Beogram 4002 (5513): Restoration of the Platter Motor, PCBs, and RPM Panel

Recently, a customer in California sent me the DC platter motor, along with the PCBs and the keypad assembly, of a Beogram 4002 (Type 5513) for restoration. I was given the additional information that the Beogram would blow fuses whenever plugged in.

As usual, I began with the platter motor. The oil infusion of the bearings under vacuum can take up to 3 days, so it was the perfect starting point for this project. This shows the motor as received:

I took it apart to extract the shaft bearings:

The bearings are the two small donuts on the black pad upfront. I submerged them in synthetic oil and pulled a vacuum. Immediately, strong bubbling started:

The bubbling is indicative of air being drawn from the pores of the Oilite bearing material. As the air goes out, the oil goes in! When the bubbling stops, the bearings are replenished and can be used again. While this process was underway, I focused on the main PCB. This shows it in its as received condition:

Here is a detail shot of the 'RPM section' with the original Siemens RPM relay and the RPM trimmers:

Due to the 'blows the fuses' warning I was given, I installed the board in my bench 4002 to see what was going on. I hooked the board up to power using a bench supply instead of the Beogram transformer. This allows ramping up the voltage slowly while watching the current meter. Indeed, already at a voltage below one volt, the current of the bench supply was maxed out, and the current limiter prevented worse things from happening.

The culprit for the short circuit was found quickly: When I removed the screw that holds one of the two Darlington power transistors that are mounted on the solder side, it became apparent that the screw was missing its insulator sleeve and the mica sheet under the package was also absent:

The sleeve and mica sheet prevent contact between the collector of the transistor and ground (via the PCB mounting post that is used for heat dissipation from the transistor in this design). Without these insulators, there is a direct short circuit between the 21V power rail and ground, i.e., the fuses will blow immediately when the Beogram is plugged in. The other Darlington was also missing the bolt insulator:

Someone clearly did not appreciate Ohm's law!...;-).

While the board was still mounted, I replaced the two Darlingtons. This shows 1IC1, which regulates the 21V rail:

I usually replace the original TIP120s with their higher current cousin, TIP102. For some reason, modern TIP devices need some additional capacitance (the yellowish component in the above picture) at their emitters in this circuit configuration. Otherwise, they can develop a high-frequency oscillation superimposed on the 21V rail, which can fool the record detection circuit into believing there is no record on the platter. This subsequently disables the arm-lowering circuit even if there is a record on the platter.

This shows the new TIP107 that replaces the original TIP125 that serves as 1IC4 to control the arm-lowering solenoid:

Then I removed the board and inspected the component side. Immediately, I saw that one of the four H-bridge pnp transistor cans had been replaced with a non-spec type:

A closer look revealed it was a S9012, a pnp transistor rated for 500 mA collector current:

The original BC143 types are rated 1 A. 500 mA may work for some time if the carriage mechanism is in top shape and there is not much mechanical resistance. Then the carriage motor runs reliably below 500mA. 

I replaced all the electrolytic capacitors and the power transistors. This shows the restored board together with the extracted original components:

The Siemens relay was replaced with a Beolover Siemens Relay Replacement for Beogram 4000, 4002, and 4004, and the RPM trimmer received an upgrade with modern 25-turn precision encapsulated trimmers for more exact adjustment:

Next, I focused on the output PCB. This board had been modified by a previous owner of the Beogram:

The circuit on this board delays the output relay so it only opens after the needle has hit the groove:

I replaced the output relay and the electrolytic capacitor that determines the delay:

The picture below shows the RPM panel that is mounted above the keypad. It contains two incandescent bulbs that I usually replace with LEDs. The panel is shown flipped on its back, revealing the two bulb covers:

I removed the covers. This shows the bulbs still installed:

The two small green PCBs are the Beolover RPM Panel LED Backlights for Beogram 4002 and 4004 (Types 551x/552x). The boards are directly soldered to the terminals that connect the wires of the bulbs:

This shows one of the boards installed in detail:

The PCBs do not obstruct the bulb covers, which can be reinstalled after the boards are in place:

Meanwhile, after about 48 hrs the bubbling around the motor bearings had stopped. I extracted the bearings from the oil:

I reassembled the motor and installed all the components for testing in my bench Beogram 4002. I ran a 24-hour RPM stability test with the BeoloverRPM device:

The BeoloverRPM has two operational modes. In 'slow' mode, it measures the RPM in 10-second intervals and relays the measurement to a serial port of a computer. This allows graphing the RPM over long periods of time using Excel or similar software. This shows the result of a 24 hrs measurement:

This result is pretty much as good as it gets with the DC motor Beogram 4002. 

In the 'fast' mode, it transmits an RPM measurement every time a platter rib passes under the sensor. This yields high-resolution graphs that show short-term RPM changes ("wow and flutter") in detail. This graph shows a measurement covering about 35 turns of the platter, representing a run time of a little more than 1 min:

The zig-zag pattern is a measurement artifact that originates from small spacing variations between the platter ribs of my bench Beogram. All Beogram platters have such variations due to manufacturing imperfections. This generates a repeating pattern every 24 measurements (there are 24 ribs around the platter), which is superimposed on the real RPM changes that are introduced by the feedback system that keeps the motor RPM stable over time. This real RPM change is essentially the sine-wave-like pattern that modulates the zig-zag pattern. An evaluation of the wavy component yields a wow and flutter estimate of about 0.1%. This is 2x of the 0.05% stated in the specs list in the service manual.

    This difference is most likely systematic due to the entirely different way wow and flutter were measured in the 1970s when these Beograms were produced. Back then, the measurement was carried out with a 1 kHz tone on a test record. In these measurements, deviations from the 1kHz center were measured with an analog spectrum analyzer and then converted into a wow and flutter number. It should be pointed out that this discussion is pretty academic since humans typically start recognizing frequency fluctuations above the 0.7% threshold, i.e., the RPM fluctuations of this Beogram are well below this threshold, whether the number is 0.05% or 0.1%. This motor is definitely ready for duty again!

This concluded my work on the received parts, and I will soon send them back to my customer in California.

 

Beomaster 2400 (Type 2902): Full Functional Restoration

This post discusses the work done restoring a Beomaster 2400 (Type 2902) for a customer located in Minnesota.

This shows the fully restored unit:

Let's see what was required to get it to this point:

This image shows the receiver as it was received. Fortunately, it was packed very securely, so it suffered no damage during shipping:

The unit needed a good cleaning, for sure! On the positive, the veneer panels are in pretty good shape with pretty good front corners!:

This image shows the unit in its 'service position' on top of my Lazy Susan:

I removed the fragile veneer panels so they would not get damaged while I worked on the unit. The blue tape covers the control panel door dampers, which are usually covered in a good amount of sticky viscous grease.

Utilizing a rotating assembly platform is highly recommended when performing work on these vintage Beomasters, as it drastically reduces the likelihood of stressing or breaking internal wiring leads while servicing the main board. Here is a closer perspective of the main PCB in its original state:

I went ahead and replaced all electrolytic capacitors on the board. I usually start out by doing those inside the shields first. As an example, this shows the detector can:

I removed the shield on both sides. This revealed the sole capacitor (C18) inside:

I replaced it:

And put the shield back on:

Then I replaced the capacitor in the remote control receiver shield. The next 'special' capacitor replacement task was replacing the four (green) capacitors on the small piggybacked board above the volume control LDR assembly:

 I unsoldered the board and replaced the caps. This shows the rebuilt board put back into its place:

Then I replaced all the remaining electrolytic capacitors on the entire board. After this was done, I focused on replacing the two big reservoir capacitors. This shows a detailed photo of the wire connections on the original capacitors:

I installed new capacitors. I use modern 10,000uF units since they have the same form factor as the original 5,000uF caps. Bigger is better in this application, so why not upgrade the circuit a bit? Modern capacitors contain more capacitance per volume. I usually tie the wires together with short bits of shrink tubing after I remove them from the original capacitor terminals. This makes the installation of the new capacitors much easier:

It is good practice to upgrade the main rectifier to a component rated for higher current. The stock factory rectifiers are generally running close to their limits and are prone to burning out over time. This image provides a direct comparison between the old component and the beefier new replacement:

Note that the replacement component requires its leads to be pre-bent so it fits neatly into the footprint of the factory unit. 

This shows the new unit in place:

Another critical area to address is the replacement of the quiescent current trimmers. The stock open-frame trimmers are almost always oxidized, which often causes erratic bias and can lead to catastrophic failure of the output transistors. This shows one of the fully sealed, 25-turn precision trimmers that I installed:

I mount these components with extended lead lengths so they can be strategically angled, allowing the adjustment screw to line up with the original access holes in the circuit board, allowing adjustment from the solder side, just like the factory single-turn components:

If you look closely, you can see two small 'hooks' below the trimmer that I soldered to the emitter resistor used for the quiescent current adjustment (R356). This makes it easy to do the adjustment without having to worry that the multimeter clips come off.

This shows the completely overhauled PCB side-by-side with all of the discarded components that were replaced:

After completing the work on the main PCB, I replaced the two electrolytic capacitors on PCB 3. This shows it in its original condition:

And with new capacitors:

There is one more electrolytic capacitor that needs replacing. It is in the tuner front end. The front end is in the shield next to the antenna port:

The first step is unsoldering the grounding wire that emerges from one of the holes in the bottom lid. Otherwise, one cannot remove the lid for access to the solder points:

Once the wire is disconnected, the lid can be removed:

I also removed the lid on the component side:

This revealed the single (orange) electrolytic capacitor on this board. I replaced it:

After I put the lids back on and soldered the grounding wire back on, I started working on the control panel PCB (#4). This shows it in its original condition:

This shows the bass, treble, and balance sliders after removing the printed indicator foils:

These Beomasters came with two types of slider potentiometers. This unit has the older style, which is failure-prone due to poor design of the slider bridge, allowing the sliding contact terminals to break free of the bridge. This meant all three sliders needed to be rebuilt. I started with the bass slider and unsoldered it:

This shows it belly up:

This shows the end with four contacts:

The process starts with bending the two tabs up that hold the slider clamp in place on this end:

Once the tabs are vertical, the clamp can be removed with a suitable screwdriver:

I pulled the slider out of the other end and put it down upside down:

You can see that one of the small U-shaped terminals detached from the slider bridge and remained on the tracks on the bottom of the slider body. This is the issue with this type of slider. The reason is that the flimsy plastic tabs that hold the contacts to the plastic bridge deteriorate over time.

Luckily, there are reproduction plastic bridges available at the DKsoundparts store in Denmark!

If you try this at home, be extra careful not to lose the small carbon plugs that are on one side of each of the contact terminals. They are only mechanically pushed into the small hole on the terminal and easily come off when the terminal is liberated. They are essential since they make proper contact with the carbon resistance track of the slider.

The potentiometer will not work properly without these plugs!

This shows both of the terminals with their plugs in place:

This shows the terminals transferred to the new part from the DKsoundparts store:

I transferred the metal bar to the new assembly:

Before I put everything back together, I cleaned the tracks with alcohol and then coated them with DeoxIT. I use F100L on the carbon resistive tracks and D100L on the metal pickup tracks:

This anti-oxide coating helps keep the sliders crackle-free. Installation of the slider assembly is done in reverse order. Once the clamp is back in place, the tabs need to be bent back to secure the clamp in place:

The restored slider:

And back in its place on the board:

I did the other two sliders and also replaced the electrolytic capacitors. This shows the completed board:

Next up was rebuilding the three indicator boards, which utilize a number of small incandescent lamps:

I usually replace these bulbs with a Beolover Set of LED Replacements for Beomaster 1900 and 2400 Indicator Light Bulbs. The primary motivation here is longevity, as LEDs will easily outlast traditional incandescent bulbs. Furthermore, standard bulbs gradually dim over decades of operation, drifting away from their original specifications. They also generate a significant amount of heat, which will fade the red and green plastic color filters over time. Moving to LEDs eliminates all of these concerns. It seems some technological advancements in modern times are genuinely for the better!...;-).

First, I tackled the program selection display board. Here it is in its factory original state after removal of the plastic filter assembly.  The 10 LED replacement boards are positioned below the board:

I desoldered all of the original bulbs and populated the board with the new LED modules:

Every individual replacement module features an integrated miniature circuit that emulates the specific aspect of the electrical signature of a filament bulb for its particular employment in the circuit.

The inclusion of these specialized matching circuits is vital because the unique cold-to-hot resistance curve of a traditional bulb filament is used as an active component within the circuit design. For instance, the Beomaster 2400 leverages this specific behavior of the display bulb resistance to momentarily trigger the muting circuit whenever a source button is depressed, ensuring that no audible switching pops reach your speakers. My custom LED modules implement a network of capacitors and diodes to replicate this transient behavior, maintaining full compatibility with the circuit design.

Once the new LED modules were securely soldered into position, I carefully snapped the circuit board back into the plastic cover

There are no clearance issues between the new LED modules and the molded plastic housing. From the outside, the assembly looks entirely stock, with no hint of the modern upgrade hidden behind the colored filters!

The next phase required swapping out the three illumination bulbs located inside the horizontal slider scale assembly:

This assembly already showed some signs of heat-induced warping of the red filters:

After pulling away the protective plastic housing shroud, the underlying bulbs are accessible:

The LED modules shown at the bottom are populated with 9 separate LEDs each to achieve a smooth and balanced diffuse backlighting effect across the entire scale. This shows the LED modules installed:

This shows the assembly with the protective plastic assembly reinstalled:

The final display section requiring bulb replacement was the volume level indicator:

This particular assembly relies on just two bulbs:

The LED boards are shown at the bottom. 

I installed the LED boards:

This display assembly is different because it leverages the nondirectional light distribution of standard bulbs. The molded plastic block is shaped like a funnel to serve as an optical waveguide. This arrangement creates the visual effect of an expanding or contracting illuminated bar that matches the volume level. To successfully preserve this visual characteristic, the new LED modules need to be mounted at an approximate 30-degree tilt, as captured here:

This shows the entire optical assembly buttoned back up and reinstalled:

With all the LED modules in place, I powered up the Beomaster to do some adjustments. First came the 15-volt rail:

An important adjustment in any thorough Beomaster 2400 overhaul is dialing in the idle current bias. The trimmer potentiometers must be set for both channels to achieve a 12mV reading measured across the emitter resistors of the main output transistors. Adjusting the right channel:

and the left channel:

Calibrating this bias properly guarantees that harmonic distortion is minimized within the output stage while ensuring the Beomaster idles as coolly as possible. Be careful if you do this adjustment at home: if this voltage gets too high, it is possible to damage the output transistors!

Next came the calibration of the tuning voltage:

I listened a bit to a local NPR station and enjoyed this now perfectly working Beomaster 2400! Then it was time to do some bench characterizations with my new QA403 audio analyzer! I recently upgraded from my QA400 since the QA403 offers remote control through a web interface and also has a much higher input impedance for more accurate and easier calibratable measurements.

I also built myself a second 8 Ohm dummy load, enabling simultaneous Total Harmonic Distortion (THD) measurements on both channels under load. This shows my setup:

For the THD measurements, I injected a 1 kHz reference signal at 0 dB (corresponding to an amplitude of 2.83Vpp) directly into the tape input from my signal generator. I then powered on the Beomaster and brought the volume up to a level just before the onset of clipping. This captures the resulting signal monitored across the 8-ohm dummy load resistors:

Nice, clean sine waves for both channels! Typically, visible waveform clipping on these units starts at around 40 Vpp output signal.

The QA403 inputs were hooked up to the dummy loads for this measurement. I integrated 100:1 voltage dividers into the dummy loads, effectively reducing the voltage at the QA403 inputs by 40dBV. This allows measurements without using the QA403 internal input attenuators. There is an issue with the attenuators since they introduce THD and cause measurement errors.

The QA403-native THD measurements came out to 0.13% for the left channel and 0.12% for the right channel. These values are confirmed by my own THD evaluation based on the measured FFT spectra measured by the QA403 analyzer:

My QA403 control code automatically generates a 'first order' THD value based on the first harmonic peaks of both channels. The values were 0.07% for the left and 0.06% for the right channels. These values are lower than the QA403 values since the QA403 software adds up all of the harmonic peaks and not only the first one, like I did. I think it is always a good idea to try to confirm 'black box' measurement values by assessing the actual raw data.

These numbers align well with the original factory specification of "<0.2%" detailed in the official service literature. Excellent! After letting the unit 'cook' for roughly 30 minutes at the above load, the heatsinks stabilized around 55 °C, which is normal for this architecture. This Beomaster operates properly at high output! I used my iPhone IR camera attachment to take this picture of the heat signature:

It is nice to see how the load resistor heated up as well. They stabilized at about 61C. A nice little space heater, this Beomaster!...;-)

Following the THD measurements, I measured the Frequency Response (FR) of the amplifier. This shows my setup:

The QA403 has a very convenient way to measure FR. One simply connects the QA403 outputs to the input of the Device Under Test (DUT) and the QA403 inputs to the output of the DUT. Switching the QA403 generators to 'Frequency Response' and clicking Start then directly yields a FR spectrum across the chosen frequency range. 

For my FR measurements, the QA403 inputs were connected to Test Points (TP) 203 and 303. First, I measured the FR via the tape input. My control software measured 10 spectra and averaged them. This is what I got:

Nice and flat and linear across the spectrum, and just a minor drop to very low frequencies! Also, both channels have similar spectra. This is how it should be!

For my second FR measurement, I connected the QA403 outputs to the phono input of the Beomaster. This measurement aimed to see whether the RIAA preamplifier had the proper de-emphasis. Due to the 100x (40dBV) higher sensitivity of phono inputs relative to high-level inputs like the tape input, I constructed a special cable with a 100:1 voltage divider integrated so the QA403 outputs would not overwhelm the input. This is the curve I measured:

These curves are very close to the ideal RIAA de-emphasis curve, which prescribes a 40dBV  attenuation across the audio bandwidth. There is a nice Wikipedia entry on this.

These measurements demonstrated that this Beomaster 2400 is ready for duty again!

Time to begin reassembly of the enclosure! First, I removed all oxidation from the contact tabs of the touch buttons:

This makes it more likely that the touch buttons will work properly after the operating panel is installed.

The Beomaster had lost all four plastic feet:

I installed a set of new ones from the DKsoundparts store in Denmark:

And then it was finally time to put the enclosure back together! Here are a few nice pictures of this restored Beomaster 2400 in all its glory!:

Beotiful!..;-)

I will run this restored Beomaster 2400 for a few days to verify long-term stability, after which it will be ready to return to its owner in Minnesota!