Beogram 4004 (5526): Adaptation of a 1985 Platter DC Motor and an Exciting New Motor Failure Mode

The Beogram 4004 (5526) that I am currently restoring came with a strange platter motor installed with some hot-wiring:

It had a 1985 production date on it:

I wondered from what Beogram this motor may have been extracted. Dillen of BeoWorld.org helped out with his encyclopedic knowledge of B&O details. He stated:

"That motor is part# 8400098

It was used in Beogram 1700, 2200, 2202, 2400, 2402, 2404, 3400 (type 5726), 3404 (Type 5727), 6000 (Type 5751,5753,5754) and Beocenter 5000".

I tested the RPM and this motor indeed ran at 33 RPM, but not very stable. So far so normal (most DC platter motors have dry bearings at this point in time, i.e. they often do run not very smoothly). I had a closer look at the platter motor control circuit and it had been modified by replacing resistors R16 and R17 with wires and the two RPM trimmers (normally 5k) had been replaced with 2.2k trimmers.

This indicated that this motor needed a different Schmitt trigger timing in the motor control circuit. This pointed to a different RPM feedback mechanism in this "younger" motor.

I decided to take the motor apart and have a look, and restore the dry bearings while in there.

This shows the extracted motor:

The first difference to note is that it has four wires instead of the normal three. These wires were connected into the regular three prong plug. I removed the metal pulley and removed the outer housing:

The plot thickened as it became obvious that the red and blue leads went into one end of the motor, while the two yellow ones into the other.

I opened the motor up:

Immediately it became clear that this motor was different. The gear like wheel at the bottom end of the rotor obviously is an updated RPM feedback mechanism, which is not based anymore on induction of a current in two small coils under the rotor, as the powered rotor poles travel across them. This motor has a more modern type of mechanism, based on a magnet disc with 39 field shaping 'poles' around it. This disc runs next to a coil that is connected to the two yellow wires. This causes a feedback signal that has a much higher frequency per RPM than that of the original motors.

Before taking it apart, I ran the motor from a bench supply and measured the feedback signal:

This shows the earlier measured signal from an original motor at the same 5V (i.e. at a comparable RPM):

Note the different time and voltage scales. The signal of the 1985 motor has an approx. amplitude of 1V and the frequency is 268Hz. The original motor has a signal of only about 0.15V and its frequency is much slower at 36 Hz. This about 7x faster frequency of the 1985 motor is a consequence of the larger number of field shaping poles on this motor compared to the two coils and three rotor poles in the original motor. And the larger signal comes from the much closer proximity between pickup coil and magnetic field from the disc. So an all around improvement.

Back to the bearing restoration: This shows the motor completely apart with the bearings extracted (two small donuts on the black pad):

I immersed the bearings into motor oil and pulled a vacuum. Immediately the usual bubbling started, indicating air being extracted from the porous bearing material. This allows fresh oil to diffuse into the bearings:

After about two days the bubbling stopped and I extracted the bearings from the oil:

Then it was time to re-install them in the motor. I had to realize that my standard tool to press the bearing retainer back into shape did not fit due to the different feedback mechanism. I redesigned the part:

And this allowed me to put the bottom bearing back into place:

The I inserted the top bearing:

In this motor the brushes are located on the top end.

I put it back together and measured its RPM stability with the circuit modification in place to see how it would perform. This is the curve I measured after about 8 hrs:

This did not look great at all. If such such sudden RPM breakdowns are still there after re-infusing the bearings, it is usually a sign of one or more bad spark snubbers on the motor coils. So I took it apart again, and replaced the snubbers with modern TVS devices. This shows the rotor with the original snubbers (devices with red dots) in place and the new ones prepared for soldering on the black pad:

On these motors one cannot just remove the entire snubber assembly and put the replacements between the solder points (see here for a comparison with the original motor types). There is not enough space between rotor and brushes due to the new feedback mechanism. So I left the wiring in place and just cut the snubbers out with a wire cutter and soldered the new ones onto the wires..

Then I put the motor back together and ran another test:

This second surgery apparently did not help the patient at all! It even got worse. So I realized that there must be a yet another motor failure mode! Exciting!!

I took it apart again, and measured the resistance across the coils. Two of them measured the usual ~20 Ohm, but the third one showed only about 18 Ohm. This seemed a bit odd. While measuring I noticed a fluctuation of the value, so I wiggled the wires a bit and was able to reduce the resistance to about 5 Ohm with that. After a bit of more messing around, it became clear to me that one of the snubber wires apparently had chafed through the insulation of the magnet wire that is on the rotor poles and that way short circuited part of the coil whenever the contact intensified. My exchanging the snubbers must have made this issue worse as is visible from the increased RPM breakdowns.

I put a piece of double sided tape between the snubber wiring and the coil to insulate them from each other:

Now the coil measured the usual ~20 Ohms. So it was time for putting it back together and run another test:

This curve now looked pretty normal! So this motor is back in business. 

What is still did not like was the way the motor control circuit had been adapted for this motor. The person who did the modification realized that the RC time constant for the Schmitt trigger needed to be sped up to account for the much higher feedback frequency, and so he reduced the R component by removing the resistors and inserting lower resistance trimmers. This led to a faster response of the circuit, but with the unwanted side effect that the user accessible RPM trimmers above the keypad were now much more sensitive i.e. their resolution was much reduced, while their range was much larger than the ±3% prescribed by the scales. This was not Beolovely at all! When we make modifications or improvements, we try to make them in a way that preserves the original user experience.

It immediately occurred to me that the 'feel' of the RPM adjustment mechanism could be preserved if the C component were to be modified instead, while leaving all the resistances original. Thereby the ratio between the RPM trimmers to the rest of the resistance would stay the same after the modification. 

At this point I felt it was time to seek out the motor control circuit of a B&O product designed for this motor. I downloaded a few service manuals from the list Dillen gave me, but they all had monolithic motor control circuits in their diagrams, i.e. no information about the capacitances in these circuits.

But finally I found a manual with more information, the service manual for the Beocenter 2800/4600 series manufactured 1978 - 1980, concurrently with the Beogram 4004. Apparently the 4004 still carried the old motor for legacy reasons, while the Beocenters already seem to have had a motor with the updated feedback installed as is evident from the much changed capacitances in the control circuit. But it still used the old discrete circuit. This is the circuit shown in the Beocenter manual:

I marked the three capacitors that are different compared to the standard 4002/4 circuit, which is shown below:

The resistances in the Schmitt trigger RC legs of both circuits are similar, but the RC capacitor (C3 in the 4002/4 diagram and C2 in the marked up Beocenter circuit) are much different (220nF vs 68 nF). This makes the Beocenter Schmitt trigger much faster.

At this point it may be a good idea to briefly review how this circuit works. The AC feedback signal (see oscilloscope traces shown above) is fed into pin 1 and amplified with a large gain into a square wave. This square wave is fed into a non-symmetrical (two thresholds) Schmitt Trigger. When the feedback square turns the trigger on its output goes high (pin 6). This signal charges C8 via R19, and this voltage is fed into the motor amplifier via pin 5, and the motor receives power.

At the same time the trigger output signal on pin 6 is fed into the RPM controlling RC elements 2R2/R17/R15/C3 (33 RPM) and 2R1/R16/R14/C3 (45 RPM). Depending on the RPM setting C3 is fed via one of the two resistance chains, which raises its voltage. At a certain voltage the the upper threshold of the Schmitt trigger is triggered and its output voltage on pin 6 goes low. Then C3 discharges and during that time the Schmitt Trigger output remains low until it can be turned on again by the motor feedback square. This essentially creates a pulse width modulated (PWM) signal that allows controlling the energy flow into the motor.

This timing mechanism is the reason why the RC time constant needs to approximately match the motor feedback frequency, otherwise the motor gets too little power and runs only weakly. This is what probably prompted the original modification. The modifier turned the trimmer pods all the way down to zero, but was not able to get the motor up to the right speed. Then he had the idea to remove the fixed resistances and put in smaller trimmers to get up to 33 RPM.

I performed a number of tests with different capacitances for C3, C7, and C10 in place to find the best values. In order to do that with the table running, I broke the three capacitors out from the main PCB and ran wires to a breadboard:

This allowed me to exchange the caps without putting the Beogram on the bench for soldering. It turned out that 39nF for C3 gave the RPM trimmers a similar range like for the original motors. C7 seems to have little influence on the RPM stability, so I left it at 1nF like in the Beocenter circuit. C10 of course needed adaptation, too, since it controls the responsiveness of the feedback of the motor drive amplifier together with C9. The Beocenter value 47nF seems to work well, i.e. I kept it, too. This shows the relevant section of the circuit with the original capacitors in place:

And with the new ones:

I think it is a nice option to have these newer motors as a go-to for keeping the 4002/4s running. Sometimes the motors cannot be restored, and there is no obvious modern replacement motor available. This shows the motor wiring in more detail after I cleaned it up:

The yellow feedback wires need to connect to the white and the red leads of the plug. The original motor only has three leads since the yellow-red connection is made already inside the motor.

While working on this motor mod, I also replaced the electrolytic capacitors and relays on the main and output PCBs. This shows the main PCB after restoring it:

And the output board:

This shows the output relay area in detail:

As usual I installed a switch that allows connecting the signal and system grounds in case there is a humming issues when using RCA adapters. We are getting close to listening to some nice vinyls on this Beogram!