Sansui 9090DB Dobly board oddities

[michael_w]

Hello, so here is a bit of a story. I picked up a Sansui 9090DB in need of servicing. It had seen prior work to the driver board and all of the output transistors had been upgraded. The driver board was recapped but all but one of the fuse resistors were original (thus needing to be replaced). However the Dolby board was all original and non functional. Only after it was recapped did I realize that the rotary switch was also part of the problem. Removing the switch to access it was difficult and, despite my best efforts, a number of the board pads lifted off. Ultimately the only solution was to replace the board with the modern upgrade (which, though tedious, ostensibly went well and all discrete components tested fine).

I now come to that rotary switch. The contacts actually looked reasonably good. What had caused the original failure was that the metal bar that turns all of the wafers had only a loose press fit into the knob and so had pulled out. There was no backing plate to keep the metal bar in place. This has now been remedied and so this failure mode will not be a problem in the future.

Now I list the three oddities, two of which still have me perplexed.

1. It was very difficult to put the Dolby board into position and screw it in. It turned out that the front plate of the rotaty switch was actually upside down. That plate has the tab which is intended to go into its alignment hole in the chassis...which it did not. There are four metal tabs anchoring the front plate to the rotary switch and so I can only surmise it was defective from the start. Adding (i.e. drilling) a second mounting hole into the chassis fixed that problem.

2. The Dolby board works and the rotary switch does as well except when in the REC 2 position. In that case the right channel recording output (TAPE 2) is completely off base (the left channel works as intended). The right channel input does not connect at all. Rather it routes the left channel input directly to the right Dolby chip without passing through the level trim pot! I really am having a hard time making sense of this.

3. The third oddity is still a work in progress. The left channel output from its Dolby chip does not match that of the right channel. At high frequency it has a larger rms value and becomes distorted. Since all the board level discrete components tested fine and I did look at the signal as it passed through both Dolby chips, the only thing I can suggest is that the Dolby chip itself is not working as intended. I've ordered a pair of replacements to test that hypothesis. At least with the new board swapping it out won't be too difficult.

So, if anyone has any ideas for the second item, I'm all ears.

Thanks.

 

[jefferylhenr]

I believe they are opamps. Check your voltages. You might have a difference in resistors that have drifted or even a switch contact mismatch. New opamps are almost always a good idea

 

[jefferylhenr]

As for the no channel. Trace it out on the schematic. I know the board is near impossible to probe while live

 

[michael_w]

Thanks for the suggestions. As I mentioned, all of the discrete components (transistors, diode, resistors, capacitors) individually tested fine (before I completed the transfer to the new board). The electrolytic caps are all new. I looked at the voltage levels between the two chips and there were some differences with respect to the dc levels of the reference (pin 13) and those of the bypass (pins 14 and 15). Those I surmised from the test circuit in the Signetics data sheet alter the gain characteristics (but I'm not very clear on the specifics). I actually tried adjusting the pin 13 reference voltage by changed the resistance to ground but that only slightly changed the distorted lineshape. At this point it seemed replacing the chip was a better option.

There aren't any opamps on my Dolby board except those internal to the NE545B Dolby chip.

Tracing out the signal path is a pain but, from what I can tell, it happens at the rotary switch. Maybe a tab has somehow become dislodged but putting it right doesn't seem a possibility. Perhaps it was a defective switch from the start given that its front plate seems to have been installed upside down (or rotated 180 deg out of position). If the buyer never really used the Dolby feature then it would have gone undetected. On occasion I've seen other factory original oddities in various components that
have crossed my bench. Not all were fixable.

 

[michael_w]

Finally had time to look into things.

After a what I term a "deep dive into the rabbit hole" the Dolby board is finally back into service. Item 2 turned out to be due to an incorrectly positioned wire (mea culpa).

However, there were more issues. The service manual calibration procedure pretty much went smoothly although there were peculiar level differences between the left and right channels of just about 20 mV (out of the stated 270 mV). In one case the left channel was too high and then the right channel was too high. More importantly the overly large distortion of the left channel was still present. I then put in two nos NE545B chips and that didn't really solve things although there were some modest differences.

Left channel: The distorted wave I referenced was still present. Another thing I noticed was that there was correspondingly a square wave output at pin 11 (from overly large gains which led to clipping of the 400 Hz sine wave) when the recording gain pot was turned up (still for reasons unknown). However, putting a 1M resistor between pin 1 (I tried a lot of things before this...) and ground somehow reset things and the chip was now behaving (but only temporarily). Switching to Dolby playback and then back to record mode caused the output from pin 11 to drop to zero. Eventually the chip would start behaving again. I then gave up on this and eventually found (again after a lot of trial and error) that adding an additional 8 nF across C23 achieved the same desired result but without the switching issue. It also slightly increased the Dolby gain on recording so that, when in playback mode, there was just under a 1 dB drop using a 10 kHz test tone when compared to the original signal without the Dolby function engaged. According to an AI query, that meets specs especially in an IC (said not to age well) some 50 years old.

Right channel: The Dolby circuit was operating but poorly. There was very little gain of a 10 kHz test tone (from 80 mV to about 105 mV) in comparison to the left channel (from 80 mV to about 130 mV). Thus there was about a 3.5 to 4 dB drop on playback. That was way too much. However adding an overly large 27 nF cap across C24 significant increased the recording level for the 10 kHz tone so that, on playback, the measured drop was only slightly larger than 1 dB. Moreover, both channels appear to have a similar response curve with respect to frequency.

As far as I have read, very few posts talk about testing the overall record/playback calibration in these Dolby boards. In fact the only one I saw was for someone that figured out how to modify a circuit for the NE545 to work with the more modern NE645 chip.

I haven't tried as yet using an actual tape deck to see how things work in "real life" but at some point I will. Before I do this there many other boards slated for a proper recap. At least the main (mostly) and the driver (completely) boards are done. Interestingly 9090DB came to me with the driver board recapped but still with all the original fuse resistors (some well out of compliance).

That's all for now.