Input needed on Tutor Pro System (Super Pro Tutorvision)

[+DZ-Jay]

1 hour ago, Zendocon said:

I know about the STIC 1A - and I even wrote Little Man Computer to autodetect it - but I didn't know there was also a CP1610A.  What were the differences?

I believe Mr. Ives also touched on this in one of his posted.  From what I remember, one is encased in ceramic, the other in … plastic?

 

He said that other than that, they were identical, and that the enclosing material allowed for it to be over-clocked, so they rated it at a higher clock rate.

 

I’ll see if I can find Mr. Ives’ comments.

 

   dZ.

[+DZ-Jay]

Here it is:

The relevant portion:

Quote

    In 1982 GI began laying out the 5-volt only but otherwise functionality identical CP1610A as part of the Big Mac/Intellivision II cost-reduction effort. That device could easily run at 5 MHz without a heat sink, and probably faster, although there was norequirement at the time to qualify the part for greater speed. If you really want to know how fast it canactually run you can obtain one by cannibalizing one of your INTV 88 boards.

dZ.

[Walter Ives]

On 7/17/2023 at 8:07 AM, Zendocon said:

I didn't know there was also a CP1610A.  What were the differences?

The CP1600 was the original device designed by GI in partnership with Honeywell which Honeywell wanted to use as less-expensive PDP-11 replacement as part of its distributed control system concept. It was fabricated using GI's n-channel nitride process and required three power supply voltages, VDD (+12V), VCC (+5V) and VBB (-3V). The device was designed to be clocked at 5 MHz, but required special attention to be given to cooling to run at that speed and so was usually run slower. The part was packaged in ceramic. The first chip sets delivered to Mattel included CP1600's in ceramic packages.

The CP1610 was the exact same die (the die is the little piece of silicon inside the package, the part consists of the die enclosed in a package), packaged in plastic. Plastic packages are cheaper. It could in principle be clocked at the same speed as the CP1600, again as long as you arranged to get rid of the heat.

The CP1610A was silicon-gate ground-up reimplementation of the CP1600/CP1610. It required 5V only; otherwise the die was supposed to be a pad-for-pad replacement for the original. By pad I mean the tiny contact point on a die, not the package pins. When you package a die, you glue it to the center of a frame that contains pins and wire-bond the die pads to the parts of the frame that connect to the pins. You don't necessarily have to "bond out" unused signals. A particular die design can potentially have several bonding options.

Every rectangular "chip" of silicon is called a die. Dice are manufactured on silicon wafers, which were about three or four inches in diameter at the time. The CP1610A used smaller transistors then the CP1610 and hence had a smaller die size, which means that you could fit more of them on a wafer. If you can fit more die on a wafer, the per-die cost drops. Secondly, every wafer has lattice defects, and dice that span a lattice defect won't work. The smaller the die, the less likely it is to straddle a lattice defect. So decreasing die size also increases yields. As you increase yields, the per-die cost of working dice drops. Since it costs a certain amount to process a wafer, increasing yields decreases the average device cost. Thirdly, smaller transistors use less power. Since here the function of the device remained the same the number of transistors didn't change, so the CP1610A uses a lot less power than the CP1600/CP1610. That means the CPU didn't need a heat sink, which further decreased the installed cost.

The original Intellivision didn't use all 40 signals available on the CP1610. The single power supply design eliminated the need for two more: VDD (+12V) and VBB (-3V). Standard 0.5"-wide DIP (Dual in-line package) sizes were 48, 40, 28 and 24 pins; this brought the number of signals needed down to where the part could be packaged one standard package-size down. The cheaper 28-pin package CP1610A was the only bonding variant actually released; it brought out only those signals actually used by the Intellivision.

The CP1600 was introduced in 1975. By 1982 technology had advanced considerably, and to remain competitive Mattel commissioned GI to produce the CP1610A as part of a cost-reduction effort. STIC 1A, designed by APh and produced by Toshiba, was the other part of the cost-reduced chip set. GI had working CP1610A samples in 1983, but by that time Mattel had so much unsold chip inventory on hand there was no reason to go into production. When Valeski finally started to run out of his chip inventory he went to Microchip, a company that had arisen Phoenix-like from the remains of General Instrument Microelectronics, which blew the ashes off the old masks and made him a few parts.

Valeski ran out of CPUs before he ran out of graphics chipsets and produced a few Master Components that paired the new CP1610A with the old graphics chipsets, hence the INTV 1987 board. How's that for compatibility! Engineers are always pleased when things like that work out.

When Valeski ran out of the old graphics chip sets he had to go to Toshiba to see if they too could locate their old design and fabricate some STIC 1A parts. They could and did. The parts didn't quite work right—they really needed one more design iteration. But APh suggested workarounds that required a few extra parts. Valeski was spending his own money and didn’t want to pay for another iteration of chip design, and so elected to include the parts for the workaround on the boards. This resulted in the INTV 1988 board that uses both the CP1610A and the STIC 1A, which many on this board like to call TutorVision boards, even though that's retro-naming (at least for those boards that didn't also use the WB Exec, but let's not get into those weeds here.)

So I guess the real answer to your question, "What were the differences" should be, "Other than the price, you're not supposed to be able to tell."

But it turns out that there is one very important difference that you can tell: the CP1610A could be clocked at twice the speed used in the original Intellivision and still maintain a reasonable temperature. It was slated to be paired with STIC 1B to create an enhanced Master Component that, among other things, could run at twice the speed of the original. The system was Code-named Aphix, the project to create a product based on APhix was code-named Coffee and the whole caboodle was on track to create what was teased as Intellivision III at the January 1983 Winter CES in Las Vegas.

The CP1610A was not supposed to be the end of the line. A specification for a next generation part, dubbed the CP16000, had been prepared and a layout team recruited. The CP16000 used a wider instruction word, as was anticipated in the original CP1600 specification. If you look at early documents you will see that the high bits of the CP1600 instruction word were specified to be zero, with other values "reserved for future expansion." That was a warning not to put garbage there. The CP16000 was made backward compatible so future systems would be able to run the original Intellivision cartridges. There was also a specification for an enhanced graphics processor, the GP16000, which was backward compatible with STIC, STIC 1A and STIC 1B.

WJI