Up next in 10
My Rebreather Failed on a 90 Metre Dive
Oct 19, 2025
What would you do if your life support started failing at 90 meters underwater? During this technical dive on an uncharted shipwreck off the British coast, one of my rebreather's oxygen cells began gradually failing. At 90 meters depth, there's no room for error - and no quick way to the surface. In this video, I'll walk you through exactly what happened: how I detected the failing cell through my HUD readings, the critical decisions I had to make in real-time, and how my CCR training and bailout procedures quite literally saved my life. You'll see the actual dive footage, my dive computer data, and hear my thought process as the situation unfolded. This is why redundancy matters. This is why we train for emergencies. And this is why technical diving demands respect. *IMPORTANT WARNING* The decisions I make in this video will not be suitable for all circumstances. If you are diving a rebreather and do not know what the PO2 in your loop is then you should bailout!! ⚠️ LESSONS FOR TECHNICAL DIVERS: - Pre-dive cell checks are non-negotiable - Recognizing gradual failures vs sudden failures - When to abort vs when to continue - Importance of linearity checks 📍 LOCATION: Unknown shipwreck, British Isles 🌊 CONDITIONS: Nearly 90 metres deep, 3-4 metres visibility
View Video Transcript
0:00
90 m down on an unknown shipwreck
0:05
and one of my oxygen cells starts
0:07
failing.
0:09
At this depth, there's no quick way up.
0:12
Your rebreather is your life support and
0:14
mine was telling me that something was
0:16
very wrong. I'm Dom Robinson and in this
0:20
video I'm going to walk you through
0:22
exactly what happened during this dive,
0:24
how I detected the failure, the
0:27
decisions I had to make, and the
0:29
procedures that got me back to the
0:31
surface safely.
0:34
Because of course, this is the reality
0:36
of this sort of diving. Equipment can
0:38
fail and does fail. And when it does,
0:42
you've got to make decisions. And those
0:44
decisions can be the only thing between
0:46
you and disaster. So let's see what
0:50
happened. But first, let's have a look
0:52
inside my rebreather. So this is the
0:54
clever bit. This contains all the
0:56
electronics and also has three of these
0:58
oxygen cells. And what they do is they
1:01
measure the partial pressure of oxygen
1:03
in and around here in the mixing chamber
1:05
before it gets supplied to me to
1:07
breathe. So they're absolutely crucial.
1:10
And this is one up close. They don't
1:12
look like much, but you can see they're
1:14
individually serialn numbered. They've
1:16
got dates on them. They do have a finite
1:18
life, though, and the recommendation is
1:21
normally 18 months. This one was just
1:23
over 12 months old, and I'd had a great
1:26
year of diving with it, so I wasn't
1:28
expecting it to go wrong. I'd also
1:30
completed the pre-dive check, which
1:32
consists of calibrating the oxygen cells
1:34
against 100% oxygen. All three cells for
1:38
this dive had gone through that
1:39
absolutely perfectly. So once again,
1:42
there was no indication that something
1:44
was going to go wrong. They were all
1:46
fine all the way through the descent.
1:48
And as you can see here, although I've
1:50
not done a linearity check per se, you
1:52
can see that they've all shown that
1:54
they're performing well outside the uh
1:56
set point of 1.3. And even for the first
1:59
10 15 minutes on the bottom, everything
2:02
was absolutely fine. For those who were
2:04
unfamiliar with the AP Inspiration
2:06
handset, you can see it in the bottom
2:08
left hand side of my screen. and the the
2:10
top row with the three numbers 1.3 1.4
2:14
those are the oxygen cell readings. So
2:16
the the P2 and as you can see at the
2:19
moment they're about 1.3ish
2:22
occasionally the middle one is flicking
2:23
up to about 1.4. Now the way that a
2:26
rebreather works is it uses something
2:28
called voting logic. So it takes the two
2:31
cell readings that are closest together
2:34
and assumes that those provide the P2
2:37
and that P2 will be used to do two
2:39
really important things. The first thing
2:41
is control decompression. But the really
2:44
important one is that controls the
2:47
injection of oxygen. So provided that
2:50
two cells are right, it is going to
2:53
everything is going to be okay. The
2:55
nightmare scenario is when only one cell
2:58
is right and the voting logic uses the
3:01
two wrong cells and that can be super
3:03
super dangerous. So what as a diver you
3:06
have to do is decide whether you have
3:09
one cell that is okay or you have two
3:11
cells that are okay. Now obviously what
3:14
you really want is you want three cells
3:15
that are okay. And that is the situation
3:18
most of the time. But on this dive, I
3:21
started to notice that there was a
3:24
problem. I started to see that one cell
3:26
was reading a bit higher than the other
3:28
two. And that's when I started to become
3:30
a little concerned. Now, this does
3:32
happen. So, it wasn't a huge concern,
3:35
but I could clearly see that two cells
3:38
were pretty similar, pretty nailed
3:40
together, and the other one was starting
3:42
to just be a bit higher. You can see it
3:45
here. It's the middle cell. Uh, it's
3:47
currently about 1.42. 42. The other ones
3:49
are about 1.3 or thereabouts.
3:53
And this, of course, is where decision-m
3:54
comes into it because if you were to be
3:58
on a rebreather course, the instructor
4:00
would tell you if you see that a cell
4:02
fails, you should do a dealing flush,
4:04
you should consider bailing out and you
4:06
should have bought the dive. I'm sure it
4:08
won't surprise anybody who's watched any
4:10
of my videos and the kind of experiences
4:12
I had to realize that that isn't what I
4:15
am going to do. And let's talk through
4:17
that decision process. Well, the first
4:19
thing, of course, is that this isn't an
4:22
unsafe situation. I just have a cell
4:24
that is reading a bit higher than
4:26
another. So, that in itself is not is
4:29
not a huge concern. Even if the actual
4:32
P2 in this uh rebreather was 1.46. So,
4:35
the the one that's reading high is
4:37
correct. Once again, that isn't uh the
4:39
end of the world. The other thing, this
4:43
wreck is 45 mi offshore. I've spent a
4:47
lot of time and effort and frankly money
4:49
to get here and therefore I want to get
4:52
the most from myment. So unless it is
4:55
something that I consider to be
4:57
absolutely uh a safety issue, then I'm
5:01
going to stay doing you know what I'm
5:03
doing, which is of course continuing to
5:05
investigate the wreck. You may have a
5:07
different view on that. You may take the
5:08
view that I should have bailed out. I
5:10
should have aborted the dive. And you
5:12
know what? That's entirely right. If if
5:14
that is what you wanted to do, then of
5:17
course that is the absolutely perfect
5:20
solution. But of course, we live in an
5:22
imperfect world. And you know, it's part
5:26
of part of doing this kind of diving is
5:28
you make those make those decisions and
5:30
you got to live with the consequences.
5:32
And what you can see has happened here
5:34
now is this is a bit later on in the
5:36
dive. I've actually come back to that.
5:38
Um, well, whatever it is, I've seen you
5:40
maybe a water filter that perhaps. And
5:43
what you can probably hear every now and
5:44
again is a beep. And the reason you can
5:46
hear a beep is cuz we've now got the the
5:49
alarm. So, the inspiration will alarm if
5:53
there is a difference of more than 2
5:56
between a single cell and whatever the
6:00
two voting cells are. So in this
6:02
particular case, there's a cell the two
6:04
voting cells are about 1.36 and 1.27.
6:08
So the the the P2 used by the unit is
6:12
probably about 1.31,
6:15
that sort of thing. So whenever the
6:17
middle cell goes more than 2 bar above
6:20
that, that's when we're going to get the
6:21
alarm. And what you get is you get an
6:23
audio alarm, you get a flashing alarm on
6:26
the head-up display and get the uh alarm
6:29
on the handset as well. loads of
6:31
indications that something is wrong. You
6:33
can do is you can suppress the alarm by
6:36
pressing and holding one of the buttons
6:37
for two seconds. And what that does is
6:40
just stops the alarm going off for a
6:43
period of time. I think it's five or
6:44
possibly 10 minutes. Going back to my
6:47
decision making process, I've been
6:49
watching this single cell rise to the
6:52
point that it's now started to alarm.
6:56
Going back to my decision- making
6:57
process. So, I'm evaluating the factors
6:59
here. I've potentially I've got two bad
7:02
cells and one good cell, in which case
7:04
the good cell is reading 1.5 P2. That is
7:08
not the end of the world. It's extremely
7:10
unlikely to kill me, although it's not
7:13
particularly ideal. Of course, what I do
7:17
believe is that I know exactly what's
7:18
happening, which is I've got one bad
7:20
cell. I've watched the one cell track
7:23
away from the other two. Potentially, of
7:25
course, it could be that um two cells
7:28
have failed at exactly the same time.
7:30
That's just really really unlikely,
7:33
which is why I'm carrying on with the
7:35
rest of the dive, as you can see. And
7:37
you can see that the the middle cell,
7:39
it's still, you know, it's not it's not
7:42
brilliant. It's not performing
7:44
particularly brilliantly, but you
7:46
probably see the uh the flashes and
7:47
lights. I'm I'm in and around the area
7:49
of the shot line, in around the area of
7:51
the strobe. So if I absolutely needed to
7:53
bail out or needed to come back, I would
7:55
be in a great position to do that. And
7:57
in fact, I spend the rest of the dive
7:59
within about 10 or 15 meters of the shot
8:01
line, which is part of my kind of risk
8:04
reduction strategy. So if I do need to
8:06
bail out, I know I've got an easy way of
8:08
getting home. I'm not miles away at the
8:10
far end of the wreck or anything like
8:11
that. So skipping forward a few more
8:14
minutes and my dive is now over. I'm
8:16
returning to the shot line and I'm going
8:19
to start the ascent. You can see that
8:21
the P2 is still it's still in and around
8:25
that 1.4 1.5 for the cell that is
8:28
failing. So, it's it's starting to get,
8:31
you know, carry on getting worse. But
8:33
even if that cell was the only one that
8:35
was reading correctly, 1.5 bar P2 is
8:40
still not the end of the world, it's
8:42
it's obviously, you know, not great and
8:44
I prefer it to be less. But the reality
8:47
is is that 1.5 is is kind of okay. In
8:50
fact, I know somebody who does their
8:52
entire dive on 1.5 bar. So, you know,
8:55
once again, this is about big boy
8:56
decisions. It's about risk. It's about
8:59
all those kind of things. And I am
9:01
personally happy or would be happy if
9:04
that was actually reading correctly and
9:06
it was 1.5 bar. But of course I'm
9:09
operating on the assumption that that is
9:11
the bad cell. The other two are okay and
9:13
therefore I am actually breathing 1.3 or
9:16
thereabouts bar P2. So skip forward
9:19
again. You can see here I've ascended
9:21
quite a bit. The next item on the
9:22
agenda, as people who have seen my other
9:24
videos will know, is that what I'm going
9:27
to do is I'm going to up the P2 to 1.5
9:30
bar to accelerate my decompression. And
9:33
I'm just in the process of doing that on
9:35
my Inspiration handset. You'll see the
9:37
P2's uh rise very shortly. And then I'm
9:40
going to do the same thing on my sheer
9:42
water. And the reason for doing that is
9:44
to accelerate my decompression. So to
9:48
obviously allow me to get out the water
9:49
quickly. And as I do that, you will see
9:52
that the P2 on the middle cell uh rises
9:54
and stays, you know, that kind of 0.1
9:57
maybe 2 bar above the uh where the
10:01
voting logic is. And I'm still getting
10:03
alarms all the time that this is
10:05
happening. But once again, I'm quite
10:08
comfortable with those. I'm happy that I
10:10
know what is happening. I'm happy that
10:13
even in the worst possible case and I'm
10:15
breathing 1.7 bar of P2, it is something
10:19
that I am I'm kind of willing to accept
10:21
the risk of. And part of the reason for
10:24
that is that I'm now on my decompression
10:25
stops. So I've reduced other risk
10:27
factors associated with P2. So I am no
10:30
longer working hard. I am shallower. My
10:33
anxiety is less and all those kind of
10:35
things. And of course for um it's normal
10:38
on open circuit to plan decompression
10:40
stops at 1.6. six bar P2. Anyway, so
10:44
even in the worst possible case, I'm not
10:46
a million miles from that. You might
10:48
say, well, why don't you just bail out?
10:50
And of course, that is completely an
10:52
option. I could I could bail out. I just
10:55
don't see at this point in time that
10:57
there's any rationale to do that. I'm
11:00
confident. I know what's happening. I'm
11:04
willing to accept that 1.7, you know,
11:06
1.8 bar P2 is okay. Now, you may have a
11:10
different view on that, and of course,
11:11
that's absolutely fine. It is increasing
11:14
the risk of an oxygen hit. There is no
11:16
doubt at all if it is 1.7 or 1.8, but as
11:20
I've already said a load of times, I
11:21
don't think it is. I'm fairly confident
11:23
I know what's happening.
11:25
This is a bit further on in the dive. As
11:28
you can see, things have changed quite a
11:30
bit. I've continued to kind of monitor
11:32
what's happened, and I've seen that
11:34
middle cell just continue to rise, the
11:37
P2 rise and rise. I've been watching the
11:39
other two sales and make sure that
11:41
they're responding appropriately so that
11:43
they're not limited. They're not they're
11:45
not staying uh static. So, I'm pretty
11:48
confident that it's the uh I've got two
11:50
good cells and one bad cell. But, you
11:53
know what? I want to absolutely confirm.
11:55
So, I'm going to do a deal flush. Now,
11:58
the very first thing is to put the set
12:01
point from high to low. And you can see
12:02
it's changed from 1.5 to 0.7. That's the
12:05
very small number in the top left hand
12:07
corner. The next thing I'm going to do
12:09
is pump dilluent through the rebreather
12:11
and I'm going to expect the cells to
12:14
fall to the ambience P2 which is at this
12:18
depth and this gas is about.3 something.
12:21
It'll only happen very briefly and then
12:24
it will rise back up to the low set
12:26
point of you can see the uh the the dill
12:28
flush is happening. You can see the uh
12:31
very briefly you can see the two oxide
12:33
cells dropped down to about.3 something.
12:36
The the middle cell basically uh didn't
12:38
do very much. And then what's going to
12:40
happen is is the cell should rise back
12:43
up to the low set point 7. And you can
12:46
see that's happened. And you can see the
12:47
the middle cell has just behaved really
12:50
really badly. So it's gone straight up
12:51
to 2.55 again. Now, if there was 2.55
12:57
P2 within the rebreather, my solenoid
13:00
would have need to been firing an awful
13:02
lot to get enough oxygen in it. I've
13:05
been having to vent gas from the loop
13:06
and all sorts of things. So, it's clear
13:08
to me that that that you the 2.55 is not
13:13
correct.
13:14
What you'd normally do in this situation
13:16
is put the set point back up to high set
13:18
point. I'm not going to do that though
13:20
cuz I want to get some uh a bit of video
13:22
footage to share with you guys. And you
13:24
can see there I'm using my dive vulk
13:26
just to record what I'm seeing from my
13:28
handset.
13:30
And there you go. You can see it really
13:32
clear. I've done the deal flush. I've
13:34
got two good cells. One cell's
13:36
completely failed. And therefore the
13:38
question is what now? What I do is I
13:41
pump the P2 back up to 1.5. I know that
13:44
I've only got two good cells and I do
13:46
the rest of the dive on two good cells,
13:50
staying on my unit, just keeping an eye
13:52
on them. Clear my decompression, get out
13:55
the water. Jobs a gooden. So, let's
13:58
think about some takeaways then. The
14:00
first thing, of course, is to check your
14:02
cells are working at different points.
14:05
The first one is clearly a calibration
14:07
before you get in the water. Second
14:09
thing that's also important is a
14:11
linearity check or at least something
14:13
that confirms that the cells are working
14:15
at the bottom of the dive and are able
14:17
to operate within the limit that you
14:19
expect for the dive. And you'll have
14:21
seen that happened when I got the oxygen
14:24
spike, I guess, on the descent. And the
14:26
next thing is understand what is
14:29
happening with your cells. And I think
14:32
I've explained through this dive that I
14:34
was on top of them. I understood what
14:35
was going on. and then consider what the
14:38
worst case of what you might be seeing
14:40
is. So for me, a 1.5 P2, 1.6, 1.7, even,
14:45
you know, 1.8, it it's not great. It
14:47
does increase the risk, but there are
14:49
ways to mitigate this. And of course, I
14:51
never thought it was correct, and I
14:53
think I had fairly good evidence for
14:54
that. The other thing that is really
14:56
important is a dilluent flush, which
14:59
provides confirmation of exactly what
15:01
was going on. And you saw me do that
15:03
when the P2 or at least the potential P2
15:07
because I never thought it was the
15:08
actual P2 rose above a level that I
15:11
considered acceptable. Now, I'm sure
15:13
that plenty of people are already uh
15:16
typing away in the comments section. I
15:18
look forward to uh your thoughts. I'm
15:20
sure people will have different views on
15:22
what I should have done. And you know
15:24
what? That's absolutely fine. I welcome
15:25
it. At the end of the day, those are
15:27
your decisions. These were mine. And
15:30
that I think is uh the context in which
15:33
it should be discussed. And you know
15:34
what? I'm really happy with my
15:36
decision-m on this dive.
15:38
So some sometimes you kiss a few frogs
15:41
and today was a frog kissing day.
15:43
Unfortunately,
15:45
uh this was absolutely beautiful till
15:47
about 50 m and then it it got got pretty
15:52
rubbish. You dropped down and as soon as
15:53
we got on the wreck, you could see big
15:54
blocks of coal everywhere. So, I knew
15:57
that this wasn't the tamper. And then uh
16:01
went along the remains of it a bit,
16:03
found a couple of boilers,
16:07
uh which is obviously wrong cuz Tab's
16:09
only got one boiler. Then found a
16:11
twocylinder compound or two-cylinder
16:13
engine. So, it's an old it's an old
16:16
Collier. And yeah, that was about it.
16:19
Other than that, I had a self failure.
16:20
So, all my cells completely died. Um,
16:23
which is quite exciting. But yeah, there
16:27
we go. Another frog. There's always
16:28
tomorrow.
16:32
For anyone interested in what happened
16:33
to the cell, well, there's only one
16:35
thing you can do with uh an oxygen cell
16:37
that's let you down underwater. That's
16:39
say thank you very much for the service
16:41
that you provided. But your time is now
16:43
done. If you enjoyed this video and you
16:46
want to see other videos where stuff has
16:48
happened underwater, you may be
16:50
interested in my bailout video after my
16:52
rebreather flooded. I've also got
16:54
another one where my handsets flooded
16:56
underwater and there's also the famous
16:59
one where my scooter went rogue and I
17:01
had to ditch it. So, loads of
17:03
interesting videos there. It would also
17:06
be great if you could do all the usual
17:07
YouTube stuff. So, give us a like, uh,
17:10
drop us a comment, subscribe, turn on
17:13
notifications, but of course, mainly
17:15
what I'd like you to do is watch another
17:16
one of my videos. So, thank you very
17:18
much for listening and watching.