Make-a-Project

DIY outdoor all-weather 3G/Wi-Fi router

2011-10-30T18:01:00.003+02:00

My father lives in the village where no broadband connection is available. To access Internet he used creepy and hell expensive GPRS connection. Once tired by absence of normal connection and impossibility to show him some video on youtube I decided to make something to resolve this nonsense. At that time I was aware few neighbors of him are using 3G USB modems with Yagi antenna attached. After small research USB modem was bought to do the same. Unfortunately first attempts to connect to the network using solely internal antenna was complete failure. But when I was outside house and connected 1/4-wavelength long piece of copper wire into modem’s antenna socket, I was able to connect to the network.

Due to impossibility to make connection from the house, first more or less working solution included wireless router with USB port (at the time one of the cheapest TP-Link TL-MR3220) that was located in attic and this fascinating wire antenna:

Surely I was able to connect modem directly to PC using only external antenna and USB extension cable, but I have reasons to go wireless. To name a few:

I and my nephew have wi-fi enabled devices (phones, laptops, ipods, etc.) we would like to use them anywhere in the premises. So this hotspot will serve not only standalone desktop PC, but any wireless device in range.
I didn't wanted use PC as a router as it often turned off because is a little loudly and located in the living room.
While drilling in the attic was OK, inside cabling will be a pain. Wireless access from desktop is a real saving and relief.

Few weeks after installation I visited father again and found that:

Iron roof coating does not allow to connect my laptop to Wi-Fi access point from most of the courtyard;
Summer happened (oops!) and under iron-coated roof heated by direct sunlight daylight temperature is more than 50° C (122° F) and router together with modem become really hot; it was promising nothing good;
Connection rather unstable and disappears from time to time; though demonstrates not so bad performance, up to 0.5/0.2 Mbit DL/UL.

Inspired by performance measurements I decided to make more reliable solution with better antenna and install it outside of the house to avoid overheating.

I started with a search of the case that will be able to accommodate all required components. But I found that cases in stores either hermetically sealed (which is unacceptable in my design because of airflow is required to cool components) or does not provide adequate protection from harsh conditions and mostly expensive (appropriate sizes starts at around US $50 equivalent for the IP55 rated box). Thus I decided to go with own case design and manufacturing.

First of all I made plastic sheet bender out of nichrome wire strained between two long screws secured in the wood plank and 12V/10A power supply with alligator clips. Heat intensity was regulated by moving alligator clips along the wire. Then I used it to bend two 45cm*70cm big polystyrene sheets into shapes pictured below.

Polystyrene was chosen because of its availability, low cost and good frost resistance as in our region frosts under -20 degrees C are rather usual during the winter.

Plastic sheets were bent according to this drawing to form two parts (base/roof and lid) of protective case. You may refer to the previous image for resulting shapes. All sizes in millimeters.

Take into account that it's impossible to bend relatively thick (mine is 3 mm) plastic to have sharp corners. So you should consider resulting slightly rounded corners when perform bending and maybe introduce some tolerances into design. I used bottom cutouts to make exact fit between lid and the base part.

After bending lid and base was put together, secured, and four 3.2 mm assembly holes was drilled on the sides to keep resulting case in one piece. Later I drilled 6 mm holes on each side to hang the box on two crews screwed into the wall.

To make assembly more convenient I screw M3 screws and nuts together through mounting holes of the base and then melted nuts into base' plastic using hot soldering iron. To completely secure the nuts in the base I used small amount of cyanacrylate momentary glue.

To protect internal space from humidity, ventilation holes are made in the places that are safe from water sprays, but everything else should be securely insulated. I have used self-adhesive insulation ribbon for doors/windows sold in hardware store.

80mm cooler from broken ATX PSU was used for forced ventilation. I had drilled 12 mm vent holes first then glued sparse silkscreen mesh from inside to protect internals from insects. From my experience bugs are more dangerous for electronics mounted outdoors than water and frosts.

To heat electronics during winter frosts I decided to use two ceramic 10W, 10Ohm resistor connected in series. Total power dissipation at 12V will be 7.2W

All ventilation and heating components mounted on the lid. I added two LEDs to indicate when heater or fan is on but it is not really necessary. I often use melting glue gun to mount different parts as it is done in this case: both LEDs and controller glued to the lid with a glue gun.

To control cooling (fan) and heating (resistors) I prepared very simple circuit consisting of comparator, few resistors, 10K NTC thermistor and dual N-Channel MOSFET transistor. Comparator forms inverting and non-inverting Schmitt triggers. I calculated resistors values for this circuit using online calculators on Random Science Tools and Calculators site to turn on fan when temperature is more than 30 degrees C and turn it off if T below 27 degrees C. Heater respectively turns on if T is below about 3 degrees C and turns off when it rises more than 7 degrees C. Numbers are approximate, but tests showed I hadn't missed anything.

Controller powered directly from 12V/1A wall PSU.

Schematics of “HVAC” controller:

BOM:

IRF7103 SO-8 dual MOSFET 50V, 3.0A
LM393 SOIC-8 dual comparator
0.1uF C0805 ceramic capacitor
1uF C0805 ceramic capacitor
3x Molex 2-pin polar connectors 0.1" step
10K NTC thermistor R0805
33K R0805 resistor
39K R0805 resistor
330 R0805 resistor
2x 10K R0805 resistor
56K R0805 resistor
22K R0805 resistor
330K R0805 resistor

To ensure drainage in case water forced inside, meshed vent holes located on the bottom of the case also were drilled. Here SMA/RP-SMA connectors for Wi-Fi and 3G antennas and cable input for power supply located as well.

To access Internet I bought Novatel Wireless Ovation™ U720 EV-DO Rev A USB modem and data plan from local provider who is working in 800 MHz band using CDMA EV-DO Rev.A technology.

I have ordered few MS-147-to-SMA pigtails for external antenna connector but all received parts had in fact DB9 connector that does not fit on connector. So finally I disassembled the modem and made small mod by connecting commonly available U.FL-to-SMA pigtail directly to PCI-E card instead of wire connector to MS-147. U720 in fact is USB to PCI-E adapter plus PCI-E EV-DO modem.

As reception of CDMA network signal in the area where box will work is very poor, I decided to create external antenna.

I had used Slim Jim antenna calculator to calculate dimensions required for CDMA 800MHz band. CDMA 800 terminal node uses following frequencies according to modem user manual:

Transmit 824.7-848.31 MHz
Receive 869.7-893.31 MHz

I made calculation for mean frequency between 824.7 and 893.31 MHz. Then I draw the shape in design software and etched it on FR4 board to ensure correctness of the sizes and shape.

I hadn't 30cm long etching tank so I used paper box and polyethylene film to make temporary etching container. I used direct toner transfer method for PCB artwork preparation and ferric chloride as etchant.

To protect antenna and ease its mounting I put antenna board inside 30cm long 1/2" PVC plumbing pipe and put stubs on both sides. There are SMA connector on one side to connect extension cable I made using connectors and RG-316 low loss coaxial cable bought on eBay. Both stub caps are sealed with melted glue.

Here you can see all router components together. Here goes list of all components and approximate prices in USD:

2x 45cm*70cm polystyrene sheets - $15
Self-adhesive insulation ribbon for doors/windows - $3
80mm computer fan - $3 (Mine is scavenged from broken PSU, no cost for me)
HVAC controller board components and FR4 - $5 (not sure as I bought only thermistor and resistors/caps and this is few cents; everything else were scavenged from old PC motherboards and FR4 I had before and don't recall the price, but it is very small piece and also should cost few cents)
Heater (2x 10W ceramic resistors) - $1.5
PVC pipe, caps, pipe clips, SMA connector and FR4 for antenna - roughly $10
RG-316 coaxial cable (~10'/3m) and connectors - $10
Cable input and pair of screw clamps - $2
2x pigtails (U.FL/SMA, RP-SMA male/female) - $8
Tripple power socket - $3
12V/1A wall PSU - $5 (I have few of them for free from old equipment)
Novatel U720 USB CDMA Rev.A modem - $40
TP-Link TL-MR3220 3G/802.11b/g/n wireless router - $40
Collinear Wi-Fi antenna for 2.4 GHz band - $12
Melted and cyanacrylate glue, double-sded adhesive tape, nylon cable ties etc. - $5

TOTAL: around $160

In total I had spent about 16 hours in total for design, production and assembly of final product.

There are notes on the flikr to help identify components.

Below you may find photos of assembled box:

Wi-Fi box was mounted on the pediment of the house on Nord-West side of the house. Top Internet speed I witnessed was 0.9/0.4 Mbps DL/UL according to speedtest.net using CDMA EV-DO Rev.A connection. In the area where box is working, connection using solely internal modem's antenna is not possible.

Debugging firmware without debugger

2011-02-07T19:06:00.004+02:00

My thermocouple sensor stuck for a while on firmware development stage until my Logic Sniffer is arrived.

This is because I faced a problem with I²C communication:

When I communicating my firmware based on Atmel’s AVR302 application note source code, it works as expected;
When I poll LM75A I²C thermometer it also working fine, but after I read value from it strange side effect appears: my firmware stops responding with address acknowledge when I trying to address it;
After few (rather random) polls to LM75A my firmware are able to process commands until thermometer part is polled again.

There are many wait loops in firmware source code originated from app note like following:

whi_dac:
 sbis PINB,PINB2 ; wait for SCL high
 rjmp whi_dac
wlo_dac:
 sbic PINB,PINB2 ; wait for SCL low
 rjmp wlo_dac

It seems to me that because of incorrect handling of a bus states firmware hangs on one of these loops.

I ordered logic analyzer to examine bus traffic but because it seems that the package delayed somewhere due to Chinese New Year I want to try today one approach that come to my mind during last week how to debug firmware inside ATTiny13 without hawing hardware debugger. Idea is pretty simple:

ATTiny13 has 8 pins;
3 is pre-occupied (RESET, VCC, GND);
On 2 pins (PB2(T0) as SCL and PB1(INT0) as SDA) sits I²C bus;
Thus 3 pins (PB0, PB3, PB4) are free and could be used for debug purposes.

Idea is to use LEDs connected to free 3 pins as indicators of key points in source code. 2³ gives us 8 possible combinations. All that I need is to set PB0, PB3, PB4 to HIGH/LOW state according to key point number in binary format.

K-Type thermocouple sensor I2C interface board

2010-12-12T23:40:00.007+02:00

Today I finished with board preparation and soldering:

Most important part is firmware to turn ATTiny13 into I2C ADC which is not ready yet. Will continue with this task on next week.

K-Type thermocouple measuring

2010-12-10T13:21:00.011+02:00

I'm designing the circuit to measure high temperatures (up to 600 °C) with at least 1 degree resolution with TWI (I2C) interface.

Literature on thermocouples:

Practical temperature measurement application note
NIST ITS-90 Thermocouple Database
Thermocouple Technical Information

Circuit design will include LM75 TWI temperature sensor for cold junction compensation, LM358 dual operational amplifier for two-stage amplification and ATTiny13 microcontroller for ADC and TWI interfacing (software slave).

Board layout supposed to be as narrower as possible to be insulated with heatshrink tube. Thermocouple will be connected using 3.5 mm screw terminal. For MCU communication and power supply 4-pin JST connector will be used (2.5 mm step). TWI address of LM75 will be configured using solder jumpers, ATTiny address will be set using TWI protocol and default address and then stored to EEPROM.

Device will support 3.3V and 5V supply.

Dual stage Non-inverting amplifier calculation is the following:

As I need 600 degrees upper level, voltage could be aplified up to 125 times to be below 3.3V for normal operation with this supply voltage as about 25mV corresponds to 602° on k-type probe.

Vo=Vi*(1+R2/R1)

1st stage: R2=10K, R1=1K, Gain=11

2nd stage: R2=10K, R1=1K, Gain=11

Total gain = 11*11 = 121

Total gain of 121 is good for us as it gives 25mV*121=3.025V at 602 °C (25 mV reading).

Measurement resolution considering 10-bit ADC will be:

At 3.3V 3.025/(3.3/1024)=938 steps, about 0.64 °C per step.

At 5V 3.025/(5/1024)=619 steps, about 0.97 °C per step.

I will work on proof of concept today; let's see the results.

Digital Soldering Station Concept

2010-07-08T14:47:00.000+03:00

Haven't written some time due to my work and ongoing projects I will explain later.

My pumping station is on hold for a while as it is not urgently required till first frosts. Actually all electronics is complete and working perfectly. The only thing is to pack it into cases and make good water insulation for immersed parts. This part I hate most because I have no suitable tools and place to do the hardware work.

Now I'm working on digital soldering station to replace my old 40W 220AC iron with tool more suitable for lead-free soldering. Features I'm going to implement:

PID controlled heating of 24V iron (I'm going to use SL-ICMC 24V/48W iron with ceramic heater that normally used with CMC series of Solomon stations with K-type sensor)
3-digit numerical display for temperature reading and setting
Two-button 100 to 450 °C temperature setting
Calibration using external thermometer
Replaceable circuit for thermocouple or thermistor sensor support
EEPROM to store temperature settings when station is switched off

I have bought switching PSU 24V/2.7A, iron and metal case for this project and now working on electronic part. Hope to come soon with first results.

Finally!

2010-03-08T13:24:00.001+02:00

Finally I got it working yesterday! The problem with TWI was not in MCU nor in software TWI Master implementation, but in differences in addressing.

In LM75A datasheet said that 7-bit addressing used for the device. MSB bits are always set to 1001, and value of 3 lower bits depends on logic level on pins A0:A2. As A0:A2 pins I have connected to Vcc, address of my sensor is b01001111 (8-bit binary) = 0x4F (hexadecimal). I have successfully used this address with Arduino and all was working as expected.

But when I started with ATTiny13 running software TWI nothing were working. As I haven’t full-functional digital oscilloscope, I had tried to make some “poor man’s logic analyzer” using codes from here. Got it working, but results were strange to me because SCL signal on TWI bus wasn’t like a strobe but more likely signal. So I decided to make more research over the net.

Fortunately I have found topic on avrfreaks.net forum where somebody discussed interaction with different LMxx sensor, and I noticed strange addressing in code. Address of device were 0x90 that corresponds to b10010000. I have maid guess that address should be not left-padded as I used it with Arduino, but right-padded. I put b10011110 (0x9E hex comparing to 0x4F I've used before) as address and now everything is working just fine!

LM75A setup

2010-03-07T22:14:00.001+02:00

I made serious mistake during design of pumping station thinking LM75A has non-volatile memory for configuration parameters. Actually this IC haven’t such ability as store configuration between power-ups.

Fortunately I have connection to TWI pins on both sensor board and wiring, so decided to create some “hardware config file” basing on ATTiny13. I have created Arduino sketch to test approach and it is working just fine, but tiny13 has no hardware TWI implementation comparing to ATMega168. After some research over the net I found nice implementation of AVR-300 (Software TWI) by Peter Fleury: http://homepage.hispeed.ch/peterfleury/avr-software.html#libs

Here are 2 PCBs, one is test board for LM75 sensor, second is ATTiny13 for software implementation of TWI:

My configuration board is still not working despite simplicity of the library. I haven’t found why, but my guess is that internal RC oscillator is not suitable for this. Will try to run with external crystal today.

Update: I had made it working!

Pumping station project

2010-03-05T20:11:00.001+02:00

3 weeks ago I started new project as a gift for my father who live in the countryside. There he has well with electrical pump and elevated water tank for housekeeping needs.

He is a good hand (despite a bit old-schooled), and pumping is more or less automated using few scrap yard relays. But we had as low as –27 °C this winter so pipes were frozen few times. My idea was to create device that will keep both water level and temperature at desired level and be more safe, more reliable comparing to existing device he uses.

Due to workload in the office I was unable to create my “pumping station controller” fast and now I’m definitely late with heater (+5 °C behind the window). But for now the device is almost complete and I can write about its features:

LM75A I²C Digital temperature sensor and thermal watchdog
2x A3212 Ultrasensitive Hall-Effect Switches for water level sensing using foam float with magnets
2x Atmel ATTiny13 MCU: one to watch sensors and control pump and heater, second for setting up LM75 during start-up
2x powerful Omron relays (230V/20A) to control load

I have a hope I’ll have a time to describe device in detail in next posts. For now you can take a look at visualized PCBs for the device I had designed before: