Non-scientific Comparison of Dive Computer Conservatism

[Divetech99]

Ron, how different is the RGBM implementation of a Cobalt vs Suunto? Is one more or less conservative than the other? Just curious.

 

[meatsim]

Funny you should post this now because I have been playing with the same thing over the last couple of weeks. The weekend just gone was my second weekend taking out my new petrel 2. The weekend before was its first trip out and I still carried my old Aeris Atmos Pro with me as well. I set the Petrels conservatism to high because i didn't think i wanted to push anything. At the end of my second dive the petrel gave me 4min NDL when i started heading up while the Aeris still showed almost 30 mins (i can't remember the exact number)

Because of the difference in time I moved the Petrel back to med conservatism and this weekend at the end of the second dive my dive buddy signalled that they were a few mins from the NDL on their Zoop while my Aeris still gave me 25+ mins again and the petrel was showing 15. We had air left so i brought us up a few meters and both my computers gave me more time but the zoop didn't change so we called it and headed up.

I know the zoop was on a different diver than the other two but we had dived a very similar profile and the difference was more than i expected, i'm now off to read "Clearing Up The Confusion About "Deep Stops" By Erik C. Baker" as suggested by petrel manual to get my head around using the conservatism setting since i now have a computer which is a bit more open and I can have a better understanding about what it is doing

 

[NAM001]

You know with all the standards we have available to us. weight, volumn, distance ect. we do not have a standard for dive computers and the point when NDL is. I would think that every computer should be able to be measured against a standard computer. That computer would be called 100% and all computers would have a rating attached to them to day that their liberal setting is 95% of the standard and conservative is 60% of the standard. It should (I guess) be no surprise why so many tend to ignore the NDL warning on their computers go into deco and skip the stops.

 

[JohnN]

You know with all the standards we have available to us. weight, volumn, distance ect. we do not have a standard for dive computers and the point when NDL is. I would think that every computer should be able to be measured against a standard computer. That computer would be called 100% and all computers would have a rating attached to them to day that their liberal setting is 95% of the standard and conservative is 60% of the standard. It should (I guess) be no surprise why so many tend to ignore the NDL warning on their computers go into deco and skip the stops.

To have a "standard" computer, you would have to understand all the factors that cause DCS. The current state-of-the-art is to take an algorithm and then muck with it by applying fudge factors so that it follows what little empirical data is available.

Good luck with that.

 

[RonR]

Ron, how different is the RGBM implementation of a Cobalt vs Suunto? Is one more or less conservative than the other? Just curious.

RGBM provides “handles” in the algorithm that can be adjusted to arrive at various levels of risk/ conservatism. Suunto is known for being on the conservative end of the spectrum. That’s not inherent to RGBM, it’s a decision made by the manufacturer when they set up the algorithm. It’s not clear to me that all Suunto computers have exactly the same conservatism, but generally the reputation is that they are more conservative than most, particularly when it comes to repetitive dives, ascent violations, or in some cases giving lock outs. How those and many other situations are handled can vary from computer to computer, even if they are theoretically running the same algorithm.


Suunto uses a “folded” implementation of RGBM, which means they run a basically Haldanian calculation with factors added that cause it to come very close to what running a “full” RGBM calculation would give. That has advantages for recreational diving- “full” RGBM, for instance, doesn’t generate no-stop times. The Cobalt also uses a “folded” implementation unless the dive is deeper than 150’, when we switch to a fully iterative (and much more computationally intensive) version of RGBM. At that point deco is probably a given.


When Atomic was determining the settings for the Cobalt, we were careful to be in the middle of the pack for conservatism, compared to a whole range of other computers and a range of normal profiles. We also determined that the Cobalt would never lock a diver out- if you violate deco you get a big warning screen you need to page through, but we leave it up to the diver to determine if the missed stop is trivial or significant. As to differences in exactly how we handle deeper than previous repetitive dives with sawtooth profiles and ascent violations in the 60 for range continued over 30 seconds when at > 80% of NDL- well, it just gets really complicated quickly, and comparisons are hard to nail down without specific profiles.


One other comment- conservatism in dive computers is hard to nail down, but differences tend to seem more extreme in shallower, longer dives, because we treat deco/ no deco as a binary function when in reality is is a gradually increasing slope, very gradually increasing at shallow depths. So on shallower dives even a slight divergence in the algorithm conservatism can translate into many minutes more or less of no-deco time. The same computers might show much less difference- in minutes of no-deco time- at deeper depths. That probably accounts for some of the subjective differences in experience of how conservative a particular computer is.

-Ron

---------- Post added August 24th, 2015 at 09:48 PM ----------

<snip> .... i'm now off to read "Clearing Up The Confusion About "Deep Stops" By Erik C. Baker" as suggested by petrel manual to get my head around using the conservatism setting since i now have a computer which is a bit more open and I can have a better understanding about what it is doing

Erik Baker does a great job of explaining decompression without a lot of unnecessary verbiage. He&#8217;s also the source of my favorite comment about decompression algorithms, which I am probably mis-remembering as &#8220;all decompression algorithms are attempts to draw a bright, clear line through a fuzzy grey area.&#8221;

---------- Post added August 24th, 2015 at 10:00 PM ----------

You know with all the standards we have available to us. weight, volumn, distance ect. we do not have a standard for dive computers and the point when NDL is. I would think that every computer should be able to be measured against a standard computer. That computer would be called 100% and all computers would have a rating attached to them to day that their liberal setting is 95% of the standard and conservative is 60% of the standard. It should (I guess) be no surprise why so many tend to ignore the NDL warning on their computers go into deco and skip the stops.

You can do this now by comparing computers to dive tables or simulators. It&#8217;s just not very informative. It can tell you about a single profile. it wouldn&#8217;t tell you much about what happens in more complex real life situations. One computer can be more liberal on a single dive, and much more conservative in repetitive scenarios. There are way too many variables to have a useful standard.
The reality is that decompression algorithms- and dive computers- are imprecise but very useful tools.

 

[Divetech99]

Thank you Ron for a very clear explanation. When you have time, I'd appreciate it if you could provide some insights on how Dive Computers like Atomic is dealing with the NEDU study on deep stops (ie that deep stops are not good as originally thought). Also, if I understand correctly, I think the authors of that study are not advocates of bubble models (RGBM). Instead, the proponents of the NEDU study seem to be leaning towards GF of 40/70 or thereabouts.

 

[RonR]

Thank you Ron for a very clear explanation. When you have time, I'd appreciate it if you could provide some insights on how Dive Computers like Atomic is dealing with the NEDU study on deep stops (ie that deep stops are not good as originally thought). Also, if I understand correctly, I think the authors of that study are not advocates of bubble models (RGBM). Instead, the proponents of the NEDU study seem to be leaning towards GF of 40/70 or thereabouts.

A deep subject. I think there are variety of opinions.


Seriously, please don&#8217;t confuse me with a decompression expert. There are people on SB who are, I&#8217;m not one of them. I can talk about the issues involved in implementing algorithms into a dive computer, and the very complex factors that need to be considered. I have a decent handle on the variables involved there, and why computers with the &#8220;same&#8221; algorithm can end up working differently in the real world. But when it comes to evaluating the biomedical aspects, I will defer to others as quickly as possible. My understanding of the consensus is that there is no clear solution for all types of diving, but deep stops are no panacea that can drastically shorten deco times. Most scientists prefer more open models that can be verified and tested independently. I&#8217;m on board with that.


There were some specific algorithm &#8220;issues&#8221; identified in this NEDU research. That is something for which I am personally and professionally very grateful- long story. It was done about the same time as we introduced the Cobalt, and we were very aware of it. I don&#8217;t think it has direct relevance to the deep stops implementation actually in any recreational dive computer. Dive equipment manufacturers are pretty careful to not stray too far from well known territory, and I believe divers can have a lot of confidence in dive computers- while knowing none are perfect, and none can know what is happening in your body. They are just running mathematical models.


I think that in most computer discussions algorithms get too much attention and issues like implementation or user interface design too little. The algorithm itself is actually a relatively small and simple part of the development of a dive computer.


-Ron

 

[stuartv]

Tbone, Can you elaborate a little. What do you mean by model? Do you have computer codes for specific mathematical models?

It would be nice if all dive computer manufacturers were required to publish their deco algorithms in open-source format. Maybe I should do a Kickstarter project for an open-source dive computer.

Already exists. A computer, I mean, not a KS campaign.

heinrichsweikamp

 

[Divetech99]

A deep subject. I think there are variety of opinions.


Seriously, please don’t confuse me with a decompression expert. There are people on SB who are, I’m not one of them. I can talk about the issues involved in implementing algorithms into a dive computer, and the very complex factors that need to be considered. I have a decent handle on the variables involved there, and why computers with the “same” algorithm can end up working differently in the real world. But when it comes to evaluating the biomedical aspects, I will defer to others as quickly as possible. My understanding of the consensus is that there is no clear solution for all types of diving, but deep stops are no panacea that can drastically shorten deco times. Most scientists prefer more open models that can be verified and tested independently. I’m on board with that.


There were some specific algorithm “issues” identified in this NEDU research. That is something for which I am personally and professionally very grateful- long story. It was done about the same time as we introduced the Cobalt, and we were very aware of it. I don’t think it has direct relevance to the deep stops implementation actually in any recreational dive computer. Dive equipment manufacturers are pretty careful to not stray too far from well known territory, and I believe divers can have a lot of confidence in dive computers- while knowing none are perfect, and none can know what is happening in your body. They are just running mathematical models.


I think that in most computer discussions algorithms get too much attention and issues like implementation or user interface design too little. The algorithm itself is actually a relatively small and simple part of the development of a dive computer.


-Ron

Ron, thank you. I understand your position. Didn't mean to drag you in an otherwise controversial topic. I also want to thank you for so generously and professionally sharing your expertise of dive computer in general -- even if you're part of the cobalt team .

Btw, is Atomic planning to come up with an H3 like dive computer?

 

[RonR]

Ron, thank you. I understand your position. Didn't mean to drag you in an otherwise controversial topic. I also want to thank you for so generously and professionally sharing your expertise of dive computer in general -- even if you're part of the cobalt team .

Btw, is Atomic planning to come up with an H3 like dive computer?

Well, there is a wrist AI Cobalt in the works. It will have a larger, higher resolution screen than the H3, and be bigger overall. I can't really say more than that at this point. This thread is one place where information will appear, clearly we have a lot of folks waiting: http://www.scubaboard.com/forums/atomic-aquatics/454057-wrist-mounted-computer.html

-Ron