From the Fountainhead - May 1977

from the May 1977 issue of Interface Age magazine

by Adam Osborne

Now that the Intel 8085 and 8048 are here, IMSAI is going for these two products; IMSAI is not planning to offer a Z-80 CPU card. I wonder why the Z-80 and the Intel products have to be mutually exclusive?

I have had a chance to examine the 8085 and the 8048— and I believe some intriguing user patterns are likely to emerge.

Let us begin by comparing the 8085 with the Z-80, both of which purport to be the “next generation” 8080-A.

These are the two major faults of the 8080-A:

a) The 8080-A is really a threechip CPU, consisting of the 8080-A CPU, the 8224 clock generator and the 8228 system bus controller.

b) The 8080-A requires three power supplies: +5V, + 12V and -5V.

Both the 8085 and Z-80 have eliminated these two problems; CPU logic has been reduced to a single chip which requires a single +5V power supply.

Now to you as a hobbyist, these two 8080-A problems may seem a bit abstruse, but chips and power supplies both cost money; these two 8080-A problems have resulted in higher cost 8080-A CPU cards.

But while eliminating two problems, the 8085 seems to have introduced a new one — multiplexed data and address busses.

The 8085 instruction set is almost identical to that of the 8080-A; the 8085 has just two new instructions. The Z-80 instruction set, as all you hobbyists know, is about twice the size of the 8080-A instruction set. Instruction sets are very important to hobbyists, but not very important to large commercial users who often pay more attention to chip counts and part costs.

Along with its enhanced instruction set, the Z-80 contains additional registers within its CPU. The 8085 registers are identical to the 8080-A’s.

The Z-80 provides logic to refresh dynamic memories which the 8085 does not do. What this means is that it is cheaper to interface large, low cost memories to a Z-80 than it is to an 8085.

The Z-80 has three modes in which its interrupt logic may operate, but it has just two interrupt request pins. The 8085 has five interrupt request pins, three of which generate their own dedicated Restart instructions.

The 8085 has a primitive serial input and output capability, which the Z-80 does not have.

Now we could go on comparing “features”, since that is very subjective. More often than not advertised features are nothing short of marketing propaganda. In fact, I define a “feature” as a design error which the manufacturers could not get out in time, so they gave it to marketing to dress up as an asset.

But what is far more important is that I believe I can identify key aspects of the 8085 versus the Z-80 which are going to direct the two products into somewhat different markets.

The Z-80 has an instruction set which people who do much programming appreciate; it also has signals and internal characteristics which make it more attractive than the 8085 in large microcomputer systems. For example, dynamic memory is cheaper than static memory, but that only becomes a significant cost factor in large memories.

However, the 8085 has a secret trump card — at least for the moment — its support devices, the 8155, the 8355 and the 8755.

The 8155 provides 256 bytes of read/write memory, three I/O ports and a programmable time, all on a single chip. The 8355 provides 2048 bytes of ROM and two I/O ports, on a single chip. The 8755 is a variation of the 8355 having erasable PROM. Now you can put together some rather interesting two-chip and threechip systems based on the 8085, the 8155, the 8355 and the 8755. Putting together similar systems using the Z-80 would require considerably more logic and expense. The 8155, 8355 and 8755 internally demultiplex the 8085 data and address busses which you must demultiplex externally if you’re going to have larger 8085-based systems — at which time the Z-80 will generate lower chip counts and costs.

In summary, I believe the 8085 will have difficulty dislodging established Z-80 users, and there are many of them, particularly among hobbyists. The 8085 will also have a hard time competing with the Z-80 in large microcomputer systems, or programming intensive microcomputer applications. But the two-chip and three-chip 8085 configurations are going to look very attractive to commercial users who are interested in large volumes and low chip counts and cannot get by with the one-chip microcomputers now appearing on the market (the Fairchild 3859, the Mostek 3870 and the Intel 8048). Since hobbyists are not part of the high volume, price sensitive market, what this means is that the 8085 will be having a hard time competing with the Z-80 for new hobby market business. Of course, manufacturers now using the 8080-A will probably switch to the 8085.

But when will one of you hobbyists come up with a Z-80 assembler that uses 8080-A mnemonics?

What about the 8048? The 8048 is Intel’s answer to the one-chip microcomputers being manufactured by Fairchild (the 3859) and Mostek (the 3870). IMSAI is the first manufacturer to produce an 8048 microcomputer card for the hobbyist. As a programmer you will find the 8048 highly restricting in its one-chip form. This one chip gives you a total of 64 bytes of data memory and 1024 bytes of program memory. The 8748 gives you an EPROM, but at a price which rivals a multichip Z-80 microcomputer configuration.

If you take the 8048 and expand it into a multichip microcomputer, with a quantity of external program memory and data memory, then you finish up with a microcomputer system that costs the same as a Z-80, or an 8085, but is much harder to use; and that makes no sense.

The 8048 is not really a programmer’s product; it is not well suited to the average hobbyist’s needs. But this is a microcomputer which you will love if you are building dedicated controllers. If you use

microcomputers as instrument controllers in a laboratory, if you are a ham radio operator who wants to use microcomputers for dedicated control functions, even if you’re running your model railroad at home with a microcomputer, then you will love the 8048 — or most specifically the 8748 for its simplicity and ease of configuration. If you are using the 8048 as a dedicated controller, you will likely want analog-to-digital and digital-to-analog converters. Check with Precision Monolithic Inc. (408246-9222) for some interesting new one-chip AID and D/A converters.

And now for a few tidbits from other manufacturers.

National Semiconductor has been amazed at the response the SC/MP keyboard kit system has generated among hobbyists. Apparently hobbyists account for the bulk of keyboard kit customers. National now has a 4K basic available for SC/MP; which they call “Nibble”. But National does not yet have a real hobby microcomputer system. Despite rumors that National Semiconductor is imminently going to enter the hobby market, no decision to that effect has yet been made.

One of the most interesting chips due later this year is the 9511 arithmetic processing unit from Advanced Micro Devices. This chip will give microprocessors high speed arithmetic comparable to large minicomputers. The 9511 makes multidigit multiplication and division and floating point arithmetic cheap and practice. I predict it will be one of the hottest chips of 1978.

Two more hot tips: Gary Killdall, who goes by the name “Digital Research” and works in Monterey, California (408) 373-3403 produces the best software we have seen yet. Mr. Morro of Morro’s Microstuff has been designing some very good S100 boards.

In my first article I explained that I had selected the name “From the Fountainhead” since the area from which I am reporting contains most of the manufacturers and important companies in the microcomputer industry. Manny Lemas of Microcomputer Associates was dismayed when his company and its products — the Jolt microcomputer system and “Microcomputer Digest” magazine — did not appear in my list of company names. I forgot to mention you Manny, so here is the mention.

And in case you want to be mentioned, either for your products or your opinion, please call me. My telephone number is (415) 548-2805.