Monday, May 31, 2010


One my search for detailed information about the Sangean ATS-909 (aka Radio-Shack DX-398, Siemens RK777 and Roberts R861), I was able to locate schematic diagrams.

The reader may ask what that's got to do with a MEPT... well, nothing up to now... BUT:

 When checking the receiver's control PCB schematics, I found something, the main micro-processor generates the PLL reference signal. Maybe it is always done like that, I don't know, but it looks remarkably simple to me.

In the upper right corner, that would be the micro-processor. The chips leads X1 and X2 are connected to a 4.332MHz crystal. This oscillator is the classical Pierce configuration, as we know it from so many digital applications.
The crystal's cold end is provided with a trimmer (TC301) for frequency tuning.
The crystal's hot end is connected to an RC low-pass filter (R389,C363) which further feeds the signal into the PLL chip.
BTW: Can someone explain the function of C359 to me? It seems to couple GND to GND.... hmmmmm(!)

So, here's the MEPT idea (actually, this was done before, sort of, by Clayton from Down Under): clock the keyer (e.g. 12F629) by its high speed built-in oscillator, using a crystal for a QRSS frequency and couple the signal out in the way the professionals do :-))
There is just one down-side to this, the controller's clock is now at a somewhat odd frequency, odd for controllers that is, so timing could be off. I will check if compilers (I am programming in C, not in assembler) are able to compensate for that...

Now that we got the signal, what shall we do with it? Well... amplify?!
For another project, I happen to study switch-mode amplifier, aka class E or F. F seems a little bit too much work, but E seems fine for a MEPT, as Hans proved with his Caribbean beacon. Hence, a class E amplifier seems appropriate; I guess, the crystal's hot end will have enough voltage to switch a FET.

The chip could serve for than just the oscillator and the keying device.
During my time at various physics experiments I experienced the difference between fast control and slow control. In our experiment slow control was taking care of temperatures...
So, with seeing the keying as "fast" control (well, sort of fast), controlling the crystals temperature could be integrated as slow control. Those PICs got ADCs, which could easily be used for measuring temperature.

My present vision for mini-MEPTs is presently:
  • one PIC clocked by a QRSS-xtal
  • one FET as PA
  • oven control by the PIC
  • telemetry?

The whole story could have some other advantage.... I realized that, depending on the TX frequency, some PICs stall when RF leaks in. If the controller is clocked with the same frequency and phase, this should not occur.

Next step: design something! Stay tuned, I will let you now as soon as I got more. For now you may click on one of the "sponsors" to find your way out of this blabla...

Sunday, May 30, 2010

ATS 909 Idea

What about using the ATS 909's frequency generation for transmitting?
As soon as time allows, I will have a look into the service manual to find out more.
Meanwhile, here's the source for the manual:

Full Duplex QRSS TRX Idea

This has been in my head for some time... What about in-band full duplex QRSSing?
Some preconditions we have to look at. For a successful slow QSO, we need to assume condx of a single band, so that both stations can receive the counterpart equally. TX during RX in one band at the same time is a challenge even on the 2m-band. So, what can be done?

The problems, I figure, are more on the receiving side of the design. So here are some preliminary ideas that came to my mind:
  • frequencies in one band, the furthest possible apart
  • a very narrow-band RX antenna, e.g. a high Q magnetic loop
  • a crystal notch filter for the TX frequency
  • a crystal front end filter
I have no idea if this would actually work...
However, for experimentation, there are a couple of possible frequencies (cheap crystals available).
  • 80m: 3.500MHz & 3.6864MHz
  • 40m: 7.000MHz & 7.15909MHz
  • 20m: 14.000MHz & 14.31818MHz
  • 10m: 28.188MHz & 28.636MHz
The high crystals for 40m and 10m may be a little difficult to find, check DigiKey. All others are either standard, or available at GenesisRadio. Remember, we need a crystal for every QRG, at least for the notch filter!

For me, the obvious choice for making a full duplex QRSS transceiver would be K1SWL's 80m-Warbler. I could show that the Warbler is quite potent on 40m when modified. I would assume that the kit could equally modified for the 20m band.
Based on the Warblers, one would need two per band. The Warblers make nice transmitters too...

Another approach could be the use of a superhet receiver with a (switchable) notch filter in the front-end, remember, it is all about keeping the TX out of the RX when going full duplex.

Monday, May 17, 2010

IC-M700D LSB update

Wow, how cool is that. This morning, the postman dropped me a padded letter through my door. The letter contained a 4922kHz crystal! Unfortunately, no sender was noted. Must be a reader of this blog though ;-)

Dus, grote dank aan de OM achter de brief! Idd, de kristal is iets groter dan het origineel. In de doos van de IC-M700 zelf is gelukkig zooooo veel ruimte, dat dit helemaal niets uit maakt.
Ik zou me ook graag personelijk willen bedanken, maar, ik weet hellaas niet bij wie...

I hope, that this crystal can be pulled downwards enough, 2kHz should not be a big deal, so that grinding would not be required. I will, additionally, invest some time in finding out how to grind up modern crystals, so that more owners of non-LSB IC-M700 transceivers will be able to modify their's.

Sunday, May 9, 2010

Penning Down a Crystal

My previous post dealt with the fact that I needed a crystal which is non standard, but close enough to a mass produced crystal frequency, very close actually.

Years ago, I came across an old technique in which material is applied to a crystal in order to lower its' resonant frequency. When I remember correctly, that was done by attaching small patches of Scotch tape.
In more recent tests, done by a couple of radio experimenters, permanent ink was used to paint dots on the crystal to achieve the drop in resonance.

When I severely experimented with surplus crystals earlier in my life, the crystals' housings were soldered and could be opened by a medium powered iron. Today, housings seem to be welded and much harder to open. Well, it is not that hard after all.

This is what I do:
Using a high speed electric cutting tool, a groove is formed along the lower end of the housing. I do not recommend to use the tool to cut through the material however. The tool leaves a nicely roughened surface, at which a metal saw finds good grip. It is tempting, but do not hold the xtal at its' main housing, hold it at its' leads. Now carefully and slowly cut through the sheet metal by means of a manual saw.
This is how it looks just before the housing comes off:
Moments later:
It is clear that this procedure does not create an all in all precise cut, but, that is not too bad. Having some edge features helps the later reassembly. In order to place the same rough edge in the same rough corner, I mark base and housing.

Now that the crystal is open, some oscillation device could be of use. Not really spectacular, this is what I came up with:
The photograph shows an additional 78L05...

What you see is the usual Pierce type gate oscillator. This is a relatively stable oscillator. However, it oscillates above the XTAL's series resonance. Sure, one could compensate with an inductor, but, for penning, grinding or matching crystals, relative measurement are usually acceptable.
Example: a crystal having a nominal series resonance frequency of 4.9152MHz should be penned down to 4.908MHz is a 7.2kHz drop. Assume the oscillator generates a 4.9170MHz signal, a 7.2kHz drop would read 4.9098MHz.
An additional consideration was, that many of my projects involve Pierce oscillators of exact that kind, hence, it would be reasonable to pull the crystal to the exact frequency it will operate later in the application.
I spare you photographs of a crystal wearing black paint...

And no, I did not start directly with my 4.908MHz project, I did some color burst tests first.

With the drop being not too far, i.e. not that much paint applied, one can actually observe the changes in frequency in the process of applying the paint. With a lot more paint applied, oscillation stops. When that happens, a funny effect could be observed: the oscillation starts again at higher harmonics and slowly drops to the fundamental frequency.
Example: the color burst crystal (3.579545MHz) with some paint applied stopped oscillating. When the counter came back to life it first showed (about) 28.48MHz, then 17.8, 14.24, 10.68, 7.12 and finally 3.56. This corresponds to the 7th, 4th, 3rd, 2nd and 1st overtones. The more paint needed to cure, the longer the oscillator remained at the individual frequencies...

OK, that was more like a fun side remark. The interesting bit is, I was able to obtain stable oscillation of a color burst crystal at 3.530MHz, which is a remarkable drop of 50kHz.

Finally, even a painted crystal deserves a proper housing:
The penned down color burst crystal now ready for A1A action on the 80m band.

Tuesday, May 4, 2010

LSB for the IC-M700D

As I wrote before, the Icom IC-M700D, the German version of the famous IC-M700 marine radio, did not come with LSB. What I was reading on the internet lately, there is a French version, the IC-M700F, which also lacks LSB. There is hope, maybe something could be done...

Luckily, my IC-M700D came with a circuit diagram. The schematics of the IC-M700 can be found on the internet.
Studying both circuits, the following can be said (corrections made, due to an error in the Service Manual):
  1. the MAIN UNIT pcb of the IC-M700D is the same as the one of the IC-M700
  2. both sidebands are using the same oscillator
  3. two crystals are switched by means of 1SS53 switching diodes (a 1N4148 will probably do)
  4. the LSB crystal's traces are present on the D's pcb, so is the LSB control lead from the MATRIX (ending at Q334's base)
  5. the D's LSB crystal and surrounding bits'n pieces (a switching transistor(2SC3402), a varactor (FC51M), an inductor and some resistors) are not populated
  6. for LSB a 4.908MHz 4.920MHz crystal is required
  7. the mode switch needs to be wired up correctly, there may be a further transistor (Q1210) missing on the MATRIX pcb, further, there could be a jumper somewhere...
Have a look at the IC-M700D PCB

and the corresponding IC-M700 circuit diagram

 The question now would be, if there is any cheap option for the crystal. Well, I figure, we are lucky on that one... there is a cheap crystal for 4.9152MHz. I intend to open one up and pen it down to 4.908MHz grind it up to 4.920MHz. A drop of 7kHz  raise of 4.9kHz should easily be doable, although, this is a little harder than it sounds.

The plan now is to pen down grind up a crystal first and think of the rest of the modification later.

Remark: Never trust a Service Manual!
Reasons: The filter center frequency is 9.0113MHz. The (USB) beat frequency is a mix of 10.24MHz and 4.908MHz divided by four, i.e. 10.24-1.227=9.013.
To get to the other side of the filter, the resulting beat frequency should be higher, i.e. 10.24-(4.92/4)=10.24-1.23=9.010MHz.
I should have seen that before going into penning down crystals, which was a good exercise however.
Now, I need to do both, grinding up, and pen down, when ground away too much crystal material....

Saturday, May 1, 2010

30m - 10m Diplexer for QRSS use

Just an idea....

Diplexers are very popular for running VHF and UHF transceivers into the same (simple) antenna. This works since 435 is three times 145. A 1/4-wave radiator for the 2m-band is a 3/4-wave radiator for the 70cm-band.

Could this trick work on shortwave too? The 10m band is about a third of the 30m band ... well... sort of. I figure, I suitable diplexer would separate frequencies below and above about 20MHz, maybe 19MHz for good measures.
The hope is that a vertical of about 7m length would form a 1/4-wave radiator for the 30m-band and a 3/4-wave radiator for the 10m-band.
The numbers tell the following story:

wavelength for 30m: 300/10.14=29.6
wavelength for 10m: 300/28.32=10.6

30m radiator length (assuming a velocity factor of 0.95): 29.6*0.95/4=7.03
10m radiator length (assuming a velocity factor of 0.90): 10.6*0.90*0.75=7.15

A compromise could therefore be a radiator length of 7.1m to suit the 30m grabber and the 10m MEPT at the same time.

Next step, think of a 19MHz diplexer design.