20 October 2017

LT Spice and GW3UEP PA

Anyone who intends to transmit on 630m can't avoid the GW3UEP PA design. It's a simple design which can't fail to operate. Dutch authorities limited the power of our 630m transmitter. It should not exceed the 100W, you can overshoot this with 6dB, but in order to to transmit with 5W eirp as most countries have, you should have more power available.
The 100W is just the design for us, the Dutch.

But on the other hand this restricting is more or less also for for own safety. The consequences of a short antenna is that the moment the tuned signal leaves your house it has a high voltage. Fluorescent lamps will burn spontaneously. The use of high quality insulators is a must, the feeder must be as free as possible from any environment. The best is to invest in a more efficient antenna/ground system.
Some solutions I can't publish because it wont work in other situations.

Being a pensionado, I have now more time to look at the radiohobby. One of the topics I discovered, was LT Spice. It's a simulation program for electronics.
You can download it from here Linear Technology.
In principle the company made it for themselves, so the library is limited for the stuff they produce. It's possible to load 3rd party data. There are several fora and places where you can retrive missing libraries, but the disadvantage is,  when you send an ASC file to someone who don't have the library, it will not work.
If you install it and start with these 3 short instruction videos you know enough to get started.
I'm currently in process to make a 630m transverter. In this design a small "broadband" amplifier was needed. In stead of doing it on the breadboard, I used LT Spice. In reality the design worked instantly as simulated in LT Spice.
I also played with (simple) filters and and amplifiers ect. Explored the program what happened when you change components. The program is rather popular so there is always someone who wrote solution for your question. This playground moment is very essential, to familar yourself with the program. At a certain moment I decided to try to put the GW3UEP PA in LT-Spice. Not all components are available in the  LT Spice library, but there some alternatives which will work.The totempole transistors and PA MOSFET are replaced by "alternatives".

In the simulator you can easily change the voltage AC/frequency or DC, but the GW3UEP PA is driven by a squarewave, so some study should be done to arrange a squarewave in the "Voltage".
You have to construct this as a pulse. Pulse is build in "seconds" not  in Hertz, so first you have to convert  475kHz in "seconds". The result is a ridiculous long figure. Ask your friend Google/DuckDuckGo for a "Frequency to second converter". I came to 2,1186440677966 µsec. You have to divide it by 2, because you a squarewave has a "low" and a "high" state. It looks complex but it is very versatile. You are able to construct almost any waveform. You can even use a WAV file to feed an audio amp design.

Look at this instruction video which makes designing a voltage pulse clear.

The "Voltage Component" is filled in as follows:
1 Choose Pulse, a parameterbox will appear.
2 Vinital start with 0
3 Von  max voltage iI used 10 (V)
Trise and Tfall can be used to compose triangle wave form. I kept them on 0
4 Tdelay(s),  here you enter the time the "state" will be "delayed" at the start, "OFF" state, in our case 1µ0593320338983
5 Ton(s) here you enter the time there will be "ON" state, 1µ0593320338983
6 Tperiod(s) here you enter the total time the sequence will take, after completion it wil repeat itself.
The LT Spice ASC file for the GW3UEP  100W PA you can download here.
If you have LT Spice installed and "click" on it, it will be automatically opened in LT Spice.

If you "run" it, and measure the output, you will notice that the output is far below the promissed 100W. It's one of the issues I encountered at first when I build the GW3UEP design. I solved that with inserting a Ferrite bar (from a broadcast radio) in the filter coil (L2 16µH).
Now try to change the value in the simulator into 35µ!. And presto 100W!

 Any idea how much current is flowing thru C4? Now you know why you have to use those "expensive" Wima's!
What is the influence of L3 12µ?
What happens when you lower the input?
All these situations you can try to explore without blowing up the PA.
Rise the PA voltage to 48V, see whats happen with the output power!
Another MOSFET? Try the IRFP4668 (It's in the standard LT Spice lib) Stunning? Now you have 150W. Change the PA voltage to 48V,  Oeps... 570W. Measure the voltage ontop C5.... Almost 600V! C4 has now a current of 35A.
Yep theoretically offcourse. 😉

12 August 2016

Kenwood TS-440 keyboard repair.

I'm using the Kenwood TS-440 as my MF station. The reception is excellent on MF and what's more the noise blanker seemed to behave superb on the MF frequency. But the TS-440 has it shortcomings in a mechanical way. The front keypad was not designed for over 30 years intensive use. The copper clickers inside will wear out. In my case some buttons didn't react anymore. But because using the memory facility it was not imminent.
But on certain moment it becomes annoying. asking for a solution. In the past I did already several clean actions based internet information like this Kenwood- ts-440s-keypad-bounce-fix
BTW: You can use this link in a How-to disassemble the keypad..

Well being in 2016 there are now cheap switches available which fit into the place of the old copper clickers. On AliExpress I found these Aliexpress-Push-Button-Tactile-Switch-Momentary-Tact-4x4x1-7mm-4pin-SMD
View pricing for 100 pcs (for 4 sets TS-440!) I ordered a pack which arrived within 10 working days.

In case you reading this and the Ali link doesn't work anymore a picture and technical specs.:

Suggested searchkeywords: Push Button Momentary Tact switch 4x4x1.5mm 4pin SMD

4x4mm is the footprint of the switch (almost the same as the TS440) and 1,5mm is the height. Which is to high , see later.

If you don't know how to disassemble the keyboard zie link above! (keybounce fix)

The main effort is to fit the switches on the board. You have to remove some PCB lines and arrange some new "lines". For "removing" I suggest to use the same method Kenwood did, just drill a small hole. Advantage is that you can make connections thru the back with thin wire. (I used wrap wire diam 0,25mm)
You understand already that for each switch you have to look for an solution. If you have experience with electronic DIY projects, you will soon see the algorithm in the circuit.
It's a bit difficult to explain in words here, but believe me, you soon see the "system".

If you look at the photo of the switch, the swithing takes place between the horizontal connections. In some cases you can use this feature to "bridge" connections.

There are 25 switches to do.
Some are easy, some difficult (at least 2 on the left keyboard "6"& "0")
Start with the right keyboard (RIT XIT A/B SPLIT ect) these are rather easy. Best is to commence with the "1MHZ" button.
You can test directly during the assembly with the TS-440 by plugging in the connectors.
Do observe the  "F.LOCK" switch. If engaged some buttons are disabled.
Don't test it everytime, because it will wore out the connector in due time.
A meter with a "buzzer" is also OK for making the check.

You can solder with a "normal small" solder iron. Start to tin one corner. Make sure that the corner is easy accessable for you. Depends if you are left or righthanded. Place the switch on its wanted location, melt the tinned corner and move the switch in such that that is precisely in the middle. The switches I used, the feet were just to the edges, so adjusting was not so difficult. When in place, tin the other 3 corners.

As pointed out above, the switches are too high, but if you remove the pin on the plastic push key it will be just right. I used a "Dremel" tool to remove them. Do check this by the first button. Maybe you have a version which is different.
The whole operation took me 6 hours, inclusive the learning curve and coffee break.


Some more pimping... 

Replace the scalelamps by "clear led". I took 2 in series with a resistor.

Unstable frequency?
The TS-440 8Mhz crystal is just in the airflow of the fan of the PA.
This will result in a dramaticly change in temperature in the unit. WSPR transmission on higher HF bands will show a banana style trace when transmitting and after transmitting the received stations are likely to be off frequency a few Hz. Just cover the crystal with polysterene plastic. This will stabilize the short term temperature change in the unit.

25 May 2013

MF station setup at PA3ABK

Just a small overview of the MF antenna and MF transmitter.

The antenna picture gives you an idea. Click for enlargement. In total 54m of wire is used. The topwires are not connected together but it's one wire which goes back and forth. Read this item to learn the background.  Both roofpoles have a length of 5m, and distance is 7,5m. The backyardpole is 10m.
Besides the "waterworks" earth, there is also an  "earthprobe" in the garden, 3m pipe driven into the soil. For reception I use the garden counterpoise which has less "noise". A reed switch arrangement takes care of switching the correct "earth"

The whole trick isn't the antenna itself but the earth system.
Having a garden with only 12m space doesn't leave much left for a decent radial system. My house is part of a terrace arrangement. Every house has his own "earthprobe"All these earthprobes are connected together, Over the whole range of 100m  I have a firm counterpoise. Needless to say that other cables, pipes of all kind of services are connected to it. In the Netherlands all cables pipes in teh city are buried.
The toppart  of the antenna poles holding the arrangement are of glass fiber. This is important to avoid that radiation is not absorbed.
The system is tuned by a "large" variometer ex merchant transmitter S-1250, which is capable to handle over 600W. The impedance matching is done by a core 3C65 material. Primary 15 windings litze secondary is a flexible wire of  5. It's easy to increase or decrease the windings by simply  adding more or less turns thru the core. The system seems to have a total impedance of 7 Ohms, which is rather low for this given situation. Most amateurs in this environment have around 50 Ohms. The variometer is placed as close a possible to the window. A feedthru arrangement of a few centimeters is done to keep leakage a as low as possible. The wiring is leaving the house almost horizontal. This maintained by a horizontal glass fibre holder. Connection to "earth" is done  by "disposed, old" fat coax cable in order to keep the skin-effect as low as possible.
Another advantage of my location is the lack of high trees which give me a reasonable  free take-off.

The system has a basic resonance of 1430kHz. I also use it for 160, 80, 60 and 40m. Other bands are covered with a  ladderline feeded 2x 7,5m Inverted V. Which is also visible on the picture above. Note that this antenna must be "isolated" when using the 630m otherwise it will suck a lot of energy to the ground.
Below is sketch of the whole technical setup. It is not updated to the present situation.

WSPR is produced with a MP3 player and a timer. The shift has been multiplied by 4 and was created with Audicity. This to survive de DIV/4 sequence of the GW3UEP PA. Details can be found here.
To avoid "hiss" during silent periods, there is a squelch to make sure that the drive has reached a certain level. Without it, the PA will behave erratic, transmitting broadband noise.

23 May 2013

Back to basic... The Marconi antenna

One of the nice side effects of working on 630m, is the aim to make use of your transmitter power in a most efficient way. Especially when the antenna footprint is small. The main principle on this band is to get as much copper as high as possible. In NL the antenna height is restricted. It's 5m above the rooftop. If you want to go higher a permit of the local authorities should be obtained. Well I opted not to apply for "higher" for several reasons.

My first steps on MF were made a G5RV antenna, a single wire spanning over the house. The feeder was joined together and tuned as T-antenna against the waterworks. This wil work, but not for everyone. I use the houseprobe as "earth" and lucky enough this seemed to be a very good one for MF.
A long list a several antenna variations followed.
It's clear that when you fatten the top the efficiency will increase.
Now I came back to basic... The Marconi T-antenna, however a variation of it.
I was inspired by the Dutch offshore radio station Radio Veronica, running 10kW on 557kHz. In the 70's I happen to be on board a vessel crossing the Pacific.
Satcom was not common in those days and traffic was cleared by HF. Best time was around 1600-2000 UTC from that location to work EU. Time difference forced me to do this very early in the morning, just before sunrise.
After finishing traffic a had a chat with the 1st mate, having a Grundig YachtBoy in the bridgeporthole producing quite good music.  The stationcall revealed that it was Radio Veronica... I went back to the radioroom and tuned the main aerial to the Veronica's frequency this was done with tuner of the aux MF (500kHz) tx. The 1st mate was very pleased the portable could now used also in his cabin. This event repeated itself for some days, until we sailed out of this propagation "tunnel".

The antenna Veronica used, was relative very small. It's design came from a Belgium antenna expert. A meandered T-antenna.
Have look at the website of Radio Veronica. It's in Dutch, apart from a lot of nostalgia, it contains technical information. Seawater was certainly in their advantage, nevertheless a remarkable job covering the Netherlands with only 10kW.

I  tried to arranged something similar, only not with 4 but with 3 wires and the length, only 7,5m.
I can extend this to my backyard, but constructionwise I have my limitations.

The transmitter and tuner is on the 3rd floor. A "fat" earthcable is run from the tuner to the houseprobe. From the tuner the antenna wire is run 12m into the backyard picked up by a pole at about 8m. Than 2m straight upwards.
This 2m spacing is necessary to avoid de-tuning during "windy" periods.
The wire returns to the house and picked up by a pole about 5m above the roof. From there 7,5m straight to the front of the house. Also picked up by a 5m pole. On both poles there is horizontal alu pipe of 60cm, just 20cm below the upper wire. The wire is going back and forth between those poles. If you add up the total wirelength is 53,5m.

21 September 2012

Poor men's WSPR generator with 4x shift..

Using a GW3UEP class D amplifier driven by 4x the frequency has it's restrictions. Using WSPR requires a different approach.
There is a simple solution for it. You must increase the shift 4 times. This can be done in various ways. The PIC solution with a separate LF drive or 2MHz oscillator. A LF multiplier VCO. The latter I tried but the settling time was somewhat unstable, so I left that idea for what it was.

The message which is transmitted is in most cases always the same, so why not just the SSB modulator with a WAV or MP3?

But how do you create such a file with the 4x shift?
Well as follows:
Install Audicity (http://audacity.sourceforge.net/)
- Within this program there is a recording facility in the top row. Start WSPR, choose MF, dial 501,500 and in the TXbox 500,500 should be entered. So a 500Hz tone will be generated when you use "Tune".
Activate "stereo mix"for the soundcard used in WSPR and select this soundcard for the recording in Audacity.
Don't overdrive the recording, just the half which is necessary is best to become a good result.
After the recording you must tidy up the pre and after silence so you will have a complete sequence of 1min50sec.
Now the trick...
- Choose in Audacity "effects" "Change Pitch". The pop-up is showing all kinds of possibilties but just look at the box "Percentage Change". Enter 200%. Click OK. Audicity will now create a sound sample with 1,5kHz and 4X the shift.
- For a test you can play the file from Audicity by selecting the sound card which drives your trx. Manual start the sequence on the even minute.

In order to finish the goods, export the file in WAV or MP3.
Now you can use that file in a timer, Windows scheduler or within SpecLab.
Both I was not very successfull but made a workaround :-)

I added 8 mins of silence to the file. Audacity has this utility "generate" and one of the options is "silence".
Make sure the total file is now 8mins or a multiple of 2mins.
Audicity can arrange this for you within the msec!

I exported this file and played in a mp3 player with the file repeating itself.
When you start manually the sequence it will be kept in sync for several hours within the few seconds WSPR can accept.
For this purpose I used STP player a no nonsense low overhead player. Just Google it most certainly it will be the first hit.

Update: Mainwhile I found a "simple" systemscheduler program from Splinterware, were you can specify on the minute when the file should be played. STP can also work on the commandline mode. So the "out-of-sync" problem is over. Times can be scheduled at intervals you wish. I will either put some pics or a PDF with "how to" later.  Tip: the STP player likes a playlist (m3u) to get started. The systemscheduler itself has a utility called "PlayWAV.exe" so the STP player is not directly needed when you are satisfied with adressing the Windows selected soundcard.