I'm trying my hand at using SMD parts for the circuitry which will drive a hand made air-core movement.

The electronics is based on a PIC16F648A micro controller. It connects to an L293DD which will drive the air-core (or a Switech stepper gauge movement). The PIC takes data from a multi drop bus. That's what the pair of DB-15 connectors is for.

I'm using "giant" SMD parts. The resistors are .12" by .06" and the ICs have a .050" pin spacing. The finished board size is 2.4" by 2.4".

Well, the circuit boards were delivered yesterday, and I've (mostly) assembled one.



This is my first project using surface mount components. Other than some issues resulting from my not properly cleaning the board before soldering, assembly proceeded okay. I used a 2X magnifying glass and a soldering pencil with a 0.8mm tip.

It's configured for in-circuit programming using a PICkit2. The PICkit recognizes the microcontroller and can read its memory.

At this point I don't have the firmware converted to program it. Actually, I don't even have the Microchip programming tools loaded on my system. Guess I better do that soon.

Anyway... Depending on what firmware the PIC runs, the board can drive an air-core movement, or a Switech stepping motor gauge movement, or a small generic stepping motor, or up to four RC servos.

*cough*ATMega328*cough*



Great job Mike.

g.

The funny thing is that I have AVR Studio 5 loaded, but not MPLAB, AND I'm building PIC hardware.

If you replaced those big honking 15 pin connectors with some DIP pins on .100 centers, you'd save even more space.

g.

I'm maintaining compatibility with the instrument IO bus in Building Recreational Flight Simulators.

Although I started this with the idea of driving the DIY air-core movements, it's evolved into a versatile movement driver/interface. A Switech stepping motor movement will solder onto the other side of the board and make a compact, one board gauge. Or, you can get fancy with generic stepping motors or even RC servos. (Two of the connector pins are for faceplate lighting, others supply power.)

I'm working on the Nth version of the DIY air-core movement now, but will shortly build up a board with the Switech movement and post pictures.

Cool. Any reason that couldn't be compressed to fit the interior of a 2" gauge body?

g.

If you wanted to squeeze the circuitry inside a standard round body instrument body, I would remove the connectors and run a wiring harness between a round board and rear-panel-mounted connector(s). One possible concern is cooling the motor driver chip. Most of the heat comes out the 8 ground pins. The chip depends on the ground plane for heat sinking. I'm sure there are other approaches, though I haven't researched them.

The biggest issue I see is getting affordable standard instrument bodies.

This board is actually configured to present a 2" round gauge face when the instrument is mounted behind the panel. The instrument internals are a stack of 2 3/8" squares.

An intent of this approach, and this particular board, is to make gauge assembly as simple and cheap as possible. By using SMDs I'm trading off a slight increase in soldering difficulty with a reduction in point-to-point wiring and mechanical complexity, at least when using the Switech movement.

BTW, I finally loaded the MPLAB IDE.

With all the talk of making it smaller and Mike's (meant to be humorous) reference to these SMDs as "giant", I just had to chime in.... Especially since SMT is what I do to pay for the sim. my day job

I had to snicker at the "giant" reference because, in the industry, that's not far from the truth.
1206 and 1210 packages (.120" x .060" & .120" x .100") were the standard about 10-12 years ago.
Nearly every passive component we place these days is 0402 to 0603 case sizes. And we do place 0201 components.
Actually the latest case size is 01005, which is only .010" x .005". (Think finely ground pepper.)

I was looking at the PCB thinking, "yeah, the 8SO could probably be shrunk to a SOT, the 20SOLs could go to SSOPs, the passives to 0805, etc.."
And while our machines help to make the job faster...re-work is still done by hand, under a microscope.
So when you talk about soldering difficulty?.....I feel your pain.

I started the board design with 0805 passive components, but they occupy such a small percentage of the available space there was no real gain relative to using the more easily soldered 1206 components. The SOIC devices went on easier that I expected. Overall I'm quite pleased. For a modest increase in soldering difficulty I eliminated almost all point-to-point wiring in the gauge. All that's left are the wires to the air-core movement and the faceplate LEDs.

I must admit that I've looked longingly at stereo microscopes.

Mike is that board single-sided?

It's doubled sided. Red is the top layer; green is the bottom.

For me, I think using SMDs is probably easier overall. Maybe it's because the soldering isn't a big issue? (That's what lunch breaks at work are for. )

When I etch my own PCBs, I try to keep it 1-sided and not have to worry about screwing up the alignment. (I just add a couple jumper wires, where I can't get around crossing 'em.)
So, when I can use SMDs, it eliminates most of the drilling. And I really hate all the micro drilling!

And when you have to go 2-sided, there are no sleeves in the vias, either.
So, you end up doing twice the soldering, with thru-hole, to tie the top and bottom annular rings.


Maybe I should get out of the PCB making game and just let a board house do it?!
I'm curious, who did you use to fab it? And did you get stuck with a big setup fee, for the small run? And if it's not too personal....how much?

I spent years in the DIY circuit board camp. Early in my engineering career I did board layout on mylar with paper tape. We'd send the artwork to a photo house for reduction then do the board production in house.

I continued to make boards as a hobbyist up until a few years ago. The low cost of board prototyping services made the DIY effort not worth it. While I can make a board cheaper, I can't make it better.

Currently I use ExpressPCB. They offer a "miniboard" service which is 3 double-sided 2.5" by 3.8" boards, no solder mask, no legend for $51 plus tax, S&H. You must use their software (which is free). The software will cost out other sizes, options, and quantities, as well.

These particular boards were done using the miniboard service. I added a small 2 chip SOIC prototyping area to fill the available area. (See the Aug 9th entry on my site.) Total cost as delivered was $65.75.

I got the firmware together and loaded into the PIC. I added a bit of self-test code since I'm currently too lazy to re-create the PC-based test code I lost awhile back. The self-test code simply increments the data for the air-core movement shaft position each time a PWM cycle completes then sets a flag indicating new data. The PIC executes its code as though it had received a command from the host computer.



As I write this, the PIC is causing the air-core movement to rotate its shaft about once a second, so mostly I'm happy.

The part I've not tested is the comm functionality. I've used the firmware in other projects, so I don't expect any big issues there, however... Something I just read in the PIC docs make me think that using the receive functionality of the USART may preclude using the TX pin for a general output pin. Guess I'll just have to crank up MSVC++ and re-create the PC-based test code.

As it turned out, I had to change the output pins driving the Air-core movement. Bummer, I had to hack my pretty SMD board. Ah well, at least now it works.

Even better, the board will also drive a stepping motor (as long as the firmware is reflashed). I have it driving a Switec automotive stepping motor gauge movement.