en-AU.vtt

WEBVTT
Kind: captions
Language: en

00:00:00.001 --> 00:00:01.000
English (AU) (Spoken) [Manually Transcribed Captions]
github.com/WizardTim/WizardTim-captions

00:00:01.000 --> 00:00:05.118
So I have here a 48 MHz crystal oscillator from KDS

00:00:05.200 --> 00:00:09.920
which dates from probably the late 90s and it's in this quite small

00:00:10.719 --> 00:00:15.919
surface mount ceramic package and I'm quite interested to know what's actually inside one of these

00:00:16.400 --> 00:00:19.600
I think we all know what's inside these typical metal can

00:00:20.140 --> 00:00:23.519
crystals without any oscillator component. It's just the

00:00:24.160 --> 00:00:26.899
bare quartz crystal that's been metalized

00:00:27.840 --> 00:00:32.159
and I think some of us have probably seen videos about what's inside those

00:00:32.860 --> 00:00:34.860
through-hole metal can

00:00:34.960 --> 00:00:42.239
crystal 'oscillators' quite often they'll have either a normal PCB or a ceramic PCB and a 8-pin

00:00:42.879 --> 00:00:45.038
driver IC and the crystal.

00:00:46.480 --> 00:00:48.480
But these smaller devices

00:00:48.879 --> 00:00:55.139
I'm, not quite sure how they do it. Is there going to be a PCB in here or is it just the whole [metal can] quartz package?

00:00:55.840 --> 00:01:02.700
And is it just going to be direct wire bonding from the die to the quartz piece?

00:01:03.200 --> 00:01:09.440
Or have I just been lied to and this is actually a MEMs device. So let's just cut it open and take a squiz.

00:01:10.080 --> 00:01:14.959
So I think I should just be able to cut the lid off this thing with a pair of side cutters

00:01:15.119 --> 00:01:17.438
Although I'm going to do it off camera because it's a bit hard here

00:01:18.560 --> 00:01:21.920
So I've got the lid off and surprise surprise there's a crystal inside.

00:01:22.480 --> 00:01:26.560
Although I don't see the driver although I suspect it's under the crystal.

00:01:27.440 --> 00:01:32.240
It was a bit difficult getting the lid off because I thought this gold bit here this fillet

00:01:32.720 --> 00:01:40.000
I thought this was what was holding the lid on but actually it's directly under the lid on this silver bead here

00:01:40.000 --> 00:01:42.380
Which I think is probably solder [or a laser weld]

00:01:42.380 --> 00:01:46.420
And that's what I had to cut. I made a bit of a mess on the right hand side

00:01:46.480 --> 00:01:52.880
So taking a close look now at this angle, we can see just under the metalized section of the crystal some

00:01:53.439 --> 00:01:57.279
gold colored wire bonds as well as what looks like a

00:01:57.979 --> 00:01:59.439
silicon die

00:01:59.440 --> 00:02:06.300
So I suspect i'm probably gonna have to break this crystal to have a look at underneath it. Okay, so I think I can just

00:02:06.400 --> 00:02:10.880
lever this thing out and break off the silver epoxy on the left here,

00:02:12.080 --> 00:02:18.719
and yeah, that did it, and yeah there's the die. So I have my macro lens on now to take a closer look

00:02:19.440 --> 00:02:21.120
and we can see the

00:02:21.120 --> 00:02:28.000
Piece of quartz crystal was resting on three places here of the package two of these on the left

00:02:28.720 --> 00:02:30.700
Are the actual

00:02:30.700 --> 00:02:32.959
gold-plated pins that are

00:02:34.060 --> 00:02:36.479
Connected through a wire bond to the actual

00:02:37.200 --> 00:02:40.720
Die and the other one's just a plain piece of ceramic to hold it up

00:02:41.040 --> 00:02:44.640
I'm quite surprised actually that it's just held on one side

00:02:46.100 --> 00:02:50.560
So zooming in and taking a look at the actual silicon now

00:02:51.340 --> 00:02:56.080
Unfortunately, I can't zoom in very much so we won't be able to figure out exactly what's going on

00:02:56.959 --> 00:03:02.079
Although these large repetitive features here are definitely transistors

00:03:02.720 --> 00:03:06.239
This one here is probably an output totem pole transistor

00:03:06.959 --> 00:03:08.959
These might have something to do with

00:03:09.840 --> 00:03:13.039
The quartz driving probably is maybe part of a buffer

00:03:14.000 --> 00:03:20.959
This one here might be part of a regulator as this thing can operate off either 3.3 volts or 5 volts

00:03:21.680 --> 00:03:22.780
and

00:03:22.780 --> 00:03:26.000
These transistors here. I'm not too sure what these are doing

00:03:27.040 --> 00:03:31.540
This random looking mess over here is probably the logic that drives everything

00:03:32.159 --> 00:03:38.339
It's probably just a divider to divide the higher quartz crystal clock down to the 48 MHz

00:03:38.959 --> 00:03:42.479
It doesn't look like it's complicated enough to be a PLL or anything

00:03:43.920 --> 00:03:49.280
These large metallized features over the left here. I suspect this is either

00:03:50.379 --> 00:03:52.239
selective metallization

00:03:52.239 --> 00:03:57.359
During manufacture or maybe fuses that can be cut with a laser later on

00:03:58.080 --> 00:04:02.159
they're probably to select how many times the divider divides the

00:04:02.879 --> 00:04:04.879
crystal frequency

00:04:06.239 --> 00:04:10.719
It does look like there's some text and logos and probably a mask copyright here

00:04:10.799 --> 00:04:14.879
Although unfortunately, I can't zoom in enough to see exactly what they say

00:04:16.239 --> 00:04:20.078
So taking a closer look at the quartz crystal now

00:04:20.639 --> 00:04:24.579
You can see it's a pretty thin piece of core that's been cut into a rectangle

00:04:25.280 --> 00:04:33.199
And it's got that quite typical frosted look to it and that quite typical overlapping and metallized pattern on it

00:04:33.840 --> 00:04:34.940
although

00:04:34.940 --> 00:04:36.080
interestingly

00:04:36.080 --> 00:04:42.240
On the outside facing side of the crystal there's this sort of deposited something or rather on it

00:04:42.800 --> 00:04:48.240
I suspect in order for the crystal to be as accurate as they claim. This is probably part of the

00:04:48.960 --> 00:04:56.239
Factory calibration and trimming. I'm not quite sure how they do this. I reckon would be a very interesting machine that does this

00:04:57.759 --> 00:05:02.959
It would also explain why this is in a ceramic package because if in order to trim the device

00:05:03.440 --> 00:05:10.739
You have to put it in a sputtering machine at a [high] vacuum. You really don't want your plastic package to be out gassing everywhere

00:05:11.280 --> 00:05:15.359
And it also explains why having a lid to your package is so useful

00:05:17.280 --> 00:05:23.539
So that's what's inside one of these surface mount ceramic crystal oscillators, it's pretty interesting