Question on decompression models?

[rongoodman]

If you haven't already done so, take a look at Deco for Divers, although it does pre-date the current controversy regarding deep stops.

[Jax]

If you haven't already done so, take a look at Deco for Divers, although it does pre-date the current controversy regarding deep stops.

Dec 2014 issue:

https://smile.amazon.com/Deco-Divers.../dp/1905492294

[jlillest]

... that is read by a machine / computer with + / - 2% error.

Can you elaborate? Are you referring to the algorithm having problems? The hardware we have these days allows us to perfectly model whatever algorithms are in use, so I'd assume any errors would be in implementation, not in the broad category of all computers. Since most dive computers are closed source, only the manufacturers could actually determine any of this to any degree of accuracy.

You're correct on the sensor accuracy, though we could account for this in software if desired.

I can't comment on the algorithms myself. I have better material to put me to sleep at night.

-Jon

[Benderr]

Coming back into diving after a 7 year break I came across an interesting video on YouTube about bubble model decompression profiles. My question is have we abandoned theses profiles all together and what are the current dive computer companies using now? When I stopped diving we were using V-planner to plan our long deco dives.

+1 for Deco for Divers. I also like Deep Into Deco

My understanding of the current consensus is that VPM is not desirable because at deep stops, the slower compartments are still on-gassing. This has allegedly caused a number of "undeserved" hits, especially with longer and deeper dives. Another issue is that VPM is proprietary, so no one knows how close you're coming to the line, and what, exactly, the conservatism settings mean (how they change the math).

Bühlmann gradient factors allow you to customize conservatism on the first stop, and when you leave the water (how close to the M value you are). It's also open source, and well-documented, so that you can understand exactly what's happening when you change those gradient factors.

[rossh]

Can you elaborate? Are you referring to the algorithm having problems? The hardware we have these days allows us to perfectly model whatever algorithms are in use, so I'd assume any errors would be in implementation, not in the broad category of all computers. Since most dive computers are closed source, only the manufacturers could actually determine any of this to any degree of accuracy.

You're correct on the sensor accuracy, though we could account for this in software if desired.

I can't comment on the algorithms myself. I have better material to put me to sleep at night.

-Jon

In MultiDeco, there is an import feature where one can load a background image of an existing dive log graph, underneath a drawn profile. The idea being is to re-plan a new dive that's drawn over the top of the image, so one can try to match up the profiles, and discover what settings or profiles are needed to match up the dive record back to the planning tool.

It's a fiddly process, but once done, you can really check/see if the dive computer is putting out the GF numbers it claims it does. And it gives some insight to the consistency of the model implementation in the dive computer, particularly when most dive records are anything but straight lines.

.

[rongoodman]

I don't believe VPM is proprietary--there are multiple different implementations of it.

[jlillest]

In MultiDeco, there is an import feature where one can load a background image of an existing dive log graph, underneath a drawn profile. The idea being is to re-plan a new dive that's drawn over the top of the image, so one can try to match up the profiles, and discover what settings or profiles are needed to match up the dive record back to the planning tool.

It's a fiddly process, but once done, you can really check/see if the dive computer is putting out the GF numbers it claims it does. And it gives some insight to the consistency of the model implementation in the dive computer, particularly when most dive records are anything but straight lines.

Sure, makes sense. I understand using tools to validate from the outside looking in, but you're never going to fully understand how accurate it is until you can look under the hood. No manufacturer in a competitive field is going to do this, unless their competitors wouldn't be able to use that information to make their product better.

So, the point I was trying to make is that saying that our computers have +-2% error is too broad to make sense and is ultimately misleading.

-Jon

[Jax]

So, the point I was trying to make is that saying that our computers have +-2% error is too broad to make sense and is ultimately misleading.

-Jon

Not the dive computer. The actual device that converts the electric signal from the sensor to produce a number for the user to see.

[oya]

Not the dive computer. The actual device that converts the electric signal from the sensor to produce a number for the user to see.

The transponder.

Weird though, that when I wear two or even three computers all those transponders are so very off by the exact same range so that they all give me the exact same erroneous reading.

[jlillest]

Not the dive computer. The actual device that converts the electric signal from the sensor to produce a number for the user to see.

So, in other words, the computer? Whether manufacturers are using external electronics or not, the same hardware (more or less) is being used and it's all part of the computer.

We can delve into details and claim that some manufacturer's are using low-bit DACs and whatnot, but I think +-2% error across the board is a wild claim. Negligible would be a better claim, unless you're going to cite specific computers.

-Jon