Date: Tue, 7 Jan 1997 02:09:56 +0000 To: analogue@hyperreal.com From: "Robin Whittle" Subject: Re: 606 Memory The TR-606 memory is comprised of two 1 k by four bit CMOS static RAM chips. They are powered via either of two diodes: 1 - from the four C-cells. 2 - from the main regulated power supply 6 volts - which is after a regulator which is after the power switch which is after the power jack which switches between external power and the C-cells. So if there are no batteries installed, and there is a power adaptor plugged in, then the memory chips will only get power if the power switch is on - and hence the machine is running. There is a 100 uF capacitor across the power supply pins of the RAM chips. How long this lasts in the absence of battery or external power is a function of: 1 - Reverse leakage in the two diodes. 2 - Leakage in the capacitor and the circuit boards. 3 - The current drained by the RAM chips. 4 - At what voltage the RAM chips start to corrupt their stored data. All these will be affected by temperature and the phase of the moon. There were at least two varieties of RAM chip used: NEC and Mitsubishi. I think there were Toshiba chips too. All these use a six transistor memory cell - two to connect the flip-flop to the pair of lines for read/write, and four to form the flip-flop. These four transistor flip-flops, with two N- and two P- channel transistors in each, have a nearly zero current drain, except when switching state. There is never a path for current to flow from ground to the power rail - except for leakage. Later chips, such as most 6116s, 6264s, 62256s 628128s etc, used a two transistor two resistor flip-flop where one or the other resistor was connected across the power voltage at all times. Even with a million such resistors drawing current, you can still get a 628128 which draws just a micro-amp or so on standby, so these resistors must be 5 million megohms or so . . Now seriously off topic. These CMOS RAM chips draw only leakage current, which will vary from one kind of device to the next, and from one chip to the next. However in my experience it is always very much less than a micro-amp. You would need a pico-amp meter to measure the current drawn by these little babies. So if someone told me that their 606 stored data for several months - and I think they have - then I would not be surprised. If one only stored it for a few hours, then I wouldn't be surprised either. If the chips did draw a pico-amp between them, then they would probably last for a few days. What data the chips retain depends on how low the capacitor voltage gets and so how much of the original data remains. What data is in the chips when they power up seems to vary enormously from one type of chip to another. Some types of chips may come up with a particular pattern, others may come up "randomly" where "random" is actually generally a different pattern for each chip. Maybe some cells are so finely balanced that the state they come up with is affected by the presence of just a handful of electrons. To get the voltage down to 0, open the machine, remove batteries and power supply, and short pin 18 of one of the RAM chips to ground. As far as I know, the original pattern data was written into the chips in the factory - maybe they put a clip over the CPU pins, put the CPU to sleep and exercised its pins to write the data. I don't think that the 4 bit CPUs had enough ROM to store a complete set of 1 or 1.5 k bytes of pattern and song data. I liberated a TB-303 recently from its sleepy mountain hideaway - it had always had batteries in it and still has some original patterns. I feel like an archaeologist and will make a cassette of the patterns. The TB-303 is the same as the TR-606, except for having three RAM chips. Maybe you can short out the RAM supply on your TB-303, pray to appropriate deities, wait for their direction on where to place your machine and when to apply power. This divine assistance will enable you your machine to be subjected to the most auspicious set of cosmic rays, which will to disturb the distribution of electrons in individual RAM cells to cause truly cosmic patterns to result. - Robin <<< See my WWW site for the Devil Fish modifications to the TB-303 >>> . Robin Whittle . . http://www.ozemail.com.au/~firstpr firstpr@ozemail.com.au . . 11 Miller St. Heidelberg Heights 3081 Melbourne Australia . . Ph +61-3-9459-2889 Fax +61-3-9458-1736 . . Consumer advocacy in telecommunications, especially privacy . . . . First Principles - Research and expression - music, . . music industry, telecommunications . . human factors in technology adoption. . . . Real World Interfaces - Hardware and software, especially . . for music .