Advice needed for (very basic) PCB design

Hi there,

For a project I'm working on I'll need to make some really simple PCBs that will be used to mount Zilog ePIR sensors in waterproof enclosures. I haven't done any custom PCB design before, so before I start ordering it'd be great to get confirmation that I haven't messed anything up.

I have two different designs; one is extremely simple, it is just a header for the ePIR, screw terminals for the wires and a resistor and a pot that will be used to adjust the sensitivity of the sensor. I decided to go for the pot soulution because I need each sensor to have field adjustable sensitivity. The board has a relatively large area free of components because the ePIR will be placed parallel to the board (it has a 90 degree header and I have used a right angle one on my board -- a backwards solution, but the straight-headered version of the ePIR in impossible to find).

The other board is a bit more complicated because it contains a shift register and a voltage regulator. The shift register is there because I'll be using many sensors (a total of 40) and running 40 separate signal wires would be uncomfortable. The voltage regulator is because the ePIRs need 3.3V -- and while I could probably have powered the whole shebang with 3.3V, I found that using voltage regulators was an okay way to avoid voltage drop problems.

There will be 40 sensors, I need a shift register for every 8th sensor, so there will be 35 of the plain boards and 5 of those with shift registers. The plan is to do as follows:

Arduino --> ePIR+74165 --> ePIR --> ePIR --> ePIR --> ePIR --> ePIR --> ePIR --> ePIR --> ePIR+74165 --> ePIR ... x7 -> ePIR+74165 ... etc.

I'll be using a push-pull line driver at the Arduino because the clock and latch wires will be around 7 metres long total (15-20 cm between each sensor board).

The sensors themselves are connected according to Zilog's application example. Pins 1 and 8 are ground and 2 is power supply, 5 is signal (low when active). 3, 4 and 6 are inputs; 3 and 6 aren't used, so they are connected to GND and 3.3V respectively, which is recommended when they are not to be used. 4, which is sensitivity calibration, takes an analogue input between 0 and 1.8V, where 0 is max sens and 1.8 is min. It's connected to 3.3V via a 100k trimpot in series with an 82k resistor, as recommended by Zilog.

The screw terminals I have used are from the Phoenix library, but I might be using TE instead depending on what's in stock. I assume they are interchangeable as long as they have the same pin spacing.

I imported the outline of the enclosure's mounting plate from the manufacturer's CAD files and placed it on layer 100, designing the outline of the board to fit it.

The plan is to order from Eurocircuits, so I used their DRC. Does anyone here have experience with them? Their prices are reasonable for larger quantities. I also considered OSH Park, but they are cheaper only for small quantities -- and quite a bit slower, and because I'm on a deadline I really need things to get moving fast.

I have attached the Eagle BRD and SCH files below -- if someone could take a look at them and point it out if there are any huge mistakes I'd really appreciate it.

Thanks in advance!

-CC

epir-reva0.brd (62.9 KB)

epir-shiftreg-reva1.brd (81.4 KB)

epir-reva0.sch (239 KB)

epir-shiftreg-reva1.sch (272 KB)

Can you attach JPGs

Sure!

CaffeineConsumer:

The first board, with some rearrangement of the traces, could be single sided.

Good point. Thanks for pointing in out! I made a single-sided version of it, but for some reason, that turns out to be more expensive at Eurocircuits ... Their pricing policy seems a bit erratic. But maybe it'll work out cheaper somewhere else?

I always consider jumpers (or use 1 ohm resistors) to avoid a double sided board.

Traces look a bit fine to me - may be within the minimum, but I always err on 'comfortable' if you really don't need minium. Also some sparation distances are smaller than they need to be - you've got traces coming off the pinheaders at cangles that take them close to other header pins when they don't need to pass that close...

CaffeineConsumer:
Good point. Thanks for pointing in out! I made a single-sided version of it, but for some reason, that turns out to be more expensive at Eurocircuits ... Their pricing policy seems a bit erratic. But maybe it'll work out cheaper somewhere else?

Why do the traces need to 'dance' around the pot tabs? Connect the centre one straight down and the left one straight left.

Thanks again for the input!

I tried to tidy up some strange stuff that the auto-router had done (including the bizarre "dance" around the pot), and increased the trace width from 12 to 16 mils (even 24 for 5V, clock and latch). At last I'm starting to figure out how this Eagle thing works (I think) ...

I still think the area around the shift register looks a little messy, but I guess it should work. I just realised that I forgot to include a decoupling cap, but I found a neat DIL16 socket with a built-in .1µF cap at a local distributor, so I suppose that's not a problem after all.

Maybe add a 100nF (surface mount) close to chip Vcc to GND

CaffeineConsumer:
Thanks again for the input!

I tried to tidy up some strange stuff that the auto-router had done (including the bizarre "dance" around the pot), and increased the trace width from 12 to 16 mils (even 24 for 5V, clock and latch). At last I'm starting to figure out how this Eagle thing works (I think) ...
I still think the area around the shift register looks a little messy, but I guess it should work.

Is it possible to move the shift register to the left to give you more space between it and pads A-G and also to move it down a little? If so, your traces can be made more logical.