.:madworm:.'s photos

The capacitive sensing wire is just pushed into the lock's cylinder. Upon startup the device measures a baseline value.

The capacitive sensing wire is just pushed into the lock's cylinder. Upon startup the device measures a baseline value.

Initially I wanted to use a PDIP ATmega168, but it was just way too large. The knock sensor needs some more space, as it needs an opamp. Luckily the latest LED Ring boards already include a 5V regulator, which saves some more space. The number of available IO pins is sufficient as well (6 including RX/TX, but only RX is shared). I added a jumper to disconnect the reed relay from the RX pin when I need to update the code. It turns out that a load of 500Ω doesn't interfere at all. I can hear the relay click happily during upload.

Initially I wanted to use a PDIP ATmega168, but it was just way too large. The knock sensor needs some more space, as it needs an opamp. Luckily the latest LED Ring boards already include a 5V regulator, which saves some more space. The number of available IO pins is sufficient as well (6 including RX/TX, but only RX is shared). I added a jumper to disconnect the reed relay from the RX pin when I need to update the code. It turns out that a load of 500Ω doesn't interfere at all. I can hear the relay click happily during upload.

The white push button switch triggers the staircase lights. I close the circuit with a small reed relay. As I couldn't make any sense of the wiring, I added a "rogue" wire from the doorbell switch. It provides 13.8V AC with enough current capability. The installation manual of this com system is quite terrible for people wanting to understand how it works. It only contains wiring examples for all the different setups you can choose from, but nothing substantial as to what the wires may do (which is power, which is signal...).

The white push button switch triggers the staircase lights. I close the circuit with a small reed relay. As I couldn't make any sense of the wiring, I added a "rogue" wire from the doorbell switch. It provides 13.8V AC with enough current capability. The installation manual of this com system is quite terrible for people wanting to understand how it works. It only contains wiring examples for all the different setups you can choose from, but nothing substantial as to what the wires may do (which is power, which is signal...).

As the 8pin LM2940 op-amp is a dual amp, I've distributed the amplification to both amps. The first stage still adds an offset and has adjustable amplification, the 2nd stage is a fixed at a factor of 11. I've also added a low-pass filter at the output to smooth the signal some more. C5 was added for AC coupling and must be at least as big as C1. D2 was added to allow C5 to be discharged and be ready for the next pulse.

x: 10ms/div y1: 200mV/div y2: 1V/div Taken from here and using a 10x probe. y1: sensor voltage (minus offset) y2: amplified and stretched

I'm still using my old scope ;-)

LM2940 non-inverting amp, no pulse stretcher yet. All this amplifier business is a black art of its own. In retrospect I must find that all the lectures I've taken at uni regarding electronics were only capable of teaching the very basics, if at all. I didn't study EE though, but some other natural science which is highly unpopular among most people, especially at school. And I still don't know sh*t about you name it. So much to know, so much already forgotten.

LM2940 non-inverting amp, no pulse stretcher yet. All this amplifier business is a black art of its own. In retrospect I must find that all the lectures I've taken at uni regarding electronics were only capable of teaching the very basics, if at all. I didn't study EE though, but some other natural science which is highly unpopular among most people, especially at school. And I still don't know sh*t about you name it. So much to know, so much already forgotten.

A single supply non-inverting amp with pulse stretcher. It's also got an offset 'knob'. The LM2940 is a dual amp, so maybe I'll add some more pulse shaping. It sucks that all the easy stuff requires symmetric supply voltages. But using just 5V also means that there won't be any over-voltage or negative voltages that could kill the AVR. The blue curve will be integrated by a microcontroller.

A single supply non-inverting amp with pulse stretcher. It's also got an offset 'knob'. The LM2940 is a dual amp, so maybe I'll add some more pulse shaping. It sucks that all the easy stuff requires symmetric supply voltages. But using just 5V also means that there won't be any over-voltage or negative voltages that could kill the AVR. The blue curve will be integrated by a microcontroller.

A single supply non-inverting amp with pulse stretcher. It's also got an offset 'knob'. The LM2940 is a dual amp, so maybe I'll add some more pulse shaping. It sucks that all the easy stuff requires symmetric supply voltages. But using just 5V also means that there won't be any over-voltage or negative voltages that could kill the AVR. The blue curve will be integrated by a microcontroller.

Roulette chips.

This time in yellow. It seems the boards came out more like a yellowish orange, similar to what a real orange looks like. Definitely not citrus yellow.

This time I ordered the RGB LEDs from JELEDHK (ebay) and got a better deal. Unfortunately the semiconducter chips seem to be reversed, so the LED packages have to be rotated by 180° to make it work again. Now the silkscreen is incorrect. I'll have to make adaptions to the code as well.