Sunday, August 30, 2020

Local interference pattern emerging

 The terrible RF interference I see here right across HF looks like this:

But for three weekends in a row it goes off at 2:35pm local time on Sunday and the band looks pretty good for a suburban block. It comes back on again at 35 minutes after midnight. So, exactly 6 hours of peace a week.


What kind of appliance would be on all week except for Sunday afternoons?

The fundamental frequency, where it's strongest, seems to be at 421kHz where it's stronger than the local ABC Radio National transmitter here.


Another interesting pattern is that every 60 seconds, with good accuracy, the noise shifts slightly and then back. There is one example on the waterfall here.



Tecsun PL-365 on Longwave

Now that I've found the base frequency it seemed that I could locate it with a radio that receives longwave. I have a Tecsun PL-365 but it didn't seem to offer longwave even though it's listed as a feature. It turns out that longwave is disabled by default and frustratingly the manual's instructions for enabling it aren't quite right. 



The procedure that works for me is:
  • Turn the radio off.
  • Press the DEL (9/10k) button until it displays 9kHz (it switches between 10 and 9)
  • Press the MW key until it displays "150-" (it alternates between 522)
  • Turn the radio on, press MW and it will tune from 150kHz
As you can imagine, I've been walking around the neighborhood trying to find the source but it's very difficult to track down, perhaps due to the long wavelength.

Noise gone!

The broad spectrum noise I've been writing about ceased on Tuesday and hasn't come back. It's been three days now. Obviously it's possible that it could return but my theory is that it was some sort of battery charger that was being used by workers doing renovations next door.

I've just had a decent contact on 40m SSB with Kevin, VK2KB, who is in Sydney 677km from me. We have been running WSPR to figure out the best times to talk.

Friday, August 21, 2020

WSPR Watch update

There is an update to the WSPR Watch app in the iOS App Store. The recently added signal to noise graph has changed to a line chart and if you search the spot list for a callsign and then switch to the graphs the search stays.

This means you can use the SNR graph to find the best time to have a contact with a certain call.

Another welcome change is that there's a setting to show times in local rather than UTC.

Here's a graph showing SNR between me in Melbourne, Victoria and VK2ATZ near Newcastle NSW.

A few people have asked me about the use of colours on the map and graphs. Each band has a colour and I took the set used by WSPRnet.org. Unfortunately these colours sometimes have low contrast against Apple maps so it might help to switch to dark mode.

The dots on the map are as follows:

  • Green is a receiving station
  • Red is a transmitting station
  • Purple is doing both
Note that I have limited the number of spots shown on the map as it gets too slow. I stop adding them at 2,000 which I know annoys at least one user (Hi Ross!).

Thursday, August 20, 2020

A Telstra or NBN internet slowdown incident.

A few days ago we noticed that the internet was very slow. Pinging Google's DNS 8.8.8.8 showed about 20% ping loss. I logged in to the Telstra Wifi router and used the diagnostic to ping the same address and it also lost packets - so it's not within my network.

Here's how traceroute using mtr looked:

Checking for NBN outages on their status page found nothing.

I checked for any known outages on Telstra's outage page. They showed no known problems.

Next I ran their diagnostic and it seemed to agree that there was a problem and told me to ring a number, commenting that the operator would know what steps I'd already taken.

I powered off the cable router, waited a minute, powered it back on and after quite a long time all four lights became solid but the ping loss was the same.

Ringing the number, I was asked by the IVR to enter my account number, then day and month of my birthday. It asked if the phone I was calling from was linked to the account - it is. Pretty quickly I got on to someone (I guess in the Phillipines), they asked me again for my date of birth and for my name. They seemed to have no information about who I was or what I'd done.

They did some tests and asked for the MAC address of the cable modem. They asked me to turn it off, wait two minutes and turn it back on. I did. Eventually the lights came back on but the ping loss remained.

Yes! there's a fault. An NBN co technician would come to the house the next day between 8am and 12pm to fix my issue. Pretty good service I thought.

A few hours later the fault was resolved and I got a text saying that NBN co have competed further testing and no longer needed to visit.

Not much to complain about except to say that it surprises me that they can count every byte of upload and download for billing purposes but can't detect that the service is suffering major ping loss. Perhaps I'm quicker than the neighbours but I seem to be the first one to report an outage.

Things are back to normal and a trace route looks like this:

I'd love to know a bit more about the network. What is a vb08? What is a win-ice301? What's a win-core and ken-core?

A few more days have passed and an SMS just came in saying that I'll get a $7 credit to compensate for the disruption.

Nice.

Sunday, August 16, 2020

Unreliable Wifi on a 2018 Mac mini - a workaround

When I bought this Mac mini (2018) it was used with a permanent ethernet connection and all was well. It's only now that we're in a new house where I must use Wifi in my office that this problem has emerged.

For periods of time, up to about 24 hours, all will be fine and I can reliably ping the Wifi router over 5GHz Wifi, but then things will slow down and I can see terrible ping loss as shown on the right.

As you can imagine, this causes all sorts of trouble using the machine on the internet.

Searching for a fault with an Apple product, or probably any product, makes you feel that the problem is incredibly widespread but I think that it's just the illusion of being able to find similar stories with great ease on a search engine.

I have found many similar stories and tried all of the suggestions. Finally, I've found a workaround that is working for me. I have tried:

  • Rebooting everything (Wifi router, Mac)
  • Moving the Wifi router to get a better signal (RSSI -65, Noise -90)
  • Using a different Wifi router (Telstra modem and I tried a Google Wifi)
  • Turning off Bluetooth (there are stories of it interfering with Wifi)
  • Wiping the Mac and doing a clean install of macOS 10.15.6 Catalina (This seemed to work for about a day)
  • Unplugging USB devices (Again, stories of interference)
  • Running the built-in diagnostic, which found no hardware faults

In the end the solution for me is simply not to use Wifi. 

I have an old Thinkpad X230 on my desk running  Ubuntu 20.04.1 LTS which is used for WSPR, Fldigi, and other ham radio software. Following the instructions at Cesariogarcia I ran nm-connection-editor (which was already installed) and (slightly varying those instructions) edited the ethernet interface to set it's IPv4 settings to make it "Shared to other computers".

Open the Ethernet interface:


Edit the IPv4 settings to set the method like this:

I also changed the name so I'd remember. A pity this option isn't in the Sharing settings.

The Mac is plugged in with a normal Ethernet cable. (All recent Macs figure out if they need to cross over.) When I saved the setting on Linux the interface changed from "self assigned IP" to 10.42.0.42 with the router as 10.42.0.1.

I'd rather the laptop used bridge mode but I haven't figured that out yet.

Speed test isn't quite as fast as I can get when the Wifi was working reliably but it's much better than it was with 30% packet loss.


You can no doubt use a Raspberry Pi for this task and there are little Wifi routers that can do this too. The Thinkpad barely idles even when running a speed test. 

I hope this helps someone else with this issue. Incidentally, the Mac mini is out of warranty and I fear that if I take it to Apple it will work perfectly while they have it. I know I should give it a shot and I have had great experiences with the Genius folks in the past.

I've been monitoring with PeakHour and you can clearly see that ping loss (blue graph) has stopped since adding the linux gateway to the mac's network.

Apple has some tips on resolving Wifi and Bluetooth interference issues but this hasn't helped in my case. It might be worth trying before following my radical path.

What about printing etc?

One problem of having the Mac off the main network is that Bonjour zeroconf can't see the rest of the network so devices like the printer aren't immediately visible. (You can still get to them but you need to figure out the IP address).

The solution is to configure avahi-daemon on the Linux machine to "reflect". You do this by editing /etc/avahi/avahi-daemon.conf

[server]
allow-interfaces=enp0s25,wlp3s0

[reflector]
enable-reflector=yes

(I got the interface names by running ifconfig).

You need to restart avahi:

sudo service avahi-daemon restart

It takes a minute or two for the names to propagate. I check them on the Mac by using the excellent Discovery app from Lily Ballard.

Update - a USB Wifi adapter

As a simpler alternatie I ordered a Hungwu USB 3.0 Wifi Adapter via Amazon


It came with a mini CD containing the driver but I don't have anything that can take these disks. I found an open source driver that works here. It's a bit weird in that you get an additional Wifi menu bar item that has found networks in a non-standard font.


Pretty bad that it tries to take over Command-W and Command-O but happily this doesn't affect the Finder or other apps.

The seller responded to my message quickly and sent a link to a driver but the one I found is working fine so I'll stick with it for now.

So far it's working well with no ping loss 24 hours later and as good a speed as I've had with the native Apple Wifi interface.


Call me an old "fuddy-duddy" but I find <2ms ping times over a wireless network pretty amazing.  For now, I've retired the Linux wifi to ethernet router described above. Note that Airdrop and watch unlock don't work with this Wifi adapter.

Saturday, August 15, 2020

Appeared on ABC RN's Download This Show

This week, to coincide with the 75th anniversary of the war in the Pacific, we talked about some of the technologies that were developed in that time of war that have shaped the world today.

As always, a pleasure to be a guest on the show. You can hear the full radio either across the ABC RN network or here.

It was (I think) an interesting show, other topics were the death of mouse inventor William English, research that seems to indicate that Tinder charges us old blokes more and the impending ban on WeChat and TikTok.

There is a segment edited for TV that goes to air on ABC News24 on Saturday morning and you can catch that on the YouTube channel below.

There's even a portrait edition for Facebook here and an Instagram TV version. Talk about multi-media!

We're all getting pretty good at making radio and TV from home. This week I used a pretty low end Logitech headset and mic and it sounded just fine.

Thursday, August 13, 2020

Experimented with EasyPal (modern SSTV)

Last Friday I watched an excellent presentation given at the EMDRC by old friend Chris Long, VK3AML. You can see it here:

EasyPal is software for Microsoft Windows that lets you send images and other files over an audio bandwidth link. The image is sent in digital form over an OFDM modem similar to that used for Digital Radio Modiale

Here in Melbourne, a group of enthusiasts have been given permission to perform tests on the VK3RML 2m repeater on Thursdays from 9pm. Here's an image I received:

The software is a little weird (to my eye) and rather technical but I was able to get it going and both send and receive images.

Getting it working over the FM repeater seems mostly about not overdriving the audio when transmitting and Chris suggested I switch from the defaults to QAM 64 and Mode A to get faster transmission over the reliable FM link.

This is a great feature, the ID button sends your callsign in a way that displays on a spectrum waterfall.

My radio, a recent Icom rig, was not in the list of supported CAT devices and the latest version of EasyPal is dated 07/Oct/2014. I found a description of the hex codes to paste into the configuration screen on a Facebook page but it had a side effect of disabling my microphone! (In the end I had to reset the radio to get it working again).

Why no updates for six years? It seems that the author, Erik, VK4AES, died in 2015. Ironically, EasyPal came about when he lost the source code to an earlier version in a lightning strike near his home and had to start again. Erik, it seems, didn't like to keep copies of source code off site and I fear that EasyPal's days are numbered unless it can be updated.

QSSTV

There is similar software for unix computers with Qt support. QSSTV 9.4.4 was last updated in August 2019. Conveniently, it can be installed with Apt on Debian based Linux distributions. QSSTV is well documented and supports DRM modes including Mode A and QAM 64 which I used on EasyPal.

Will they interoperate? I hear that Terry VK3FTJS uses QSSTV on the Thursday contacts and says it's better at digging a signal out of noise than EasyPal.

So far, I've only been able to decode one transmission with QSSTV and I suspect it was sent with QSSTV. I've been recording the contacts on the repeater and pipe the audio over to EasyPal on Windows using VB-Cable which works well playing the audio in Audacity.



 

Tuesday, August 11, 2020

Why ECG isn't enabled on the Apple Watch in Australia

Recently I’ve experienced some unusual heart rhythms and was referred to a cardiologist by my GP. The rapid beats only happen about once a month and so, despite carrying a 24 hour “halter” monitor a few times, the event hasn’t been captured.

My Apple Watch does notice a rapid heart beat at the times I notice the events so I know it can be detected by a simple wearable device.

The Apple Watch has the ability to record a two-terminal ECG and this feature has been made available in many countries including US, Singapore, Hong Kong, UK and now New Zealand. 

The feature is not available in Australia where it must be approved by the Therapeutic Goods Administration. The Medical Software Industry Association, in a submission dated 22 May 2020, mentions Apple and opposes approving their devices to a (presumably) more lenient standard that the competing devices which their members market.

“Apple is becoming a health company along with many other large software and technology companies. The benefit of this are many, but the risk to Australian industry is that if these companies do not have the same imposts imposed upon them as our own Government places on Australian companies, the price signals will prejudice our market. Without question the cost of this regulation would be prohibitive and is avoidable”.

The Apple ECG feature has been approved by the US FDA but it’s interesting to note that they explicitly exclude health software which they call “software as a medical device (SaMD)”.

Not all medical software is regulated by the TGA. Current exclusions include software limited to managing and presenting information such as dosage calculators, software that is a source of reference information for professionals or consumers, software that has a role in diagnosing or managing an illness, and software that analyses clinical data such as results of blood tests or ECG. Note that the software can’t record the ECG and analyse it.

My cardiologist recommended a device that can record a simple two terminal ECG called KardiaMobile that is sold in Australia by “alive technologies”. 

The AU$199 device measures the 10mV voltage differential between fingers on each hand, modulates this on a high frequency audio tone and the app records and graphs the waveform.

I’ve read that they use audio for the link rather than bluetooth to save a bit of money but this means it needs a quiet environment to work. Recording the audio and zooming in you see this waveform:


The spectrum shows a big peak at 19kHz:

The app does a little analysis of your ECG, but my cardiologist warned me that it would either say “normal” or “unclassified” which is my experience.

It’s not clear if Apple has applied for approval in Australia but it is clear that there is local industry opposition and because the TGA’s scope is different to the FDA perhaps Apple is holding back.

My cardiologist mentioned that there is a dispute between AliveCor and Apple where AliveCor has been granted a patent for proactive notification of possible heart arrhythmias, further AliveCor worked with Apple to link their KardiaBand device to the watch back in 2017 but subsequently that functionality was built right in to the watch, which no doubt upset them.

The KardiaMobile device is reasonably priced and easy to carry but I would rather just use the watch. There are claims that the analysis Apple does is superior to that of the earlier KardiaBand. I hope that whatever is preventing the ECG function in the Apple Watch from being released in Australia can be overcome soon.

Thursday, August 06, 2020

Logging noise floor with hamlib

From time to time I see terrible interference right across HF and even up to VHF here. Sometimes it comes on for under ten minutes and while I make a note in the log when I see it, automatically logging the noise level seems like a good way to find out if there is a regular pattern.

At this point I haven't satisfactorily visualised the problem but I wanted to leave a note about progress so far.

Here's a graph of one minute samples of the S meter reading on 40m for the past few days. Times are local.

The values are read via a USB CAT interface to a Yaesu radio tuned to 40m. The values seem to be roughly S4 = 400.

I use the hamlib command line utility rigctl to do the real work.

To log these values I have a cron job that executes every minute and runs this script:

The resulting log file looks like this:

There's a tab between the date/time and the value.

To draw the graph, I use Gnuplot running this little script.

So far this isn't clearly showing when the interference is "on". Possibly the logarithmic Y scale doesn't help. I think it's interesting to see the noise floor measurements but they vary widely and I think the next step might be to take multiple measurements each minute and perhaps log the average or minimum.

Here's the graph of an additional day but with five quick readings taken and averaged each minute. Still not a clear result.


Monday, August 03, 2020

Graph of SNR now in WSPR Watch app

I've just updated the iOS and macOS app WSPR Watch to include a new graph of signal to noise ratios. So far I'm not seeing any useful information on this graph at my location. It's been suggested that to be valuable it would need to work with the callsign search feature so you can plot SNR over time between just two stations.

This version, 3.11.1, also fixes a surprising (to me) bug that affected users in the UK. As dates and times are in UTC and so is the UK for some reason my date parsing was failing for those users. I'm astonished that I hadn't heard about this before now.

My thanks to Paul, G0DIH, for the bug report and for working with me to test builds until the issue was sorted out.

Here's how the SNR, and other, graphs look now.


HF horizontal loop antenna in a small back yard

The back yard in our new house is very small and there is an intermittent interference source that has plagued me since setting up here. Phil, VK7JJ, is a great advocate for horizontal loop antennas which he uses to attain world-beating WSPR reception. Phil encouraged me to fill the available space with a horizontal loop.

The area between the back of the house and the front of the garage is about 5m x 3.6m and at the right you might be able to make out a loop made with thin black hookup wire. This antenna was fed from coax from the operating position via a common mode choke.

There's a small verandah with a wooden pole which is convenient for mounting baluns etc out of the rain. Here's the first attempt.


Later I did a better job and used an electrical junction box. 


The base nails to the wood and then the top is screwed on to that. Not waterproof but fine for my use. 


The loop works well on 20m but wouldn't do 40m. I'm not sure why at this point, but the loop's signal to noise seems indistinguishable from that of the 20m dipole along the fence line.

I've been improving the dipole by attaching one end to the tripod on the garage roof and I think the height advantage evens things up. One idea is that I might put more poles on the garage to hold up one end of a loop and as well as increased area I'll get some more height.