DIY Research and Development

[gello]

This query is aimed mostly at engineer and programmer types.

Over the years I have built probably 3 dozen hardware/software
devices for the boat to perform automated or unavailable commercial functions.

The projects includes circuit design, board layout, production, build, and programming and software (most have single chip micro processors onboard).

Is there anyone out there interested in collaborating or contributing
to R&D on public domain hardware/software for cruisers?

For example, my last product was a 3 to one nema multiplexer with
buffering, flow throttling, and filtering (lifted from the unix kernel and modified).
It allows me to combine all the NEMA sources on the boat onto a
single wire/stream that may be shared by many instruments and a computer.

It involved about a dozen components and about 1000 lines of software.
Total cost was under $10.
Commercial units are around $200 and do no buffer management or throttling.

I am interested if there are others doing such design and is there any future in publishing instructions, diagrams, and software for the DIY
crowd on the forums. I know OpenCPN seems to operate in this manner
for software. I am looking for like minded individuals to do hardware/software projects to be released akin to the GNU license
into the public domain for the benefit of cruisers.

Anyone interested? I have an idea for a set of wireless
touchscreen instruments to integrate
with OpenCPN as well as boat hardware.

cheers

g

[Microship]

yes.... please see nomadness.com for context. I've just been writing about the need for more open source and DIY-geekery action in this community... I'm using about a dozen Arduinos on my boat and publishing the designs.

Cheers,
Steve

[gello]

i've been using mostly pics due to their peripheral density and
internal clocks. Many things i can do on a single 8 pin dip with
a simple zener power supply.

I write in c and assembly for the micros and use
a mini cooperative multitasking OS i wrote
and a built in debugger/downloader so they can
download new software and "burn themselves".

i write on windows or linux.
use apache webserver, javascript, and php with native
c/c++ methods where needed.

currently debating whether to use the ipad/iphone/itouch or android as
my standard handheld platform. (comments?)

i am not overly attached to any platform
though as much as i hate windows i find dealing with apple
has become cumbersome in the extreme.

I am looking at microscopes and tools and techniques for doing
low volume surface mount as many of my favorite chips have become unavailable in through hole. (any suggestions on smd?)

i have working designs for a number of things.
1) spectra watermaker ekg type monitor (amps, pressure, flow, voltage all
plotted on moving graph)
2) multi level tank monitors
3) controllers for fans to monitor and vent various devices (such as propane water heater)
4) nema combiner
5) SSB/modem isolator
6) power supplies of all kinds
7) "retrofit" wireless controller for old autopilots
8) wireless windlass control
9) LED driver circuits (constant current, NO dropping resistors)
10) hall effect current shunt monitors
etc etc etc


i have a few ideas about where i want to go with a new instrument design
but want to discuss platform and native versus hosted code alternatives
with someone.

do you think these forums are a good place for the actual discussions
to develop requirements or is that asking for chaos?

thanks for responding

g

[JRM]

Originally Posted by gello

I am looking at microscopes and tools and techniques for doing
low volume surface mount as many of my favorite chips have become unavailable in through hole. (any suggestions on smd?)

<SNIP>

do you think these forums are a good place for the actual discussions
to develop requirements or is that asking for chaos?

thanks for responding

g

Generally speaking, a steady hand combined with a good iron and a lapidary headband magnifier will accomplish 75% of what can be done in SMD construction without true fabrication facilities. Solder paste and a toaster oven probably gets you another 15%. I've built several ham radio projects from SMD kits for myself and others and it just takes a little practice.

I think this would be a great place to start. I'm no engineer, but I'm a fair hand with an iron and love hacking on stuff. My boat is converted to electric propulsion, so I'm definitely not above experimenting on her. If needs require, there are other development sites that can be tapped, and probably will need to be for repository services.

JRM

[gello]

Yes, that looks like a workable solution.

Trouble is my steady hand is a lot less steady and after
more than a decade of living on the boat before LED's my eyesight
is shot with all the low light.

I am looking at a 10-30x top lit microscope for around $135.
A 1/32" tip temp controlled iron.
Solder paste and lots of liquid flux.

smd is the start, but the goal
is to get to TQFP footprints which gets me substantial power.
Microchip sells 16bit processors with all the standard AtoD/timer peripherals, including ethernet and CANBUS(NEMA2000 basis), with an internal oscillator.
There is a sourceforge available embedded linux port for this processor line.

Pretty much a one chip
solution for everything but level shifters and line drivers.
Not bad for $6 bucks.

ever done TQFP with the magnifier solution?
looks possible if you are steady.
don't know if i could do it on a boat.

which brings us back to DIY.
Do you think SMD solutions are beyond the typical DIYer?
They require a board but there are lots of places that will
print and mail a board for $2.50/square inch if we publish the board files.
Its the soldering on that tiny form factor that worries me.
Should we limit to through hole solutions for things published to a public forum?


thanks

gello

[carsonbaker]

YES. I'm very interested in this.

My background is in computer science and I have some experience with embedded programming (AVR & PIC architectures). I can also program for Android and iOS. I'm handy with a soldering iron, but I don't really have the experience to design circuit boards. I'm also no good with mechanical engineering. But programming? I can do that.

At the moment I'm working on retrofitting an old Autohelm tiller pilot with a custom head unit and wireless control.

Count me in.

[JRM]

Originally Posted by gello

Yes, that looks like a workable solution.

Trouble is my steady hand is a lot less steady and after
more than a decade of living on the boat before LED's my eyesight
is shot with all the low light.

That can be a problem, although I think you'll have more problem with part identification than placement (that type on those things is ***small***)

Originally Posted by gello

I am looking at a 10-30x top lit microscope for around $135.
A 1/32" tip temp controlled iron.
Solder paste and lots of liquid flux.

I use really thin solder, not paste for hand soldering. The paste is for the oven, although I'm not yet comfortable enough with the oven method to try a more expensive kit yet.

Originally Posted by gello

smd is the start, but the goal
is to get to TQFP footprints which gets me substantial power.

<SNIP>

ever done TQFP with the magnifier solution?
looks possible if you are steady.
don't know if i could do it on a boat.

There's no need to try and solder each pin of TQFP (and yes, I have the SDR Cube kit which has a TQFP dsPIC33). Here's a link to the sparkfun site which has a bunch of tutorials for SMD construction. They use the flood and wick method on SOIC, which I think is a bit much, as they're easy to do individually, but is really the way to go for finer pitches. Check out the videos, the guy is quick!

Originally Posted by gello

which brings us back to DIY.
Do you think SMD solutions are beyond the typical DIYer?
They require a board but there are lots of places that will
print and mail a board for $2.50/square inch if we publish the board files.
Its the soldering on that tiny form factor that worries me.
Should we limit to through hole solutions for things published to a public forum?

I think if you're looking to prototype, then the SMD adapters are the easiest to allow for standard breadboards, but there's no reason to limit the DIY stuff to that. I'm not much of a prototyper so I'll defer to those more familiar with it. I generally cheat and have an Arduino for messing about with a breadboard shield. It has limitations, but they're less than my own so it's always been fine

No, I don't think SMD solutions are beyond the typical DIYer, because I pretty much consider myself average or a little less in this arena, and I manage the SMD just fine. You'll probably find that you'll have a four or ten board minimum, so you'll be selling off some boards anyway, you could just add $10 to the price and sell it with the PIC premounted.

JRM

[conachair]

Originally Posted by gello

9) LED driver circuits (constant current, NO dropping resistors)

Now I think quite a few people would be interested in that one

[goboatingnow]

Diy SMD is very doable. I'd suggest a hot air solder de solder station. It takes a pit of practice to get right. Also low cost solder ovens are quite cheap even if the heat profile management is crude. Unfortunately as you pointed out we don't have a choice as many parts are not available anymore in tru-hole. I'm trying to get to grips with solder paste masks at the moment.

I find the biggest issue is placing SMD chips is placing them without disturbing things, wobbly hands.
Dave

[dpons]

I support free software, but I can't help with hardware. Time permitting, I could lend a hand with c, c++, java, web services, etc.

I have to confess that I prefer a low-tech boat , except for electronic charts for planning.

[gello]

I'm big on low tech boat with high tech toys.
Essential systems should be simple and field repairable.
Any non-essential system should have at least some fun factor to it.
Gives you something to tinker with in the anchorage at 3am.

Thanks for all the info on smd soldering.
I have done bits and pieces but never any homebrew SMD
large scale (in the commercial labs it was easy, we sent out for printed boards and machine stuffed them for everything).
All my home stuff has been thru hole and often point to
point till i worked out the kinks and printed a board.

The adapter boards are great but they seem to ask a premium for them.

I worked with a friend helping a bit with his homemade 3 axis CNC mill.
I have sent him eagle files and copper boards.
Lay the circuit out as isolates rather than traces and he uses a cutting
head to make circuit boards in front of your eyes. Way cheap but
a bit large for the boat. I send him designs and he mails me boards.
great for prototyping.

Since we have some programmers checking in now, I am looking for some guidance.
One of the hw/sw projects i am trying to start is a set of gateways
between NEMA2000/NEMA183 commercial sensors and handheld devices.
Apple and android devices are so cheap now, i want to put them at the helm as instruments (like the raymarine crap i have now). The gateways
would burp out UDP packets with the data from the NEMA buses onto wifi.
I want CONTROL as well as reporting. For instance, i want to run my
anchor windlass from my itouch over the wifi. (another cool gateway to build, and yes it needs bigtime security with active control).

So here is my initial debate which will play a big role in the architecture.

1. Do i write native apps for the handheld platforms and let them pick off the data they want based on screens,

OR

2. Do i use a embedded server and place all the intelligence there so the handheld devices are just javascript enabled browsers.

I love distributed architectures for their fault tolerance and speed, but
at the rate the platforms change it seems i will be rewriting code constantly with each new release and new platform.

A browser based solution will be simple to write and will work on any enabled platform. But now there is that central server forking up pages and
the associated bottlenecks as well as the single point of failure.
Lots of advantages though, including the fact it is portable enough to be tunneled over the internet so i have boat control and monitoring remotely as well.

So, any opinions? I am leaning toward the page server because i don't want to spend my life supporting this and the R&D will be so much quicker.
However, most ajax type apps look kind of clunky on the net and don't
seem very responsive. I am trying to imagine a wind indicator jerking about or a windlass that waits seconds to shut down.
The last web apps i wrote in an embedded environment were 8 years ago and had much the same issues. Of course it was in VZ and my server was a 386.

I would love some guidance to avoid wasting a bunch of time.
(I had chosen airplay and the native method as it abstracts the
underlying hardware so write once deploy multiple but then after, 3 months work, they decided to go from free to $400/copy so here i am addressing the question again).

thanks

gello

[gello]

sorry to be so wordy.
but one conachair mentioned LED controllers.
I have built some chained multiple LED controllers with the national chipset,
but i cannot build them small enough.
so here's an idea:

Why not build an LED controller capable of running all the LED's on the
boat and mount it in the panel where the heat dissipation is not such a problem
(some of those LED controllers in the bulbs get damned HOT). Then your
lighting wiring becomes an LED bus. Since it is constant current all the way from the panel all you have to do is plug in LED's, no hot controller per bulb so they get
cheap as well (i buy 3watt white leds for about $5 without the controller)
Now you have central dimming and there is no need for breakers or fusing on
the output side as a constant current source is self limiting.
Its also a heck of a lot more energy efficient than all those small scattered supply.

I still have some parts around for a switcher LED controller, perhaps its worth a try.

g

[goboatingnow]

Like the idea about the central LED controller. Good idea. Only drawback is that with boat wiring all interior lights would have to go LED at once But that's not a drawback

Other things that might be of interest, well to me anyway

One was a smart fuse panel board replacing all the breakers with electronic ones. The panel could re then re situated away and controlled by a small controller. Free up a he'll of a lot of space at most nav tables.

As to nmea 2k. Unless ( we, it,one) can acquire a license from nmea that whole area is a no go. And it's expensive. I'd like to develop a simple (!) star bus nmea 2k hub to provide redundancy like Ethernet systems but again the damm license cost

I'm an embedded systems designer. I'd like to help, not too much time available at moment though.

Dave

[goboatingnow]

One of the things that would be a useful building block would be to develop an open source NMEA. 2k protocol stack. That's a big hurdle currently to low cost projects.

Dave.

[Whimsical]

Originally Posted by goboatingnow

As to nmea 2k. Unless ( we, it,one) can acquire a license from nmea that whole area is a no go. And it's expensive.
Dave

Why?
It is just Can Buss with a proprietry set of PGN's. Nothing wrong with reverse decoding them, many already have been. I saw a site that has done a lot of work on it but i can't find it just now, i found it originally from an article on Panbo.

I have a few wants for my own instruments and bought an Arduino to play with but i don't think they have enough HP So am eagerly awaiting the Ras Pi Raspberry Pi | An ARM Linux box for $25. Take a byte!

Mike