View Single Post
Old 9th February 2015
Gear Addict
M.Retra's Avatar

OK, I'm back! Sorry for the delayed response! As you can see from my post above, the LCD screen is working! When I first saw the splash screen pop up, I nearly fell out of my chair!

And as you can see below, the software is running and getting past the splash screen!

I apologize for the crappy images above and below. I was so excited that I decided to make a video. But then after listening to myself, I decided I don't do very good voice-overs, so I axed the project! Anyway, the video capture below shows my test arrangement.

Overall, what we're looking at here is an old Wavetek 182A function generator, which I had to get working after years of storage, generating and supplying a 3.0MHz square wave (albeit a distorted square wave). This square wave is fed into the LCD module. Here's why:

While troubleshooting and looking for data streams on the ribbon cable, I came across an important discovery. On the back of the LCD module is a Toshiba T6963C LCD driver IC. This IC handles all of the data communication between the LCD itself and the 87C52 microcontroller. I looked up the datasheet for the chip, and I learned that the T6963C has an internal oscillator. This piqued my interest, so using the pinout diagram in the datasheet, I put my oscilloscope probe on the X0 pin, which is supposed to be the clock/oscillator output. I wasn't getting anything. I then moved the probe over to the X1 pin (clock input). Again, nothing. So either the T6963C was defective, or the clock is coming from somewhere else.

Enter the 3.0MHz ceramic resonator.

That little guy was found to be dead! Basically, a ceramic resonator functions the same as a crystal oscillator, but it is a cheaper and less precise. The ceramic resonator was here on the LCD module board:

Why 3.0MHz? According to the T6963C datasheet, the clock frequency is determined by the number of rows and columns in the LCD screen. The more rows and columns to populate with a "pixel", then the faster the clock frequency. Think of the clock signal as the refresh rate, like in TV's and LCD monitors for your computer.

So I had an idea--what if I unsoldered and removed the bad resonator and hooked up an external clock source to the X1 pin of the T6963C? Well, you guess it...the LCD module came to life! What's interesting to note is the fact that my Wavetek function generator is completely unstable and nowhere near being dead-on accurate like a crystal oscillator. Well, I guess the T6963C doesn't care! It just needs a clock!

So the fix now will to order a 3.00MHz ceramic resonator. I will also order a 3.00MHz crystal oscillator and install it instead, as I prefer them more, but I will need to swap out SMD capacitors also in order to provide stability.

Now here's the kicker--without the LCD clock source, the Andy will not, I repeat WILL NOT boot., IMO, and it could be an explanation as to why many Andromedas out there do not get past the splash screen! Perhaps once a big graphic is displayed on the screen, like the splash screen, then the T6963C craps out (and hangs), perhaps due to a bad/weak clock signal.

Certainly the above, last statement is plausible. I find it truly unbelievable that the Andy will not boot with this clock signal removed. This contradicts typical LCD module operation and the Andy schematics. By looking at the Service Manual schematics, there is no feedback path from the LCD back to the 87C52. So is there bidirectional data going on instead? Maybe. Further investigation may be warranted.

Apparently, a bad ceramic resonator on the LCD module can stop an Andromeda dead in its tracks.

The above sentence is worth reading again.

So there you have it. More than you expected, I'm sure. I am very happy now, but not getting my hopes up 100%. I still need go back and:
  • Replace the bad ceramic resonator with a crystal oscillator for more reliable operation. COMPLETE. SEE POST #35.
  • Fix the pull-up resistor on the SRAM chip to be more permanent. COMPLETE. SEE POST #49.
  • Go thru ASIC osc/filter tuning procedures. COMPLETE. SEE POST #27.
  • Replace all of the electrolytic caps on the ASIC and Master Analog board. COMPLETE. ONLY REPLACED CAPS ANALOG MASTER BOARD. SEE POST #50.
  • Add a static-grounding wire to the LCD module if none of the screw posts tie the board to the digital-ground plane/pin/wire. NOT NEEDED; SCREWS TIGHTEN DOWN ON GROUND PLANE.
  • Redo tuning procedures, as needed, after capacitor replacements. NOT NEEDED
  • Clean and lubricate all pots. DECIDED NOT TO. MORE ON THIS LATER...
  • Tighten the right end cap/cheek. COMPLETE.
  • Replace memory backup battery. COMPLETE. SEE POST #24.

Last edited by M.Retra; 10th February 2015 at 01:11 PM.. Reason: Added: "and hangs..."; New Edit: Should be "T6963C".