[updated:LAST EDITED ON 04-Jul-02 AT 07:36 AM (GMT)]If you buy the kit from Grandata, they come with a photocopied diagram with brief instructions showing which capacitors to replace. They're assuming you've already got a certain amount of repair expertise to tackle a tuner.
First you need to remove the rectangular metal cover from the tuner on the side where the components are. It clips on and can be eased off with a wide blade screwdriver but go gently as you don't want to bend the metal tabs too much. The component side can be identified by the fact that the metal cover has a couple of holes in for tweaking coils during the factory set-up and you can see the components through the holes. The solder side is covered with a cover with no big holes in it (from memory).
As for Rolf's question, it is difficult to unclip the tuner cover without first removing the tuner from the main board as there are some other components mounted right next to the tuner. It may be possible to unsolder these components first but I haven't tried it myself. You will also need to be fairly dexterous to unsolder the tiny capacitors if the tuner is still on the board. I found it much easier to change the capacitors on the tuner once it was off the board and on the bench so that I could use a magnifying glass. Also it is much easier to test the tuner on its own and I wanted to test the tuner before putting it back on the board. (If you're going to try doing the repair with the tuner in place, make a note of which component was where and the correct orientation before removing anything else).
OK, the next step is to identify the 16 pin quadrature downconvertor IC. It is made by Zarlink (which used to be Mitel Semiconductor and before that GEC Plessey Semiconductors). It will have SL1710 marked on it and either Mitel or GPS depending on its age. It is in the same section as the big round metal can marked R2663 (a 478MHz SAW resonator from Epcos, which used to be called Siemens-Mashushita, hence the S+M marking!!).
The first capacitor that you need to change (let's call it 'Cap A' for future reference) is the one connected to pin 13 of the SL1710 and pin 2 of the NEC uPC1688 wideband amp (the four-legged device with C1C marked on it). The existing capacitor is a surface mount size 0603 with a value of about 1pF. The Grandata kit replacement seems to be about 3.9pF (obviously still the same size 0603). Before I had the Grandata kit, I didn't change Cap A, so it is not as critical as Cap B. Now though I change it because it only takes a couple of minutes.
By the way, I used two fine pointed soldering irons to remove the old capacitor, then I clean the pads with some desoldering braid before soldering the new capacitor in place.
The second capacitor (let's call it 'Cap B') that has to be replaced is connected to Cap A and ground (the metal case of the tuner is ground as well if you want to check it). I make a point of saying ground because there is another capacitor which is also connected to Cap A but its other terminal is connected to the resonator and not to ground. Cap B is also a 0603 size and seems to be about 3pF in the original tuner design. The Grandata replacement has a value of about 1.5pF. Before I had the kit, I used 1.8pF and this worked just as well.)
Having done this, the tuner should work again. If you have access to test equipment, you can connect the tuner to a +5V supply and use a spectrum analyser to verify the presence of a clean 479.5MHz oscillation. If the tuner isn't working, I'm afraid you will need to try increasing the values of the other capacitors such as the 22pF connected to Cap A, CapB and the resonator or the 1.8pF between the resonator output and the coil.
For the technically minded, these changes reduce the amount of 480MHz signal that is shunted to ground via Cap B and increases slightly the signal that is fed into the quadrature demodulator via Cap A. Overall this increases the loop gain in the 480MHz oscillator circuit and tends to reduce slightly the signal level going into the resonator. The basic fault in the Amstrad design is that the resonator insertion loss increases with time because the signal level is outside the R2663 specification; eventually the oscillation stops as the insertion loss grows bigger than the loop gain.
Let me know if my instructions aren't clear enough.
NB. The quoted capacitor values have been obtained by measurement using a network analyser. As it is difficult to measure capacitances of a few pF without special test equipment, I don't guarantee that I got it 100% right. But the values I quote worked for me!