Open Source Dive Computer

TR300

Hello to all,

for all of us who want to know exactly was is in the dive computer, there is now an opens source dive computer available.

All information are open source, including shematics and source code.

the design is very nice and handy.



the pricing is also "smart"

and it is "made in germany"

here is the english link to the webpage

HeinrichsWeikamp GbR

what is mising today ? just an O2 Sensor interface, but I think is it just a question of time when it will available. (I'am just thinking about to make it)

regards

Holger

__________________
If it makes bubbles it means its brocken
Visit: www.tr300.com, Missions: www.rebreatherteam.de

Christoph

Danke für diese Information!

I just wrote them a mail with a couple of technical questions and got the reply only 2 mins later. Impressive

What´s currently not on the homepage:

• the housing is rated down to 100msw (tested to 120m)
• there is a piezo buzzer built in
• as of mid-february they will install a watertight 5 pin connector on demand, at the moment the fisher plug can not be used for underwater connections (e.g. O2 sensors)

__________________
Cheers, Chris

When my basement became full, I believed to have enough divegear. Truth is, I had to get another basement.
http://www.sharkforce.at

tdzao

Unfortunately the CPU has only 96K of flash and almost no RAM, making it useless for technical diving. I'm uncertain as to how the designers thought you could fit any type of useful decompression program in such a tiny amount of memory! It's really too bad -- otherwise it would have been a nice unit.

Maybe they will change their mind about the CPU.



Eric

Dutchy

INHO an uneducated comment. There's no OS and most is written in assembler (in a powerfull RISC set). One can create multimedia devices using this thing. I once wrote a PLL synthesized tranciever with keyboard input and (smart) LCD display is just under 1k !! It's really that compact.

iow 96K is a helluva lot!!

Assuming you have some 2k of RAM left and use the *program accessible* EEPROM you could store profiles and ppO2 in *major* intervals (1min) in EEPROM (because no battery is needed) and minor intervals (20 sec) in RAM.

How much data?
depth 10 bits
ppO2 8 bits per channel = 24 bit ( can be reduced by averaging and storing as deltas thereof -> 16 bit (or less) add to that storing of alarms etc and you'll end up with 32bit or 8byte.

Even when you don't store things efficiently it will last for 6.4 hours.
With some compression (or just recording average levels of pPO2) my guess is that you can multiply that by a factor 2-4.

I wouldn't write it of that fast... That is if the pPO2 input becomes available... The proc has the necessary input channels so from a hardware pov it's covered.

__________________
= This post is environmentally friendly. It is composed of 100% recycled electrons only. =

fin

I guess tdzao is not so wrong by means that assembly programming is not that 'open' as it should, having in mind the unnumbered technical diving source codes that are already out there.
I must agree that todays hardware is so cheap that it would make linux's usability worth sacrificing assembly's absolute stability.

Taucher Raucher

Eric is the same Eric as X1 computer no? shame on you!

I have already the Open source computer you are quick to criticise. I can confirm that is very good decompression device. It come with GF/Buhlmann software preinstalled. I have a test model with bubble model and CCR implementation already for dive testing and soon will have the external o2 monitor option installed.

for deine information. This open source computer has the very same OLED display as your own X1 but with this model the battery is replaceable by the user and not simply discarded like your version maybe even just 1 year of use. The OSTC has also push buttons unlike your very strange tap system which is not popular and useless with handschuhe.

I think you have now seen a superior computer alternative to your X1 and most importantly it is CHEAPER!!!

If the X1 computer is so good why is it now revision 3 of the OS already!!!

Taucher Raucher deine Smoking Diver

__________________
I smoke therefore I dive!

tdzao

For your information, V-Planner Live takes up more than 350K of flash memory and so it could never fit in 96K. Further, a real bubble model requires a very fast CPU (32-bit, 20MHz+).

The open source computer is a good device, and useful for some divers.


Eric Fattah
Liquivision Products

jradomski

with efficient coding and a good compiler there is more than enough ram - 3.3K is more than 50% more than I processor I have implemented a full GF program in.. IF you need alot more than this you need to learn to code optimally or switch to a better architecture..

The biggest chunk is the tissue loads.. they can be done with 128 bytes ( 32bit floats * 16 compartments X 2) for the real load and another 128 for a working copy..

if you use an intersema pressure sensor then you need to keep a copy of the cal data.. after that, there really is not a big need for ram..

__________________
Joe Radomski
CCR Trimix Instructor Trainer
ANDI Instructor Trainer Director #10

All posts are personal opinions and DO NOT reflect any affiliated agency unless specifically stated.

jradomski

The normal GNU library is HUGELY ineffeicent and makes code much larger than it need be.. (switch development to Crossworks and you will see a code space reduction) Plus the instruction set of most 32bit chips wastes ALOT of code space.. A good 16bit chip will have similar performance and easily use half the resources..

__________________
Joe Radomski
CCR Trimix Instructor Trainer
ANDI Instructor Trainer Director #10

All posts are personal opinions and DO NOT reflect any affiliated agency unless specifically stated.

tdzao

Of course it is possible to squeeze certain algorithms into 96K of flash and 3.3K of RAM.

However, any open-source / open-platform computer should be designed to make it very easy to program, even for someone who is not an expert at generating tiny, ultra efficient code.

The X1 was designed to use the inefficient GNU toolset, and it comes with way more flash (512K) and more RAM (32K) than required, and the CPU can run 32-bit instructions at 60MHz, allowing for very inefficient code to work smoothly. This allows programmers who are used to PC programming to still generate usable code, even by doing 'cut-and-paste' from Microsoft Visual Studio.

In contrast, the F1 had an 8-bit CPU with 32K flash and 2K of RAM. With such limited resources, it was not feasible for anyone other than the designer to actually implement code, as there was hardly room for anything, and everything had to be super optimized to work at all....

The X1 also keeps a RAM shadow of the entire OLED display (8192 bytes), allowing very easy RAM based graphics operations with a single 'update display' function which copies the RAM shadow to the actual display.



Eric Fattah
Liquivision Products