COBALT-2 wishlist

[HIGHwing]

The Cobalt doesn't "lock out" users who miss or don't complete a deco stop. It will display a prominent warning for 24 hours saying that a schedule was not completed and that the computer's schedules might not be valid, but it's up to the diver to determine if the violation was trivial or not. It basically keeps running the algorithm no matter what- at some level of violation it's no longer possible to generate a schedule that doesn't involve going deeper than the current depth, but I don't think massive violations are what you are concerned about.

Correct. I'm not at all concerned about life-threatening violations, I'm looking at those skipped a minute because the computer won't clear and I'm fine by RT. I was not aware the Cobalt would not lock out given an ascent to the surface with deco remaining.

As to matching the Cobalt's RGBM to another algorithm, the Cobalt has an onboard simulator that allows for very easy running and saving of profiles in a log of simulated dives, these could be matched up to other algorithms and settings tweaked to get general alignment. Cobalt uses a "folded" RGBM on dives shallower than 150' (calculation-wise, this would be more like GF), and switches to fully iterative RGBM on doves deeper than 150' (which would be closer to VPM). But the planner on the Cobalt will generate schedules for planned dives very easily, and will display the stop schedule onscreen so you can write it down, just as one would using a desktop computer. Since it knows your detailed actual previous dive profiles, history, and tissue state, it is arguably going to be more accurate than a desktop planner for generating tables. And it will certainly allow you to plan schedules the Cobalt will agree with.

OK! So I knew I wasn't crazy. I did not know what the Cobalt was doing, only that when one is the group of divers we get a different results after a single dive. I agree with your summation we can tweak our VPM and GF to match the proprietary RGBM magic running on the Cobalt and we do have to take this course of action in practice for everyone to play nicely. However, I noticed to make this work we have to make GF changes on our Tec computers and desktop software to match the proposed profiles generated for 1) deeper dives 2) repetitive decompression dives within a single day. My initial guess was the proprietary RGBM model appears to be somewhat close to 20/85 if you added a deep stop and a 3-minute safety stop. However, if you make a second deep decompression dive or a 150+ft plus dive there appears to be an accumulated deco penalty on the shallow stops. [I've actually wondered if the projected deep stop is at/near/below the off-gas depth but that might be an entirely different discussion.] It's as if you go from say 20/85 to maybe 30/70 + deep stop + safety stop for the repetitive dives. My actual solution was to purchase an iPad mini and load up V-Planner and Baltic Deco and I just take it on the boat with me viola. We use the software to cut tables and the computers become advisory only fully admitting we sometimes just put 30/70 in the Shearwaters and that just about guarantees we're behind the Cobalt's deco model for sure.

My next thought should be prefaced by agreeing with your thesis and adding that any reasonably competent technical diver can handicap to the Cobalt as valid, as is the assertion the decompression profile for the repetitive dive might be more accurate. What I'd like to submit for consideration is that the profile the Cobalt creates could be the most aggressive for the repetitive dive. I don't know because I don't know where the model is in relation to the M-Value line. I'm a GF guy at heart, what can I say. Consider if you will non-Cobalt dive planners would have their prior obligations governed by fundamentally square/multi-level dive translated to a straight table (by using V-Planner/GAP/Baltic). Accordingly those profiles are going to assume the maximum possible loading versus precise load tracking the Cobalt performs unless the divers flawlessly executed the profile (we'll assume Tec divers never go deeper or miss the RT for sake of discussion). Does it matter? Maybe not, but this is a pretty productive conversation and I think it's worthy of a thought. Of course I realize the Tec computers have the ability to plan a dive right on your wrist, but I generally don't bother.

Say I'm using Baltic with 20/85 to plan and 20/85 on the Tec computer (we can of course switch models) if I swim the profile accurately, I can plan and cut tables with crazy accuracy. As you've surmised, I do not have a solid feel for how to equate this experience with a proprietary RGBM for all situations.

Agree. The Cobalt algorithm is fully capable of supporting trimix "under the hood" (or constant PPO2, for that matter). There are some mitigating factors. Atomic specifically wanted a version that was recreational diver friendly, that means a simpler interface- of course, we could provide two modes, much as we do now with gas switching, which must be enabled to become accessible during a dive. Right now, we don't have a user interface enabled that allows a fraction of He to be entered. And the general feeling was that a console version would not be that popular as a trimix computer. I wonder if you think this might be changing with changing standards?

The Cobalt isn't going to be super-popular as a Trimix computer out of the gate, but I was suggesting giving my clients a path forward with their existing investment. Let's put some real thought toward how a computer like the Cobalt could stay with a diver for a longer period of time without being outgrown. There's little doubt the Cobalt could do more than it does, or in different ways to make an additional (I won't say broader to recognize the Tec market is niche) appeal.

This is a very interesting discussion.

 

[Scared Silly]

Some rambling thoughts:

Let me ask this question that is not really specific to the Colbalt but drives some of the above discussion. There seems to be a move towards having "recreational" computers to allow multiple gases. I am not sure the reasoning other than for the convenience - i.e. program in 21% plus a couple std. nitrox mixes. Further, if the models allows mixes > 40% then you have "technical" divers looking at the computer. Especially because the Colbalt has such a fine display and other nice features.

When I started tech diving I bought a computer that allowed for 3 three O2/N2 gases. The model was fairly conservative as was the deco software I was using. But then again the best part was that I could simulate my dive profiles on my laptop look them over then easily tweak the profiles. Once done I could set my computer and the simulate the dive and compare noting the differences. I note the above process, i.e. laptop simulation first, then dive simulative because while I could do the simulation on my dive computer I would not want to do so unless really needed. That is because the interface on all dive computers take too much time for repeated trials. I have done the later when I did not have my laptop it worked but do the way I would want to do it.

I bring this up because once the "tech" side gets a hold of a computer - it is a different ball game. Simply because of the need to simulate and the desire to utilized models that are common within the community. This may or may not have been that was thought about when the Colbalt was designed as the "rec" side is more of plan-dive-log, rather than plan-simulate-dive-log. So different needs. Especially if He2 is brought into the mix (pun intended).

 

[HIGHwing]

Replace "Multi-Level Diving" (old model) with "Extended No-Decompression Diving" (new model) which allows a diver to take two Nitrox blends of less than 40% to extend their available bottom time significantly by switching gasses. Maybe a tank of 28% to get on down there with an AL40 of say 40% for the second level and back to the boat. I don't see why this isn't manageable at a recreational level.

What I'm learning is the Cobalt will do this handily out of the box if the user pre-programs the switch depth.

 

[RonR]

....I note the above process, i.e. laptop simulation first, then dive simulative because while I could do the simulation on my dive computer I would not want to do so unless really needed. That is because the interface on all dive computers take too much time for repeated trials. I have done the later when I did not have my laptop it worked but do the way I would want to do it.....

HIGHwing and Scared Silly, you bring up some interesting questions, I have limited time today but wanted to reply to some of them.

re. Planning:
Have you used the simulator on the Cobalt? If so, I'd be interested on your impressions as to functionality relative to desktop planners. If you haven't, there is an interactive demo here: Cobalt Guide: Simulation Menu . Is there something missing that you would like to see? We designed it to practically fill the same functions as desktop planners- it's graphic, it allows for a wide variation in setting variables- gas mixes and switches, exertion levels and conservatism levels, the altitude can be current pressure or somewhere else, saturation can be current as measured or after any specified period on the surface, it allows multi- dive simulations and deco, including generating surfacing tables. It also uses either your historic breathing rate or allows you to set a specific RMV for the simulation, and gives gas requirements for the simulated dive for all of your mixes. Simulated dives can be stored and compared to one another. It displays a facsimile of the dive screen during simulations, with the cf or liters of gas used in the place of the pressure. You can try and compare different scenarios- the Simulator is essentially a standalone program running on the Cobalt, it uses your personal data and history, but settings in the simulator don't change settings for "real" diving.

The advantage in this approach, it seems to me, is that you could be in the middle of a multi-dive trip, on a boat, and the Cobalt knows exactly what you have done and what your tissue calculations are. So when you start a simulation with an additional two hour surface interval, it has a better starting point.

The one limitation is the "video game" type interface, where each minute of a dive simulation requires a button press. It goes quickly, but is not probably the most elegant solution for longer dives. Some more technically oriented divers have requested a numeric input interface in the simulator, so someone wanting to simulate a long, square profile or multi-step dive could do it with just a few button presses. This is something we want to implement, but haven't gotten to yet.

re. Algorithms:
These kind of questions can get almost theological pretty quickly, I'd just say don't place too much emphasis on the "proprietary" nature of the Cobalt's RGBM model. All RGBM implementations are by definition proprietary, but mostly what that means is that the baseline conservatism factors may have been tweaked a bit to align with the individual companies' desires. Atomic is in the middle of the pack here, intentionally. There are two basic things that get called "RGBM"- one, in all other recreational computers other than the Cobalt, is "folded" RGBM. This basically means a Haldanian algorithm with bubble factors added to limit factors that the full RGBM calculations show to be tending to generate bubbles. In that sense it has more in common with the GF type models. The "full" RGBM, which the Cobalt uses when a dive extends below 150', is far more computationally intensive, doesn't use M-values, and involves actually running bubble calculations iteratively, honing in on the best solution- why it is sometimes referred to as "iterative". In that sense it has more in common with VPM. To my knowledge the only other "full" RGBM implantation that runs on a real time dive computer is one that can be purchased for Liquivision (though that may change with time). RGBM is generally going to penalize multiple dives, deeper than previous, and multi-day diving because they are factors that the bubble calculations show lead to bubble formation. GAP, I believe, has full RGBM planning software for Windows and Android that allows direct comparison of identical profiles, or you could do the same thing with a Cobalt and a desktop planner. When we implemented the full RGBM version that is in the Cobalt, values for some bubble factors used in the equations were fixed within a range- they are what you adjust with the conservatism settings on the Cobalt. On the desktop software, these ranges are also adjustable, so the settings should be able to be aligned quite closely.

BUT- let me insert a huge caveat here. The actual algorithm calculations, what a desktop planner runs- are only one part of what is needed to implement an algorithm in a real time computer, and by far the easiest part. How each computer- even ones running the exact same algorithm calculations- handles things like time away from a stop (how much is too much- at what depth?), allowable stop ranges, ascent rates, how they handle the algorithm implementation when the diver does something unexpected, like re-descending after starting an ascent profile or over staying a stop, even small variations in depth on the dive or higher or lower surface ambient pressure- these and literally hundreds of other questions have to be answered and unambiguously hard coded into the computer's firmware. They are questions desktop software doesn't need to deal with, and one of the reasons why it is much, much easier to come up with desktop planners than dive computers.

Interesting to hear what more tech oriented users would like to see. We originally conceived of and developed the Cobalt as a dual mode, recreational and tech, computer. The recreational side has dominated as the biggest market, but as things evolve other options could open.

Ron

---------- Post added March 31st, 2013 at 02:34 PM ----------

Replace "Multi-Level Diving" (old model) with "Extended No-Decompression Diving" (new model) which allows a diver to take two Nitrox blends of less than 40% to extend their available bottom time significantly by switching gasses. Maybe a tank of 28% to get on down there with an AL40 for the second level. I don't see why this isn't manageable at a recreational level.

What I'm learning is the Cobalt will do this handily out of the box if the user pre-programs the switch depth.

It does. Just to clarify, the reason that pre-programming a switch depth is required on every dive in order to enable gas switching is to avoid the situation where a recreational diver could hit the UP button, trigger a manual gas switch dialog, and then, not understanding it, inadvertently switch to a nitrox mix they were not using. So on the Cobalt, gas switching is disabled in a dive unless the diver has intentionally programmed at least one planned switch.

We have considered also enabling an "expert" mode that provides more controls over settings, possibly including always-on gas switch capability.

Ron

 

[HIGHwing]

Ron -

I must say this is a far more informative thread for me than the 4-digit post level thread over in the other section. I'm going to use the online simulator and help myself to a comparisons.

Again, I'm excited to learn the model really does morph in the realms of Tec diving. I don't want to devolve a discussion about features into a bubble model debate as this will crash into the gutter before midnight.

I think you're right on top of how to bring the Cobalt v2 into the market space be that with an expert mode, a Tec "skin", or perhaps you follow the path plowed before you by manufacturers and offer a pin activated upgrade for qualified divers/instructors. I know Atomic spent a lot of time figuring out how to position the Cobalt in the market place and I recognize there may be limited interest in expanding that risk envelope. In the mean time, I appreciate candid conversation.

 

[Scared Silly]

Ron, I have not run the simulation hands on but there are lots of features like the table summary that other recreational oriented computers do not have. I would need to do some hands on to give you substantive feed back. But from my limited cyber diving with Colbalt it has far more than more "recreational" computer and some tech computers. Let me as naive question. What simulation is possible via the laptop software.

My comments regarding planning is more from the interface perspective. Having access to everything visually vs several screens as well as keyboard/mouse vs video controller. So no matter how many features are there unless there is a keyboard and screen real estate there will be limits.

One of the things that I was taught when learning my tech diving was get on a simulator and see what different mixes do for you. I still do that. So I look at the planning on a laptop for what I would do at home for planning depths mixes etc. The fact that Cobalt has many of those same options like RMV / tables is great so when out in the field and a dive gets blown and no laptop one can still plan a new dive with relative ease.

 

[RonR]

Ron, I have not run the simulation hands on but there are lots of features like the table summary that other recreational oriented computers do not have. I would need to do some hands on to give you substantive feed back. But from my limited cyber diving with Colbalt it has far more than more "recreational" computer and some tech computers. Let me as naive question. What simulation is possible via the laptop software.

None, really- it's just dive logging and updating of the firmware for Windows. For Mac we supply a firmware updater, but rely on Mac Dive or Dive Log Manager for dive logging. We generally felt that the Cobalt's onboard simulator would be more capable, and our agreement for full RGBM is for use on a dive computer only, not for a desktop simulator. And anyone wanting to do serious simulations is likely to get dedicated software for it.

My comments regarding planning is more from the interface perspective. Having access to everything visually vs several screens as well as keyboard/mouse vs video controller. So no matter how many features are there unless there is a keyboard and screen real estate there will be limits.

That is the limitation- fewer pixels, limited interface- but with good design you can make the simulator fairly easy to use. I think we have room for improvement there, particularly for inputting square or stepped profiles. The biggest interface limitation I think we have is the need to go to alternate screens to change mixes or switch depths, where with desktop programs you could change mixes more easily, probably on the same page.

One of the things that I was taught when learning my tech diving was get on a simulator and see what different mixes do for you. I still do that. So I look at the planning on a laptop for what I would do at home for planning depths mixes etc. The fact that Cobalt has many of those same options like RMV / tables is great so when out in the field and a dive gets blown and no laptop one can still plan a new dive with relative ease.

That was the idea, to have a waterproof planner you could take with you, one that knew your history and status. Also there is the way that the no-stop time calculator is implemented- you can change the mix or the surface interval on the same screen that lists your current no-stop times, and it will update no-stop times on the fly based on what you input. Good for recreational dive planning, but also for illustrating the effects of different mixes. Cobalt Guide:No-Stop Time Calculator

Ron

---------- Post added April 1st, 2013 at 10:25 PM ----------

<snip>
Again, I'm excited to learn the model really does morph in the realms of Tec diving. I don't want to devolve a discussion about features into a bubble model debate as this will crash into the gutter before midnight.

<snip>

Thanks for that. If we meet one on one I can tell stories, but for here I'll just say that pretty much all currently used algorithms have good records, and appear to be converging around generally comparable results. You can have different paths to the same end. As the people responsible for coding and implementing one of the more complex algorithms in a real time dive computer (fully iterative RGBM), as well as doing the same for it's more conventional "folded" variant, we often feel there is much silliness and inappropriate ego involvement when these subjects come up. And often limited understanding. As I said above, how the algorithm is used in a real time, messy, not-always-following-the-rules diving environment is a very important and under appreciated factor- that is where most of the complexity and difficulty in developing dive computer firmware lies.

My favorite observation about algorithms was from Erik Baker, who said (I'm paraphrasing) that "all decompression algorithms are attempts to draw a bright, clear line through a fuzzy gray area." That sounds about right to me.

Ron

 

[Scared Silly]

Ron, I have been doing a bit more reading of the manual on the Cobalt and came up with a few thoughts. Gauge mode, while I am sure it was considered it might be worth looking at again given the above discussion on He.

The other is the QD. From previous discussions I know the male side is built into the computer whereas the female side can be decoupled from the hose. I would suggest that computer be built with a std 7/16 thread which could if needed be attached directly to a HP hose. While adding the male QD would increase the length 3/4" and while some would argue another o-ring failure point it would be more for the ah crap situation where someone drops a cylinder on the male QD damaging it. At that point the computer can not be utilized fully. And putting into a BCD pocket is asking for it to be lost.

FWIW another well known company sells their QD separately and it is one of the best on the market. I mentioned that because Atomic's swivel hose is the best in the market and they make their universal swivel hose for others. So if available separately I am sure a few QD kits would be sold. It would also facilitate those who might want to use an SPG while teaching in the pool (going back to the person who made the male portion for that purpose).

 

[charlier]

Ron, very nice discussion. I have only a few thoughts for a refresh. 1) reduce the blocky size and make it thinner. Please do not reduce the screen interface size. My wife uses a magnifying glass for the small fonts. I know she need prescription reader lens. 2) dive planner/simulator thoughts. Here I feel that the Liquidvision x1 has a superb tap interface and a wonderful simulator. One can control and change all variables. I realize that this computer is not designed for the recreational diver, but do consider its features/interface for dive planning. The open source for dive algorithm is nice. 3) as a scientist that dives for work that uses computers everyday on the boat, consider integrating Cobalt software more tightly with the Mac OS and/iOS. 4) expert/tec mode might have some merit, but this will take some beta testing.
Thanks for a great product! Charlie

 

[RonR]

Ron, I have been doing a bit more reading of the manual on the Cobalt and came up with a few thoughts. ...

Gauge mode is something we have thought about, and that often comes up as a question. For a recreational computer the main problem we see is that if it is entered for a dive, It invalidates any subsequent deco calculations for a significant time afterwards. It's kind of the equivalent of a lock out, which is something we wanted to avoid. I have wondered what is really the goal in having a gauge mode on a computer- is it just to turn off deco calculations? To avoid alarms? To have a different, simpler display? The computer is, obviously, always a gauge- just one that does extra calculations. I'd like to understand better what divers want to accomplish (or not accomplish) with a gauge mode.


The observation about the QD is also something we discussed in development, knowing most QD's are sold as add ons. We felt it was something most Cobalt owners would want, considering the nature of the computer, and it would be more expensive and add size to have it as a separate bolt on. So that, like downloading capability, was just included from the get go. But you are right, the way we did it means that if the QD is damaged, the air integration is gone. If it were a standard fitting on the Cobalt, with the male QD attached, it gives a way out if the QD is damaged, or (more likely) the internal O-ring goes. It's worth reconsidering this as we rework the design.


We have a dealer who has made up adapters that allow a SPG to be attached to the Cobalt hose with a QD- it's something Atomic should probably make for those who teach and don't want to log hundreds of pool dives on their Cobalt. Your point about aftermarket QD sales is well taken.


Appreciate the ideas.


Ron

---------- Post added April 18th, 2013 at 07:44 PM ----------

Ron, very nice discussion. I have only a few thoughts for a refresh. 1) reduce the blocky size and make it thinner. Please do not reduce the screen interface size. My wife uses a magnifying glass for the small fonts. I know she need prescription reader lens. 2) dive planner/simulator thoughts. Here I feel that the Liquidvision x1 has a superb tap interface and a wonderful simulator. One can control and change all variables. I realize that this computer is not designed for the recreational diver, but do consider its features/interface for dive planning. The open source for dive algorithm is nice. 3) as a scientist that dives for work that uses computers everyday on the boat, consider integrating Cobalt software more tightly with the Mac OS and/iOS. 4) expert/tec mode might have some merit, but this will take some beta testing.
Thanks for a great product! Charlie

Charlie,
Thanks. We have no intention of reducing the screen size- we want to move in the direction of more legibility, rather than less. One of the changes in the recent firmware update was increasing the font size of the numeric compass heading- that was a customer request. The case is probably overkill strong, and we may well be able to make future generations smaller. The thickness comes mostly from the battery being in its own watertight compartment.


I would really like to hear what planning/ simulating features you would like to see that the Cobalt doesn't now have, or would like to see implemented differently. One thing we have planned for a future update is to allow entering depth and time numerically, to facilitate the rapid simulation of square or stepped profiles. It just hasn't made it into the firmware yet.


Also wondering if you could expand on what you mean by greater integration with Mac OS- I can imagine bluetooth integration with iOS devices- is that the sort of thing you mean?


Would like to hear more of your thoughts.


Ron