Saturday, 29 July 2017

New Mode Time - MSK144, and Random MS

With propagation down, and the weather rather hit and miss, I thought i'd try my hand at something new.

To make things difficult for myself, I picked a little known datamode protocol - MSK144, and an underused band for which I have only a low, fixed azimuth beam - 6m.

And to ensure I really did have the hardest time possible, this mode is used for a difficult propagation path - Metero Scatter!

The principle behind meteor scatter is fairly straightforward. Everyday, thousands of minute particles of space dust, usually no larger than a grain of sand, collide with the earths atmosphere, and are burnt up by the protective layers above us. Those grains, while out in space, are Meteoroids. As they pass through the atmosphere and burn up, they become Meteors. Should any part of it survive its encounter and reach the earths surface (which would need to have been a rather big bit of rock!) the remains are Meteorites. On any clear night from a suitably dark location, you can see a few random meteors. Those of you of a Disney fairytale princess persuasion may wish upon these shooting, or 'falling' stars - those with a hundred watts or more of throbbing RF and suitable software can make contact over them!

As the meteor burns up, it leaves behind it a trail of ionized particles. It is against this trail, a fleeting moment lasting from a few tenths of a second to a handful of seconds for a big, pebble sized meteor (which visually would produce a fireball or 'bollide') that we fling our radio messages, hoping to scatter enough signal from the trail to reach our partner station.

At certain times of the year though, the earth passes through streams of dust left behind by the passage of comets through the solar system. At these times, there is a marked increase in the number of meteors, and these are known as Meteor Showers. As with any event that affects propagation, these bring the latent operators out of the woodwork. However, the next shower, the Perseids, is several weeks away.

At times like this, you either set up a 'sked' with a particular station, or call CQ to try and make a random QSO...

As you can imagine, A random MS QSO is hard work! Your facing a combination of limited beamwidth (the same trouble that we have photographing meteors - your limited lens field of view means the meteor you see with your eyes is usually not in the cameras view!), rare and fleeting propagation, and few, if sometimes indeed any, other stations using the same mode and same frequency.

But, there are ways around! And thanks to the internet and fast CPUs, its a bit easier! We can announce on dedicated chat forums that we are calling, there are strict rules and message formats, and there are modulation protocols that run fast messages to increase the chance of a usable reflection. In the old days, MS was done by high speed morse! Now, its done using fast data modes such as MSK144, JT6M and FSK441.

But, despite all that, a random MS QSO is still a challenge! To state how much of a challenge, well, Ive probably spent 50h on it in the past few days, transmitting for maybe 2h total key-down time, and ive made two contacts!

The screenshot above shows a meteor 'ping'. The two halves of the window are 15 seconds long, and my transmit slot was 15 seconds between them. In 30 seconds then, only a single brief meteor passed at the right altitude, and position, between my station and that of Jurgen DK4AN. But, it was enough!

Sometime in the previous five minutes or so, a meteor had also passed that point, during one of my 15s Tx slots. Its decaying trail lasted long enough to reflect enough of my MSK144 message to Jurgen. The meteor in the image above, reflected his reply!

The software Im using is Joe Taylor W1JT's WSJT-X v.1.7. Joe is the chap who brought us WSPR plus a host of other modes for weak signal or fleeting propagation use - JT65, WSPR, JT9, JT6M etc etc. A new mode from him FT8, is rapidly gaining popularity! My antenna in this case is a 3 el yagi for 6m at the low height of about 15ft, fixed in direction to 120 degrees (roughly SE), the radio my Alinco DX-70TH running 100W. I am also using the VHF/UHF chat room of ON4KST online, in order to help co-ordinate my effort with others!

Around five minutes later, the above meteor 'burst' occurred. This was probably a slightly bigger grain of dust, maybe about the size of a bit of gravel, and it seems to have burnt up in pulses, a bit like when you see a naked eye meteor that seems to blaze up a couple of times. During this burst, I copied DK4AN again, decoding his message a couple of times. Now, as you can see, the messages are very short and direct. No data is wasted. The format is strictly controlled to ensure the right messages are sent and theres no confusion. As it happens, the WSJT-X program allows you to select 'Auto Seq' which means once you choose to reply to a signal, the computer can do the rest!

This last screenshot shows the completed QSO. Over a period of about 15 minutes, enough meteors were in the right place at the right time to allow their fiery destruction to do good and reflect our messages. A final confirmation via the online chat once the QSO was complete was made, to say thanks to Jurgen for his patience.

So, if you've always thought that meteor scatter was a mode for those with 'big gun' money for huge antennas, expensive low noise GaAsFET preamps, big linear amps and costly dedicated reel to reel high speed morse tape recorders, think again, and perhaps give it a try. If your in an IO or JO square, send in the 2nd slot! (convention - South and East sending in 2nd slot, North and West sending 1st). You only need a modest station, and a lot of patience!

Wednesday, 12 July 2017

'Scope repaired, and some fun with the Rectum Paralyzer

After some time spent blasting every moving contact in sight with Servisol, and resoldering any connection that didnt look pin sharp, my Hitachi V-212 20MHz Dual Trace Oscilloscope is now back working how it should be.

Inside the scope is very tidy, but clearly many years of grime had built up on the controls, causing all manner of irregular operation. Since I had to remove many of the plugs, these were all numbered with a Sharpie type pen so I knew where to put them back!

These two photos might seem pointless - actually, they are visual references as to where the correct control knobs and nuts were to be refitted! I also found that the protective film was never removed from the front panel, and was now in a poor state, Removing the film has brought the front panel back to near new looking condition!

After putting the scope back together, and replacing all the test gear on the bench where it should be, I started to have a play with the Spectrum Analyzer. (the weight of my analyzer is why it gets the name in the title! If your not careful with your manual handling these things can give you some interesting ruptures). I had heard that it was possible to get demodulated audio out, but wasnt sure how. Several youtube videos 'almost' helped, except that the described outputs were not as described on my instrument!

I sent a post in to G-QRP forum for assistance, but in the meantime, Sam came home from school, so I commandeered him for a few minutes to swap BNC connections around on the back of the analyzer, whilst I looked at what was on them on the 'scope. My connections are marked X, Y, and Z. One of these, Z I think it was, was tried first, resulting in a pulsing DC level. Clearly not that then. X was next up and showed a large triangle wave with a long 0v period, clearly the timebase! So the final Y connection was tried - yes! The 'scope trace showed just what the zero-span display of the analyzer did. Swapping the connection from the 'scope to the audio input of the 2955B, we had demodulated audio out of the test sets loudspeaker.

Using the 2955B isnt really convenient for this though, so I found a general purpose AF amp, based on an LM386, that was kicking about the bench, and connected that up. Perfect. I now need to find a box to put the audio amp in, so I can connect it neatly to the spectrum analyzer Y output for monitoring.

I hope eventually to us it for narrowing down airband transmissions, but im not accurate nor fast enough on the analyzers controls yet to pinpoint a transmission!

Car Stereo fault - Beat 485 repair

Well, marvelous isnt it? The Marconi 2955B goes on the blink, but before I can fix it the oscilloscope requires an overhaul. Get the 'scope dismantled for cleaning, lubricating and repair, and the flipping 2955B decides to start working again, leaving me with no idea whether something ive done, like reseating the cards, has fixed it, or whether the fault is indeed temperature dependent and likely to be a real twat to narrow down,

And while all this is going on, to really piss me off, the volume control on my car stereo fails! It randomly decides whether to go up or down whichever way I turn the knob!

I have a cheap after market DAB stereo, a Beat 485. This does DAB and DAB+ (why I bought it!) plus CD, USB etc. The volume control is not a traditional potentiometer, but a rotary encoder.

No photos for this fix im afraid! But it really doesnt need any, if your capable of doing this repair then you dont need a pictorial guide!

Opening the head fascia unit up is fairly simple - four screws and case clips, easily separated using a finger nail or similar thin object. The encoder turns out to be a mechanical type. This was to be expected as optical units are far more expensive. It also is a type with a central push button action (which I didnt know about its function on the stereo - turns out it selects the tone and balance options).

Once inside the head unit, the next job was to remove the encoder. This isnt particularly easy - there is a sizeable metal casing to it soldered down at either side, plus the five switch contacts (three one side, two the other). Ideally use a good hot iron here with plenty of heat capacity! I just plodded on with the 18W Antex, but the 45W iron would have been more sensible! Once desoldered, there are four folded tags on the underside that have to be lifted to dismantle the device.

Inside, the main encoder track could be seen, with two sliding contacts either side of the shaft. The pattern of the track could barely be seen it was so dirty! A blast with Servisol and a good scrub with a brush had it clean and shiny again. A gentle push with the end of a terminal driver reset the sliding contacts.

Reassembly was just the reverse of the above, ensuring of course that everything stayed in alignment, and then giving the securing tabs a good push down. After cleaning up the terminals, resoldering the encoder to the board was simple, as was rebuilding the head unit. I removed any other muck and dust at the same time.

A quick test back in the car shows that I now have a properly functioning volume control again!

Thursday, 6 July 2017

And now the Marconi 2955B is on the blink again!

Well, this just about takes the biscuit!

Not only do I discover that my understanding of the use of my spectrum analyser is probably woefully inadequate, and that my oscilloscope is playing up and needs a service, but now my 2955B test set is on the fritz! Again!

This time, I know im at serious risk of screwing the firmware if I dont fix it. So, its time to open her up.

The 10MHz Oven Compensated Crystal Controlled Oscillator (OXCO) unit, is mounted, horrifyingly, on the same board that generates the EHT for the CRT!

Rear of the Reference/EHT board
I very much dislike playing near EHT supplies! I know from experience what a belt from the anode cap of a brand new never been connected TV tube feels like - and dont fancy experiencing the same from an actual supply! So I very carefully ensured (with a pair of long nose pliers) that the reservoir capacitors were properly discharged!

Inside the Marconi 2955B. Note the warnings!
The actual OXCO's six pins desoldered very easily, and the whole module came out smoothly. I opened it up to see if there was anything obviously wrong inside

The 10MHz OXCO... bloody complicated inside!
But other than the trimmer capacitor (which I seem to remember being told could be the problem) it all looks ok, albeit extremely complex! So, I connected it up on the bench. 12v at 120mA steady (450mA initial) for the heater, 5v via a 22ohm current limit for the oscillator. 

On the bench...
 And the damn thing seems to be working perfectly!

...and seemingly working perfectly!
 I have had it powered now for about an hour. Ive even abused it - heating it with a hot air gun until it was too hot to touch, whereby it crept up to 10.00015MHz, and freezing it until the heater current maxed out again, and it barely dropped below 9.99998MHz!

Im now at a loss, but my suspicions center around either a power supply fault, or something as insidious  as a dry joint or leakage due to old flux! Where I go from here, im not sure, and ive a ton of work piling up that requires a working test set!

Spectrum Analysis - Am I doing it Wrong?

After a lot of puzzling and pondering over the transverter, ive started to come to a slow realization that much of the problem might actually be with me!

What I mean is, im starting to think that my knowledge of how to correctly read the spectrum analyser, and hence my interpretation of the results of tests on the transverter, may be lacking in understanding.

Sadly, when I try and find tutorials online, they are either far too in depth, such as guides to measuring phase noise, or based on much more modern and advanced LCD FFT instruments, rather than the 'classical' CRT superhet instrument I have.

So I guess what I need to do is get to very thorough grip with my SA instrument and its operation, before I progress with the transverter.

Tuesday, 4 July 2017

FARS Rally - Sunday 23rd July

A quick and blatant plug for this exceptional radio rally!

The Finningley Amateur Radio Society rally is fast approaching!

Located at the club HQ near Sandtoft, this rally is one of the best component and 'junk' sales around! I myself, with Bob M1BBV, will have a stall again this year, where I hope to be able to sell on some interesting items, big on our stall this year will be batteries for the Clansman PRC-349 radios (12v in a handy sized screw in pack), 'covert' diplexers, 'Panorama' professional antenna mounts, and Kenwood TK-349 PMR handhelds. Plus all manner of other junk!

I only hope to make enough money to cover my stall fees and the cost of whatever junk I buy!!!

More Transverter Trouble

Coming back to the 4m transverter after a short while doing other things, I found the bias current still unstable! So, I bit the bullet and added a 10ohm emitter resistor with a 1nF bypass. I can now get either an immediately stable standing current, or, it quickly (a few seconds) reduces down from around 40mA to a suitable 20mA as the diode stabilizes the bias.

Me, My Boys, Beer, Camping, and Radio

After a bit of playing around with Clansman radio in the garden, I moved on to build the PA section of the transverter. This involves the use of my one and only MRF237 RF power transistor... and it didnt go well!

First go at the 1W PA

 In the above photo,  the PA is built, and the transistor heatsink attached. On switching on - the PSU pegged the current limit!!!

Oh heck, not good! Lots of things were tried, resulting ultimately in nothing being connected to the transistor other than the collector to the tank coil and the emitter to ground - and it still pegged the PSU! Yet the transistor tested out ok every time! What on earth was going on?

Checks of the datasheet showed I had it connected properly - or so I thought! I asked on the G-QRP club forum, and they agreed the datasheet. Yet a check of the old Marconi RC-690 RF board the transistor came from, showed otherwise! It turns out that every manufacturer builds these with the expected 'tab to emitter' device layout - all apart from the original designers Motorola! Genuine Motorola parts, which this is, have the tab on the can next to the collector!

So the whole layout of the PA has to be redesigned! Luckily it seems the transistor has survived, possibly because I was very quick in turning the thing off before it got too hot.

BBQ fried eggs - new uses for old equipment panels
 There are a few other issues with the transverter though which im not so happy about. Firstly, I seem to be getting far too much leakage of the local oscillators 42MHz signal through the mixer and into the Tx chain - enough that it is present as an amplified carrier! Likewise, when the 28MHz carrier is fed in for testing, that shows too much leakage. Either there is a mixer problem (doubtful with an SBL-1 DBM) or I need better filtering before the PA stages. Im also worried about correctly terminating the PA during testing, before I have the LPF built. Im not sure of the correct impedance to add a temporary load.

Another issue, noticed yesterday - the Tx buffer transistor gets too warm! This should be a BF199, but as I didnt have any ive been using a 2N3904. Ive now ordered some BF199's to put the correct specified part in place.

The past few nights have seen amazing displays of noctilucent cloud over Europe. The image below was taken at 00:30UTC


Ive also done a little bit towards finally finishing the TDOA direction finder, by making the main boom section of the antenna array. 


Whilst working on this boom, I discovered two of the elements of the UHF section of my tape-measure crossed Yagi for satellite work were off-center. Ive now fixed that, and tested the UHF beam with my MVT-7100 receiver on a number of passed of SO-50. Next step is to add a pair of headphones, couple the UHF beam to the VHF beam, attach the VHF transmitter - and see if I can actually get into the satellite!

Tuesday, 13 June 2017

Clansman Power Indication mod proposals

One of the problems Ive seen that users of ex-military Clansman radio have, is that of inadvertently leaving the set on, and ending up with a flat battery.

Clearly, in its original role, the radio would rarely be unattended. Either an operator would have it on his back and be wearing 'phones, or it would be in a position where the low battery alert (breaking of the squelch three times a second) would be heard.

Not so in amateur use. Besides, in amateur use, its unlikely that any other form of power indication, a light for instance, would cause any hassle, as it might under battle conditions. So, it seems sensible to look at the possibility of providing these radios with a 'Power On' indication.

Electrically, such a thing is about the simplest circuit you can think of! An LED, a series resistor, and a source of power that is constant when the set is switched on. Physically, this could be a challenge! It will involve finding a suitable place on the case of the radio to drill a small hole and mount the LED. It also requires finding where the connection can be made.

Although the PRC-350 is likely to be the easiest of the radios to add a power LED to physically, I dont have the service manuals and EMERs to hand for it! But I do have them for the PRC-351, so have had a look...

There are several supply lines in the PRC-351 - 3v, 6v, 9v, 17v and 100v.  The 3v and 6v lines seem likely to be quite sensitive, and the 17v and 100v too high. So the 9v line would seem to be the supply of choice. I dont want it to take too much current though, but limiting the current to much less than the LED manufacturer specifies will only result in a dimmer light, which is likely to be preferable anyway. Ideally, a low current 3mm LED would be used, to minimise the loading.

I suspect that a usable brightness can be had from a modern 3mm red LED for very little forward current, but I will rig one up on the bench with a 9v supply and a potentiometer and find an acceptable brightness, then measure the pot to find what series resistor is needed.

On the PRC-351, the main source of the 9v line is module 19 pin 3. But, this line can also be found on several other modules - 20 pin 23 (DC switching), 13 pin 10 (Tx AF), 7 pin 9 (Rx AF) and 8 pin 2 (Non linear amp). Which to select would be the one that is nearest to the LEDs position!

I do need to open my -351 up at some point to perform the L/W volume mod described earlier in this blog (to make the Tx modulation level the same in both settings), and to finally add the rubber boot to the 10m conversion switch, so will look at adding a power LED at the same time.

Monday, 12 June 2017

T4 class AB bias - sorted?

As I said I would, I replaced the bias resistor R15 (1k8) for T4 with a potentiometer, which just happened to be on the bench and a value of 4k7. Monitoring the base voltage with one eye, and the current drawn with the other, I adjusted the value of this pot. At some high value, the current actually dropped!, but as the resistance was lowered, I hit the point where the transistor would go into thermal runaway, the current climbing, slowly at first then exponentially faster, up to several hundred mA!

Several attempts later, and I hit on a setting where the meter needle stayed put even after watching for longer than it took to start climbing before. Measuring the pot (and the original resistor, they were in series) gave a value of 2800 ohms.

I replaced the pot and original R15 with a 2k7 unit, but this was too low and the current started to climb. A value higher, 3k3, and the current stays where it is on the meter needle, a total circuit current of around 75mA. I had already measured the circuit up to but not including the driver at 60mA, so the Tx driver is now set taking about 15mA. This might still be too high, but at least now seems stable! Ive asked for confirmation and advice on the G-QRP club forum, that ive done the right thing here.

One ZTX327 left - Taking it steady!

The old Marconi RC-690 PA board had one remaining ZTX327 transistor on it. I have very carefully removed this (not easy on public safety grade equipment designed for shock and vibration immunity), taking care to ensure it doesnt get too hot. On my component tester, the freshly removed device reads a base voltage of 723mV and beta 52.

Testing... Testing...
The removed transistor from the transverter reads 714mV and beta 85. Oddly, this read much higher on the first pass, but then stabilised at 85. I suspect that is down to the tester. But, it seems both transistors are functional. The 1N4148 diode in thermal contact also tested out fine. I can only think that the problem is the bias voltage. I shall replace the 1k8 resistor at R15 with a potentiometer and slowly increase the bias, watching the current drawn as it goes.