Beogram 4002 (5513): Replacement of the Sensor Arm Bulb with an LED and Improvement of the Record Detection Circuit

On the way to an all-LED Beogram 4002 the replacement of the sensor arm LED is usually my final step. And so, after restoring the DC motor, it was time to implant my recently developed flex-PCB based LED assembly to replace the incandescent light bulb. This shows the sensor arm with pulled out detector assembly:

If you click on the picture to get the full resolution and magnify it a bit on your screen, you will see that there is already a small black spot on the glass bulb, indicating that significant tungsten evaporation has taken place from the filament. This points to a likely near term demise of this bulb, which would incapacitate the Beogram. The installation of a LED promises a much better long term stability, and this is the main reason for LED upgrades outside the scarcity of fitting small light bulbs - they are hard to find these days.

The challenge with this particular LED upgrade is of course the small space and the need to replicate the power draw of the light bulb to not confuse the record detection circuit (see below).

My LED replacement therefore contains an additional resistor that compensates for the much lower power draw of the LED (which is a high brightness low color temperature model running at a very low emission level). This shows the assembly:

It is built based on a flex-PCB that can fold into the bulb compartment:

The glued on 3D printed red wedge on top assures that the LED sits in a place comparable to the bulb filament.

This shows the assembly in action:

Note the accurate color of the B&O logo, which is a result of the use of a low color temperature LED which has a high red component.

Any light source replacement in the sensor arm should be accompanied by measuring the sensor response when the arm is over a spinning empty platter to verify that the signal is strong enough to precisely and reliably trigger the protection circuit that prevents arm lowering in absence of a record. This is the trace I measured on the collector of TR3 after installing the LED:

On first glance it looks textbook, the dips going all the way until they bottom out at 0V. However, closer examination yields that the amplitude of the signal is only about 3.7V, while the circuit diagram prescribes something closer to 6V. This amplitude is too small to reliably trigger the protection circuit.

This shows the pertinent part of the circuit diagram:

This is how it works: The base of TR3 is biased by the voltage divider formed by R26 and the BE-diode in the transistor. The signal from the photocell in the sensor arm OPH1 is coupled in via C12. When there is no signal (like over a record), the voltage at the base is such that the collector of TR3 is at 4V. This means that the transistor is slightly ON creating a voltage divider with R27 (and R30). When OPH1 delivers its AC signal of about 20 mV amplitude, the voltage at the TR3 collector starts oscillating between 6 V and 0 V, i.e. TR3 operates as an amplifier. This 6Vpp signal is then fed into the base of TR4 via C16, which takes out the DC component. Since the base of TR4 is pulled up to 21V via R32 and protected against negative voltage (relative to its emitter) by D15, the resulting signal at the TR4 base is an oscillation between 21.6V and 20.4V. While at 20.4V the transistor is on, and C18 charges. This in turn biases it towards 21V. This pulls up the base of TR6 via R34 as current limiter and the collector of TR6 is pulled to ground. This signal then prevents the arm lowering circuit to work and the arm stays up.

There is a second way to pull up the base of TR6, which is via the D18/R37 link. This connects to the collector of TR5 via the Zener diode D17. This is the circuitry that can invoke the 'no record present' signal at the collector of TR6 if the light bulb is broken. If the light bulb is open circuit, no current flows in R36, causing the voltage at the collector of TR5 to go up. This pulls up the base of TR6 and the collector of TR6 goes down and the arm cannot be lowered anymore under any circumstance. This is the reason that the LED replacement needs to be designed in a way that the current through TR5 is similar to the current caused by the original light bulb (~50-60mA).

Back to the signal on the collector of TR3, which measured to be less than 6Vpp. I long ago noticed that some Beograms have issues with the detector circuit and that sometimes they do not recognize that there is no record and happily lower the arm on the platter even though everything seems to be in working condition. If that happens the stylus' life depends on the proper adjustment of the arm lowering limit keeping it from hitting the platter ribs.

I addressed this problem in the past by soldering in a 10M resistor between the base of TR4 and GND. This biased TR4 slightly on and even a weaker signal from TR3 would be enough to stop the arm lowering circuit. At that point I did not understand the root-cause of the problem, but rather discovered the fix by accident: The circuit tended to work whenever I connected my oscilloscope to the base of TR4 and I realized that the internal resistance of the scope fixed the problem...a bit of trial and error yielded that a 10M resistor to GND tugged enough on the base of TR4 to keep things working after disconnecting the oscilloscope.

But as I know now, there is a much better solution: Double R26 to 2M. This causes a reduction of the current through the BE diode of TR3, and so the transistor is less on, and the voltage at the collector goes up to the prescribed 4V. This fixed the problem in this Beogram:

The amplitude of the oscillation is now close to 6V and can trigger TR4 reliably.

An interesting question is: Why did B&O let Beograms leave the factory with 1M resistors in place and a too low TR3 collector voltage?

One of the issues with the design around TR3 is that the collector voltage depends on the DC current gain of the particular transistor used as TR3. The current gain is the most variable parameter of any transistor series, i.e. even within one production run it can vary considerably by up to a factor two or three. It would have been better to bias the base with a stiff enough voltage divider to GND, and to implement a feedback for setting the gain of the amplification. With the design as it is, every Beogram has in essence a different TR3 working point depending on the particular BC182 that is soldered in. With a new light bulb in place in the sensor arm this variation may not have been a problem due to the high initial light intensity making enough signal even with a non-spec working point of TR3. But as the bulbs aged the intensity gradually went down due to deposited Tungsten on the bulb glass. And after a while the signal became weaker and at some point the mechanism stopped working. So they may not have noticed back then that there is a problem when they manufactured these Beograms.

Interesting stuff (at least to the Beolover...;-)!