't make

170
00:06:10,560 --> 00:06:14,160
any solder stick i tried abrading the

171
00:06:12,960 --> 00:06:15,759
surface

172
00:06:14,160 --> 00:06:17,440
probably with the right flux and the

173
00:06:15,759 --> 00:06:20,000
right solder i could have made it work

174
00:06:17,440 --> 00:06:22,639
and i could have soldered closed the

175
00:06:20,000 --> 00:06:24,240
apertures that i don't want anymore

176
00:06:22,639 --> 00:06:26,319
but i failed

177
00:06:24,240 --> 00:06:28,160
so there was then a suggestion of how

178
00:06:26,319 --> 00:06:29,919
about just covering up apertures with

179
00:06:28,160 --> 00:06:31,840
tape and that's what i'm going to show

180
00:06:29,919 --> 00:06:33,280
you now so you can see on the back of

181
00:06:31,840 --> 00:06:35,360
this stencil

182
00:06:33,280 --> 00:06:38,319
there are some of the apertures closed

183
00:06:35,360 --> 00:06:39,120
up with kapton tape which is very very

184
00:06:38,319 --> 00:06:41,919
thin

185
00:06:39,120 --> 00:06:43,199
i was quite skeptical when i first heard

186
00:06:41,919 --> 00:06:44,639
of the idea

187
00:06:43,199 --> 00:06:46,800
and i thought well let's just give it a

188
00:06:44,639 --> 00:06:48,080
try we'll see what happens

189
00:06:46,800 --> 00:06:50,160
what i was worried about was the

190
00:06:48,080 --> 00:06:52,000
thickness of the state the tape holding

191
00:06:50,160 --> 00:06:53,919
the stencil off the pcb and allowing

192
00:06:52,000 --> 00:06:55,680
leakage underneath

193
00:06:53,919 --> 00:06:57,120
so

194
00:06:55,680 --> 00:06:59,199
if i

195
00:06:57,120 --> 00:07:01,199
spin over here and go to the overhead

196
00:06:59,199 --> 00:07:02,400
camera

197
00:07:01,199 --> 00:07:04,479
you can see here

198
00:07:02,400 --> 00:07:05,919
oops the the stencil

199
00:07:04,479 --> 00:07:07,840
and on the back

200
00:07:05,919 --> 00:07:10,240
you can see the kapton tape

201
00:07:07,840 --> 00:07:11,919
so what i did was use a scalpel and some

202
00:07:10,240 --> 00:07:13,840
tweezers and cut off tiny little pieces

203
00:07:11,919 --> 00:07:16,479
of tape and then just covered up the

204
00:07:13,840 --> 00:07:17,840
holes that i did not want paste to come

205
00:07:16,479 --> 00:07:19,919
through

206
00:07:17,840 --> 00:07:21,520
so that way it

207
00:07:19,919 --> 00:07:23,199
it wasn't putting paste down on the

208
00:07:21,520 --> 00:07:25,599
board where i don't want it

209
00:07:23,199 --> 00:07:28,319
and what you can see here is

210
00:07:25,599 --> 00:07:31,520
the um the vacuum bed this system was

211
00:07:28,319 --> 00:07:33,680
designed by ceon the unexpected maker so

212
00:07:31,520 --> 00:07:35,120
it's a really cool system the idea is

213
00:07:33,680 --> 00:07:37,120
that you connect a vacuum cleaner and

214
00:07:35,120 --> 00:07:38,639
this acts as a vacuum bed and sucks the

215
00:07:37,120 --> 00:07:41,120
stencil down

216
00:07:38,639 --> 00:07:43,680
onto the target board

217
00:07:41,120 --> 00:07:45,199
so and there is a cut out there

218
00:07:43,680 --> 00:07:46,879
this particular cutout was done by

219
00:07:45,199 --> 00:07:49,039
exporting a dxf

220
00:07:46,879 --> 00:07:49,919
out of the original design that andrew

221
00:07:49,039 --> 00:07:52,240
did

222
00:07:49,919 --> 00:07:55,360
and then bringing it into fusion 360

223
00:07:52,240 --> 00:07:56,479
applying the dxf as a shape onto this

224
00:07:55,360 --> 00:07:59,520
and then

225
00:07:56,479 --> 00:08:00,639
recessing it enough for the pcb to fit

226
00:07:59,520 --> 00:08:04,000
in there

227
00:08:00,639 --> 00:08:05,199
so what i was thinking of doing if i

228
00:08:04,000 --> 00:08:07,120
get this in approximately the right

229
00:08:05,199 --> 00:08:09,039
place now this is not at all how this

230
00:08:07,120 --> 00:08:10,400
would normally be used

231
00:08:09,039 --> 00:08:12,479
this is just sitting on my bench because

232
00:08:10,400 --> 00:08:15,039
i happen to have an overhead camera here

233
00:08:12,479 --> 00:08:17,199
and please excuse the back of my head

234
00:08:15,039 --> 00:08:18,319
but what i'm going to do

235
00:08:17,199 --> 00:08:20,720
is

236
00:08:18,319 --> 00:08:21,520
just put some solder paste on here

237
00:08:20,720 --> 00:08:24,080
and

238
00:08:21,520 --> 00:08:25,199
show you the end result

239
00:08:24,080 --> 00:08:26,720
so

240
00:08:25,199 --> 00:08:28,000
this is not actually going to be

241
00:08:26,720 --> 00:08:30,639
populated

242
00:08:28,000 --> 00:08:33,440
so i'm not really all that concerned

243
00:08:30,639 --> 00:08:34,640
this is just a test of the whole process

244
00:08:33,440 --> 00:08:37,599
of

245
00:08:34,640 --> 00:08:39,680
applying solder paste with some of the

246
00:08:37,599 --> 00:08:43,039
apertures masked out

247
00:08:39,680 --> 00:08:46,480
so grab my spatula

248
00:08:43,039 --> 00:08:48,080
scrape it across the stencil

249
00:08:46,480 --> 00:08:50,560
i'll just put that aside for now and try

250
00:08:48,080 --> 00:08:52,160
not to lose any of the expensive solder

251
00:08:50,560 --> 00:08:54,320
paste

252
00:08:52,160 --> 00:08:57,200
the pcb is stuck on the bottom there we

253
00:08:54,320 --> 00:08:57,200
go it's fallen off

254
00:08:58,399 --> 00:09:03,279
so there we have the pasted pcb and

255
00:09:01,920 --> 00:09:05,200
because i was just doing this without

256
00:09:03,279 --> 00:09:06,880
any proper alignment

257
00:09:05,200 --> 00:09:08,480
i suspect that these apertures are going

258
00:09:06,880 --> 00:09:09,920
to be all over the place

259
00:09:08,480 --> 00:09:11,519
but at least you can see there under the

260
00:09:09,920 --> 00:09:14,399
microscope

261
00:09:11,519 --> 00:09:17,440
we'll get the focus correct

262
00:09:14,399 --> 00:09:18,800
that there is paste on the pcb

263
00:09:17,440 --> 00:09:20,160
and if i

264
00:09:18,800 --> 00:09:21,760
move over

265
00:09:20,160 --> 00:09:25,040
where is an area

266
00:09:21,760 --> 00:09:26,480
well you can see there are some examples

267
00:09:25,040 --> 00:09:28,080
these particular

268
00:09:26,480 --> 00:09:30,160
pads

269
00:09:28,080 --> 00:09:32,880
just here don't have any solder paste on

270
00:09:30,160 --> 00:09:34,480
them because i masked that out on the

271
00:09:32,880 --> 00:09:36,399
stencil

272
00:09:34,480 --> 00:09:38,800
and you can see the alignment there is

273
00:09:36,399 --> 00:09:41,360
quite bad but also there are some parts

274
00:09:38,800 --> 00:09:43,120
down here that have no solder paste

275
00:09:41,360 --> 00:09:46,080
and down here

276
00:09:43,120 --> 00:09:48,160
there were also some parts that

277
00:09:46,080 --> 00:09:50,800
were on the old design that are not on

278
00:09:48,160 --> 00:09:51,839
this new one and what you can't tell is

279
00:09:50,800 --> 00:09:54,560
that they

280
00:09:51,839 --> 00:09:55,839
were also masked out because on this

281
00:09:54,560 --> 00:09:57,600
current system

282
00:09:55,839 --> 00:10:00,320
on this current version

283
00:09:57,600 --> 00:10:02,320
there is no footprint there anyway

284
00:10:00,320 --> 00:10:04,640
but if i had used that stencil without

285
00:10:02,320 --> 00:10:06,480
having that tape applied

286
00:10:04,640 --> 00:10:08,720
there would be blobs of solder in

287
00:10:06,480 --> 00:10:11,120
different random places on here

288
00:10:08,720 --> 00:10:13,839
where the old parts were

289
00:10:11,120 --> 00:10:15,680
so what this has left us with is a pcb

290
00:10:13,839 --> 00:10:17,360
that has solder paste on the vast

291
00:10:15,680 --> 00:10:19,040
majority of the pads

292
00:10:17,360 --> 00:10:20,000
horribly misaligned as you can see in

293
00:10:19,040 --> 00:10:21,760
this

294
00:10:20,000 --> 00:10:23,519
but at least it's on there and there are

295
00:10:21,760 --> 00:10:26,000
only a couple that are missing

296
00:10:23,519 --> 00:10:29,200
so then what i did was use a syringe to

297
00:10:26,000 --> 00:10:31,920
apply solder paste just to the pads

298
00:10:29,200 --> 00:10:35,519
that had been added in the new version

299
00:10:31,920 --> 00:10:37,040
and were not existing in the old stencil

300
00:10:35,519 --> 00:10:39,680
so

301
00:10:37,040 --> 00:10:41,600
that was a bit of a hacky way to do it

302
00:10:39,680 --> 00:10:43,440
but it actually worked out surprisingly

303
00:10:41,600 --> 00:10:44,720
well now i'm being a really bad

304
00:10:43,440 --> 00:10:47,200
advertisement for the system here

305
00:10:44,720 --> 00:10:48,160
because i got the alignment so wrong

306
00:10:47,200 --> 00:10:50,399
but

307
00:10:48,160 --> 00:10:51,279
at least you can see the general concept

308
00:10:50,399 --> 00:10:53,279
so

309
00:10:51,279 --> 00:10:55,360
if you have a situation where you need

310
00:10:53,279 --> 00:10:57,680
to make some changes to a stencil that

311
00:10:55,360 --> 00:10:59,760
you've already produced

312
00:10:57,680 --> 00:11:03,600
then you can possibly do that by closing

313
00:10:59,760 --> 00:11:06,640
up apertures that you don't need anymore

314
00:11:03,600 --> 00:11:08,240
so that was one of the issues we had to

315
00:11:06,640 --> 00:11:10,480
overcome

316
00:11:08,240 --> 00:11:11,680
so one of the other problems

317
00:11:10,480 --> 00:11:13,440
and this is coming back to mark's

318
00:11:11,680 --> 00:11:15,519
comment earlier which was

319
00:11:13,440 --> 00:11:18,160
if you do not design a fuse into your

320
00:11:15,519 --> 00:11:20,240
circuit the circuit will choose

321
00:11:18,160 --> 00:11:22,240
a fuse for you

322
00:11:20,240 --> 00:11:24,160
and

323
00:11:22,240 --> 00:11:25,920
that also comes down to an issue with

324
00:11:24,160 --> 00:11:27,920
something that we've seen

325
00:11:25,920 --> 00:11:30,000
on these boards as we started testing

326
00:11:27,920 --> 00:11:30,959
them

327
00:11:30,000 --> 00:11:33,360
so

328
00:11:30,959 --> 00:11:35,360
this particular board has

329
00:11:33,360 --> 00:11:37,920
failed initial testing so i know that

330
00:11:35,360 --> 00:11:40,320
this one is not going to do good things

331
00:11:37,920 --> 00:11:41,360
and if i turn on my

332
00:11:40,320 --> 00:11:43,040
handy

333
00:11:41,360 --> 00:11:44,480
multimeter here

334
00:11:43,040 --> 00:11:47,519
so this

335
00:11:44,480 --> 00:11:49,040
on-screen display is the multimeter that

336
00:11:47,519 --> 00:11:51,200
is on my bench

337
00:11:49,040 --> 00:11:53,120
and the power that is coming in here

338
00:11:51,200 --> 00:11:55,040
this usb cable

339
00:11:53,120 --> 00:11:56,399
is actually coming from my lab power

340
00:11:55,040 --> 00:11:58,079
supply

341
00:11:56,399 --> 00:12:00,000
and the multimeter is measuring the

342
00:11:58,079 --> 00:12:02,800
current that it's delivering

343
00:12:00,000 --> 00:12:04,240
now when these boards power up what they

344
00:12:02,800 --> 00:12:06,880
really should do

345
00:12:04,240 --> 00:12:08,399
is pull a tiny amount of power maybe a

346
00:12:06,880 --> 00:12:11,200
few milliamps

347
00:12:08,399 --> 00:12:14,000
because when they initially begin the

348
00:12:11,200 --> 00:12:15,600
fpga does not have anything loaded on it

349
00:12:14,000 --> 00:12:17,200
there is nothing actually running

350
00:12:15,600 --> 00:12:20,399
there's nothing executing

351
00:12:17,200 --> 00:12:22,480
so it's a trivial amount of power draw

352
00:12:20,399 --> 00:12:25,519
and this particular board one of the

353
00:12:22,480 --> 00:12:27,519
things we discovered in fact the way we

354
00:12:25,519 --> 00:12:29,279
tracked this problem down

355
00:12:27,519 --> 00:12:31,839
is we had a couple of these boards that

356
00:12:29,279 --> 00:12:33,839
failed testing but we didn't know why

357
00:12:31,839 --> 00:12:35,760
and andy took

358
00:12:33,839 --> 00:12:38,480
several of the boards home

359
00:12:35,760 --> 00:12:40,320
and put them in a bag and transported

360
00:12:38,480 --> 00:12:42,480
them home and then when he got home

361
00:12:40,320 --> 00:12:43,680
opened the bag and put the pieces out on

362
00:12:42,480 --> 00:12:46,480
the table

363
00:12:43,680 --> 00:12:48,000
and something small fell out and there's

364
00:12:46,480 --> 00:12:49,680
little silver part and i was wondering

365
00:12:48,000 --> 00:12:52,240
where does that come from

366
00:12:49,680 --> 00:12:54,079
and it turns out that it is the 48

367
00:12:52,240 --> 00:12:55,519
megahertz

368
00:12:54,079 --> 00:12:57,360
crystal

369
00:12:55,519 --> 00:13:00,959
which is

370
00:12:57,360 --> 00:13:02,079
where is it it's on the board

371
00:13:00,959 --> 00:13:03,200
right there

372
00:13:02,079 --> 00:13:05,279
it's over there

373
00:13:03,200 --> 00:13:07,600
um yeah so just here you can see that

374
00:13:05,279 --> 00:13:10,240
little silver part above my finger that

375
00:13:07,600 --> 00:13:12,560
is the 48 megahertz crystal which is

376
00:13:10,240 --> 00:13:14,160
used by the fpga

377
00:13:12,560 --> 00:13:15,920
and it had fallen off

378
00:13:14,160 --> 00:13:18,800
and our first thought was well there's

379
00:13:15,920 --> 00:13:20,399
been some terrible assembly error like

380
00:13:18,800 --> 00:13:22,320
i've simply failed to

381
00:13:20,399 --> 00:13:24,320
do the reflow properly

382
00:13:22,320 --> 00:13:25,680
and it's fallen off

383
00:13:24,320 --> 00:13:27,519
and then we started testing some more

384
00:13:25,680 --> 00:13:28,959
boards and we discovered

385
00:13:27,519 --> 00:13:30,320
uh to our surprise that they were

386
00:13:28,959 --> 00:13:32,320
pulling a large amount of current and

387
00:13:30,320 --> 00:13:35,920
getting very hot very quickly

388
00:13:32,320 --> 00:13:37,519
and demonstrating that fuse principle so

389
00:13:35,920 --> 00:13:39,199
with this particular one

390
00:13:37,519 --> 00:13:41,440
what i'm going to do now

391
00:13:39,199 --> 00:13:42,800
is attempt a recovery i don't know if

392
00:13:41,440 --> 00:13:44,320
this is going to work

393
00:13:42,800 --> 00:13:46,880
but let's try it

394
00:13:44,320 --> 00:13:48,480
so i'm going to turn on my lab supply

395
00:13:46,880 --> 00:13:50,560
just for a moment and there might

396
00:13:48,480 --> 00:13:52,399
actually be some smoke here

397
00:13:50,560 --> 00:13:54,880
so let's just see what happens

398
00:13:52,399 --> 00:13:56,399
watch the reading on the multimeter all

399
00:13:54,880 --> 00:13:58,160
right turn it on look you can see it's

400
00:13:56,399 --> 00:13:59,600
hidden amp

401
00:13:58,160 --> 00:14:02,800
and

402
00:13:59,600 --> 00:14:04,800
i don't see any smoke

403
00:14:02,800 --> 00:14:06,480
let's see if we can arrange some smoke

404
00:14:04,800 --> 00:14:08,320
one way to get smoke

405
00:14:06,480 --> 00:14:10,880
now because i happen to know where this

406
00:14:08,320 --> 00:14:12,240
heat is coming from is to put some flux

407
00:14:10,880 --> 00:14:14,639
on there

408
00:14:12,240 --> 00:14:17,360
so if i now turn the power supply back

409
00:14:14,639 --> 00:14:18,720
on again it hopefully will heat up the

410
00:14:17,360 --> 00:14:21,040
flux

411
00:14:18,720 --> 00:14:24,160
and we will get some smoke maybe i need

412
00:14:21,040 --> 00:14:24,160
to turn up the current limit

413
00:14:25,040 --> 00:14:27,680
yes i think i need to turn up the

414
00:14:26,240 --> 00:14:31,839
current limit let's

415
00:14:27,680 --> 00:14:31,839
give it two amps and see what happens

416
00:14:33,120 --> 00:14:36,480
no i think the current is being limited

417
00:14:34,639 --> 00:14:38,720
maybe in the cable

418
00:14:36,480 --> 00:14:40,800
but i can feel that it's getting warm

419
00:14:38,720 --> 00:14:43,760
so what was happening is that when we

420
00:14:40,800 --> 00:14:45,360
connected this to a laptop

421
00:14:43,760 --> 00:14:46,800
to andy's laptop

422
00:14:45,360 --> 00:14:48,560
it was supplying enough oh there is a

423
00:14:46,800 --> 00:14:50,160
little bit of smoke coming off

424
00:14:48,560 --> 00:14:52,000
yeah hang on i'll see if we can capture

425
00:14:50,160 --> 00:14:53,199
that i'll stuck it stick it under the

426
00:14:52,000 --> 00:14:54,720
microscope

427
00:14:53,199 --> 00:14:56,320
you can no longer see the multimeter but

428
00:14:54,720 --> 00:14:58,720
that doesn't matter

429
00:14:56,320 --> 00:15:00,720
but the um

430
00:14:58,720 --> 00:15:02,720
there for yeah you can just see a bit of

431
00:15:00,720 --> 00:15:03,440
smoke coming off there

432
00:15:02,720 --> 00:15:06,399
so

433
00:15:03,440 --> 00:15:08,480
the 48 megahertz crystal is getting hot

434
00:15:06,399 --> 00:15:10,000
enough that it is heating up that flux

435
00:15:08,480 --> 00:15:11,440
that i put on it and there is a bit of

436
00:15:10,000 --> 00:15:14,399
smoke coming off

437
00:15:11,440 --> 00:15:15,920
yes you need to see it drifting up there

438
00:15:14,399 --> 00:15:19,040
i do but i don't have an easy way of

439
00:15:15,920 --> 00:15:19,040
getting it onto the stream

440
00:15:19,680 --> 00:15:22,560
so it's still sitting there pulling an

441
00:15:21,120 --> 00:15:23,279
amp

442
00:15:22,560 --> 00:15:25,040
so

443
00:15:23,279 --> 00:15:27,519
what i was thinking of doing was maybe

444
00:15:25,040 --> 00:15:28,880
getting the hot air running

445
00:15:27,519 --> 00:15:30,639
and

446
00:15:28,880 --> 00:15:32,639
just to demonstrate that this is where

447
00:15:30,639 --> 00:15:34,399
the power is going

448
00:15:32,639 --> 00:15:36,240
so we saw that there was an amp of

449
00:15:34,399 --> 00:15:38,480
current being pulled then

450
00:15:36,240 --> 00:15:40,160
i am going to put this under the

451
00:15:38,480 --> 00:15:41,920
microscope

452
00:15:40,160 --> 00:15:45,360
run the hot air

453
00:15:41,920 --> 00:15:48,880
and remove that

454
00:15:45,360 --> 00:15:50,480
oscillator where is it right there

455
00:15:48,880 --> 00:15:51,600
so there will be a little bit of noise

456
00:15:50,480 --> 00:15:54,600
when i'm running this hot air

457
00:15:51,600 --> 00:15:54,600
unfortunately

458
00:15:58,000 --> 00:16:02,160
so effectively what was happening was

459
00:15:59,600 --> 00:16:05,360
that these were self-disassembling

460
00:16:02,160 --> 00:16:07,279
circuit boards they got hot enough

461
00:16:05,360 --> 00:16:09,199
that it melted the solder and that was

462
00:16:07,279 --> 00:16:12,399
what made the

463
00:16:09,199 --> 00:16:12,399
the oscillator fall off

464
00:16:14,880 --> 00:16:18,160
there's a very interesting project being

465
00:16:16,800 --> 00:16:19,759
done by

466
00:16:18,160 --> 00:16:20,959
that was worked on a little while ago by

467
00:16:19,759 --> 00:16:23,199
someone at the hackspace to do

468
00:16:20,959 --> 00:16:25,440
self-assembling pcbs

469
00:16:23,199 --> 00:16:27,440
which worked on a simple like principle

470
00:16:25,440 --> 00:16:29,360
you passed a lot of current through

471
00:16:27,440 --> 00:16:31,360
a layer of the pcb there we go so we've

472
00:16:29,360 --> 00:16:32,720
removed the crystal

473
00:16:31,360 --> 00:16:34,480
it passed a lot of current through the

474
00:16:32,720 --> 00:16:36,480
pcb and made it get hot so that you can

475
00:16:34,480 --> 00:16:38,000
have it self-assemble

476
00:16:36,480 --> 00:16:39,759
and now if i

477
00:16:38,000 --> 00:16:42,000
move this back over

478
00:16:39,759 --> 00:16:43,839
under the overhead camera

479
00:16:42,000 --> 00:16:46,880
and we can see the current

480
00:16:43,839 --> 00:16:50,079
the lab supplies currently turned off

481
00:16:46,880 --> 00:16:52,000
i'll plug this in turn on the lab supply

482
00:16:50,079 --> 00:16:54,959
and you can see there we're getting

483
00:16:52,000 --> 00:16:56,560
about 16 milliamps being pulled by the

484
00:16:54,959 --> 00:16:59,519
board

485
00:16:56,560 --> 00:17:00,560
so the short is passing through the

486
00:16:59,519 --> 00:17:02,399
crystal

487
00:17:00,560 --> 00:17:04,559
which is one of the last things that i

488
00:17:02,399 --> 00:17:06,880
would have expected to be happening

489
00:17:04,559 --> 00:17:08,720
but there you go sometimes you just got

490
00:17:06,880 --> 00:17:10,640
to look at where things get hot

491
00:17:08,720 --> 00:17:13,199
find that

492
00:17:10,640 --> 00:17:14,240
self-assigned

493
00:17:13,199 --> 00:17:16,000
fuse

494
00:17:14,240 --> 00:17:17,679
and then use that as a clue to guide you

495
00:17:16,000 --> 00:17:19,120
to where the problem is

496
00:17:17,679 --> 00:17:20,799
so the board itself still feels quite

497
00:17:19,120 --> 00:17:21,760
hot to me but that's just because i've

498
00:17:20,799 --> 00:17:24,240
been

499
00:17:21,760 --> 00:17:26,559
putting it under the the hot air

500
00:17:24,240 --> 00:17:27,679
but now if we put a new crystal onto

501
00:17:26,559 --> 00:17:29,360
here

502
00:17:27,679 --> 00:17:30,160
this board may well

503
00:17:29,360 --> 00:17:31,200
be

504
00:17:30,160 --> 00:17:33,360
uh

505
00:17:31,200 --> 00:17:34,320
may well be functional so how am i going

506
00:17:33,360 --> 00:17:36,960
for time

507
00:17:34,320 --> 00:17:38,320
how long have i been talking so far

508
00:17:36,960 --> 00:17:40,000
17 minutes

509
00:17:38,320 --> 00:17:42,720
i've got ages

510
00:17:40,000 --> 00:17:44,480
okay so what we could try doing is put

511
00:17:42,720 --> 00:17:48,160
another crystal on this

512
00:17:44,480 --> 00:17:50,559
i did not plan this far ahead

513
00:17:48,160 --> 00:17:54,080
so i don't have one handy um they are in

514
00:17:50,559 --> 00:17:55,440
a box over in the corner i believe

515
00:17:54,080 --> 00:17:57,919
start to check that i wasn't going to

516
00:17:55,440 --> 00:18:01,160
rip the microphone off something when i

517
00:17:57,919 --> 00:18:01,160
walked away

518
00:18:01,440 --> 00:18:06,640
so this is the box of parts

519
00:18:04,960 --> 00:18:07,360
that go into

520
00:18:06,640 --> 00:18:09,200
the

521
00:18:07,360 --> 00:18:11,440
rockling

522
00:18:09,200 --> 00:18:13,360
it's many many different parts and this

523
00:18:11,440 --> 00:18:16,080
isn't all of them either many of the

524
00:18:13,360 --> 00:18:18,080
really common jelly bean parts like 10k

525
00:18:16,080 --> 00:18:20,400
resistors and 1k resistors and 4k sevens

526
00:18:18,080 --> 00:18:22,400
and those sorts of things those are all

527
00:18:20,400 --> 00:18:25,120
reels on my pick and place machine

528
00:18:22,400 --> 00:18:29,120
so these are some of the unusual ones

529
00:18:25,120 --> 00:18:32,799
so in here we should have a crystal

530
00:18:29,120 --> 00:18:32,799
it's a microphone inverter

531
00:18:33,520 --> 00:18:40,919
and a regulator power switch

532
00:18:36,559 --> 00:18:40,919
no surely there is a crystal somewhere

533
00:18:42,720 --> 00:18:45,600
there

534
00:18:44,240 --> 00:18:47,600
not there

535
00:18:45,600 --> 00:18:49,120
no perhaps i don't have it here it might

536
00:18:47,600 --> 00:18:50,559
be somewhere else

537
00:18:49,120 --> 00:18:52,880
so

538
00:18:50,559 --> 00:18:52,880
also

539
00:18:53,440 --> 00:18:57,679
sure

540
00:18:54,400 --> 00:18:59,840
so let's ah let's put that aside for now

541
00:18:57,679 --> 00:19:01,760
i won't proceed with adding that extra

542
00:18:59,840 --> 00:19:02,880
part onto there

543
00:19:01,760 --> 00:19:03,919
okay

544
00:19:02,880 --> 00:19:04,880
so

545
00:19:03,919 --> 00:19:06,320
the

546
00:19:04,880 --> 00:19:07,200
third

547
00:19:06,320 --> 00:19:09,520
uh

548
00:19:07,200 --> 00:19:10,720
little area of failure that i want to

549
00:19:09,520 --> 00:19:12,720
talk about

550
00:19:10,720 --> 00:19:14,640
and problems that we have to resolve

551
00:19:12,720 --> 00:19:16,400
is in relation to

552
00:19:14,640 --> 00:19:17,360
the party button

553
00:19:16,400 --> 00:19:18,960
so

554
00:19:17,360 --> 00:19:20,640
this is the

555
00:19:18,960 --> 00:19:21,600
add-on

556
00:19:20,640 --> 00:19:23,840
which

557
00:19:21,600 --> 00:19:25,919
steph piper designed

558
00:19:23,840 --> 00:19:28,080
steph from elky education

559
00:19:25,919 --> 00:19:29,760
and this is based on a project that she

560
00:19:28,080 --> 00:19:30,559
did a i think it was a couple of years

561
00:19:29,760 --> 00:19:32,240
ago

562
00:19:30,559 --> 00:19:34,320
where she built a very large art

563
00:19:32,240 --> 00:19:35,360
installation which was like a pedestrian

564
00:19:34,320 --> 00:19:38,480
crossing

565
00:19:35,360 --> 00:19:40,480
and that's why it has this shape and the

566
00:19:38,480 --> 00:19:42,480
pedestrian crossing had a button on it

567
00:19:40,480 --> 00:19:44,000
the regular pedestrian crossing button

568
00:19:42,480 --> 00:19:46,080
but the idea was that when you walked up

569
00:19:44,000 --> 00:19:47,840
to it it was a party button and you

570
00:19:46,080 --> 00:19:49,600
would press it and lights would start

571
00:19:47,840 --> 00:19:52,559
flashing and things would happen on this

572
00:19:49,600 --> 00:19:54,240
big pedestrian crossing installation and

573
00:19:52,559 --> 00:19:55,440
so what we have here

574
00:19:54,240 --> 00:19:57,280
is a

575
00:19:55,440 --> 00:19:59,600
miniature version

576
00:19:57,280 --> 00:20:02,080
of her full-size

577
00:19:59,600 --> 00:20:03,280
two plus meter high i think it is art

578
00:20:02,080 --> 00:20:05,440
installation

579
00:20:03,280 --> 00:20:08,080
but it's in the form of a simple add-on

580
00:20:05,440 --> 00:20:09,120
that can go into a conference badge

581
00:20:08,080 --> 00:20:11,200
and there are a couple of interesting

582
00:20:09,120 --> 00:20:16,080
things about this design

583
00:20:11,200 --> 00:20:17,760
one is that you can plug it into a badge

584
00:20:16,080 --> 00:20:18,640
and it will receive its power from the

585
00:20:17,760 --> 00:20:20,640
badge

586
00:20:18,640 --> 00:20:22,640
and be controlled by it

587
00:20:20,640 --> 00:20:25,760
so there is a transistor here which

588
00:20:22,640 --> 00:20:27,679
turns on leds which will appear through

589
00:20:25,760 --> 00:20:29,280
the pcb

590
00:20:27,679 --> 00:20:32,000
but you can also

591
00:20:29,280 --> 00:20:34,799
use this little add-on stand

592
00:20:32,000 --> 00:20:37,200
which provides power

593
00:20:34,799 --> 00:20:39,039
to the sayo connector

594
00:20:37,200 --> 00:20:41,679
so if i plug that in and also it is

595
00:20:39,039 --> 00:20:44,000
mechanical so it allows it to stand up

596
00:20:41,679 --> 00:20:46,320
by itself so you can put that on the

597
00:20:44,000 --> 00:20:48,000
bench and it becomes self-supporting

598
00:20:46,320 --> 00:20:50,080
so if you press the pedestrian crossing

599
00:20:48,000 --> 00:20:51,760
button and hold it down

600
00:20:50,080 --> 00:20:54,000
you'll see that there are leds on the

601
00:20:51,760 --> 00:20:56,159
back that shine through the pcb so they

602
00:20:54,000 --> 00:20:58,640
are rear mounted leds

603
00:20:56,159 --> 00:21:00,159
and they are changing color over time

604
00:20:58,640 --> 00:21:01,440
these are leds that have built-in

605
00:21:00,159 --> 00:21:03,679
drivers

606
00:21:01,440 --> 00:21:05,360
that just cycle through different colors

607
00:21:03,679 --> 00:21:07,200
you don't need any logic or control

608
00:21:05,360 --> 00:21:09,520
system to run them

609
00:21:07,200 --> 00:21:11,760
so it's a pretty cool thing

610
00:21:09,520 --> 00:21:13,200
and by taking that off and applying it

611
00:21:11,760 --> 00:21:15,440
to the badge

612
00:21:13,200 --> 00:21:17,600
what you could do is have this turned on

613
00:21:15,440 --> 00:21:20,080
by logic within the batch because it can

614
00:21:17,600 --> 00:21:22,080
be controlled through these pins

615
00:21:20,080 --> 00:21:23,360
now one of the difficulties that we had

616
00:21:22,080 --> 00:21:24,559
with this one

617
00:21:23,360 --> 00:21:26,640
is that

618
00:21:24,559 --> 00:21:27,840
there were a couple of iterations of the

619
00:21:26,640 --> 00:21:31,200
design

620
00:21:27,840 --> 00:21:34,240
and the layout of the pcb

621
00:21:31,200 --> 00:21:36,400
and the schematic at some point i think

622
00:21:34,240 --> 00:21:38,640
may have diverged

623
00:21:36,400 --> 00:21:40,320
we're not quite sure yet what the

624
00:21:38,640 --> 00:21:41,760
the sequence was but it doesn't really

625
00:21:40,320 --> 00:21:43,840
matter at this point

626
00:21:41,760 --> 00:21:46,799
but what ended up happening was that we

627
00:21:43,840 --> 00:21:48,240
had a design that we thought was final

628
00:21:46,799 --> 00:21:49,520
based on testing that had been done on

629
00:21:48,240 --> 00:21:51,919
prototypes

630
00:21:49,520 --> 00:21:54,000
and then the pcbs were ordered and when

631
00:21:51,919 --> 00:21:56,720
they arrived they didn't work

632
00:21:54,000 --> 00:21:59,760
and it turned out that there was a

633
00:21:56,720 --> 00:22:00,799
misconnection on the pcb

634
00:21:59,760 --> 00:22:02,960
so

635
00:22:00,799 --> 00:22:06,720
going back across to

636
00:22:02,960 --> 00:22:06,720
the handy dandy microscope once again

637
00:22:08,559 --> 00:22:13,120
what we can see here are the tracks on

638
00:22:10,640 --> 00:22:15,200
the pcb you can see there

639
00:22:13,120 --> 00:22:17,679
there are some points here for

640
00:22:15,200 --> 00:22:18,720
an led there's a serviceman led option

641
00:22:17,679 --> 00:22:20,559
as well

642
00:22:18,720 --> 00:22:22,000
two options for transistors which is

643
00:22:20,559 --> 00:22:23,600
just connected in

644
00:22:22,000 --> 00:22:25,760
direct parallel so this is a dual

645
00:22:23,600 --> 00:22:27,039
footprint you can either use a surface

646
00:22:25,760 --> 00:22:29,360
mount part

647
00:22:27,039 --> 00:22:30,960
or you can put a through-hole part on

648
00:22:29,360 --> 00:22:32,159
there whichever you're more comfortable

649
00:22:30,960 --> 00:22:34,480
with

650
00:22:32,159 --> 00:22:36,240
and likewise these footprints are dual

651
00:22:34,480 --> 00:22:38,480
footprints

652
00:22:36,240 --> 00:22:41,039
it has holes in them here so that you

653
00:22:38,480 --> 00:22:43,039
can put a regular through-hole resistor

654
00:22:41,039 --> 00:22:44,720
and it has pads so that you can put a

655
00:22:43,039 --> 00:22:45,840
surface mount resistor on it if you want

656
00:22:44,720 --> 00:22:46,799
to

657
00:22:45,840 --> 00:22:48,960
now

658
00:22:46,799 --> 00:22:51,200
the difficulty or the change that we

659
00:22:48,960 --> 00:22:53,360
eventually track down is this trace here

660
00:22:51,200 --> 00:22:55,120
which comes off this connector

661
00:22:53,360 --> 00:22:56,960
of that pin on the connector

662
00:22:55,120 --> 00:22:59,360
right there comes down this part of the

663
00:22:56,960 --> 00:23:01,760
board comes across and then it goes

664
00:22:59,360 --> 00:23:03,039
through a via or via

665
00:23:01,760 --> 00:23:04,159
someone needs to tell me the correct way

666
00:23:03,039 --> 00:23:06,240
to pronounce that because i still don't

667
00:23:04,159 --> 00:23:07,520
know after all those years they go

668
00:23:06,240 --> 00:23:08,960
through that to the other side of the

669
00:23:07,520 --> 00:23:11,600
pcb

670
00:23:08,960 --> 00:23:14,400
it wanders up here and it comes out

671
00:23:11,600 --> 00:23:15,280
and joins on the other side

672
00:23:14,400 --> 00:23:17,520
up

673
00:23:15,280 --> 00:23:19,360
here somewhere anyway

674
00:23:17,520 --> 00:23:21,520
the issue is

675
00:23:19,360 --> 00:23:24,559
that this track

676
00:23:21,520 --> 00:23:25,919
goes to the wrong place so in order to

677
00:23:24,559 --> 00:23:28,240
fix that

678
00:23:25,919 --> 00:23:31,280
we need to do a couple of things

679
00:23:28,240 --> 00:23:32,400
now i have here my handy dandy little

680
00:23:31,280 --> 00:23:33,760
jewelers

681
00:23:32,400 --> 00:23:37,360
grinder

682
00:23:33,760 --> 00:23:39,280
so the way to apply this fix

683
00:23:37,360 --> 00:23:41,380
is that first

684
00:23:39,280 --> 00:23:43,600
we need to cut this track

685
00:23:41,380 --> 00:23:45,440
[Music]

686
00:23:43,600 --> 00:23:47,440
like that

687
00:23:45,440 --> 00:23:50,320
so that there is no longer a connection

688
00:23:47,440 --> 00:23:52,400
running up to the transistor and then

689
00:23:50,320 --> 00:23:55,120
the next thing we want to do

690
00:23:52,400 --> 00:23:56,880
is redirect that because we still need

691
00:23:55,120 --> 00:23:59,200
that connection to go through

692
00:23:56,880 --> 00:24:00,799
it's just that it was going to the

693
00:23:59,200 --> 00:24:03,039
source

694
00:24:00,799 --> 00:24:04,720
pin and we want it to go to the gate

695
00:24:03,039 --> 00:24:05,679
instead

696
00:24:04,720 --> 00:24:06,880
so

697
00:24:05,679 --> 00:24:08,960
what

698
00:24:06,880 --> 00:24:11,120
what i'd do to fix that

699
00:24:08,960 --> 00:24:13,360
just change the tips on my

700
00:24:11,120 --> 00:24:16,559
handy little grinder

701
00:24:13,360 --> 00:24:16,559
and just buzz off

702
00:24:16,640 --> 00:24:22,400
a little bit of the solar mask like that

703
00:24:20,400 --> 00:24:24,640
and then we've got some exposed copper

704
00:24:22,400 --> 00:24:28,720
that can be soldered onto

705
00:24:24,640 --> 00:24:31,039
so then a jumper can go from that point

706
00:24:28,720 --> 00:24:33,279
and it can come up and be soldered onto

707
00:24:31,039 --> 00:24:34,320
this pad here

708
00:24:33,279 --> 00:24:36,320
now

709
00:24:34,320 --> 00:24:38,480
you might be wondering why

710
00:24:36,320 --> 00:24:40,240
i would bother scraping back the solder

711
00:24:38,480 --> 00:24:41,919
mask and putting a jumper on from this

712
00:24:40,240 --> 00:24:43,360
point doing it right out here at the end

713
00:24:41,919 --> 00:24:45,279
of this track

714
00:24:43,360 --> 00:24:48,080
instead of doing it from here where

715
00:24:45,279 --> 00:24:50,080
there is a nice big pad

716
00:24:48,080 --> 00:24:51,760
and what i could do is run a jumper from

717
00:24:50,080 --> 00:24:53,440
that point

718
00:24:51,760 --> 00:24:56,080
up to here

719
00:24:53,440 --> 00:24:58,240
and not have to do that fiddly work of

720
00:24:56,080 --> 00:25:00,240
attaching it just on here

721
00:24:58,240 --> 00:25:02,720
now the reason for this is that we want

722
00:25:00,240 --> 00:25:04,559
to send these out as kits

723
00:25:02,720 --> 00:25:06,080
so that people can assemble them

724
00:25:04,559 --> 00:25:07,679
themselves

725
00:25:06,080 --> 00:25:09,279
now if you look at one here which has

726
00:25:07,679 --> 00:25:11,520
already been patched

727
00:25:09,279 --> 00:25:12,400
this one has the little jumper that runs

728
00:25:11,520 --> 00:25:14,640
from

729
00:25:12,400 --> 00:25:18,080
this point on the pcb

730
00:25:14,640 --> 00:25:19,760
and it runs up to this pad

731
00:25:18,080 --> 00:25:22,159
what we could do

732
00:25:19,760 --> 00:25:23,520
is solder from here but

733
00:25:22,159 --> 00:25:25,679
when it comes time there are two

734
00:25:23,520 --> 00:25:28,000
problems firstly when it comes time

735
00:25:25,679 --> 00:25:28,960
for someone to put the header into this

736
00:25:28,000 --> 00:25:30,640
point

737
00:25:28,960 --> 00:25:32,960
and then to

738
00:25:30,640 --> 00:25:34,000
solder it on the jumper would probably

739
00:25:32,960 --> 00:25:35,679
come off

740
00:25:34,000 --> 00:25:37,840
and that is if we hadn't filled that

741
00:25:35,679 --> 00:25:39,360
hole anyway in putting the jumper in

742
00:25:37,840 --> 00:25:40,880
place

743
00:25:39,360 --> 00:25:42,880
we would need to put the jumper in place

744
00:25:40,880 --> 00:25:44,559
in a way that doesn't fill the hole

745
00:25:42,880 --> 00:25:46,960
and allows the

746
00:25:44,559 --> 00:25:47,919
pin header to be installed

747
00:25:46,960 --> 00:25:50,480
so

748
00:25:47,919 --> 00:25:53,200
in planning this fix

749
00:25:50,480 --> 00:25:55,520
my one of my objection my objectives was

750
00:25:53,200 --> 00:25:57,440
to put the fix in place

751
00:25:55,520 --> 00:25:59,600
and not have it thermally

752
00:25:57,440 --> 00:26:01,440
impacted by other soldering that is

753
00:25:59,600 --> 00:26:03,360
going to be done on the board after the

754
00:26:01,440 --> 00:26:05,600
fix has been done

755
00:26:03,360 --> 00:26:07,279
so what we're going to be doing

756
00:26:05,600 --> 00:26:11,200
over the next little while

757
00:26:07,279 --> 00:26:13,840
is applying this patch to all of the 50

758
00:26:11,200 --> 00:26:15,679
or so of the party buttons that we have

759
00:26:13,840 --> 00:26:17,600
so that when they are shipped out to

760
00:26:15,679 --> 00:26:19,520
those who receive them they will get

761
00:26:17,600 --> 00:26:21,360
them in the patched state

762
00:26:19,520 --> 00:26:22,960
and then you can just put on the extra

763
00:26:21,360 --> 00:26:26,080
parts and assemble it as you normally

764
00:26:22,960 --> 00:26:28,159
would and when you put the parts on here

765
00:26:26,080 --> 00:26:30,720
and solder these connections

766
00:26:28,159 --> 00:26:33,200
because this solder joint is at the end

767
00:26:30,720 --> 00:26:36,320
of this long thin track

768
00:26:33,200 --> 00:26:38,559
the heat of doing this solder joint

769
00:26:36,320 --> 00:26:40,799
will begin to flow down this track

770
00:26:38,559 --> 00:26:43,200
but it's isolated enough that it should

771
00:26:40,799 --> 00:26:45,760
not melt this joint at the same time

772
00:26:43,200 --> 00:26:47,679
so that jumper will stay in place

773
00:26:45,760 --> 00:26:48,880
even while you are soldering this header

774
00:26:47,679 --> 00:26:51,520
in

775
00:26:48,880 --> 00:26:54,320
so that might be a bit of a strange

776
00:26:51,520 --> 00:26:56,000
design objective in terms of how that

777
00:26:54,320 --> 00:26:57,440
this patch was

778
00:26:56,000 --> 00:26:59,679
designed

779
00:26:57,440 --> 00:27:02,720
but you just have to work with

780
00:26:59,679 --> 00:27:04,559
whatever the circumstances are

781
00:27:02,720 --> 00:27:07,679
um so a couple of minutes is it all

782
00:27:04,559 --> 00:27:10,080
right okay in that case i think i had

783
00:27:07,679 --> 00:27:12,960
better jump to some questions what

784
00:27:10,080 --> 00:27:14,400
questions do people have

785
00:27:12,960 --> 00:27:15,919
we're talking about whether you could

786
00:27:14,400 --> 00:27:18,960
drill

787
00:27:15,919 --> 00:27:22,000
ah yes okay

788
00:27:18,960 --> 00:27:24,640
yes so this is referring to the solder

789
00:27:22,000 --> 00:27:25,520
stencil i assume

790
00:27:24,640 --> 00:27:26,720
yes

791
00:27:25,520 --> 00:27:28,880
so

792
00:27:26,720 --> 00:27:30,000
on this

793
00:27:28,880 --> 00:27:34,520
yeah

794
00:27:30,000 --> 00:27:34,520
on this pcb which

795
00:27:34,960 --> 00:27:38,480
has some pads that need solder applied

796
00:27:37,440 --> 00:27:41,200
to them

797
00:27:38,480 --> 00:27:42,559
one of the issues is then how do you put

798
00:27:41,200 --> 00:27:44,480
that solder on

799
00:27:42,559 --> 00:27:46,240
the way i have been doing that is just

800
00:27:44,480 --> 00:27:48,720
with a syringe

801
00:27:46,240 --> 00:27:50,799
so i use the stencil to apply the

802
00:27:48,720 --> 00:27:52,399
majority and then i syringe on for the

803
00:27:50,799 --> 00:27:54,880
extra parts

804
00:27:52,399 --> 00:27:56,880
but one thing that i was considering

805
00:27:54,880 --> 00:28:00,559
was drilling extra apertures into this

806
00:27:56,880 --> 00:28:03,520
pcb out onto the pcb into the stencil

807
00:28:00,559 --> 00:28:06,399
i have not actually done that

808
00:28:03,520 --> 00:28:07,919
and um part of that was a bit of a risk

809
00:28:06,399 --> 00:28:08,960
analysis thing

810
00:28:07,919 --> 00:28:10,720
because

811
00:28:08,960 --> 00:28:12,880
the benefit of doing it means that it

812
00:28:10,720 --> 00:28:15,360
saves me a little bit of time a very

813
00:28:12,880 --> 00:28:17,600
small amount of time with the syringe

814
00:28:15,360 --> 00:28:20,559
but the downside to getting it wrong is

815
00:28:17,600 --> 00:28:22,799
ruining the stencil and so i did not

816
00:28:20,559 --> 00:28:24,799
want to risk

817
00:28:22,799 --> 00:28:27,039
putting this under the drill press

818
00:28:24,799 --> 00:28:28,480
and drilling a couple of 0.8 millimeter

819
00:28:27,039 --> 00:28:30,880
holes or something in it

820
00:28:28,480 --> 00:28:32,480
and messing up the stencil

821
00:28:30,880 --> 00:28:34,480
but modifying the stencil is something

822
00:28:32,480 --> 00:28:36,880
that i would like to try in future

823
00:28:34,480 --> 00:28:39,200
so once we are beyond the

824
00:28:36,880 --> 00:28:41,679
mission critical aspect of this i'm

825
00:28:39,200 --> 00:28:43,279
probably going to grab an old stencil

826
00:28:41,679 --> 00:28:45,039
i've got a collection of stencils that

827
00:28:43,279 --> 00:28:47,279
are from obsolete designs

828
00:28:45,039 --> 00:28:49,440
and just try drilling and

829
00:28:47,279 --> 00:28:52,080
cutting different shapes into it

830
00:28:49,440 --> 00:28:54,720
and see if there is some way that i can

831
00:28:52,080 --> 00:28:58,159
modify the stencil to add apertures

832
00:28:54,720 --> 00:28:58,960
as well as to remove them

833
00:28:58,159 --> 00:29:01,919
so

834
00:28:58,960 --> 00:29:03,279
are there any other questions

835
00:29:01,919 --> 00:29:06,720
could you not cut on the other side of

836
00:29:03,279 --> 00:29:06,720
the board nearer to the transistor

837
00:29:07,679 --> 00:29:14,000
uh quite possibly let me have a look

838
00:29:12,080 --> 00:29:16,720
section okay

839
00:29:14,000 --> 00:29:18,320
so could i cut up closer to there on the

840
00:29:16,720 --> 00:29:20,159
other side of the pcb

841
00:29:18,320 --> 00:29:21,200
yes i can't remember where that goes oh

842
00:29:20,159 --> 00:29:24,799
still there

843
00:29:21,200 --> 00:29:24,799
so i could cut from there

844
00:29:24,840 --> 00:29:29,440
yes but

845
00:29:26,880 --> 00:29:30,320
i would still need somewhere

846
00:29:29,440 --> 00:29:32,080
to

847
00:29:30,320 --> 00:29:35,120
well actually no that would

848
00:29:32,080 --> 00:29:35,120
sorry just looking at this now

849
00:29:35,760 --> 00:29:40,640
if if we jump back to this view and i'll

850
00:29:38,399 --> 00:29:41,600
try zooming the microscope out so that

851
00:29:40,640 --> 00:29:43,760
we can

852
00:29:41,600 --> 00:29:45,520
get a better overall view of what's

853
00:29:43,760 --> 00:29:47,440
going on

854
00:29:45,520 --> 00:29:49,440
and turn down the brightness what's

855
00:29:47,440 --> 00:29:52,240
happening is that the

856
00:29:49,440 --> 00:29:53,840
the cut from here goes through this wire

857
00:29:52,240 --> 00:29:57,120
which comes through here and runs up

858
00:29:53,840 --> 00:29:59,919
this track to this pin on the transistor

859
00:29:57,120 --> 00:30:01,919
so what we could do is cut the track on

860
00:29:59,919 --> 00:30:03,919
this side and further up

861
00:30:01,919 --> 00:30:06,320
but the problem is that we still need

862
00:30:03,919 --> 00:30:07,520
somewhere to take a connection off this

863
00:30:06,320 --> 00:30:10,240
track

864
00:30:07,520 --> 00:30:12,080
and solder it onto

865
00:30:10,240 --> 00:30:13,360
either one of these surface mount pads

866
00:30:12,080 --> 00:30:15,520
or up here

867
00:30:13,360 --> 00:30:18,080
so we need a destination for it as well

868
00:30:15,520 --> 00:30:19,919
as a source for the connection

869
00:30:18,080 --> 00:30:21,279
and by changing the source to this side

870
00:30:19,919 --> 00:30:23,520
of the board it would actually make it

871
00:30:21,279 --> 00:30:26,000
harder because we would not have easy

872
00:30:23,520 --> 00:30:27,919
access to the destination which ideally

873
00:30:26,000 --> 00:30:30,159
is this existing surface mount pad right

874
00:30:27,919 --> 00:30:30,159
here

875
00:30:30,480 --> 00:30:34,399
last question was people who only know

876
00:30:32,960 --> 00:30:37,679
eagle pad do you think it's worth

877
00:30:34,399 --> 00:30:39,679
switching to keycard how do you find

878
00:30:37,679 --> 00:30:42,240
okay so

879
00:30:39,679 --> 00:30:43,279
for the design on this this was all done

880
00:30:42,240 --> 00:30:45,200
by bob

881
00:30:43,279 --> 00:30:48,240
and bob is probably a good person to

882
00:30:45,200 --> 00:30:51,360
have an opinion on this because he uses

883
00:30:48,240 --> 00:30:53,120
he doesn't use kicad as his main

884
00:30:51,360 --> 00:30:54,960
uh software he's used it for a few

885
00:30:53,120 --> 00:30:57,279
projects in the past but for his regular

886
00:30:54,960 --> 00:31:00,080
day-to-day contracting i think you use

887
00:30:57,279 --> 00:31:01,600
altium is that right bob yeah

888
00:31:00,080 --> 00:31:03,919
yeah so

889
00:31:01,600 --> 00:31:06,399
for me personally i use eagle for all of

890
00:31:03,919 --> 00:31:07,679
my day-to-day work bob uses ltm for his

891
00:31:06,399 --> 00:31:11,919
day-to-day work

892
00:31:07,679 --> 00:31:14,399
so neither of us are native kicad users

893
00:31:11,919 --> 00:31:15,519
and i've used it a little

894
00:31:14,399 --> 00:31:17,039
playing around with other people's

895
00:31:15,519 --> 00:31:19,919
designs

896
00:31:17,039 --> 00:31:22,320
i haven't made the jump into using it

897
00:31:19,919 --> 00:31:24,399
i've found in i know that there has just

898
00:31:22,320 --> 00:31:26,799
been a big new release that came out

899
00:31:24,399 --> 00:31:28,880
only you know a week or two ago

900
00:31:26,799 --> 00:31:30,799
so a lot of things like keyboard

901
00:31:28,880 --> 00:31:33,440
shortcuts and things have just changed

902
00:31:30,799 --> 00:31:35,919
in kicad so experiences that people have

903
00:31:33,440 --> 00:31:37,679
had with kicad up until about a week ago

904
00:31:35,919 --> 00:31:39,360
may not be valid anymore apparently

905
00:31:37,679 --> 00:31:40,960
things have improved significantly with

906
00:31:39,360 --> 00:31:41,840
that change

907
00:31:40,960 --> 00:31:43,679
and

908
00:31:41,840 --> 00:31:45,360
from my personal experience has been

909
00:31:43,679 --> 00:31:47,840
that i found the

910
00:31:45,360 --> 00:31:49,840
disjointed nature of the way the pcb and

911
00:31:47,840 --> 00:31:52,480
schematic tools work to be quite

912
00:31:49,840 --> 00:31:55,679
frustrating because i'm used to the

913
00:31:52,480 --> 00:31:58,640
highly integrated system that eagle uses

914
00:31:55,679 --> 00:32:00,080
and the idea of separating out those two

915
00:31:58,640 --> 00:32:01,039
things and having them very loosely

916
00:32:00,080 --> 00:32:03,039
coupled

917
00:32:01,039 --> 00:32:05,120
i've i found quite frustrating when i've

918
00:32:03,039 --> 00:32:07,200
been using kicad in the past

919
00:32:05,120 --> 00:32:08,559
but it is certainly a very capable piece

920
00:32:07,200 --> 00:32:10,799
of software

921
00:32:08,559 --> 00:32:12,640
and if anyone is wanting to get started

922
00:32:10,799 --> 00:32:13,840
with pcb design from a hobbyist

923
00:32:12,640 --> 00:32:17,519
perspective

924
00:32:13,840 --> 00:32:17,519
that is the thing that i would recommend

925
00:32:17,760 --> 00:32:19,840
okay

926
00:32:18,799 --> 00:32:21,679
so

927
00:32:19,840 --> 00:32:24,159
are we about out of time

928
00:32:21,679 --> 00:32:28,200
i think we're out of time so uh i'm

929
00:32:24,159 --> 00:32:28,200
going to wrap myself up now

930
00:32:32,559 --> 00:32:38,960
ah that is that is a really good point

931
00:32:35,120 --> 00:32:42,399
so i am not 100 sure yet but i think

932
00:32:38,960 --> 00:32:44,960
that the problem is that the pads

933
00:32:42,399 --> 00:32:46,559
under the crystals or the apertures for

934
00:32:44,960 --> 00:32:48,320
the crystal footprint is allowing a

935
00:32:46,559 --> 00:32:49,360
little bit too so much solder paste to

936
00:32:48,320 --> 00:32:52,159
go on

937
00:32:49,360 --> 00:32:54,399
and it is making contact with the body

938
00:32:52,159 --> 00:32:56,399
of the crystal so there is a short

939
00:32:54,399 --> 00:32:58,640
circuit that is happening around the

940
00:32:56,399 --> 00:32:59,440
edge of the crystal body

941
00:32:58,640 --> 00:33:00,960
and

942
00:32:59,440 --> 00:33:03,120
that's where the

943
00:33:00,960 --> 00:33:04,640
all that heat is coming from so once the

944
00:33:03,120 --> 00:33:06,000
crystal is removed there is no more

945
00:33:04,640 --> 00:33:09,279
short circuit

946
00:33:06,000 --> 00:33:10,799
but why that would be a problem is also

947
00:33:09,279 --> 00:33:11,760
a little bit of an unexplained thing at

948
00:33:10,799 --> 00:33:14,480
the moment

949
00:33:11,760 --> 00:33:16,399
because the way they are connected to

950
00:33:14,480 --> 00:33:18,960
the fpga

951
00:33:16,399 --> 00:33:21,279
uh typically with crystals they in this

952
00:33:18,960 --> 00:33:23,120
sort of format they have four pins and

953
00:33:21,279 --> 00:33:25,840
two of them will be ground pins and two

954
00:33:23,120 --> 00:33:27,200
of them are the pins that go to the rest

955
00:33:25,840 --> 00:33:28,240
of the circuit

956
00:33:27,200 --> 00:33:30,960
so

957
00:33:28,240 --> 00:33:34,240
those pins are normally very very low

958
00:33:30,960 --> 00:33:37,360
power they are just the

959
00:33:34,240 --> 00:33:40,080
signal that comes from the fpga

960
00:33:37,360 --> 00:33:42,000
which excites the crystal and makes it

961
00:33:40,080 --> 00:33:43,679
oscillate so

962
00:33:42,000 --> 00:33:46,000
we should be talking about

963
00:33:43,679 --> 00:33:48,159
nano amps or micro amps of power through

964
00:33:46,000 --> 00:33:51,519
those connections anyway

965
00:33:48,159 --> 00:33:54,480
so why shorting those pins in particular

966
00:33:51,519 --> 00:33:56,320
causes a dead short and very high

967
00:33:54,480 --> 00:33:59,279
current drain i don't yet have an

968
00:33:56,320 --> 00:33:59,279
explanation for that

969
00:33:59,360 --> 00:34:03,519
okay so thank you very much we're going

970
00:34:01,919 --> 00:34:04,559
to have a short break and then we're

971
00:34:03,519 --> 00:34:09,550
going to be back with the next

972
00:34:04,559 --> 00:34:14,439
presentation so we'll see you all soon

973
00:34:09,550 --> 00:34:14,439
[Music]