Beogram 8000: Strange Behavior Caused by an Intermittent 5V Supply

**********************This post is a sequel to this restoration.***************************

I declared 'mission accomplished' after restoring a Beogram 8000 recently, and I sent it back. After about 2 weeks I received an email describing some strange behavior, where it would shut down during play, but sometimes only very briefly and then start up again etc...

The Beogram was sent back to me and I had a look. First I was not able to reproduce the issue, and after playing more than 20 albums, desperation set in. So I gave it a bit of impatient rapid play and stop and other buttons, hoping to provoke a reaction. And after a while, indeed, it gave me a signal. The display showed this after it returned home after pressing STOP:

The dot on the right should not be on! This encouraged me to continue pressing buttons and I got this:

and this:

After each of these display readings I was only able to go back to normal after pulling the power plug and reconnecting. I continued working it for a while, and finally I got it to loose power briefly, and then it came back on. So I started thinking there must be a problem with the 5V supply. The 800x often has power issues, but mostly due to bad capacitors or bad solder joints on the headers that connect to the transformer block. Of course I already had re-soldered these headers and installed new capacitors, i.e. these standard root causes could be ruled out.

Since the issue seemed to be based on very brief outages or brown-outs that confused the processor, I needed to find a way to monitor circuit nodes for brief voltage dips while playing with the buttons. I decided to use the external interrupt pins (2 and 3) of an Arduino nano board, which I outfitted with a npn transistor and a 20k resistor on one pin to be able to sense higher than 5V signals, so I could also go between the transformer and the 5V regulator. After the 5V regulator the voltage is stabilized, i.e. Arduino pins can directly be connected. LEDs were used to signal power drops on each pins. This shows the board:

First, I connected the two pins to the green marked nodes in the circuit diagram to test if the 5V chain was interrupted between regulator and the plug that connects to the micro controller can:

After I was able to produce another failure the Arduino told me that on both green points the voltage briefly dropped to zero. This verified my power outage hypothesis, but did not show me where it occurred. Suspecting a bad regulator, I connected the transistor-buffered lead to the red marked spot right at the transformer-facing side of the fuse holder.

I played the buttons again, and again, and after a long time, I was still not able to reproduce the problem anymore, while before it took me always less than 5 min! This indicated to me that I might have had accidentally fixed the issue when soldering the test lead to the solder spot on the board:

It turned out that this solder spot serves to connect a black jumper wire between the fuse holder and the 5V rectifier:

So I suspected that the wire had a bad connection, and re-soldering the solder point while tacking on the measurement jumper accidentally fixed the problem. I replaced the back jumper with a new one, and had a closer look at the connector:

Indeed, the wire had pulled out a bit from the solder terminal, but was now soldered to the terminal. I suspect that some of the solder I put on the point during attaching my Arduino jumper managed to go inside the terminal, which fixed the intermittent contact.

Therefore, in conclusion, I think that the wire was loose, but still stuck inside the solder terminal, and the intermittence of the 5V supply was caused by vibrations when I pressed the buttons in quick succession etc...I will play it for some longer, but I am pretty confident that this problem may have been fixed with the new wire.

Beogram 8000: Test Drive with Miles Davis

After performing the mechanical and electronic adjustments of the Beogram 8000 that gave me some fun insight into aging transistors recently when I troubleshooted its record detection circuit, it was time to reassemble the unit and do a test drive:

I selected "Water Babies" by Miles Davis. One of my favorite records of all time. In fact I bought this one in 1979 as my 29th record (as you can see from the number label in the top left corner...;-)! This Beogram 8000 came with a decent MMC20CL cartridge and Miles sounded very smooth! It seems this Beogram is working well again. I will play it for a few more days and then it will be time to send it home to California.

Beogram 8000: A Fun Evening With the Record Detection Circuit (the Conclusion)

This is a follow up to this post. After I made the post about the failed transistors in the record detection circuit of a Beogram 8000, I received this forwarded input from the owner of the 8000 who has a friend who understands circuits:

“I think he [the Beolover] replaced every transistor he came across until he got to tr16.  Then he measured it, it was OK, so he didn’t replace it.  I think he should replace TR16.  The reason is TR16 operates with 6 ua of base current, maybe his curve tracer is a different current.  Clearly he has the right waveform going into TR16, and noise is still coming out.  With the beta he measured this should not be happening, so something is wrong with the measurement.  This thing where all the transistors wear out is disturbing.  Of course this circuit calls for very high beta in every stage, but it shows us how transient our technology is.  A transistor is only a transistor because of the sandwich of 3 different types of impurities, which we know move around. After 40 years, it seems they move around a lot.  Clearly these parts worked at one time, but they all changed enough to stop working.  It’s rare to see something this old with good enough documentation and people willing to keep it going.”

Point taken. I also wondered why this circuit was still not working properly even with TR16 measuring o.k., and I have no idea what exactly my little transistor tester does in terms of applied voltages etc...when it does its magic.

Only one way to find out: I restored R72 to 1M and put R70 back into the circuit, and then replaced TR16 with a 2N2222 that yielded an hfE of 225 on my transistor tester. It still did not detect the absence of a record. So I rummaged through my parts collection and finally found a new BC547B that yielded an hfE of 430, and put it into the circuit, and that fixed it:

The yellow trace is the measurement point between R72 and R71, and the green trace is HH (collector of TR16). This qualifies as a good logic LOW, and the micro controller apparently thought so, too. It swept the arms across the platter in search of smaller records, and then returned home and turned off, as it should!

So the conclusion is, yes, the circuit can work as designed (indeed, a satisfying result!...;-)

One question remains: Why does TR16 need such a high hfE? Sure, the base current is pretty low due to the 1M resistor. In this particular case, the minimum voltage in the sawtooth at R71/72 is about 6V, i.e. the minimum base current into TR16 should be about 5.3uA. The current across CE should be limited to 5V/33kOhm=151uA if the transistor is fully on. That is only about a factor 28, i.e. a hfE of 225 should be plenty to pull the collector of TR16 down with 5.3uA.

This got me motivated to look a bit deeper into measuring hfE values. It turns out they are strongly collector current dependent. This figure is taken from the BC547B datasheet provided by ONsemi:

So we see that at the low 151uA current that TR16 sees, the current gain is only about 60% of the maximum value at about 20mA. The 2N2222 has a similar curve (see here), and that (together with the strong variation of hfE values in general between individual transistors of the same series) may explain why the 2N2222 was not able to perform properly in this circuit. 

Beogram 8000: A Fun Evening With the Record Detection Circuit

*************************This post has a follow-up. Please, see here*****************************

I recently started working on a Beogram 8000 from California. As the usual, as a first step, I replaced the electrolytic capacitors and the motor AC cap in the power block, then re-flowed the solder points of the headers, after which it was time to give the unit a first try. I pressed start and the carriage started moving, and without a record on the platter the unit activated the tonearm lift at the LP setdown point. Not what it was supposed to do. I concluded there was an issue with the record detection circuit. 

This is the relevant part of the Beogram 8000 circuit diagram:

I should point out that the Beogram 8000 featured here did not have the trimmer R64 at the sensor input, i.e. is one of the earlier models.

The first spot to check is the collector of TR14, which amplifies the signal of the sensor 5PH1 as the platter ribs pass through under the sensor arm. I drove the arm manually further onto the platter to get a stable signal, and then put the oscilloscope probe to the collector. This is the signal I measured:

The dips correspond to the ribs passing under the sensor and blocking the light reflection. Unfortunately, there is no prescribed wave form in the manual, but judging from the very similar circuit in the 4002/4 models this signal was way too low.

the next step was to measure the sensor signal directly at C15, which yielded about the waveform shown in the manual:

Therefore, I suspected that TR14 had a too low gain (hfE), which is often encountered in the Beogram 4002/4 models. I extracted and measured it:

It only showed with a gain of 94, which is well below the minimum gain of 200 specified for the BC547B. I replaced it with a 2N2222 that demonstrated a gain of 215, and measured again:

Much better. The dips go to zero and the amplitude is almost 10V. Unfortunately, there is no spec in the manual for this measurement, but judging from the 4002s the above qualifies as a decent signal. I hoped at this point this would fix the issue. But no cigar. The Beogram again dropped the arm at the LP setdown point in absence of a record. 

If the sensor signal is good, then the next measurement spot is at the junction between R71 and R72. I got the yellow curve in the screen below:

The triangles peak at about 7V and dip down to about 3V, in contrast to the specified 10-to-7V drop shown in the manual. This means there is a too small voltage at the base of TR16, which may mean trouble pulling the 'Disc Detector' signal (HH) properly to a logic LOW state below 0.2V.

I measured HH and that is the green curve in the graph above. At this point I though, bingo, there must also be an issue with TR15 that does the tugging at the base of TR16. I extracted it for measurement:

It came up with an hfE of 158, which is also a bit low. I replaced it with a new BC557B with an hfE of 275, above the minimum 200 spec. So far so good. I measured again at R71/72 and now the curve was according to specification:

Unfortunately, record detection still did not work, and the green curve was a bit better, but the spikes, while a bit weaker, were still present. Ok, I thought, TR16 must also be out of spec. I measured:

...and got a happy hfE of 434 (it is normal that the hfE varies a lot, and all is good as long as it is above its minimum spec, at least for most circuits.

To make sure that the chain of command was intact between TR16 and the micro controller input I grounded the collector of TR16, which should permanently disable record detection, and it should always behave as if there were no record. I pressed Play, and indeed, the arms ran across the platter in search of a smaller record, and then turned around and went back home. So that was working.

At this point I was a bit at a loss. For some reason, the pull at the base of TR16 was not strong enough, and it did not fully turn on. A mystery!

I decided to play a bit with the circuit and I replaced R72 with a 100k resistor, increasing the current by a factor 10 into the base of TR16 (or so I thought). This did not fix the issue. It turns out that the higher current was depleting C19 too quickly, reducing its voltage too fast between the passing of the platter ribs, again causing spikes in HH. This meant I needed to increase the current into C19 also, which was easily done by removing R70, which partially drains the collector of TR15 away from C19. After eliminating R70 from the circuit, it started working properly with HH going fully to zero when the ribs were visible to the sensor:

Luckily, HH also quickly restored itself to 5V when covering the sensor (simulating a record), i.e. the record detection mechanism of this Beogram 8000 is doing what it is supposed to be doing.

At this point, I am not fully satisfied with this result since I do not fully understand what is going on. I guess another 8000 where this circuit works 'as is' will need to show me the light some other day. The B&O excitement just never ends!...;-)