Because the frequency of a free running oscillator is unpredictable, changing with temperature etc, and could cause interference to the ships radio apparatus with the antennas being directly above the cockpit, I have chose a frequency of 16 MHz in a crystal type colpitts oscillator. This will only interfere with 16 MHz on the shortwave which isn't used by anything (probably because it's a microprocessor fundamental frequency) and a few kc's up and down from there.
By putting an LC (inductor-capacitor) circuit in series with the crystal, it can be varied, slightly, by a finger in close proximity.

Here's where it gets strange. As I did tests using 2 oscillators, one fixed, the other one variable by proximity, I noticed that the 2 oscillators will fall into phase lock when on the same power bus. This is to say that because they pull current during a certain part of their cycle, as their pulling power lines up, they tend to "see-saw" with each other, pulling together.
Because this wasn't what I had planned, I had to sway away from trying to use this effect because it would require extensive R&D.

The theory of operation is as follows:
The PIC chip's crystal will run at 16 mhz, and will be used as the reference for the other oscillators.
Each "button" has it's own crystal and LC. ***
The outputs of each of the 6 oscillators are fed into a diode switching network, controlled by the PIC, to be mixed with the reference crystal's frequency.
This mix is then filtered to only pass the lower frequency resulting from the difference between the two.
The op-amps output is turned into a trigger style pulse that can be measured by the PIC chip, which takes an average change period over time calculation.
If this change varies beyond preset delta parameters, the PIC reads that oscillator as "button pressed" and forwards the data to the MIDI box.

It seems to work thus far, and the PC board layout's "coily tracework" seems aptly spaced. The true test will be once behind 1/4" Lexan.

The "buttons" must also emit an audible "beep" from a little piezo buzzer on the main board. So one knows the "button" has been pressed.

**Other methods I researched were:
  Optics-using IR photo-transistors and emitters to sense finger proximity. Rain drops, dark gloves, bird poo, & intense sunlight made this a bad choice.
  Piezo-pressure sensing-Using piezo beepers properties to sense a finger tap in the vicinity. Problems were once again, rain, mechanical aspects of design ( I built a rather awful drum machine using these ) and temperature variation.
  Hall Effect Sensors- Would work great until you forgot to wear your magnetic finger thimble! I toyed with little magnets in rubber sheaths but it's too design intensive, plus they'd probably all fall off he he.
  The Tablet Pen - The newer Tablet PCs use a magical pen that will pull the mouse around even if you are 1" away from the screen! Pretty cool but.... I have a hard enough time finding my lost tablet pen in the living room or car, forget the whole ocean during a gale!
  Stupid Ideas - I'm too shy to admit online

***Originally, I was going to just use one crystal and switch inductors with the PIC chip, but because the proximity sensing is so delicate, any electronic switching method at that end would cause great variance, thus reducing the ability of the PIC the decided whether or not the signal was true. Also, all oscillators require a stabilization period that would vary and cause all sorts of intermittent problems.

After doing a corner test ( 2 pieces glued & screwed together) with 1/2" Lexan and Gorilla glue, I am confident that things will hold together well.

Routing in the sides 1/8" to receive the edges is a good method as it offers strength, alignment, and a rough surface to bond the glue with. Gorilla glue is waterproof, but has bubbles in it, from expanding during setting. as long as the bubble size is smaller than the join size there will be no problem. I will line all of the inner corners with silicone anyway. The face will be 1/4" clear Lexan.

Small SS screws inset will hold it together initially, but will be left in as they don't create a leak hazard if left alone after curing.

The bottom removable panel will be gasketed with rubber and silicone on the rough edge. The smooth edge will create the seal with some grease on it. Several screws will hold the bottom in place.

Attached to the bottom will be a vertical support that will hold the tablet PC, MIDI box board, Power Supply Board, and Interface boards in place. This way, to service unit, the case can remain in position while the "guts" can be taken in.

The bottom will have a waterproof 8 pin plug, and a fill nipple ( a standard tire style one should work ) to pressurize the box to, say 5 psi.

The power considerations don't end there though.
Several issues to powering up the unit, powering it down, and keeping data have arisen.

(i) The Midi box USB must be fully initiated when coming on from standby. The USB doesn't always recognize the fact that the thing is plugged in for whatever reason, so the MIDI box must be physically switched out (latching relay?) until after the startup time has passed.
From a "cold start" everything works fine, but takes a minute or so to get everything loaded and running. Not very good in a hurried situation.

(ii) The tablet won't run reliably on 12 volts alone. It must have it's little 12 volt - 18 volt charger thing. When I originally tested this, I was fooled as the voltage I was using was 13.8 volts. It works fine there, but drop that to 13 volts and sometimes it won't start up. Because the house batteries vary from 14 volts to 12.5 volts, this is unacceptable. Luckily we were able to find some Fujitsu 12 volt adapters on eBay.

(iii) When the unit is shut down, the PIC board must remain powered to occasionally check the "on button" and also to maintain the DRAM chip connected to it, in the form of a refresh cycle, so it doesn't forget options set, data tracking etc.(see iv)

(iv) Because Macromedia flash can't seem to actually write anything to the hard drive, ( Yay! One more annoying security feature!) once the system was shutdown to a cold shutdown, all configuration data will be lost. i.e. display type ( I have themes haha! ) user preferences, etc. as well as any tracking data that needs to remain.
Tracking barometric/temp/humidity may be done in the dashboard unit anyway, but there still needs to be some DRAM somewhere in the unit to store all of the junk that will be going back and forth.

PIC Unit will fire up PC if:
Engine is started
An emergency happens i.e. radar perimeter, overcharge, low battery
Dead Helmsman switch is not pressed within (5) minutes of warning beep while under way if helmsman had previously "checked in" before shutting down unit without "checking out"

Display light will shut off after preset time, if activated, and can be lit back up by waving near any of the "buttons" - like a mouse move screen saver.
 

I am now seriously thinking about placing a low profile aluminum heat sink on the back of the unit. This will help dissipate the hot air inside, without the air actually passing through the case. This will require some careful sealing considerations as aluminum, bare, can be corroded quite quickly in salt air.

Another solution would be to use a Peltier junction and small heat sink. This would consume a large amount of power though.

I can only hope that wind blowing by while underway will keep things cool enough. Only time will tell.

This'll work if
(a) the frequency of the transmission doesn't matter, &
(b) both stations do the same thing.

The best solution would be to include an RF sensor input to the control PIC that would cause any following instructions to be disregarded within a certain time period. This could be added on later.

*A lesson in integrated systems seems to be nosing it's way into the design. An RS-232-like CTS (clear-to-send) would sure be nice here!

My main concern is that the fingertips won't allow close enough proximity to the sensor array. This could be solved by increasing sensitivity with a start up test routine, but seems tedious.

A temperature sensor reading would work, but who's to say which gloves one might be wearing on that particular day.

The obvious solution is to NOT cruise northern lats he he!

Forum <- - -> Boatcam