Repairs are not always easy

 Being at somewhat of a loose end, and having finally stopped my regular lift repairs back in April this year (2024) I decided to try and fix a cinema surround sound amplifier I'd bought a few years ago. I'd actually got the amplifier and its remote working satisfactorily but the HDMI output was bad. I'd searched fruitlessly for a repair manual so had decided to just put it on one side for the moment (3 years ago).

 

Read more about this Teac amplifier

As I knew what the problem was I needed a way of checking HDMI pictures so went off in search of a suitable monitor. The best option was a TV I'd bought many years ago but where was its power supply? The TV was helpfully marked with its requirement of 12 Volts 5 Amps.. but where was the original PSU? A 30 minute search revealed it in a box with others, mainly 19 volt.

 

 

  I plugged it in and found it was dead. It actually has a green LED that should be lit when working properly but this was not illuminated so I opened the case to figure out why. Only two ordinary philips screws held it together but the circuit board was encased in a wrap-round copper clad cover. I unsoldered a pair of grounding wires (mains safety earth and negative output), detached the cover and studied the components. Nothing seemed amiss so I powered it up and measured across the thick wires emerging from the mains smoothing capacitor. Nothing registered on my voltmeter so I unplugged the IEC mains lead. Oops... I got a shock (the case does have a warning! I peered at the thing and realised the thick wires from the capacitor were insulated with transparent sleeving and the capacitor voltage was actually present (circa 320 volts hence the shock as I'd grasped simultaneously the metal shields on either side of the circuit board).

 

 

 Not to be put off I carried on fault-finding, but all the components tested OK... at least the diodes, FET, resistors and capacitors all measured fine.

I looked at the controller chip and spotted it was marked "UC3843". I looked at the UC3843 device spec then checked the pins... all the voltages seemed OK with the Vcc pin sitting at around 9 volts. As far as I could tell the FET wasn't receiving any pulses hence the reason for zero volts output.

 

 

At this point things went from fair... to middling... then pretty ropey. The chip was an SOIC8 but my spares box had yielded only a DIP8 version of the UC3843. Shall I order a new chip? This would mean a long wait so I decided to remove the old chip and solder in its place the larger dual-in-line device on the grounds that, if it worked, I'd order an SOIC8 version, knowing my time and money wouldn't be wasted.

Leaving the thicker device in-situ wasn't an option because the board wouldn't fit in the case. Anyway.. I fitted the DIP8 (below) and I saw 12.48 volts output so the old chip had been faulty (not a common fault).. so I need a new surface-mount UC3843.

 

 However, logistics are against me as a I found new SOIC chip would cost nearly as much as a brand new 12 volt 5 Amp power supply. Although these chips are around 50 pence each they're either supplied in tens with high postage or at ridiculous prices unless one opts for a Chinese supplier and a fortnights delay. So.. an option would be to try a near equivalent or hunt through my scrap board collection. I found and fitted a new UC3844BD1G and it tried to work but failed because it was turning on and off, probably because it had a different Vcc requirement. I actually tried to measure the pin voltages but as I approached pin 4 with my fat meter lead there was a quiet "pop" and the randomly changing 12 volt output sank to zero.

I put it on one side.

 The next day I checked the continuity across the 320 volt HT circuit and found zero ohms. After thinking about this for a millisecond unsoldered the FET and found it had a dead short across all three pins. I also found the mains rectifier bridge had a shorted diode and the mains fuse was blown. So, instead of a new chip, I now also needed a new FET and a new bridge rectifier.. certainly double the price of a new PSU because of postage costs. What I needed to do was to make a more thorough search of my scrap boards. There must be more than 500 of these of which about half use surface-mount components. Finding a PSU chip was made easier because all I needed to do was to check anything using a switching transformer. Not too easy though because chip markings are hard to read and often truncated. My search did turn up a new FET and a new bridge rectifier before noticing a chip marked "13844". The FET was marked "K2645" and will be a good option for the original 2SK2996. I hunted around and discovered the 13844 was probably an FA13844. The data sheet carried a picture showing that its pin connections had the same names as the UC3843 which seemed promising, at least worth a try.
 

 

 Disconnecting the replacement parts was straightforward and I was able to fit the new FET and the new SOIC8 without difficulty. Unfortunately the new rectifier was a millimeter too long so I had to mount it vertically and wire its pins to the board. I then unsoldered the old fuse and soldered the thinnest wire I could find across it before replacing it. Checks on the HT circuit proved the short had gone and it was now worth plugging it into the mains.

Before going further I'll mention a strange feature I'd discovered. Ordinarily these control chips are initially triggered from a voltage derived from the HT supply via a high value resistor and smoothed by a small capacitor in the range 10 to 47uF, but this design used a strange alternative. The Vcc pin was connected to one of the raw mains connections at the bridge rectifier via a pair of high value resistors in series. This theoretically supplies a rather roughish voltage of about half the fully rectified 320 volts.

Once the control chip has started driving the FET an isolated voltage is fed back from the chopper transformer to maintain operation. One drawback in the odd scheme was that my accidental short of the chip pins had destroyed the FET, the bridge rectifier and the chip rather than much lesser damage.

A rather funny lapse as a tail end to the story... I connected my voltmeter to the output lead, picked up the end of the IEC lead and plugged it into the circuit board only to instantly see 12.37 volts output. I'd missed the correct mains lead (unplugged from the mains supply) for one that was already live and was rather lucky not to have repeated the shock I'd had earlier. Anyway... all's well and I can now (hopefully) fire up the TV to commence tests on the audio system.

 

But first I should put it back in its case..

The TV is a Digimate LTV-1929WHTC and a quick check revealed it supported up to 1440 lines. I started by checking to see if i was serviceable as it hasn't been used for over a decade. I connected a short length of wire and it picked up digital Tv with perfect sound and vision. But could it respond to the HDMI input driven from the Teac? The result was a screenful of noise and a legend 722 x 481 (= Standard Definition). I expected information much as provided my Teac AG-15D except perhaps in higher definition, but no ... just noise at SD. Was the HDMI feature defunct? My plan was to now first connect my Asus Mini PC to the TV VGA input and prove that was working... it was fine. Then to connect the Asus to one of the HDMI input sockets and see what was sent to the TV. The answer was fairly surprising as a display popped up with a 1920 line notification (=1080). In fact, each HDMI socket provided the same results but wasn't being decoded by the TV as the screen was just a mass of lines.

This is interesting because it indicates that the HDMI circuitry is working OK at 2K or with HD definition. No "on-screen-display" of information may merely be due to my lack of understanding of the Teac or a fault. I need to study the Operators Manual in more detail. Click to read this.

Thankfully I've now learned enough to abandon my plans to buy a new AVD7623BSTZ HDMI chip which are available for a few pounds but would mean a pretty tricky replacement as it has no less than 144 pins ! What is interesting though is its TV spec extends only to 1080 lines with up to 1600 line capability and is therefore not compatible with the 1920 image I fed into the HDMI socket. It's use is therefore restricted to nothing better than High Definition TV and, unless downscaled, my PC HDMI 2K output will not work with it.

Reading the TEAC manual appears to say that the signal at the HDMI output socket will be commensurate with whatever HDMI input source is selected. In other words the output may well be 720 (or SD) with noise if the selected input has no signal. I would hazard a guess that the noise shouldn't be present but, instead a message indicating something.. mabe "no signal" or similar? Reading the ADV7623 spec seems to indicate that a message can be overlaid on the raster.

 

 pending
 
 
 

 Return to Reception