Door control system/Logbook

All changes/problems reported with the door entry system. See also http://github.com/londonhackspace/Doorbot/commits/master

2011-3-10

Doorbot stopped seeing rfid cards at approximately 23:00 the night before. Unlocking from the inside works, with the expected info shown in the SCREEN session, using openDoor.py also works, as does the doorbell. Ciemon 06:40, 10 March 2011 (UTC)

2011-2-26

Pulled the refactored-glados branch from git to test it before merging with master. Restarted glados. The branch adds auto-reloading the wavetable.dat file, and the ability to have a sound playing in the background while being greeted (docs in the wavetable.dat). Also removed soundplayer.py it uses solely mplayer for sounds now .--Eb4890 01:40, 26 February 2011 (UTC)

2011-1-19

Doorbot stoped working at midnightish on the 18th, i joined the screen session and hit ctrl-C. doorbot was restarted and afterwards worked fine. JasperWallace

Interesting... try stracing it if you catch it happening again Ms7821 10:30, 19 January 2011 (UTC)

2010-12-20

Doorbot got jammed on Tuesday, so we're going to rework it so that only unlocking is done realtime. Everything else can subscribe to network broadcasts.

2010-10-26

Tried adding a card as per the instructions. It failed with the following reason:

babbage:/usr/local/bin/Doorbot$ sudo ./addid.sh
Name to authorise:
softhook
ssh: Could not resolve hostname bell.lan: Name or service not known
ssh: Could not resolve hostname bell.lan: Name or service not known
lost connection
Card EAFE09B1 is now authorised to softhook

Using the IP address for bell manually also failed. What's the new procedure for adding members' cards?

The door is no longer being run by bell, references should be changed to connect to flowers instead. Robert 22:23, 26 October 2010 (UTC)

2010-10-12

• People are reporting the solenoid not triggering properly. This could be the bolt getting caught against the strike plate, so will test tonight.
  • This is due to people not disengaging the door wedge before closing it. If this is done correctly then the bolt disengages correctly. Solexious 00:58, 16 October 2010 (UTC)

Added a diode across the solenoid (there was already one over the relay to protect the transistor). This stops a large negative voltage spike from being transmitted down the wires and potentially burning of the relay contacts, but note that this spike only appears on the opened relay contact - it does not reach the circuit except via capacitative coupling or contact arcing. The diode was fitted after the reports of the solenoid not triggering, but if it needs to be removed it is fitted to the chocblock just above the letterbox and can be disconnected easily. - artag

2010-10-06

Operator: Mark

• Doorbot died at 10:04. Apparently shorting the pins wasn't it.

2010-10-05

Operator: Mark

• Swapped out the hub with Oni's. It's much nicer, so we should try to find another similar.
• At Jonty's suggestion, swapped out Bifferboard for his spare
• While doing this, we suddenly realised what might have been causing all the problems: the tape beneath the bifferboard had been scraped away by the pins, so they were touching metal case. This explains all the random failure issues below.
• We've put it in its own plastic case

2010-10-02

Operator: Jasper

• noticed doorbot & door not working
• as per instructions form irc power cycled the usb hub and bifferboard
• door now working.
• the door stopper needs a big arrow or something, it's easy to miss it and that it's down which makes the door hard to close
• the speaker wires need a better way to get out of the door bot box, as the case is opened and closed it will damage the wires.

2010-09-30

Operator: Solexious

• Rootfs went away at 10:30 when samthetechie tried to enter the space. Bifferboard restarted, but didn't boot.
• Reboot of usb hub fixed biffer booting
  • The USB hub has been unplugged from its own power supply, in the hope that this will let the bifferboard reset the USB drive on failure. Left plugged in, because the reader seems to fail.
    • maybe we should power it off the main 5V supply, but arrange to switch it off when USB power goes away. Then it would reset with the bifferboard.
• Phase 1 of the doorbell completed
• Arduino sends a 1 over serial when the door bell is pressed
• LED lights green during solenoid pulse
• Red/Green LED can be controlled by sending numbers to the arduino
  • 2 Green on
  • 3 Green off
  • 4 Red on
  • 5 Red off
• There is also a speaker in the doorbell, this is not yet implemented

2010-09-29

Operator: Solexious

• Worked out why avrdude stopped working - Mark fixed Doorbot to keep the serial port open!
• Wired the door bell up

2010-09-28

Operator: Artag (write-up by Mark)

• Replaced the chocblock with a lovely 25 pin D-Type connector, with soldered joints and heat-shrink.

• Started testing on the scope, but couldn't find a suitable way of using the trigger feature.

• Suggestions for improvements:
  • See whether PWM/AC requires less current to trigger the solenoid
    • PWM the relay? What? Russ 23:20, 29 September 2010 (UTC)
    • It followed on from a discussion about solid-state relays (>500Hz SSRs exist), but there's probably a suitable mosfet. Ms7821 08:33, 30 September 2010 (UTC)

                 10:16:40 < artag> russss: re your query on the doorbot solenoid driver, I was 
                 thinking that if we replaced the relay with opto+fet, we 
                 could drive the solenoid with chopped dc or even ac (h-bridge 
                 drive). Because of the inductance, this might turn out to be 
                 significantly less current than 12V dc.

• • See whether the 5v supply is having difficulty when the reader is powered
  • If so, isolate it and use a hefty capacitor to keep that going (rather than supporting the solenoid)

2010-09-27

Operator: Mark

Problem 1

Identified by Solexious at the weekend

• Arduino resets on serial open() from bifferboard
• Arduino flashes digital outputs high on reset
• Digital output flashing high triggers door open (this seems to be getting worse over time)

Tried:

• Disabling DTS (didn't work)
• Pull-down resistor (10k to 1k)
• Physical reset button on arduino (it also makes the pins flash)
• 120 ohm resistor between 5V and RSET on arduino - works :D

The board no longer flashes the outputs on boot, but still does when the reset button is pressed.

Workaround:

• Open serial port on boot
• Hope it doesn't restart

To try:

• Check voltage during boot, so see whether either line is spiking
• Replace diode on solenoid
• Resistors to reduce sensitivity of darlington
• Use a capacitor to smooth over the reset pulse, requiring > 1 second to trigger relay
• Rebuild darlington/relay board
• Use a solid state relay (no darlington required)

Problem 2

• Chocolate block "plug" is loose, sometimes causing loss of power at slightest touch

To try:

• Replace with header/plugs (we'll try a D-type on Tuesday)

Fixed by Artag

Problem 3

• Flashing arduino has stopped working (stk500v1: no response)

To try:

• Another arduinio (Solexious will bring one on Tuesday)

This was because doorbot had the serial port open

Problem 4

• Rootfs went away at 4:04am

• This might have been the sensitive chocblock again, as there were people in the space.

Previous problems

• Progressive failure overnight before Sunday of the Arduino talk - very bad. After investigation it was found that the original alarm charger was outputting 6v, so the Nneil donated a wall charger to replace it.

• Numerous bifferboards were destroyed by cigarette lighter regulators which spontaneously become zero ohm resistors when a large (> 0.6A) current is pulled through them.

• The original spec for a 7805 did not take into account the amount of current passed. 7V * 0.8A = a small radiator.

• The original transistors we tried were not enough to trigger the relay, so were replaced with Darlingtons.