Sunday, 29 August 2010

Multi-Mode Beacon & Blogger Image Upload Failure

Some more programming for the PIC 16F628 in its hi-rise accommodations has seen my 50mW 30m beacon go truly "multi-mode". It now transmits WSPR, QRSS and Sequential Multi-tone Hellschreiber in a 10 minute cycle.

I have stayed with the basic timing framework offered by Gene Marcus w3pm's excellent audio WSPR generator code, but added my own routines for the QRSS and Hell. This was easy-enough to do, as I'd already hacked the code to run with my VXO-based architecture and all three modes are just FSK.

Here's a screen-shot from Argo, captured locally, showing all three modes...

You can see the closing "d" of my call sign in QRSS, a S/MT Hell version of my call and (most of) my WSPR message. The frequency difference between the QRSS/Hell and the WSPR signals exercises pretty much all the range of my VXO - I have to work the bottom of the WSPR band, which places the QRSS/Hell emissions around 10.140050 MHz.

I turned the beacon on late yesterday afternoon and got good WSPR spots through the night (once again bucking the trend that had been established last week). Ususal dx to Douglas w3hh and very solid propagation to Al, ra6as...

OK - I give up - Image Upload isn't working. It is producing garbage like this (which I'll leave to shame the providers of this "service")...

I would like to have shown you the WSPR spots, but Blogger disagrees. I've tried different image formats, uploading from different browsers (I usually use Chrome), re-starting the computer, uploading from another computer - all to no effect.

It is now three hours later and - guess what - everything is working FB now. No change this end - of course. Just working now when it wasn't before. So here's the WSPR spots I wanted to show you earlier...

In between each of the 5.6 Joule flashes from WSPR, there are now also the other modes to entertain me. Here's a nice spot of the QRSS from pa3tab's "Tabber" grabber..

Here's Joachim's spot of the Hell (preceded by some of my QRSS) ...

I wonder if anybody else is running multi-mode?

...-.- de m0xpd


WSPR spots coming in at a pleasing rate (more overnight spot stats to follow) and I was particularly pleased to receive a report from Pierre, on5sl, through the knightsqrss mailing list...

Thanks Pierre!

To:- the people at "Blogger" (/Google)
The point of blogging is diluted if the supporting infrastructure doesn't work (as in my image upload issues above). We didn't turn to our paper diaries and journals to make an entry, only to be frustrated by a book that wouldn't open or a page that wouldn't accept ink.

Blogging needs to work NOW in real-time.

Forget all the "too-clever-by-half" facilities you're trying to add and just make the system WORK reliably, please.

Saturday, 28 August 2010

Just when a pattern seemed to be forming...

After describing the silence which fell on my WSPR spots during (local) nightime last weekend, I continued to gather overnight data through the week.

The pattern of activity fading during the evening to an overnight quiet, followed by a re-awakening with the dawn seemed consistent (although there was a general fall-off in spot frequency throughout the week). Then, just as I was getting complacent, last night's data threw up a surprise - spots throughout the night!

The week's data are seen here (with apologies for the AWFUL 3-d graphing capability of Excel)...

I wonder if this is a reflection of a real change in propagation last night - or a reflection of greatly increased WSPR reporting activity at the weekend? There is a slight suggestion of weekly periodicity in the "spots per hour" data at - although the peak seems to be on Sunday/Monday as opposed to Friday/Saturday.

Despite this collapse of what I believed to be a consistent pattern, one truth does seem to be preserved throughout the last week's observations; I have good chance of DX on 30m just before midnight (UTC). Aside from Douglas, w3pm's spots reported last week, I was delighted to be seen by John, w3prb to the West and Al, ra6as to the East in this "hot time" around midnight (UTC) - pity I'm fast asleep!!

Incidentally, the manual collation of numbers of spots of each of my five-and-a-bit Joule flashes was becoming a bit of a bore - so I wrote a program to convert the data from WSPRnet into a count. The program was written in MATLAB (because of familiarity and convenience).

Here's a typical section of data downloaded from WSPRnet...

and here's the associated section of the "counts" output file...

Logging the actual spots (such as those at 17:24 and 18:04 in the example above) is pretty trivial - but padding out with the flashes which weren't reported (such as the zero spot count at 17:34) is a more interesting programming problem - especially as I wanted the program to be robust to "resets" of my transmitting schedule (which meant I may have been transmitting at 4,14,24... minutes past the hour but then reset and changed to 8,18,28... minutes past).

Either that or my senility is advancing faster than I first thought HI HI

...-.- de m0xpd

Monday, 23 August 2010


I put the new WSPR beacon to use over the weekend, concentrating on the over-night period (I want to use the station antenna at other times!).

I was up ridiculously early on Sunday morning (the result of drinking too much on Saturday evening) and I noticed the beacon wasn't getting spotted. I checked that it was indeed working - which it was - and I resolved to watch the "start-up" of spots. What I saw was an interesting pattern of fading reception in the evening, silence through the night and a re-awakening with the dawn.

I decided to count the number of spots reported for each "flash" from the beacon. Each flash is 112 seconds long and (at 50mW) that is 5.6 Joules of energy.

I present the counts for the last two nights as a graph - The night of Saturday 21st in blue and the night of Sunday 22nd in red...

The additional "authority" which graphing confers to the data isn't to be taken seriously... I realise that this is at best a pseudo-scientific statistic! However, the number of times my five-and-a-bit Joules are reported is interesting - even if it doesn't account for relative location of the reporting station, the received SNR, etc., etc..

Also, it doesn't account for any differences in WSPR operating practice - I wonder if there are just more WSPR stations active on a Saturday evening than on a Sunday (explaining the peak just after sunset on Saturday)?

Whatever the shortcomings of the "data", it does suggest some patterns of 30m propagation from my QTH...
  • First, there's an orderly fade down to overnight "Quiet" (see the postscript below) and a more abrupt fade up with the dawn.
  • Second - and more interestingly - there's a chance of some DX overnight. Douglas, w3hh, in Ocala, FL (who has spotted my emissions before) heard the new beacon at almost exactly the same time during both the last two nights. He's hearing me at the end of a path achieving 85000 miles per Watt (with reported receiving SNR of -24 dB).
Those (many) of you who know more about propagation than I can comment on how ordinary (or otherwise) these observations are.

I'll try and collect some more overnight data to make this study (slightly) more meaningful.

...-.- de m0xpd

p.s. "Corcovado" is the name of the mountain on which stands the famous statue of "Christ the Redeemer" above Rio de Janeiro. It is also the name of a song by Antonio Carlos ("Tom") Jobim. The song was made famous by a number of artists who've recorded the English version, the lyrics of which were contributed by Canadian writer Gene Lees.

That version is entitled "Quiet Nights of Quiet Stars".

Saturday, 21 August 2010

Hi-Rise Development

There comes a point (as Tokyo's town planners realized) when simply spreading out in two dimensions stops making sense - it it better to exercise the third dimension. Thus it was with my new controller which allows the VXO beacon system to generate WSPR signals.

I had demonstrated how Gene w3pm's software could be modified to generate contol codes for the MCP4922 ADC in my VXO system, firing these control codes down a temporary SPI funiculus umbilicalis. But I wanted something more robust...

It made sense to build an "add-on" board, which literally plugged into the socket which once held the PIC16F676 of the original system - hence my foray into the third dimension.

The board was quickly designed in Eagle...

... produced using Alberto's magazine method and my new laminator and quickly stuffed...

The cut-out seen at the top-right corner is designed to leave access to the RF output from the VXO/mixer board.

Here you can see the underside of the new add-on board ready to plug into the vacated 16F676 socket...

... and here is the new, three-dimensional, "Hi-Rise" assembly...

Now the VXO-based WSPR system is rather neater than before...

All ready to fit into the nice new box I saw at Maplins today.

...-.- de m0xpd

Wednesday, 18 August 2010


No sooner had I smoothed out the wrinkles in my WSPR audio generator, I decided to return to the original plan: generating WSPR signals with a voltage controlled crystal oscillator.

I had enjoyed success with my QRSS and S/MT Hell Beacon using a reverse biased Zener diode to trim the frequency of a crystal oscillator. Ever since building that system, I planned to use the same approach to generate the 4-level FSK signal at the heart of WSPR.

I had Gene Marcus, w3pm's PIC code running in the AF generator, so all the timing problems were solved for me. All I had to do was add the ability to send control codes to the ADC on my VXO system relating to the four FSK frequencies. This was achieved by inserting a subroutine call within each of Gene's "symbol send" code segments...

You can see these additional subroutine calls highlighted by the red lines.

The called subroutines are really just individual entry points to the same routine - one entry for each of the four frequencies...

The actual Data send to the MCP4922 analog to digital converter is performed over a three-wire SPI interface and the driving code is embarrassingly "brute-force" - but it works...

With these simple modifications to w3pm's code for the PIC 16F628A (which, conveniently, had four unused I/O pins in Gene's application), I could apply the three required SPI lines to VCO/mixer board, via the pins in the socket that originally housed the PIC 16F676 in my system...

I also used the existing Tx keying line in the k3bm code to mute my VXO-derived WSPR signal, as required.

The system worked first time, delivering a nice clean 50mW (17dBm) into 50 Ohms...

I didn't even need to change the control codes (mapping to the VXO control voltages) that set the FSK frequencies for WSPR - they're the same step sizes I used in generating the S/MT Hell signal, obviously close enough to the required frequency deviation of 1.43 Hz in WSPR.

Talking of my measured 17dBm, I wonder just how many of the WSPR signals one decodes are from transmitters really running 37 dBm. Certainly, the majority I spot seem to declare themselves as pumping out 5 Watts. On the one hand, they are likely to be received at such power levels (to my mind, 5W isn't a "weak signal" by any means). On the other hand, 37dBm seems to be the "default" setting for the Power in the "Station Settings" dialog of the WSPR PC application - perhaps many people don't change this setting.

This marks the end of my production of DSB WSPR signals. What would have been my "lower sideband" (when running on 30m) is down near the 80m band and it is well attenuated by the 10MHz bandpass in my beacon system, as evidenced by the clean sinusoid in the (poor) photo above.

Here's an idea of performance of the new system...

No spectacular dx, just good solid propagation into Europe.

What now? Well, there's obviously a chance to replace the breadboard with something a little more robust. But the latest number of RadCom has just landed on the doormat, in which Richard, g0vgx, describes a QRP PSK31 beacon for 30m. That could be fun!

...-.- de m0xpd

Sunday, 15 August 2010

Back on track with WSPR

A few weeks back I reported problems with my attempts to transfer the PIC-based WSPR signal generator to PCB - I was getting very poor frequency stability. More recently, even the breadboarded version started to display exactly the same symptoms (after having been working perfectly). I decided to take another look at the PCB-based WSPR generator.

I discovered that my "keying" arrangements were not just muting the audio signal, as intended. Rather, they were also inadvertently shutting down the oscillator. Thus, each time the system went into "transmit" mode, the oscillator started from cold - little wonder I wasn't getting the best stability.

I disabled the faulty "keying" stage and observed the frequency change after a cold start, seen in this trace generated by the WSPR PC application...

You can see how long the thermal time constant is and appreciate why I was having such trouble with frequency stability when switching on the Tx every ten minutes!

I modified my version of Gene Marcus' circuit again, this time using a quarter of the analog switch to explicitly gate the audio signal, rather than switch my faulty "keying stage". This left the oscillator running all the time - et voila - the desired stability. I quickly finished off populating the PCB and connected it up to the rest of the system...

I hooked up to the rejuvenated g5rv and - although it wasn't the best time for 30m spots - started making my presence felt in a quiet, WSPR-ed way...

It has been very satisfying to get these totally homebrewed, PC-free, double sideband WSPR signals spotted. Perhaps its time to start thinking about putting all the beacon boards into a permanent enclosure - certainly it is a nuisance having them clutter up the bench!

...-.- de m0xpd

Friday, 13 August 2010


Flushed with the success of recent modifications to my straight key, attention turned to the m0xpd g5rv antenna.

The antenna has be "misbehaving" over the past few weeks - first symptom was a sudden change in the required settings on my MFJ-969 ATU, along with a sense that performance was poorer than in "the good old days". But things came to a head when the antenna suddenly started presented a changing load, such that I could tune for great VSWR one minute and the next, the indicated reflections were enormous. I had to do something!

After checking the feeder and all the cabling between rigs, switches and the ATU, I finally dropped the g5rv and had a look at the centre...

Sure enough, the left hand connection in the photo above was not only rusty - but LOOSE! I replaced the machine screw and nut and tightened up - and I'm pleased to say that the g5rv is back to its old ways!

I had a few more QSOs on 80 and 40m with the "trusty rusty" FT101ZD and then I decided to try something which has been on the "to do" list for a few months...

I fell for an old HW-7 at this year's Red Rose QRP Rally, but hadn't yet been able to try it in anger. I had discovered a few "idiosyncrasies" about the rig:

First, the jack socket for the key is isolated from ground (you can just see the fibre insulating washer in this photo)

I subsequently discovered that this strange arrangement is such that the rig can be used with the "Heathkit Electronic Keyer, Model HD-10". That's all very well - but it couldn't be used with my "Funky Keyer" until I made up an adaptor.

Second, the RIT control (part of the g4zqk suite of mods to which my example had been subjected) needed to be set a long way from the indicated "neutral" position to net anybody...

After I'd cracked these little "issues", the rig was ready to rumble...

I hooked her up to the newly rejuvenated g5rv and answered a CQ from fellow FISTS member Brian, m0brb. He gave the HW-7's 2 Watts a 559 in Devon. Thanks Brian.

The HW-7 works well enough, but I'll be using it more as "novelty" than as a serious rig. I is stable and generally easy to use - but I HATE the buzz-saw sidetone (regular readers will know I'm fussy about sidetone). It is good to own a nice piece of history and it certainly makes me feel "at home" when I leaf through old numbers of SPRAT!

...-.- de m0xpd

Thursday, 12 August 2010

Str8 Key Modifications

A change being as good as a rest, I decided to have a bit of a do with my "trusty rusty" FT101ZD. My home-brewed straight key wasn't keying the rig cleanly - at first I thought this was due to dirty contacts, so I cleaned them (with paper - I don't use abrasives). No effect.

Further consideration led me to try what I thought was the unlikely possibility that the electrical connection through the "hinge" was less-than-perfect - simply tested with a temporary cable...

Sure enough - and surprisingly enough - this sorted the problem.

I've seen little electrical bridges over the "hinge" on Kent straight keys and always thought this over-kill. However, I made up a flexible link (from some solder tags and some copper braid)...

... and fitted it to some conveniently located screwed fittings...

Result: the key is infinitely better! I can't see why the electrical path through the bushings (Phos. Bronze), the axle (Silver Steel) and the key's "arm" (Brass) should be so "iffy" - but the new strap has made all the difference.

Not bad for ten minute's work!

...-.- de m0xpd

Monday, 9 August 2010

Where have all the grabbers gone?

Where have all the grabbers gone?
Long time passing
Where have all the grabbers gone?
Long time ago
Where have all the grabbers gone?
Gone to graveyards every one
When will they ever learn?
When will they ever learn?
(after Pete Seeger)

Various distractions and responsibilities have kept me away from the radio (and this blog) for a couple of weeks - but I was determined to put some time in over the weekend. Initial looks at the CW band segments on 80 and 40m suggested little satisfaction was to be had there - so I fired up the 30m QRSS beacon.

On opening up i2ndt's grabber aggregator page I was met with something of a surprise. 30m grabbers are becoming an endangered species (over last weekend, at least)!

Many QSYs to other bands (including a temporary shift to 20m from Guenael and his precipitating pachyderms), some talk of astronomical observations - what's going on? With my h/b 30m beacons a QSY to another band isn't a matter of flicking a couple of switches!

Now - let's be quite clear... this isn't a complaint. I have already explained on this blog the debt of gratitude we owe to the grabber operators, saying (with deliberate hyperbole)"those who only transmit are much to be despised". However, an unexpected drought of reporting receiving stations is worthy of comment!

Once again Joachim, pa1gsj, was my salvation - here's a spot from his "Suburban Subharmonic QRSS Grabber"...

(you can just make out my "m0xpd" in S/MT Hell followed by QRSS CW [bottom left at 10.14002 MHz]).

Andrew, g4cwx, also did me some service - thanks!

Frustrated by the poverty of grabbers and my poor propagation, I shifted to my h/b WSPR signals. After a weak start (that poor propagation again), eventually I started to get noticed...

Note here that Joachim, pa1gsj, has spotted BOTH my QRSS AND my WSPR emissions. WOW! Thanks Joachim! (I just added Joachim's blog "de draaggolf" (The Carrier Wave) to my links).

I have a few issues to resolve...

First, I am beginning to believe there's something "different" about the m0xpd "home" antenna (a half g5rv with inductors and extensions for 80m). There has been a small but noticeable "shift" in the ATU settings required on all bands. Also, I notice a difference in the wet and (as a newcomer) I don't know if it is within "normal" limits. I am not sure my antenna isn't the origin of my "poor propagation" noted above. I'm having a lot of trouble upholding my faith these days!

Second - and much more important - I want to encourage and support the 30m grabber operators. Obviously, one of the clearest means of supporting the grabber role is to put my money where my mouth is and operate a grabber myself - that is now under consideration. Until the time I make such a tangible sign of collaboration, I simply ask all fellow transmitting QRSS enthusiasts to join me in my expression of thanks and support for the remaining 30m grabber OPs.

...-.- de m0xpd

Sunday, 8 August 2010

Codebreaker is broken

Thoughts today turn to the fickle fragility of faith – how easy it is to burst the bubble of belief. It is Sunday, after all!

Regular readers will remember how I have been playing with the cryptanalytical challenges posed in the back of Stephen Pincock and Mark Frary’s book “Codebreaker – the history of secret communication”. I had been sailing along, happily revealing the messages in Challenges I – IV until I hit an obstruction at Challenge V.

The subject of Challenge V is Public-Key Encryption and, some two weeks ago, I followed the authors’ advice and referred back to Chapter 5 for help, before tackling the challenge. This was a big mistake – this was where all the corrosive doubt set in – here faith faltered and belief burst. For the example which seeks to illustrate the workings of PKE on pp 134 – 135 contains an error.

The error looks harmless enough – it simply gives the product of 11 and 17 as 181 (when, in truth, 11*17 = 187). It may even be that the mistake crept into the process during transcription of a hand-written 187 into a type-written 181 – the character “7” can look like “1” in some fists. (Indeed this is precisely the reason I use the affectation of putting a line through my scripted sevens, in the continental style).

Unfortunately, this potentially innocuous error propagates through the entire example, which is then totally incorrect – the mathematical transformation through the encryption “function” is not undone by the inverse function, such that we do not have a viable encryption scheme.

All might have been fine if this were just an error in the body of the book – but I had learned from previous Challenges that the examples in the text often couple directly to the configuration of the challenges (as, for example, where the Polybius Square in the text was used verbatim in Challenge IV) – so I began to suspect that the error in the example might contaminate and undermine Challenge V. As soon as I began to suspect, the cascade to doubt, faithlessness and disbelief followed.

My suspicions were compounded when I looked at the “Solutions” page

This solution resource is interesting because it ONLY gives the solution – not the “working” that explains how to get to the solution (which I have included in my posts about Challenges I – IV, Blogs passim).

In looking at the “solution” to Challenge V, you will see that there is not a unique mapping between cyphertext characters and plaintext characters (of course it is possible and desirable to make ciphers which are not “character based” in that sense – but my suspicion and belief is that the present Challenge is to be interpreted in that "character-based" manner – certainly that is the authors’ strong implication).

Looking at the first three cyphertext characters, each has a multi-valued mapping into plaintext (according to the "solution")...

Cyphertext Plaintext
N T, T, S, E
C A, O
W E, E, N, I, E

Sorry – but I don’t buy it!

Of course, there are lots of other “issues” in the solution of this Challenge that I haven’t alluded to here (particularly in relating alpha characters to the numerical codes that represent them), but I won’t bore you with those.

At this point, I wrote to both authors (using that familiar contemporary “code-breaking” method of finding peoples’ email address when they prefer to remain anonymous), thanking them for the book and pointing – politely and constructively - to their error in the description of PKE. I did not mention my crumbling belief or identify myself with this blog. To date, I have had no reply. Evidently Stephen and Mark are too busy penning their next bestseller/contribution to the remainder bin (delete as appropriate). Pity.

I’m not going on with the remaining challenges at the moment – in fact, I’m setting one myself...

If anybody has “solved” Challenge V (I am NOT saying it is wrong or intractable – I’m just saying that I have currently lost faith in it) PLEASE contact me. (What do you mean - "you haven't got my email address"?) I would be happy with just a clue which re-instates my faith and I would even pick up the challenge again!

All this nonsense and frustration has, at least, had one good outcome – I was tempted to buy another “code-related” book -

Perhaps I'll share my impressions of B. Jack Copeland et al's book when I've had a chance to open it. Until then, here's an end to all this Cryptic distraction - let's get back "on message" with some radio-related content!

...-.- de m0xpd