They should both be SPI devices and actually sharing SCK, MOSI and MISO. Each should be on a separate Slave Select pin. This is usually labelled CSN on the nRF (Chip Select Not, as it's Active Low). The nRF also needs another pin for CE. This controls whether teh nRF is in Transmit or Receive mode.

The libraries being used should take care of asserting/de-asserting the respective SS pins, but it's not guaranteed. If in doubt then manually set their SS pins before and after using each device, to be safe. Only one device should be active at a time and then that will be the one to respond to any SPI data sent by the Master (the Arduino).

22AWG shouldn't present any issues in short runs for 1-3 standard 20mA LEDs in parallel.

Just received another order back, bang on 15 days again.

This one was 100mm x 100mm panels with 8 up 48mm x 28mm PCBs panelised.

They tested each individual PCB and marked two on one panel + one on another panel as bad. That's three PCBs out of 80. They actually sent me those panels (with 13 good PCBs) PLUS the full 10 perfect panels, so I've ended up with 93 PCBs instead of 80.

Another order of 5 up tiny PCBs on 50mm x 50mm also arrived with a friend.

These guys are still impressing me. It's not very often nowadays you can be absolutely happy with a service so it's a refreshing change. :-)

It sound like you have some short when your fan is switched off.

My guess is that your 9v battery is a PP3, in which case the reason you're not seeing the same overheating/melting issues with the 9v is nothing to do with the voltage but actually the batteries inability to supply high current.

I also suggest all grounds get commoned as you could also have a path back through other components. Strange things can happen when you forget to reconnect all your grounds together.

Without a drawing of what you actually have then your not going to get any further wit help.

1) Is it a single phase transformer?
2) DO you have 4 wires coming out of the transformer?
3) What colour are your wires?

In theory yes, you can, but you need to be very careful about what you are doing.

If you have 4 wires then colour coding may give you a clue.

What you need to do is find the correct polarity for those two windings as you want to connect them in phase and not out of phase. Connecting out of phase will be like applying two power supplies shorted together and you DO NOT want to do that.

I would advise you DO NOT continue unless you are a qualified electrician AND totally understand the safety precautions, hazards and risks associated.

If I was going to do this then I would look for wiring to indicate polarity of the windings.

I would then join two wires of the secondary together.

I would ONLY connect TWO wires, to leave me three ends, which I would ensure are all separated and safe for energising.

I WOULD NOT connect the other two wires yet.

I would then power up the primary and test the ouput of the secondary. I would be looking for stated voltage between each single wire and the connected two. Across the outer single wires I would be expecting to get double the voltage or zero, with zero being what I want.

IF I had double voltage then the secondaries are in series and I'd de-energise and alter my commoned connection.

IF I had zero between the OUTER wires then I'd go ahead and connect the remaining two wires to put the secondaries in Parallel as required.

*** AGAIN, I ADVISE YOU NOT ATTEMPTING THIS UNLESS YOU ARE QUALIFIED AND COMPLETELY COMPETENT TO DO SO ***

You don't need to run the LEDs at maximum current. Even dropping to 0.5 or 0.25 times will result in a very similar brightness level as the human eyes response to intensity is logarithmic.

It wouldd be advisable to learn how to use a transistor to switch higher loads from an Arduino output pin as you'll need this sooner or later.

However, in this case, you could just wire your LEDs in parallel and run each at less than 10mA to stay within the limits of the pin output. Running at 5mA each would be best, to give a maximum 20mA on the Arduino pin.

Set a variable as a timer record
Set timer to current millis()
Generate a random integer between 1 and the number of patterns
Do your Case statement to determine Which pattern to use
On each loop() check if the difference between current millis() and your timer is more than 10s
If greater than 10s then reset timer record and generate a new random number

Pointless, non-conforming, post of the week???

The non-P 328 has a different Signature to the P version.

This isn't present by default in boards.txt or averdude.conf.

You need to modify both files to add support for the non-P 328, in order to burn a bootloader and set fuses. Once you have a bootloader installed the you have to revert back to the standard 328P board in IDE to upload sketches over serial.

I believe that Nicks programmer sketch might support the non-P and save you needing to make the modifications. I haven't tried it personally but I think I saw in the code that it looked like it supported non-P 328.

I've successfully burnt bootloaders to non-P using Arduino ISP with the mods to boards.txt and averdude.conf

It's a DF Robot LCD Keypad Shield, or one of many clones, and everything you need is on teh DF Robot site

First link if you use Google

http://lmgtfy.com/?q=df+robot+lcd+keypad+shield

And, as said, you don't need to do all that conversion to a voltage for the comparison. Just read the analog value as between 0 and 1023 and act on that.

A copy of your circuit would be good information.

You can probably make it run longer by using power save modes.

if you flash an LED at a 1:2 duty cycle then you only need to do something every 330 ish milliseconds. That is still a long time in uC land at even 1Mhz (each cycle is 1 millionth of a second, or 1uS)

So...you could probably go into low power or sleep mode for the rest of the time. I've not played with power saving modes myself yet, so I'm not sure how it affects the ability to hold pin at a set state, High or Low. If that's not a problem then you could sleep most of the time and maybe extend that run time by quite a large factor.

Someone with experience of using power save modes will give more advice I'm sure.

This thread might help you:-

http://arduino.cc/forum/index.php?topic=139675.0

How about just using some hi-visibility reflective tape as a collar? That is passive as it uses the light of car headlights and street lights to make the wearer visible.

Why re-invent the wheel? This is what hi-vis reflective material is for.

If you really want to use LEDs then why not just have an LDR to detect when ambient light is low and then flash some superbiright LEDs. The superbright LED's can be used at very low current and will still be visible at night. The human eye is very sensitive to light in darkness; even more so if it's flashing.

As an example I use some blue SMD LEDs as power indicators on my PCB and run them at 5v via a 10k resistor. That's less than 0.2mA and is visible in daylight. You'd need a minimum of 3 LEDs for 360 deg visibility. 0.6mA or less, flashing 1:2 is an average of 0.2mA PLUS whatever the uC uses.

You could run an ATTiny at 1Mhz and it should use about 300uA in Active mode. That's 0.5mA total consumption.

Eking out the power is the biggest challenge, as I'm sure you'll want to use a small power source, like coin cells.

Something like an LR2450 is 120mAH 3.6V. That would run everything for approximately 240 hours or 30 days of 8 hours per day darkness.

It does mean the cat will be lit up all night, but you'll need to decide which is more important to you, the safety aspect or glowing cat.

You may well have all your neighbours reporting strange glowing alien beings in your neighbourhood for a while though.......

Nick, the version of your programmer I have is using 0xFF as Low Fuse byte for the 2560.

The change we made to 1284 to fix the Serial timing problems was to change from 0xFF to 0xF7 for Full Swing Oscillator.

Could you have possibly already made that change as you said you're now using 0xF7?