This section is in "The Queens Language", since it might be relevant to people who do not master the Swedish language, so bear with me. (Tja, jag hade inte uppmärksammat den fina knappen tillhöger när jag skrev detta.)
I bought a TH9800 quad band transceiver from a web-shop. It arrived after the mandatory customs delay, but once it showed up, it was immediately put on the air. It seemed a nice radio, however after building a two band antenna for 10m / 6m I discovered that "someone was fooling around with me" ( I thought).
This showed up as random silent carriers interrupting my traffic on 10 meters, usually a S9 silent carrier blocking out the response to my transmissions. Changing QRG it would sometimes follow but now always.
When I was listening in on a DX QSO, I found that moving my hand on top of the radio caused the signal to change from barely readable to clear. So, no-one was messing up my QSO's I had discovered an instability.
I caught the signal on a 29.6XX frequency. It was S9, no modulation. Tuning either VFO would not change it, except on 29.6XX (naturally), so it was not dependent on settings.
Now it was beginning to feel exciting, so as a true radio amateur / engineer I got a golden excuse to dismantle the rig and have a look at its guts. Before pulling out the tool kit, I decided to measure sensitivity on the radio. All four bands showed excellent sensitivity, below 0.2 uV for 20 dB quieting, so everything was sort of working.
Next step was to remove the lid, and i sniffed the inside with a small capacitive probe connected to an Advantest R4131 spectrum analyzer. Sure, no weird signals on 432, 144 or 50 Mhz, but a distinct signal on 27.560 or so. Tuning the radio (no antenna ) to the indicated frequency, a S9 carrier was heard.
Now I connected the spectrum analyzer to the antenna jack of the receiver. A -40 dBm signal on 10 m. So, I am actually transmitting while receiving. The conclusion of this experiment clearly indicated that it was actually the first stage, the RF preamp that was causing the problem.
Now, this is a radio with modern manufacture, so all components are so small that they are anonymous, i.e. no visible text on the tiny tiny chips. Being curious though, the board was removed from the chassis, not hard, but 22 phillips screws to remove + the soldering to the SO239 (on 432??) antenna connector.
Of course, the 29 Mhz preamp is on the BOTTOM of the PCB, so there is no access to it unless the board is removed. Furthermore, after doing anything to the radio it has to be all assembled again (I think) before testing.
Even though I had no schematic or service manual, a small marking on the top, "30" beside a gold pad gave me a clue of what was the 10 m preamp. The RF transistor was marked R25, and I decided it was probably an 2SC3356S, a 7.5 Ghz "cellphone" transistor, with an extremely high gain at 30 Mhz, my guess 33-35 dB, the datasheet stopped at 100 Mhz on the low end.
So, how do you tame a wild microwave BJT? Well, negative feedback might be the answer. So, a small network similar to the ones used by YU1AW on his BJT preamps might be one solution. Since the transistor was CE, the only idea I came up with was a resistor from base to collector, DC blocked with a 1nF capacitor in series. Experimenting with the value, I seemed to achieve stability even with 2.2 kohm, which would result in a gain of 2200/50 IF the s11 of the transistor was around 50 ohms, for a maximum gain of 44 or 33 dB.
Assemble for the fifth time, connect spectrum analyzer, all clean, reconnect signal generator, well sensitivity lost a little after the MOD, now 0.3 to 0.4 uV for 20 dB quieting, as opposed to 0.2 uV on the other bands, (50/144/432).
A final air test, and now I could clearly recognize the modulation of the *voice* of my QSO partner as opposed to understanding what he said. So, radio is probably fixed, however, I am waiting to get a chance to discuss this with an engineer of the manufacturer. Also, a single day with a few QSO is not enough, but in the coming weeks I will decide if the rig is actually permanently cured.
So, why did this happen in the first place? Well, with a good signal generator and a precision 50 ohm 10 db pad in line, the radio sees *exactly* 50 ohms, and the preamp does its job. Thus sensitivity is according to specs, so radio is "good", and possibly passed final testing without fault.
However, connected to a typical home-brew antenna perhaps the impedance is 47 +j11 or somesuch, and also different att different frequencies, remember the antenna is typically only a "perfect" match at ONE frequency. Since my mobile antenna is a "quadband" rod, it probably has a number of interesting complex impedance points.
Perhaps my 2SC3356S is one of the "best" selection as far as gain goes, perhaps producing up to 40 dB of gain at 29 Mhz. Perhaps there is some *other* problems in the loading of the stage, or in the layout of the components, hard to say. However, just *slightly* reducing gain of the wild BJT with feedback takes closed loop gain down to a level where the amplifier does not become an oscillator, and the box behaves as intended.
So, what about the title of this rant? Well, if you do import things from the other side of the planet you have to decide if you want to send it back, and incur the delays of returning, repair, receiving, customs and snailmail. In my case I just decided, if they could make it surely I can fix it.....
More on this if I come to other conclusions.....
73 de SM6FBD (now on 10, 6, 2 and 70 cm) mobile.
Since I feel the (slight) loss of sensitivity a little too much, I will attempt to find a better component value that will kill the self oscillation with less sensitivity drop..
Also, another TH9800 owned by a friend exhibits the exactly same problem, so it looks more like a design problem than a manufacturing problem...