SANSUI SV2916 TV Repaired

As I have said in my About page, I did not do any electronic repairs in a long while so it took me a while to fix this TV. It started out with a smaller fault, I tried something which caused the TV to have no picture. Then also it took me much longer than necessary, because of my stubbornness - my mentor suggested I should try something, but I said no, I want to find the fault first.The original problem was that the picture became too wide for the screen, losing some 20% of the picture I'd say. I opened the TV and found the H-Size pot, VR350. It did not make any difference to the picture, neither did VR351 (East-West correction). So I got a schematic. I thought from there on it should be easy.

I do not have an oscilloscope, so I was trying to find the fault by looking at voltages at certain points and trying to see if it agreed with my understanding of how the circuit works. To follow my discussion, get the diagram as SV2916-P1 and SV2916-P2. (Or look right at the bottom)

Obviously I started looking at the circuits around Q350, Q352 and Q353. On Q353 my multimeter showed 0 V on all three terminals. I unsoldered the transistor and tested it, it was OK. In the process of unsoldering it, I broke a PCB track and had to use a short piece of wire to repair it. Now I wasn't sure if everything was OK where I soldered back the transistor . Still 0 V on all three terminals.

I thought OK, if this transistor is 0 V anyway, let me try to isolate this part of the circuit and concentrate on it. I decided the horisontal deflection coil is driven by the LOPT-driver Q402 from the bottom, the top goes via L401 to the collector of Q353 via R365 with a low resistance of 3.9 ohm. I found a jumper between L401 and R365, unsoldered the side going to R365 and earthed it. Swiched on. No picture on the TV.

OK, undid the earthing and soldered back the jumper. Switched on. Still no picture. In addition to having no picture, I now had a new problem: When switching the mains on, the standby LED was on. I could use the RCU or Ch+ button on the front to switch it on, but not off again. So I was in for a completely new round of fault finding.

The so-called "video jungle" chip is a TDA8842, so I started looking at the datasheet of that chip. Somebody told me that the CPU, IC001 (M37221EASP) could be waiting for something after I told it to swich on. I studied it and could not find any inputs to it, except for Hsync an Vsync, which it needs to sychronize the OSD information. I thought it might be possible that the CPU were waiting for these syncs and refused to do anything further until they arrived.

I kept on looking in the region where I grounded the circuit and found something: C410 (0.022 uF, 400V) turned into a short circuit. I was lucky to borrow one quickly from a monitor awaiting repair work. Soldered it in and hoped to get a picture. Still, no picture, turn on with RCU but not off again. The TV also has A/V-out, so I connected this CVBS output to another monitor and found that the set was actually receiving the channel it was last tuned to. The RCU also would not bring any OSD on this monitor.

By then I realized that the CPU reads certain status bits via I2C from the TDA8842, so I had to look for inputs on the TDA8842. The ones that I really considered as possibilities were EHT-PROTECT (50) and BEAMCURR VGAURD (22). If you trace it out on the circuit diagram, you will find they come from the circuit at Q404 and Q403. I saw that the emitter of Q404 goes via resistors to the bottom of the secondary winding of the flyback transformer. With the base of Q404 on ground, you would need a negative voltage for it to ever turn on. I was reading 0 V on R430 (100 ohm, numbered differently on the board), so I suspected something wrong here. I told my mentor I am reading 78V on the screen voltage, but I suspected the bottom diode in the flyback to be blown. He said no, he did not the that diode could be blown and 78V is a bit low for the screen voltage, suggesting I use Tippex to mark the position of the screen adjustment on the flyback transformer. No, said I, I don't want to go adjusting, I must have blown something else than C410 and wanted to find the fault first.

Fortunately I was able to borrow an oscilloscope at this point and also my mentor told me to make sure, since I could feel static on the screen so we expected EHT to be present, that I get RGB through right to the CRT board. I told him that I already saw 3 V on each of the three going into the CRT board, I could not see any variation. The oscilloscope confirmed: no RGB coming out of the TDA8842. Just to be sure, I also checked for HSync and VSync on the CPU - they were present.

Now I wanted to follow up my theory that a diode was blown in the flyback transformer. I read that people use two multimeters in series to test it, but I only have one. I read that somebody else said, use a DC supply and put the multimeter on volts. I tried this and was able to read 0.9 V when the supply was in the direction to forward bias the diodes in the flyback transformer. But I wasn't very sure, since I was testing it in circuit, I was not sure if the current was maybe going via the resistors in the flyback.

To take this measurement, I of course had to remove the EHT lead from the tube, for which I very safely used resistors tied to a screwdriver to discharge it first. But the excercise almost led to catastrophe! I used long nose pliers to put it back again, but when I switched on, it was not properly hooked up, fell out and started arcing on the aquadag! I cannot stress it enough: Be very carefull, if you need to remove the EHT lead, make sure you know how to safely discharge it, when putting it back, make very sure it has clipped in properly. At least, now I was VERY sure I had EHT!

I now decided to inject a negative voltage on the emitter of Q404. I used the 24V DC supply again, tying it's + to the ground, a 10k pot across it and it's wiper to the junction of R403 and R404, first having made sure of course that the pot is turned on the + side, so it would not pull towards negative. I then started turning it, checking the voltage going negative and also on pin 50, the EHT protect of the TDA8842, which I think initially was over 3 V, pulling it to 1.6 volts or so. But to no avail. No RGB coming out. I started looking for any more inputs on the TDA8842 and found pin 18: BLACK CURRENT IN. Now it dawned upon me. The TDA8842 wants to see a little bit of current passing through the tube before it will switch on the RGB. I measure 0V coming out of the CRT board to this point.

So now I though - maybe the guy is right. I have EHT over the tube. The RGB amps should be biased normally, so if there is no current, why? Could be the screen voltage. So I found Tippex, marked it and started turning clock-wise while watching in the mirror. At first I started seeing a blueish glow on the screen - there is life! Then I started seeing flyback lines with something of a moving picture. I realised I turned it to quickly - it is a fine adjustment. I was then able to fine tune it, so the picture brightness was normall. Now the picture was actually too narrow. I tried the H-SIZE and EW pots. They worked now!

I measured the screen voltage again and found 79V, just 1 V higher! With the picture back now, I noticed that the moment I touch the meter on the grid, the picture dies. Only now did I notice that the moment I put the meter on, there is momentarily a higher reading, I thing I saw 150 V and 125 V, it is a bit difficult to see using a digital meter. Finally my Tippex marking showed a difference of maybe less than a millimeter!

My final conclusion was that C410 was busy developing a problem, causing the picture to go too wide, and it blew completely when I grounded part of the circuit.

I used the H-SISE to adjust the picture just to almost reach the sides. I then used the EW pot to make the sides exactly straight vertical lines. Then the HSIZE again to hide the slight unevenness you get there.

My idea was to take several oscillograms for future reference, but it was late in the evening and after more than 20 days I just wanted the bulky TV off my workbench. So I only took 3, here they are:

All the oscillograms were done with a 10x probe and 0V DC was vertically in the centre.

R420

Any of the R, G or B out of the TDA8842, on R202, R203 or R204.

D201 Cathode.

I used the opportunity to fix something else that I have been putting off for a long while. The set has an AV2 input on the side which was broken. The S-video connector was completely loose, so I resoldered it. I also noticed that it had a hole were it was intended to be mounted with a screw, which the manufacturer chose not to do. It was a bit difficult to mark the hole on the case side, but I did it, drilled a hole and fixed it from outside with a small self-tapping screw.

In this picture you can see the S-video connector has a hole where it was supposed to have been mounted with a screw.

While working on this TV, I noticed something that made me not very happy with Sansui. I picked up several short bits of wire, ends of component leads that have been cut off during manufacture. I took several pictures of the board, finding it easier to look for certain components sitting at the computer than standing over the board. I think I removed about 10 in total. I only noticed later there are about four in this picture! Look around the bottom. I don't know how many of them have remained inside the TV.