Type Faster

A collection of thoughts on things attempted, and why they did or did not work.

[typefaster]

IRLP & WIA News

So I’m the administrator of my HAM clubs IRLP node. One of the nodes tasks is to re-broadcast the Wireless Institute of Australia’s news service once a week. All going well it’ll do this on Sunday morning and maybe 1/2 dozen people listen in. (Who really knows how many people listen on the radio?) That means that each week the node has to download the latest version of the news, which is an MP3 file. Now for some reason this has been quite problematic. 

There is a Yahoo group for the news service, however only the owner of the group can post to it, so zero support. :-(

Happily the owner does say that we’re welcome to write to him if we see issues. So I shot a note off detailing my problem and asking if he was aware of anyone else seeing issues. 

I never saw a reply, but I did see that the message published in group when the news was published each week change to say that the method I was using to get the news files was never to be used, other than as back up. All done in caps, i.e. internet speak for shouting. Again, zero support. :-(

Perhaps if there was some two way comms I could have got some guidance on how it should all be set up when I took the role over and moved the node to Pi-IRLP, and everyone would be better off. But no, I had to invent the wheel all over again.

I concluded that my download stalling issue was related to a Linux problem with the SACK packet, and made a change recommended in point 2 here. Why the node had this issue and none of my other systems ever have I don’t know.

That still left the news being downloaded from the backup site. While the previous administrators didn’t seem to fussed about pulling from the backup site, I would prefer to do the right thing. 

The issue with the primary WIA site is that the file name includes the intended date of broadcast, and so changes every week. Whereas the back up site has a standard name and location, and the file is updated each week. Simply trying to grab the same file each week, which is how the node was historically set up, simply won’t work from what’s intended to be the primary site.

Since the news is published in an RSS feed as a podcast I figured I should try using an RSS Podcatcher. After some hunting and a couple of false starts I found the quite excellent ‘Greg’. Greg is capable of doing lots of clever things, but as a command line tool it’s capable of some manipulation, perfect for my application of grabbing a single file. 

I did come across an issue with Greg. The author was very responsive and helpful, for which I was grateful. Ultimately the issue was fixed by upgrading the node from Wheezy to Jessie, which also upgraded Python.   

The Jessie upgrade introduced an issue with the USB sound card, which was fixed following the tips in response 2, here. Specifically:

There is a file called aliases.conf in /lib/modprobe.d which contains the line options snd-usb-audio index=-2. That line overrides the /etc/modprobe.d/ files, so you need to change that one. Comment out with a # the line options snd-usb-audio index=-2

Lastly I wrote a BASH script that utilizes Greg to grab the news each week. The script checks that it’s been successful. If it hasn’t then it calls the ‘old’ news download script to grab from the backup site, as is intended. The script sends out an email to let me know if it worked. Giving me time sort any issues out ahead of the broadcast time. The script is up on Github for anyone that’s interested.

https://github.com/TypeFaster2/IRLP

I’m sure anyone who has real programming skills will groan, but it’s worked for me so far. I still have no idea how anyone else does it. 

The WIA news team do a great job of putting together a very high quality news broadcast every week. I’m sure it’s time consuming and tough when there’s not a lot in the way of contributions. But perhaps if they were a bit more communicative I could have benefited from work already done by them or other re-broadcast stations, and I wouldn’t have needed to do any of this. Or just maybe my work might be of use to others.

Hopefully this post might be of use to someone else.  

[typefaster]

Power Supply Rebuild

Picked up a 12 volt 5 amp home brewed power supply at a local hamfest. I was assured that the guy who built it knew what he was doing, so thought it was worth the few dollars I paid for it. I’d been looking for a PS for awhile anyway.

It’s a nice, heavy, linear supply, with an LM723 as the regulator and 2N3055 as the PA. I looked at the case and thought there’d be room for a meter on the front. When I opened it up it I could see the mains wiring wasn’t that well insulated and overall the layout was a bit messy. Otherwise it looked fine.

The mains fuse and switch are on the left of the picture

Off to ebay where I purchased a combined volt / amp meter. There’s lots of different listings. Mine was a 100v, 10amp unit with no need for a shunt.

Once the meter arrived in the mail I quickly cut the hole in the front panel and wired it up. Then disaster struck. I’d not been careful reading the schematic, wired it up all wrong. And paid the price as the magic smoke escaped.  

In order to be able to sort it I really wanted to find a schematic of the supply. Fortunately the PCB had a number etched on to it, and I was able to work out that it was actually a short form kit from Electronics Australia. The article’s here:  http://messui.the-chronicles.org/cb/ps_k3449.pdf

After quite a bit of mucking about I found that I’d blown the IC and the IC socket. I also blew the 1st stage transistor and the over voltage zenner diode. Surprisingly the thing that was supposed to blow, being the fuse, was dirty but otherwise fine. Happily all the replacement parts only cost me a couple of dollars and were all available locally.

I made sure all the mains connections were fully insulated, and generally cleaned up the internal wiring, including adding brass bus bars on the three front connectors. 

The last step was to calibrate the meter, and set the voltage. It’s done many hours of service since then. Even when on all day it barely gets warm. 

[typefaster]

Arduino Battery Load Tester

I’ve got hold of a Yaesu FT-817nd which I’d link to be able to use out in the field, in particular for SOTA. Working in the field needs a power source. The radio has some internal power, but that limits it to 3 watts. The best external power seems to be LiPo batteries. These are lite and powerful. But these have a fully charge voltage higher than the radio can tolerate so you need a voltage controller. And they are reasonably expensive. Another alternative would be LiFePo4. These are still very lite though they have the benefit of a max charge voltage that’s lower that the FT-817 max voltage. Then there’s Sealed Lead Acid Batteries, or SLAB’s. These are cheap, easy to charge, rugged. But they’re heavy.

To be able to compare how the different batteries perform we need to test them. Taking them out into the field is one way. But every trip is different. A better solution would be to have a standard load and track the voltage over time. Even better would be a simulation of the radio load on receive and transmit. Which seemed like an ideal task for the Arduino.

Load

As it happens, a 21 watt / 5 watt car brake / tail light globe has about the same current draw as the FT-817, and only cost a couple of dollars.

Switching

A relay board was used to switch the load in and out. These are available from ebay for about the same price as the light bulb. These boards turn the relay on with the the Arduino control pin LOW, which tricked me for a bit.

Schematic

Fritzing didn’t have a symbol for a lamp, so that’s shown as LEDS. Similarly, the SLAB is shown as a AA battery in the diagram only. And I didn’t show the Arduino power connection. If fact I just ran it from the USB lead. 

Note the voltage divider to bring the sense voltage down below 5v. I should add a 100 Ohm trim pot into the divider to enable easier calibration. 

It’s important to tie the battery ground to the Arduino ground i order to get the divider to work. And test the voltage on the sense lead prior to connection.

Code

The Arduino switches the high and low current load in and out, so that the high load has a 25% duty cycle. I guess that’s about what it’d be like in the field. The voltage and time readings are sent to the serial port and written to the SD card.

Prior to running the load it’s necessary to calibrate the sense voltage. At this point I’ve done that by adjusting the calculations in the sketch but a trim pot would be much better way of doing it. There’s a calibrate sketch. It doesn’t write to the SD card, doesn’t switch in the load and writes out to the Serial port very regularly.

All the code’s up on GIT

Results

Using this I was able to work out that one of the batteries I had wasn’t going to cut it, but the other was fine. In fact it holds the voltage above what’s needed to operate for nearly 2 hours. So far the experience in the field has reflected that too. So overall this has been worthwhile.

[typefaster]

IRLP Node

I recently took over management of an IRLP node at my local club. The node has a weather announce function that had long broken when the BOM changed their web site set up. I couldn't get the old code working, so looked around and found a script that was working and then rewrote it for my own needs.

Also changed it so that weather reports are grabbed on demand rather than every hour. The reality is that I’m probably the only one that ever uses the functionality, and only then to confirm everything’s running.

The code’s been loaded onto GIT. It’d be great if it helps someone else.

https://github.com/TypeFaster2/IRLP

[typefaster]

Beach 40 - Part 3, RF Amplifier

Another wet weekend so made some more progress on the Beach 40. Built RF amplifier stage and learnt a few things, or more correctly was reminded of a few things, along the way.

Lesson 1: Amplifier stages need to be connected to the power rail.

The PA power is switched by the transmit relay. The layout of which isn't that clear on the schematic I'm using, especially on the printed version where the line runs off the side of the page.  Doh!

Lesson 2: The Emitter and Collector leads are not interchangeable.

I installed the 2N2222 round the wrong way, which unsurprisingly didn't work, and  fried the transistor.  Doh!

Lesson 3: To test the stage it needs to be complete. 

I tried to test the stage once the coil was installed, with no luck. Once I added the 100n decoupling capacitor and the 470 ohm load resistor it worked as expected.  More Doh!

I also added another PCB as I was going to run out of PCB real estate. The final product wont fit into the nice little box Peter had, but I don't think that's an issue. While I was replacing the fried transistor I changed the layout so that the RF stages run from front to back of the PCB. I had originally placed them all on the front PCB. I think that will work much better.

I've realised I don't have any BD139's, which is annoying as I thought I had everything and could easily have bought a couple over the last few weeks. So the bench has been cleaned up and building has stopped for this week. 

[typefaster]

Beach 40 - 40m DSB Transceiver - Part 2, Mixer.

A wet weekend meant a few productive project building hours. The mixer and the microphone amplifier are the next stages. 

Peter recommended listing to the radio at this point to adjust the trimmers to knock out the carrier. I didn't want to get set up for that, so just used the scope.

Before and after adjusting the trimmers. (Same voltage but different time base on the scope)

I've got an old  dynamic mic to use with this, where the circuit is really for an electret. It took a bit of fiddling about to work out the pin out and the PTT switch was a bit rough But it seems to be working now. I tested it out by playing a tone on the PC and holding the mic to the speaker. I'm not sure I if it's getting the amplitude it should from the mic. I might still have to try a higher value coupling cap. 

Here's the board. I can see I'm going to run out of PCB real estate, so I'll have to put in a vertical board for the final stages and AF amp. The microphone amplifier is hard wired on here as I hadn't put the relay in yet. 

[typefaster]

Beach 40 - 40m DSB Transceiver - Part 1

I've decided to have a go at building a little QRP transceiver designed by Peter Parker, VK3YE. He's posted a collection of videos describing the design on YouTube. It's been around for a couple of years, so there's a bunch of other builders with notes and some schematics. I thought this one was easiest to read.

I started off collecting all the bits. Mostly the parts are generic and readily available, so a bulk order with Futurlec was in order. These guys are cheap, and I've never had any issue with the product. Though they are slow, and have zero proactive communication skills. So if a parts on back order you won't know unless you ask. (If you shoot a note to them a couple of days after placing the order they're quite happy to tell you if there are issues, and if needed make changes)

Saddly there were a few parts they didn't have, in particular the 7.2 MHz ceramic resonator. In fact they have stopped making the 2 leg version and it was only after a LOT of Googling I managed to finally obtain the last 2 that a supplier less than 100km away had.

Once I had the bulk of the parts I needed to make up a base to mount it in. As I haven't tried a project like this I've opted to build it on a bread board. Once working I'll move it into a more practical container.

Here's the blank PCB with a front face plate that has the tuning cap, power switch, etc, mounted and ready to go. I really don't know how much room I'll need. If I have too I can mount a second PCB behind this one. Hopefully I won't need it.

First stage is the oscillator and buffer. I think they call this the Ugly build style. I hope I don't have to rework to much as it won't be the easiest to follow later.

I've read others notes where getting the oscillator running has been a trick, but mine ran first time. Here's highest frequency. The wave form is starting to collapse.

 Lowest frequency.

Wave form with the tuning around the mid point:

I'd read that the peak to peak voltage of the wave with no load should be more than 6 volts, but mine was showing on the scope as less than a volt. I was starting to wonder how I might address that when I realised that the scope probe was set to x10, so the real voltage was more like 8.8v.

Next is the Balanced Modulator and Product Detector. I need a 200 Ohm pot and a 4-40 uF trimmer cap, neither of which I have. Not sure how that happened.  

[typefaster]

Setting up the Copy Cloud Agent

Follow this process to set up Copy on Ubuntu. I was using 14.04. The folders I copy up to the cloud are outside the Copy folder, so there's some fiddling with Symlinks

Step 1: Get Copy from the website : www.copy.com/home/

Step 2: Unpack the zipped file: $ tar zxvf Copy.tgz -C $HOME

Step 3: In the unpacked folder there will be multiple versions of copy for x64 and x32 platforms. I'm running x64, so move that folder to where ever you want it to live. (I used ~/bin)

Step 4: In the x64 folder, run CopyAgent.

Step 5: CopyAgent will open a console. Log in and it'll start Copy running.

Step 6: I replicate bin, Documents and sketchbook. If they exist on the new machine, delete or move them from the Home folder as we'll be importing the replicated Cloud version.

If this is the first machine for this instance of Copy, move the folders to be replicated into the Copy folder at this step.

Step 7: Create symlinks from the Home folder to the Copy Folder. To do that for the sketchbook folder, the command will be: ~$ ln -s ~/Copy/sketchbook ~/ This can be done even if the sketchbook folder hasn't yet been downloaded from the cloud onto this machine.

Step 8: If the "Documents" folder was moved, we need to change it's location. Edit the file "~/.config/user-dirs.dirs"

That's it!

[typefaster]

Ubuntu One is No More

As of June 1st 2014 Ubuntu 1 is turning off. Bummer. I'd been using it for some time and found that it worked really well for me. I use multiple machines at home, and my work just moved (mostly) seamlessly between each machine with out me even thinking about it.

So what to do? Google Drive with Insync seemed good. I think it has 15Gb of free storage, though I've seen articles refer to 25Gb. The cost of extra storage is very low. And it looks like Goole is in it for the long haul. But I already use Google for other stuff and didn't want to mix Drive with my home stuff. (Mixing different types of "Stuff" can be bad!)

Dropbox is a market leader, which is a plus. But they only offer 2Gb of free space and compared to Google, the extra space is expensive. It also has the limitation that everything needs to be in the Drobox folder. A big "No Thanks" from me.

Lots of folk seem to like OwnCloud. If you run your own always on, always connected server then it looks great. And if the server is hosted in some far off data center so there's no risk of local fire or theft or other data loss disaster, then all the better. But that doesn't work for me either.

Then I found Copy. Copy has 15Gb of free storage, which is expanded to 20Gb if you sign up via a referral link. Both you and the referee get an extra 5Gb for free.

But the best thing about Copy is that you can use symbolic links, or in Windows speak, short cuts. That means you install the client, create the copy folder somewhere convenient, but not in a folder you want synced to the cloud. Then create a symbolic link (create a short cut) for each folder you want synced inside the Cloud folder. That's it. It'll start syncing and, depending on the speed of your connection, if a short while you'll be done.

In Ubuntu, that means your documents, Music and Pictures folders can be left as they are, and still be synced. That's really the same as the Ubuntu One service worked. Perfect.

When I first set up Copy it seemed a little slow to Sync, once running it quickly completed and has been fine since. Copy also has a local peer service. So if multiple clients are on the same local network the sync traffic doesn't need to go via the cloud, which can really wind up the data usage counts for those ISP's that are somewhat limited in metered allowances.

There's a client for Windows, Mac and Linux, but pretty much all the Cloud services have that now.

All round Copy looks like it'll be a contender in the Cloud Storage space. I'm sure they'll be getting a lot of Ubuntu One customers over the next few months.

Here's my link to referral service https://copy.com?r=vI3gAD . Get yourself an extra 5Gb and help me out with an extra 5Gb at the same time.

[typefaster]

Fridge Monitor Update to Version 3

Coming back to the Fridge Monitor as the basis of another little project I found that it wouldn't run. After some mucking about realized that the IDE has been updated to 1.0.5 and seems this has caused this sketch to run out of memory. Specifically, it'd run out of SRAM, which is where the variables are kept. After some research, particularly this page on the Adafruit site I took actions to reduce the amount of SRAM usage. Mostly using the  F() macro. I tried to make use of PROGMEM to move an array out of SRAM, but couldn't work it out. Pity as it would have freed up a bit more memory.

This process also added a free memory checking routine, which is enabled for debugging only.

I've recently acquired a Leonardo board as perk of funding Boarderless Electronics via an Indiegogo Campaign. The I2C pins on the Leonardo are in a different location, so a change to the pin assignment to enable the sketch to be used on Uno or Leonardo along with an Ethernet shield.

Apart from the changes to the pinouts, the schematic and outputs are the same, with the exception of the initialization indication written to the SD Card.

The code's up on Github

[typefaster]

Arduino GPS GridSquare Decoder

For some time I've been looking at how to make use of a GPS module with Arduino. Working out what my Grid Square or Maidenhead location for which is used when making VHF contacts with Amateur Radio seemed like a good little project. Of course there are easier ways to do this, such as apps loaded on your GPS smart phone, and then there's a range of web apps, assuming you have internet access and a pc of some sort. But where's the fun in that.

First thing to get is a GPS module. Most of the regular Arduino sites sell them. I particularly like LittleBird and Freetronics. Ebay is always worth a look too, and that's where I got mine. I bought a "SkyNav SKM53 Series with embedded GPS" that came with a small breakout board. The breakout board enabled me to use the standard 0.1" spaced pins. In hindsight, this probably wasn't needed. The GPS module cost just under $30 Australian.

For the prototype I used an ATmega328 chip in a breadboard. I was going to make up my own PCB and created an Eagle file that was optimized for a single sided board. I broke out all the LCD, FTDI and GPS pins as separate pins, which allowed me to move them around on the PCB. In the end it all seemed too much effort so I bought an Arduino Pro Mini off ebay for less than $10 Au, which is less than the cost of all the components for a custom build.

The GPS data sheet showed that VCC is 5v and RX and TX is LVTTL, which needs 2.85v to switch. As the Arduino I/O pins are 5v some voltage conversion is required. A voltage divider network was used to get a 3.3v reference which, along with a pair of 10k resistors, was used as a pull up on both the RX and TX lines.

For the GPS TX pin, the GPS drives the line and the Arduino only listens. I'd planned to add a diode in line to prevent the Ardunio accidentally driving the GPS RX, but never did.

On the GPS RX pin, a 1k resistor was used to drop the voltage which is driven by the Arduino. If you were to use a higher value resistor you risk not being able to get the pin down to zero. Sticking a multimeter on the line doesn't show any movement, so I'm not even sure the line is needed.

I also needed an LCD display. For the prototype I've used my stock blue 4 line one. For the final one I'm thinking of a 4 line green one, so it's not as dependent on the back light.

The code avoided making use of any floating point functions. This made it both smaller and faster. I made a point of building up strings of data to be printed to the LCD. Whilst this makes the code a little bit bigger, it's quite a bit faster than multiple LCD writes as opening and closing the LCD for writing is quite slow. Writing out a whole line of data in one write saves some time.

I displayed the location in WGS84 format. This seems to be the most common format these days, though it needs some calculations to get that format out of the decimal format provided by the NEMA string.

The display cycles through two formats. The second shows various pieces of information that's able to be extracted from the GPS. The only key piece of a data I didn't show is speed and course. I really didn't think I'd need it.

The Grid Square is shown on both screens. Given that's the whole point of the project, it seemed worthwhile.

The local time is calculate via some offsets that are hard coded. So if I travel to another time zone I'd need to be able to update the code. Same thing when daylight savings kicks in or out.

FCS is a count of the number of strings that have failed check sum.

I think it would be easy enough to adapt to a 2 line screen. That would make the whole package smaller, which would be good. The variation in cost seems to make it not worth the effort.

Only a prototype at this point. I've ordered the Arduino pro mini for the final build. Next thing is to order the LCD. I'm thinking about adding a temperature monitor and also a battery voltage monitor. Watch this space I guess.

The Arduino Sketch and the Eagle files are on github.

Any questions, just ask.

[typefaster]

Arduino Fridge Temperature Monitor

The poor old fridge didn't seem to be keeping things as long as it had used to, so I wanted to find out what temperature it was actually running at inside. A perfect opportunity for a small and practical Arduino project:

Goals:

Arduino Controller.

Dallas One Wire temp sensors to monitor temperature inside the fridge.

LCD display showing temperature.

Logging of temperature to an SD card.

Real Time Clock, so data is logged with a time stamp.

Ethernet connection, so data can be viewed remotely. (Missed)

Main Material:

EtherTen

5pcs DS18B20 1-Wire Digital Thermometer Dallas DS1820

RTC I2C Chip  (PCF8583T I2C Clock/Calender IC)

Construction:

The one wire sensor was mounted onto a small piece of strip board. The Dallas one wire can run with only 2 wires, but it sounded as though this would be a bit of messing about and I only had a short run, so decided to go with 3 wires. Power, ground and signal. Two female 4 pin headers and a male header were added to the board to allow connection to and from the board with different pin types. The male header was on the outside, but should have been on the inside to minimize risk of shorting. Hot glue was applied to the base of the board to remove the risk of shorts.  

I stripped some wire out of a cat5 cable to connect the sensors up. My home made RTC stamp (Post to come) was utilized again. The LCD was just plugged into a breadboard. A LCD I2C backpack is on the to do list to make it easier to deploy an LCD in future.

A battery pack was used to run the unit on the fridge. I really need to get a small A/C supply. Whilst there are lots of small plug packs kicking round, they all seem to be unregulated, so a small regulator stamp is on the to do list also. With good batteries (NiMH) the pack lasted about 8 hours. I left the back light off and I'm sure leaving the LCD off would add a lot of life.

Three sensors were daisy chained. One on top, outside, of the fridge. One at the bottom inside and one near the top inside. The one wire sensor bus (middle pin) was connected to pin 2 on the Arduino, along with a 4.7k pull up resistor (connected to 5v). I found that if a 5th sensor was added to the network the 4.7k value was too high. You can have up to 127 sensors on the one bus, so not sure what value pull up you'd need with a full bus!

Data was logged to the SD card on the Etherten. There was not sufficient memory to add the Ethernet port, so that goal was dropped.

The code from Arduino site was used to interrogate the sensors and get each serial number.  My final sketch then had the serial numbers coded into an array so the sketch can scan each sensor.

The procedure to get the temperature is to address a sensor and ask it to perform a temp conversion, which is dumped into it's scratch pad. Then we reset the bus, address the sensor again and read the scratch pad. The conversion takes ~750ms, which means scanning 3 sensors takes ~2.5 seconds. With parasite power you can't do anything else on the bus whilst a conversion is in process. As I have power and sense I think I could ask all three sensors to perform the conversion, wait 750ms, then scan all three. I might be able to scan all three in under a second. To be tried out.

The data logged to the SD card was pulled out via the serial port display, then copied and pasted into Libre Calc so that a chart of the temperature change over the lat 24 hours can be created. I logged the temp to the SD once a minute. The LCD is updated every scan.

The fridge was found to be running at or above 10C during the day when the door is being opened, and dropping down to only 7.5 overnight. This is way to warm. After some adjustment it's now running at around 8C during the day and about 3C at it's coldest, though the motor is on a lot more (of course). Opening the door almost always sees the temp go up by at least 3C.

I think a new fridge is probably needed. Given this one is about 25 years old I guess it's done okay. The refrigerator is a single door unit with manual defrost. so it's as simple as they come.

I'd like to be able to flag when the door was opened and also when the motor is running. The door could be monitored with a light dependent resistor, though would mean more wires into the fridge. Monitoring the motor would be trickier. something for another day.

The following sites were useful when developing the code:

http://www.arduino.cc/playground/Learning/OneWire

Arduino sample code in the IDE

My code is on github: https://github.com/TypeFaster2/Fridge-Temp-Monitor

Hopefully code has enough comments to make it easy enough to follow. Any questions, then just ask.

[typefaster]

Add a Text Post

Not really anything to see here yet.