This is a follow up to my recent post about the redesigned Beogram Commander remote control board, which now works in both (DC-motor) Beogr...
Tuesday, March 21, 2017
Beogram 4002 (5513): DC Motor Restoration and Yet Another Exciting Aspect of Achieving RPM Stability
The restoration of the platter drive system of a DC motor Beogram 4002 (5513) requires the installation of modern multi-turn RPM trimmers, a new encapsulated RPM relay, replacing the light bulbs that back-illuminate the user accessible RPM trimmers with LEDs for better thermal stability, and rebuilding the motor with freshly oil infused oilite bearings. See here for RPM measurements taken after each of these steps to see how they individually improve the RPM stability.
The Beogram 4002 that is currently on my bench already received most of the above treatment except the motor restoration. And so here we go:
This shows the extracted DC motor:
To get to the bearings the motor needs to be fully taken apart:
The bearings are the two small donuts on the black pad.
Once the bearings are out, they need to be oil-infused. They are made from porous "Oilite" brass that is factory oil infused. After some run time the oil in the bearings is depleted and it needs to be replaced. This needs to be done under vacuum to draw out the air from the pores that over time replaced the oil. This shows the bearings in a mason jar under SAE30 oil after pulling the vacuum:
The air bubbles show that the oil infusion process has started. When the bubbling stops after 12 to 24 hrs, the process is finished.
Then it is time to put the bearings back into the motor housing.
This shows the bottom bearing re-inserted into the brush carrier plate (which also contains the white feedback coils for the tacho control of the motor):
This shows the process putting the top bearing in:
I use special 3D printed fixtures to press the tabbed ring back onto the bearing to hold it in place securely:
Once the motor is back together it is time to give it an initial test. Run it at ~5V and measure the current. It should be <30mA. If it is higher then the brush carrier plate needs to be loosened and retightened until the current is low enough. This ensures that the motor spins easily for good long-term stability.
Then it is time to put the motor back in and do a multi-hour RPM characterization. I usually run them for 12-24 hours. This usually shows if there are any problems left to tackle or if the motor is good to go. This shows the measurement process with my Beolover RPM device:
It clamps on to the Beogram frame and delivers a stream of RPM measurements to a computer serial port, which allows graphing the RPM stability over time. It is available to other enthusiasts. Just send me an email or use the contact form on the right. The blue curve is what I measured after the above procedure:
And that was a pretty disappointing result! Pretty big RPM spikes towards higher RPMs. Absolutely Beounlovely!
After a bit of head scratching I decided to put the motor into my own Beogram 4002 5513 and run it there. It performed flawlessly as expected after the oil infusion procedure. Then I ran the motor again with my main PCB installed in this Beogram. Again, flawless performance! This pretty much narrowed the problem down to an issue with the main board of this Beogram. I compared the two boards and I realized that C10 on my board had a 10 uF capacitor installed, while the board in this Beogram had a 0.47 uF capacitor installed. I had noticed in the past that some Beograms have 10 uF, while others have 0.47 uF or even 0.33 uF installed as C10.
I replaced the 0.47 uF capacitor with 10 uF and voila! I was able to measure the red curve in the above graph. So we can conclude that it is a good idea to replace C10 with a 10 uF capacitor if there still are RPM issues after doing all the other tasks outlined above to make the system stable.
Of course I wondered why a smaller C10 causes these RPM fluctuations. This shows the control circuit of the DC motor on the PCB:
C10 is the smaller capacitor to the left of the C1003 motor control IC. This shows the circuit that controls the motor:
Essentially, the motor induced feedback ("tacho") signal from the small coils in the brush carrier plate shown above is fed into an amplifier at pin 1. The overdriven amplifier changes the sinusoidal feedback signal into a square wave, which is compared with a time constant network formed by the RPM trimmers and C3 in the Schmitt trigger. This results in a square wave with a duty cycle that now depends on the motor RPM. If the motor RPM is too low the duty cycle increases. If the motor is too fast the duty cycle becomes smaller. That way C8 is charged to different voltages, which changes the DC voltage at pin 5 that controls the amplifiers that control the current that goes into the motor.
The sensitivity of this process is controlled by the feedback fed from pin 4 into pin 5 via the C10/C9/R20 voltage divider. Depending on C10 this feedback gets stronger or weaker. Since the feedback voltage divider is build from capacitors it is essentially only active when the motor voltage changes at pin 4. The impedance of a capacitor is 1/wC (w being the frequency). This means that the impedance of C10 is smaller for 10 uF than for 0.47 uF. This means that the voltage divider tries to reduce voltage spikes across the motor more resolutely, tamping down more resolutely on sudden changes.
In other words the system becomes slower and less severe in its reaction to RPM changes, which we clearly see in the red curve. It appears to me that 0.47 uF leads to an overreaction of the system and even the establishments of a number of unstable oscillatory states. This can be inferred from the tendency of the blue curve to always jump to similar RPMs. Essentially, it flip-flops between several RPM states in a random way.
The fact that some Beogram 4002 were already factory-fitted with 10 uF (while the above diagram from the service manual shows 0.47 uF) indicates that B&O at some point realized this problem and took corrective action. I'd love to get my hands on this service bulletin! It was probably not so simple in the 70s to quantify long term RPM aberrations due to the absence of digital tools like my Beolover RPM device.