Beogram 8002: Strange Behavior - After PLAY Arm Just Moves 2 cm, Then Stops

A Beogram 8002 from Australia arrived for some TLC. The problem at hand was that after pressing PLAY the arm just traveled in for about 2 cm and then it stopped. While this was very reproducible on my bench the previous experience was more intermittent in that sometimes the deck would start playing normally, but then after a few minutes it would suddenly stop. 

I opened the deck up and my first guess was rotary encoder trouble. If the encoder on the carriage spindle does not work properly, the normal response of the control system is to stop playing after about 2 sec. And 2 sec is about the time it takes to travel 2 cm. This shows the relevant sections of the circuit diagram:

The rotary encoder section is in the top left quadrant. The IR diode (OPE1) is powered via the 5V rail and a 150R resistor. This results in about 1.3V at the diode. When I measured I got 5V, indicating that no current is flowing through R1, meaning that the diode had gone open circuit. This is the typical failure mode of these devices.

This shows the rotary encoder assembly removed from the aperture wheel section:

The bottom unit is the IR diode, the upper two units are the two photo diodes. Their voltages are fed into IC1/IC2 and after amplification are fed into processor pins 28/29 ("Slide Tacho"). When this signal is flat while moving the carriage (slide) the processor stops the movement after 2 sec. I replaced the IR emitter with a Optek OP240, which can be had in the same package as the original unit:

This restored the 1.3V at the diode and I was now able to measure the proper encoder signal at plug 6 Pin 13/14, but the problem did not go away. Frustrating!

My initial guess was that the encoder signal did not arrive at the processor pins. So I removed the PCB, opened up the processor EMI can and ran the unit, while measuring directly at the processor pins. Micro grabber jumper cables are really nice for such measurements! Here is an impression:

Well...the signal arrived at the processor. And I also noticed during these experiments that the deck would sometimes work, especially when freshly plugged in! The plot thickened! Working on B&O vintage units for some time now, I immediately put my money (I lost...;-) on a cracked solder joint or a hair line fracture in the PCB. So I set out to re-solder all relevant connections. Here is an impression of the processor board, solder side:

Well, al this effort did not fix the problem!

So I hooked up my oscilloscope and watched the relevant signals on the processor inputs as the issue unfolded. Finally, when monitoring pin 31 ("lift manual") I got these traces:

the green and yellow ones are the encoder signals on 28/29. The red one is the signal at pin 27 (<<), which essentially controls the carriage servo. The blue line is the signal on pin 31. And what we see is that the signal on this pin increases, and at about 3.5V (which is about where a logic HIGH starts on 5V systems) pin 27 caves turning the servo off, causing the carriage to stop as we see from the petering out signals on the encoder diodes.

This immediately suggested that something gave the processor the impression that the arm needs to be lifted and the carriage stopped.

Considering the relevant sections of the circuit diagram, it was clear to me that somehow the signal on the input opamp IC2 (same opamp package as the IC2 that amplifies one of the encoder diodes) that is hooked up to pin 31 must go high. The only way that this can happen is when the resistance of the photoresistors R9 or R10 in the << >> control housing gets too large without pressing one of the <</>> buttons. The way this control system works is that when these buttons are pressed an aperture is driven in between the light bulb IL1 and the either R9 or R10. This gradually increases their resistance, thereby raising the voltage at the corresponding IC2 turning on the carriage servo OM1 in either forward or reverse direction. While this happens the motor control signal is also fed into the opamp that connects to pin 31 via diodes D5/D6. 

All this suggested that one of the photoresistors increased its resistance by itself without pressing one of the << >> buttons. The service manual prescribes that the voltage at the resistors as measured at plug 5 pins 4 or 6 needs to be 0.61V when the buttons are not pressed. I measured 1.2V at pin 6 and 0.6 at pin 4. My first response to that was trying to adjust the screws that are in the housing allowing the reduction of the light that falls on the resistors. But I was not able to get the voltage below 1.2V. And that suggested that the photo resistor was either the wrong type or broken. This also explained the 2 sec delay before the deck would shut off, since these resistors are temperature sensitive, i.e. the lamp would heat up the resistors and the bad/wrong one then crossed the threshold at which the opamp started triggering pin 31.

I took the control panel out:

The two screws on the bulb housing are used to adjust the light intensity that arrives on the photoresistors. Then I removed the PCB from the front panel. Here is a look at the panel backside:

This shows one of the photoresistors. 

They are mounted in compartments on either side of the light bulb. Here is a picture with the bulb illuminated.

I measured the resistance on the photoresists on the 'good side' of the assembly. I found about 20k when the button was not pressed. The other side had about 75k, and there was no way to get that down.

Since the manual does not specify what type of photoresistor they used for this assembly, I needed to experiment a bit. So I bought a bunch of photoresistors with different resistances for a 'shootout'. 

A few days later, I received six different types, and I measured their resistance in the cavity with the light bulb on. I finally settled on GL5549, which is widely available on ebay. It is specified to have a dark resistance of 10M and 100-200 under illumination. But that seems to be a fairly weak light intensity. In the Beogram assembly it showed 20k with the aperture fully opened. 

While in there I decided to also replace the light bulb with a white LED (Newark 14N9428) to ensure better long term stability. I used a 1k resistor to limit the current. Here is an impression:

This shows the backside:

After verifying that the light intensity was high enough to yield the 0.61V I put everything back together, adjusted the brightness screws on the bulb housing to get the 0.61V on both sides and then fired up the deck. And it was back in business. So this appears to be fixed!