Thursday, November 02, 2017

OB-Sx Voice Cards

This one came in on the AH list from Tony of Oakley Sound. The AH list is archived and public so I thought I would capture it here in the MATRIXSYNTH archives as well, and of course to share it for anyone not on AH that might be interested.

"A long somewhat geeky post...

Oberheim had a history of creating vast numbers of 'engineering change orders' (ECO). Essentially changing their designs during the lifetime of the product, ECOs are notes for official modifications done by the factory during manufacture or authorised repair technicians in the field. ECOs are useful tools to allow the manufacturers to improve on their original design, perhaps because of updated specifications, new found parts or deal with component obsolescence. However, quite often they are used to fix things that should have done properly in the first place.

There must have been many ECOs regarding the Oberheim OB-Sx but not all of them are available online. I have an original service manual but the ones in there are the same as these:

In addition to these this is what I've found out about the voice cards on the OB-Sx.

1. There are at least two types of voice card. The primary difference appears to be the addition of a -15V supply. This requires a different voice motherboard as the power is now admitted via a 5 way connection and not the original 4 way one. The -15V seems to power only the post VCA op-amp. I've not come across any clipping issues in the post VCA amplifier in early cards so I'm not sure what the reason is.

I think newer cards can be used in OB-Sx units with older motherboards simply by linking the two outermost pins of the 5-way connector that takes the power onto the card. Any new card will then behave as an older one.

2. The early voice card was originally designed to use the CEM3340's linear FM input for the 'cross mod' feature. All the older voice cards I have seen have a mod whereby one of the tracks is cut and a bit of wire added to link the cross mod path to the exponential FM input. This was corrected on the newer boards with a different track layout so the modification is not required.

3. All the early voice cards I have seen have an additional 100pF ceramic capacitor mounted across one of the two diodes in the feedback path of the VCF. The capacitor was added to the board layout in the later version of the card.

4. All of the early voice cards I have seen also have an additional 330pF ceramic capacitor connecting pin 11 of the CEM3320 to 0V. This appears to make the unused and externally unconnected gain cell in the 3320 behave the same as other two in the SVF. Why I'm not sure but it could be to prevent the unused cell from oscillating and bleeding through into the audio. The 3340's datasheet seems to be happy leaving it off. This additional capacitor was added to the board layout in the later version of the card.

5. I have seen one early version of the card, dated August 1982, where another modification is added to the underside of the board. This connects pin 8 of the CEM3320 to 0V with a 1nF ceramic capacitor. This does not seem to have been applied to the second version of the card so it is possibly fixing a problem only detected after the second voice card was designed, or perhaps it only affects the early cards. It is probably a problem caused by the physical layout of the PCB.

On preset B1 and with the cut off set to full the problem can be clearly heard, but it is present on other sounds too. At high cut-off frequencies and at low resonance the CEM3320 in the OB-Sx oscillates in nasty way and creates a unpleasant roughness to the sound. Adding the 1nF capacitor prevents the unwanted oscillation. However, I have found that one needs only 56pF to work rather than the 1nF Oberheim used. Adding this capacitor does have a slight knock on affect on the sound even when the filter is not unpleasantly oscillating - it now sounds slightly more polite. The pros outweigh any of the cons in my opinion.

What is interesting is that the same filter circuit is used in the OB-Xa and no such capacitor is required. I guess then that it is PCB layout and not the circuit itself that is misbehaving. It may also be dependant on the batch of CEM3320 used.

6. Later versions of the early card and all the later cards I have seen have another possible layout correction. The tracks going to the 470R/10nF series combination on the CEM3340's linear FM input on each VCO have been cut. A clumsy bodge on the top side of the board then recreates the required series combination but this time physically closer to pin 13 of the CEM3340. I can only guess that this problem was found only after the later card was designed as it is not corrected there. That the mod appears on both many early and all later versions of the voice card means that Oberheim were clearly producing both cards at the same time.

7. The later version of the voice card has an extra trimmer used to set the maximum resonance of the filter. On the older board trimming is set by swapping out resistors.


No comments:

Post a Comment


©2018 Matrixsynth - All posts are presented here for informative, historical and educative purposes as applicable within fair use.
privacy policy