Wednesday, 4 January 2012

Reengineering alee650's SID Symphony II clone


For Christmas I got a SID Symphony II cartridge, at least what I thought was a SID Symphony II cartridge, being sold by an ebay seller by the name of alee650.

http://www.ebay.co.uk/itm/HOLIDAY-OFFER-Commodore-64-128-SID-Symphony-II-6581-Cartridge-/310366044461?pt=UK_VintageComputing_RL&hash=item484342012d

In fact these are clones, not of the final design but of an earlier pre-production design and unfortunately there are errors in the design as produced by alee650.

The genuine SID Symphony II cartridge can be seen in the following True Chip Till Death article.

http://truechiptilldeath.com/blog/2011/10/10/sid-symphony-ii-cartridge-appeared/

First off, let me state that the basic SID function of the clone cartridge does work. What doesn't work is the access LED and there are problems with cooling the SID chip in the clone mounted in a transparent cartridge case.

The designer of the SID Symphony II, Vanessa Ezekowitz, was kind enough to furnish me with circuit diagrams of her SID expansions. There are internal and external expansions available, from her website Digital Audio Concepts. (back online after a recent SOPA protest)

http://digitalaudioconcepts.com/

The following diagram shows the difference between the access LED circuits of the alee650 clone and the genuine Digital Audio Concepts cartridge.



There is an additional problem with the alee650 clone, which is that the PNP transistor (2N3906) is inserted the wrong way round. The silkscreen on the PCB shows the transistor with the flat side on the top, i.e. turned away from the edge connector. In fact the transistor needs to be turned the other way, with the flat towards the edge connector. I confirmed this by studying the circuit diagram, the PCB layout and the transistor datasheet, which is available here.

http://www.fairchildsemi.com/ds/2N/2N3906.pdf

Once this is turned round, the access LED starts to light although extremely faintly. A 1uF capacitor should be added between the collector of the transistor and ground. This stores enough charge to keep the LED lit for approximately 1ms (based on an RC time constant of 1K * 1uF) after each access. 6502 write cycles to the DExx area last half of one clock cycle (500ns) and there will be several of these accesses each frame (16ms or 20ms) for a stereo tune but not enough to make the LED visible without the capacitor.

At this point my cartridge basically worked as specified, but I was having some problems with dodgy 1541 disks so for the avoidance of doubt I decided to implement the rest of Vanessa's changes too. She reported instability with the original 2.2K base resistor and it went away with a 4.7K. It is possible the 2.2K could draw too much current from the 74HCT688 address decoder chip, and thus the chip select line to the SID might not be recognised as logic 0 during an access. The timing of this signal is fairly critical; it goes low on the processor's half of the clock cycle during an access to the selected address range. Swapping this for a 4.7K resistor reduces the base current to below 1mA which should be well within HCT specs.

The LED is not very bright with a 1K collector resistor, so to make it brighter Vanessa replaced it with a 47 ohm resistor. Myself, I thought 47 ohms was rather low for a current limiting resistor on an LED so I decided to take my inspiration from the C64 itself. When the cartridge is plugged into a standard breadbin C64 there is one LED on the cartridge and another for the C64's power light. I decided these should be of similar brightness when the cartridge is being accessed. The C64 uses a red LED and a 390 ohm current limiting resistor connected in series between 5V and GND. alee650's SID Symphony II clone uses a yellow LED so to compensate for a higher forward voltage drop I chose 330 ohm. Replacing the 1K collector resistor with 330 ohm made the LED significantly brighter, and you could now also see the LED pulsing slightly depending on the access pattern of the stereo SID being played.

The final problem with the alee650 clone is that of heat. The SID chip generates a large amount of heat and needs at least passive convection cooling not to run too hot. Some C64s have metal shields above the PCB with tabs contacting various chips including the SID. Others leave it unheatsinked but there is at least some space above the chips allowing convection cooling. alee650's clone mounts the cartridge in a transparent plastic cartridge case; very sturdy and convenient but unfortunately lousy for cooling. The chip is in direct contact with the top of the cartridge case when the two halves are screwed together; there is absolutely no space for convection cooling at all. On my clone there is now a melted spot of plastic on the top half of the case directly above the SID chip die (in the centre of the chip) and I only ran it in the cartridge case for an hour or two before noticing this.

So to preserve the life of my valuable working SID chips and avoid more melted plastic I decided to run the SID Symphony II clone as a bare PCB rather than in the cartridge case. This required unsoldering the signal wire of the RCA phono connector, taking the connector out of the case and soldering the wire back on again. The final pictures show the finished PCB with a 6581 2384 and all mods fitted.




Oh yes, and as everyone knows, the Commodore 64 turns 30 this year. Here's an article from a UK perspective.

http://www.reghardware.com/2012/01/02/commodore_64_30_birthday/

7 comments:

Shareware Plus said...

Thank you for reviewing the product I made and I have taken onboard your corrections & suggestions. I will ensure these are done for all future sales.

I would just like to comment that I produced this product from the Digital Concept website before Vanessa put any restrictions on the design or use. The latest version is (was) available on the new Digital Concept webstore.

Best wishes - Tim

Richard Atkinson said...

Hi! Thank you for the cartridge! I think the main changes needed are:

1. Rotate 2N3906 transistor 180 degrees.

2. Add 1uF capacitor between collector and GND.

3. Change 1K collector resistor to 330 ohm.

I'm not convinced the 2.2K base resistor needs to be changed for 4.7K. Vanessa says it solved some instability, but a 74HCT688 should be well capable of sinking 2mA. The datasheet I looked at said it was capable of sinking or sourcing up to 25mA.

I'll put the 2.2K resistor back in and do some more testing.

Richard Atkinson said...

I put the 2.2K resistor back in and it's running fine so I don't think there's any need to replace these. Should be better than 4.7K because it provides more base current to turn the PNP transistor on - i.e. the LED can shine brighter and with less flickering.

Shareware Plus said...

I have posted you a PCB mount Phono connectror so you can do away with the wires. Should look alot better!

Richard Atkinson said...

PCB mount phono connector arrived this morning, have fitted it and it looks great! Much more sturdy. Now just need to finish the work on that jumperless internal board.

Bieno64 said...

I've the same card and possibly the same problem. Can you put some higher quality pics?
Thanks.

albertspinner said...

The great thing which inspire me in this blog is the way of sharing some of the useful and usual facts, so I really like it, thanks for sharing.

Flexible PCB