RGBM Algorithm for Technical Diving

[Sevenrider860]

In respect to using an RGBM for recreational diving...I know...deviating from my own thread topic!!


I spent a week in Bonaire in November of last year and made 22 recreational dives in 6 days. I wore a Suunto DX and my Shearwater Petrel for all the dives. The Suunto was set at P0 and A0. The Petrel was run in Recreation mode with Medium Conservatism (Bühlmann ZHL-16C 40/85) with "Adaptive" Safety Stops. I found both computers to be very close on the NDL numbers. There was never a case where the Suunto cut a dive shorter than the Petrel. The Suunto called for Deep Stops, on some deeper dives there were two deep stops, the Petrel does not implement Deep Stops. The Petrel would indicate a 5 minute Safety Stop on some dives, the Suunto always a 3 minute stop. I personaly did not find the Suunto "too conservative" for that week of diving. My son was with me on every dive except one and he uses an Atomic Aquatics Cobalt running RGBM and our NDL times were very close to each other...no large deviation between the two. I don't take the "Too Conservative" complaints too seriously. I "can" dive in a way that will make the Suunto too conservative, but I don't dive that way and I don't find that it limits my dive time.

I am associated with an AquaLung shop, so with the arrival of 2016 it was out with Suunto and in with the AquaLung branded PPS computers. All of our rental gear had Zoops and they will be changed out with the i300. I started brushing up on the i300 and noticed the PZ+ algorithm used is actually more conservative than the Zoop RGBM...at least for the first dive. The i300 does not look like it will penalize the next dive because of bad habits (rapid ascents, short surface intervals, etc)...or at least is is not documented.

 

[tomfcrist]

Furthermore, one could argue that it's riskier to dive the RGBM deco model on a computer because of the fact the user has no way to predict what it's going to do, whereas with other options you know if you do this your computer is going to do that.

That's not an RGBM thing....That's a Computer programmer thing. Like I said earlier, the tables do work....follow the schedule. I understand that the tables are limiting, requiring long SI's and limiting the number of dives per day, but they do work.

The Abyss program that Mark used is similar to V-planner, in that its not an on the fly algorithm. Its a custom cut table. Abyss had issues, Abyss is gone. I have used GAP on smaller dives, but am unwilling to shell out the money they want for the program. I have used the NAUI RGBM tables, and they work fine.

 

[Sevenrider860]

Andy,

I think that is an excellent assessment. I know what the Suunto "dislikes" on recreational dives and I know the likely results.

Ascend too fast you get a "Mandatory" Safety Stop with a 10 Foot Ceiling...Depending where the "violation" occurred this Stop may clear during your slow accent or it may clear while on your safety stop. Get out of the water without honoring the stop and you get the "Attention" Symbol. Starting your next dive before that clears and you will see the famous (infamous?) Suunto conservatism in action. For the computers that have a Deep Stop, don't bust that stop either...that gets you a penalty too. For recreational diving, these penalties can be considered an inconvenience and following the Suunto triggered "deep" and "mandatory" stops is not that difficult should they occur...none of them add more than just a few minutes...most of which is in the 3 to 5m depth range and spending extra time at that depth is not always a bad thing.

Transition to a technical dive...

Does a rapid ascent generate a "Mandatory Safety Stop" that alters the time on my final stop? I don't know.
Does the Suunto throw in Deep Stops that are different than my planned stops? I don't know.

As you point out...that is a lot of "I don't know".

Does that mean the RGBM algorithm is bad or does is mean the Suunto implementation of the RGBM algorithm is bad?

Is RGBM fundamentally a bad decompression algorithm for technical diving or is the implementation of RGBM by Suunto bad for technical diving? With GAP dive planner generated tables and the XEO running RGBM I did not have those triggers and I never felt diving using that algorithm was bad. The tables generated by the GAP software were very close the VPM tables generated by V-Planner. I have a few deep trimix dives where my buddy was using a Petrel with VPM and I was using the XEO with RGBM. When comparing the tables for each and the computers on the dive itself, there really was not that much of a difference.

 

[100days-a-year]

Anything Suunto is useless in technical diving unless you are old+out of shape+a smoker and have a PFO.
The way they implement algorithms is to conservative to start and gets worse from there.Fine for one dive per day where optimal conditions exist.

Buying one to use a gauge is like buying a Yugo to use as a shopping cart.

I've used RGBM on deeper divers and mix dives and didn't get bent but have also used VPM-A-0 conservatism on the same dives and not gotten bent.

 

[rongoodman]

The full implementation for the Xeo doesn't seem to be available any more, and there's no way of knowing what Suunto is providing.

It turns out that the full Free Phase implementation for the Xeo is still available. The link on the Liquvision page had been broken for a couple of weeks.

 

[abscuba]

http://archive.rubicon-foundation.o...23456789/10269/NEDU_TR_2011-06.pdf?sequence=1
Here is the link to the NEDU study one of several that showed using deep stops increases the chance of DCS. No they were not using Suntos they were testing the popular theory of using deep stops. There have been others and they do draw the same conclusion. An easier approach is look is what being used on dives that are pushing limits right now, you will find very few using suntos or RGBM. As a technical diver you do have to draw your own conclusions as it is your life you are dealing with and after reading study after study and book after book all come out saying RGBM was a cool idea but started to have issues when started to be heavily used I made it a point to avoid. I took it to the point of talking with one of the doctors from the NEDU and he was surprised it was still a discussion. This study was from 2011, but more work has been done drawing the same results.

 

[dmaziuk]

I'll add, is it that difficult to reverse engineer an algorithm on a dive computer? I keep hearing about the secrecy of the RGBM, but I would think once the product is released someone smarter than me could figure it out.

A: very. Also expensive: they will sue you to kingdom come for violating DMCA, MPAA, COICA, RIAA, SOPA, PIPA, and whatever else acronym the current crop of copyright-chasers will come up with. There are recent rulings to the effect that FBI will probably not going to drag your a** to jail for being an Evil Hacker(tm), but you're still getting buried in legal fees.

On the technical side the rule of thumb is: if you have the code, figuring it out is ~80% of the effort it takes to write your own from scratch. For a "clean room black box" hack -- just roll your own, it's way faster and easier.

---------- Post added January 8th, 2016 at 12:37 PM ----------

This, for me, is one of the major concerns. For technical diving, the algorithm needs to be predictable. Putting 'triggers' into an algorithm which unilaterally modify the mathematic function based on unexplained parameters.

There's assertions that need to be true for the algorithm to work. Your profile could potentially push a parameter out of bounds and get a deco stop on Mt. Everest for -10 minutes. Then there's implementation details. Like how much oomph does it take to run the model near real time: a cheaper low-powered device may well be running some approximation or even a pre-calculated lookup table. So yeah, even if understand the theory, you still don't know what mathematic function your DC is actually running.

 

[muzikbiz22]

In respect to using an RGBM for recreational diving...I know...deviating from my own thread topic!!


I spent a week in Bonaire in November of last year and made 22 recreational dives in 6 days. I wore a Suunto DX and my Shearwater Petrel for all the dives. The Suunto was set at P0 and A0. The Petrel was run in Recreation mode with Medium Conservatism (Bühlmann ZHL-16C 40/85) with "Adaptive" Safety Stops. I found both computers to be very close on the NDL numbers. There was never a case where the Suunto cut a dive shorter than the Petrel. The Suunto called for Deep Stops, on some deeper dives there were two deep stops, the Petrel does not implement Deep Stops. The Petrel would indicate a 5 minute Safety Stop on some dives, the Suunto always a 3 minute stop. I personaly did not find the Suunto "too conservative" for that week of diving. My son was with me on every dive except one and he uses an Atomic Aquatics Cobalt running RGBM and our NDL times were very close to each other...no large deviation between the two. I don't take the "Too Conservative" complaints too seriously. I "can" dive in a way that will make the Suunto too conservative, but I don't dive that way and I don't find that it limits my dive time.

I am associated with an AquaLung shop, so with the arrival of 2016 it was out with Suunto and in with the AquaLung branded PPS computers. All of our rental gear had Zoops and they will be changed out with the i300. I started brushing up on the i300 and noticed the PZ+ algorithm used is actually more conservative than the Zoop RGBM...at least for the first dive. The i300 does not look like it will penalize the next dive because of bad habits (rapid ascents, short surface intervals, etc)...or at least is is not documented.

I gotta be honest here; this could arguably (because that's what happens here on Scubaboard), be the most comprehensive, realistic, diplomatic & informative post EVER, with regards to the whole Suunto/Shearwater/Conservative/Dive Computer "centuries old conflict". My hat is tipped to this person for providing real deal data & experience.

 

[KenGordon]

So... I planned my dive. But, for instance, I trigger an ascent warning by moving my arm too quickly and alerting a hair-trigger depth sensor. The computer whips me for it and now I'm diving on on am algorithm the computer adjusted, not me. And that has consequences on my team deco, my gas plan etc etc.

It adjusts your safety stop. You can choose to ignore the extra time as it is only a safety stop. So your original plan stands. No surprises. Most of my downloads are full of little red triangles saying I misbehaved.

On the slate in front of me I have a Suunto plan for the Duke, a popular channel dive, it is the deeper longer plan 58m, leaving the bottom at 30. Comparing it to the closest plan I can get from multideco by using gf 40/65, DM5 insets the first (1 minute) slightly deeper but then spends less time in the teens and arrives at 12m about 8 minutes ahead of the multideco schedule. A bonus is I spend longer on the rich deco gas.

VPM-b 0, gf 45/95 and gf 95/95 take about 20 minutes out of the ascent. 75 vs 96 run time.

For sure this is not real RGBM, reading the manual (helo2)

"While the Suunto Technical RGBM model is based on M-values, the use of deep stops brings the model closer to the full RGBM model. The way of implementing deep stops is conducted by Dr. Bruce Wienke."

I am not sure multi deco is deriving its stops directly from the gf limit either, it really seems to want you to be going up at 3m/s and likes to add stops to make that so. I may try planning that dive on a petrel (in the absence of a planner in Shearwater Desktop) but that may drain my will to live...

I really dive these plans. I am very aware of the bottom gas plan in particular as it is the limiting factor.

I have both Suunto and shearwater computers. Neither is perfect but neither is useless for either deco diving or no stop diving.

Ken

 

[DevonDiver]

So the 'technical' Suuntos add a safety stop to the deco stop?

I must admit, I'd otherwise have guessed they would tweak their hiGF to cause longer shallow stop.

Its proprietary.... So guessing seems the order of the day... LOL

What of repetitive dive adjustments? For instance, if you choose to ignore that mandatory 'safety stop'? It won't lock out, as that's only for incomplete deco... but it's a fair assumption that a Suunto would want to spank you on subsequent dives.

Suunto: "bwahahaha.... now you can sweat on your gas plan puny human, that'll teach you to ignore me...!!"

Are their algorithm penalties if your surface interval isn't long enough? How long an SI do they dictate for technical dives? What about repetitive or multi-day dive missions? Does the algorithm insidiously mutate?

For technical diving, we shouldn't even be asking these questions... It should all be clear to the diver. As it is with other, open source, algorithms.

Also, which computers have 'technical RGBM'? I'm guessing Helo2, Eon Steel and D9tx?

But not Vyper2, D4, D9, Zoop etc... as their deco computations are for ' emergency deco' only.... very haldanian , it seems, with an impressive 'tekkie' Pyke stop unilaterally added because.... well, just because...