Discrepancy between my computers during the dive

[lexvil]

But the difference was not 3 minutes. It was 10+ minutes. Same algorithem, same dives, same everything. And simple dives as can be seen in the graphs.

It's not a minor difference and shouldn't be dismissed.

The question is; which one is right vs which one do you want to be right?

 

[pisauron]

The question is; which one is right vs which one do you want to be right?

Seeing as 5 other buddies in the dive didn't go into deco, I'm guessing what was right with high certainty.

 

[pisauron]

Hi @pisauron

No, CNS% is time of exposure and ppO2. It doesn't matter if there was deco or not.

The discrepancy in CNS% is interesting in addition to the NDL difference. I would alert Ratio to both in your communication.

If your Ratio computer is operating at the wrong atmospheric pressure/altitude, it might explain both the shorter NDL and the higher oxygen exposure. Can you find the surface pressure for your dive on the Ratio as I am able to do for my Teric in the Shearwater Cloud?
View attachment 839339

I am able to find the surface pressure for the Ratio but not for the Garmin. The Garmin doesn't show in the logs.

The Ratio Atm. Pressure was 1018 mBar.

 

[lexvil]

Seeing as 5 other buddies in the dive didn't go into deco, I'm guessing what was right with high certainty.

No one got bent either right? There are no absolutes on these, its software guess-o-rythems, the best way to avoid this issue is to get two computers from the same company and era, and if you don’t want to veer to far off from your buddies you should probably all use the same computers, sell the Ratio and buy another Garmin or whatever.

 

[pisauron]

No one got bent either right? There are no absolutes on these, its software guess-o-rythems, the best way to avoid this issue is to get two computers from the same company and era, and if you don’t want to veer to far off from your buddies you should probably all use the same computers, sell the Ratio and buy another Garmin or whatever.

Or we can just say the Ratio algorithem is broken somehow and point this to the company and maybe help others in the community.
How about we be more helpful...

 

[scubadada]

I am able to find the surface pressure for the Ratio but not for the Garmin. The Garmin doesn't show in the logs.

The Ratio Atm. Pressure was 1018 mBar.

So much for the pressure being an explanation for the short NDL and higher CNS%.

It will be interesting to see what Ratio says. I've exhausted my guesses.

 

[lexvil]

Or we can just say the Ratio algorithem is broken somehow and point this to the company and maybe help others in the community.
How about we be more helpful...

It may be broken but unless you get a few more of their computers involved it may just be this one. It’s still just a software difference otherwise.

So far to many treat algorithms as a religion in my opinion.

 

[dmaziuk]

Not anymore as the Heinrich Weihkamp died.

This means there is a difference between the programmers about the 'offgassing'.

As a programmer I promise you there isn't. It's a (relatively) simple function of delta-pressure and time. That why we like ZH-L so much: it's simple.

Since binary CPUs operate on integers and the formula: on real numbers, there is a CPU and implementation-dependent round-off/truncation error that can accumulate over time and result in a noticeable discrepancy, but that would a) take a very long dive and b) never result in one implementation "not off-gassing".

 

[eja12a]

Please share what you find. The discrepancy is definitely concerning.

I also dive with two different branded dive computers running the same Algorithm. A Garmin on my left wrist and a Shearwater on my right. I've never seen much of a difference between the two. NDL, Depth, and Surf GF are pretty much spot on. Occasionally, I may see a slight discrepancy in Depth, NDL, or Surf GF, but I'm only talking about 1', 1', or 1%. And that could be attributed just to differences in where my arms are / have been, as well as sensor tolerances. Certainly nowhere near what you are seeing. Mine are close enough that I'm confident that they are right enough.

That’s the setup I use and I have pretty much the same experience. My Garmin G1 usually reads slightly deeper (maybe a foot or so) and generally clears my deco faster than my perdix. But we’re talking maybe a minute faster than my perdix, so they’re close enough I would feel comfortable surfacing when it clears.

 

[pisauron]

Updating the Ratio response:

Hello Mark,
We have analyzed the logs you sent us, from Friday to Monday, with particular attention to the dives made on Sunday.

It is important to remember that we are talking about the "theory" of decompression. Although there are commonly accepted guidelines, there is no "exactly correct" answer regarding the profile to follow. For a single dive, two computers using the same model with the same settings will give similar results, but the calculations and factors to consider increase exponentially in the case of repetitive diving, or multiple repetitive dives (being indeed an algorithmic calculation).
Even minimal differences in the implementation of the model to obtain an "algorithm" (consider that the original Bühlmann model is written in prose and that Baker's gradient implementations also have margins of interpretation), along with necessary rounding, can lead to having "different" results in cases like this. (Although at the end of the dive both computers did not report "mandatory" deco stops having recalculated the non-necessity to perform deco stops.)

In addition to the above, in this specific case, we can see that the two computers start the repetitive dives with different CNS values. The CNS value does not influence the decompression calculation (at least it does not on the RATIO, but I believe Garmin does not consider it either), but it might indicate that the two computers off-gas the tissues slightly differently.

To have a third term of comparison, we set up Subsurface with BUL settings and gradients 70/85. The result from Subsurface would have been even more conservative compared to both dive computers. (in the images, the deco suggested by Subsurface is shown in green, the one by RATIO in red).

So, in summary, even if two implementations follow the definition of a decompression model to the letter, they might process "different" decompression plans. In practice, these differences should be small (like a small percentage of the total decompression time) and therefore be obscured by rounding and the systematic error margin that the implemented algorithms necessarily adopt. These differences become more visible with multiple repetitive dives that "stress" the algorithm more. (after all, we do not have precise information on everything that is happening in the body of a diver).

At your disposal
Best Regards
Mario
Ratio® Computers - Support