How to properly assess the relative conservatism of various algos/computers for Rec divers?

[Jay]

Q: How to properly assess the relative conservatism of various algos/computers for Rec divers?

————————————————————————————————————————————————————

By Rec I mean NDL dives.

It often seems computer related questions branch off to this question but don’t get resolved or throughly discussed and resurface like Groundhog Day. So, here it is in its own thread.

Q1/ What are wise metric(s) to objectively judge a computer algo’s conservatism by for Rec divers? I can only think of NDL.

Q1a/ If NDL, then at what stage(s) of the dive should NDL be compared between two computers? Definitely when NDL is more stable than not, probably towards the end of a dive when still at depth. I guess this is more of a common sense approach and answer than a prescribed one.

Q2/ If the two above questions are answered then what data do we have to compare algos/computers?

The only data I’ve seen is ScubaLab’s annual test: Scuba Diving Magazine It’s merely a snapshot for those dive profiles. It’s not statistical proof, or a scientific test. But it’s certainly handy to have some data to reference than none.

I’ve used their data and selected part of each dive as per my Q’s above. That info is pic’d below.

Why does any of this matter? a/ Knowledge & Information. b/ Dive Planning - e.g. diving with a new buddy, or new group (dive), or a new computer, then you have an idea of what to expect. If my Teric (GF x/95) says NDL=5 then I know that non DSAT & ZHL 16-C (same GF) computers are probably nearing or just into deco depending what dive # we're on (and of course how close our profiles actually are). The data also tells me that if I’m doing 4 dives/day I need to watch dive #4’s NDL as my aglo seems to behave relatively aggressively on that dive.

There is an apparent quirk in the data in Dive #3 across both years. The differences in that year don’t seem smooth relative to Dives 2 & 4. I don’t know why, but that was also apparent when the data was sliced a different way in this thread: Place of dive tables in modern diving

I’ve noted some rough Rules of Thumb on the attachment. I would be curious if other divers feel those are about right (for NDL Rec dives).

Please note, this thread is not about non-Rec diving, light deco, or deco diving. It’s also not about which is better or worse, just what is. It’s also not to say or imply that ZHL 16-C divers aren’t going to have pleasant dives with RGBM divers, or more explicitly Shearwater and Suunto divers!

A list of Dive Computers and their Algos can be found post #11 from @scubadada here: Question about dive computers

 

[CandiveOz]

Hi Jay,
I'm curious to know why you are restricting your assessment to NDL dives? Hypothetically, say there is a computer which gives the longest NDL but once in deco, it provides the longest decompression. Overall, would you consider it liberal or conservative? To give a fair evaluation, you need to compare both NDL and deco dives. I've attached two articles from Divernet. In the Sep 2018 article, the diver compares a Suunto to a Scubapro. On the second dive the Suunto provides the longer NDL, but when it goes into deco, it recommends a longer decompression profile! The second article from 2009 reads like a horse race (Melbourne cup?). The computers are not necessarily current but the algorithms are the same today. Its very interesting how the "liberal" Oceanic DSAT performs in deco.

The conclusion is; you use the computer/algorithm best suited for your type of diving. Oceanic DSAT has been optimised specifically for multilevel, repetitive NDL dives. This algorithm wasn't really tested for decompression dives and its not encouraged. In fact they don't even offer any form of decompression planning software. Suunto on the other hand, does offer decompression planning software for all their computers and even with their basic model; the Zoop, you can plan and execute a decompression dive. But to be fare to Oceanic, it also offers PZ+ where its test data set consists primarily of repetitive deco dives at altitude.

 

[CandiveOz]

Jay,
Attached are two additional articles from the Journal of Diving and Hyperbaric Medicine. The 2014 article ranks 43 computers based on NDL and 30 mins of decompression at different depths. The 2016 article subjects 4 manufacturers; Mares, Suunto, Oceanic and Uwatec to square profiles, multilevel profiles and multi dive series.

BTW, I expect Scubalab will be coming out with their 2018 computer review very shortly. Time to update your tables!

 

[tursiops]

There is an apparent quirk in the data in Dive #3 across both years. The differences in that year don’t seem smooth relative to Dives 2 & 4. I don’t know why, but that was also apparent when the data was sliced a different way in this thread: Place of dive tables in modern divin

Isn't this because the various computers treat surface intervals differently? Dives 1 and 2 are in the morning, then there is a long break, then dives 3 and 4.

 

[kelemvor]

Excellent idea, @Jay_Antipodean! I hope this results in something useful. I suspect there's a manufacturer or two that might try and get in the way. Hopefully I'm wrong.

 

[Jay]

Isn't this because the various computers treat surface intervals differently? Dives 1 and 2 are in the morning, then there is a long break, then dives 3 and 4.

Thanks, yes, it does look like that; the SI between dives is 1hr except for pre dive #3 where it's 2hrs. Interesting. That also means my last 'Rule of Thumb' (Diff between DSAT & 16-c (x/95) is more initially a function of SI for the next dive unless it's dive #4 ... would be interesting to see how much dive #4 differences would be narrowed if the SI was 2 hours there too.

 

[Jay]

Hi Jay,
I'm curious to know why you are restricting your assessment to NDL dives?
...
The conclusion is; you use the computer/algorithm best suited for your type of diving.

Pretty much for the same reason as your conclusion; many Rec divers will almost always dive NDL and for them this is what's relevant. Light-deco, RecTec, Tec probably/do (etc) have a different view of what's relevant, and different ways to measure it. You may have seen a bit of that occur in your other thread (and many more on SB!) and that, whilst interesting and to a degree informative, doesn't answer the question in this thread.

Thanks for the other links too. Still digesting

 

[Jay]

Overnight the Scubalab dive computer review came out The Best Dive Computers Diving their usual chamber profiles on the computer's least conservative setting.

Below's the updated data.

Observations:
re the Teric. It's hard to tell what's going on. On dives 2 & 3 it looks like it's on the wrong conservatism setting (medium instead of least conservative) as it ~matches the Perdix on that setting in the prior year, but on dives 1 & 4 it seems to sit between both settings. For those with both a Perdix and Teric (@stuartv et al ) I'm assuming you've not noticed any NDL discrepancies, and in theory I can't see how there could be any.

The Descent ~matches the Perdix from previous years which makes sense.

The EONs running Fused are interesting. They are generally in line with what @Diving Dubai mentioned in another post. They prefer the longer SI (pre dive 3 of 2hrs) than the 1hr ones relative to DSAT & 16-C. It would be really interesting to know where Fused would be on dive 4 with a 2hr SI relative to DSAT and 16-C to see which algo (if any, or both) is going too aggressive/conservative.

Mares, SEAC, and Aqualung are all sitting where they 'should' be based on prior year's observations.

 

[CandiveOz]

If they missed the liberal setting on the Teric because medium (40/85) is the default, I wonder if they left the Eons at their default setting too, missing out on the two additional liberal settings offered by the Fused algorithm.

 

[stuartv]

I have not yet noticed any difference in NDLs between my Perdix AI and my Teric. But, I have not really dived them enough yet.