Computer Corner by Bill Kibler

As I was going through my old mail, I came across a couple of catalogs from DYNACOMP. They had been sent to me some time back, one with a note on it saying "your readers might like to know that all the programs in this catalog are still available."

So I went through the catalog #29 and the newer one #38. The old one did have lots of CP/M, NorthStar, Apple, Atari, TRS80, and just about anything (even 8") listed. The newest was mostly MSDOS and MAC, plus a few Apple programs. I noticed they carry the Toolworks C/80 Version 3.1 for $49.95. That got me curious if this was true and so I called the 800-828-6772 number and asked. Sure enough they are still able and willing to pull out anything from the warehouse and ship, including NorthStar. The person said he still uses his Altair. I quickly sent them a complementary copy with ad rates and figured everyone should call them and get one of their old catalogs.

Now some of the old programs are public domain, but they also have many of their own items as well. They may be one of the few remaining original dealers still selling programs. They do have a little bit of everything and so if your looking for programs to run on the old machines here they are. I also noticed they have the CP/M collections, but what was unusual was the Atari, Antic, C64, and Piconet libraries. I am not sure anybody else is still offering those collections. What a find.

PLC Hacking

I have been testing PLC transfers and operations, mainly with a GE 90-30. This is their small end unit and works ok. I say ok, since their performance at moving data into and out of their ladder area is a bit slow. What our application needs is to be able to move a fairly large amount of data into the ladder from a non-standard remote platform.

Now all PLC vendors have some method or other for doing this. We used the Omron version, and have looked at the PLC-Direct as well. My analysis so far is the "Basic" expansion cards all seems to have the same problem, slow. What the problem appears to be is the way in which they talk to the regular ladders memory.

I assume the designers wanted a pretty secure and crash proof design and as such have chosen to use some form of serial communications on the back plane to transfer data. If the back plane was memory mapped and thus any application could read or write the memory location, errant programs might turn all outputs on and thus cause major problems. A slower more difficult process probably seemed safer and easier to limit.

Those limits also make applications like ours almost impossible. A number of Standard Bus vendors have PLC expansion cards that do p73 sit on the memory Bus and from what I gather they don't seem to have any problems. Understand that these applications have several CPU's with each running their own program. Having one of the CPU's go astray is possible, but would normally be found in the startup and development stage.

To me the problem is more of the proprietary concept that many vendors have about their products. If too much about how it works is known, others would make cards and products that might fit into and on their product. The idea is lost sales, but in our case that design limit might mean we look elsewhere for the complete system.

PC Invasion

What I am starting to see is the invasion of the PC based systems into the PLC market. No longer is a manufacturer wanting just a standalone PLC, instead they want PLC functions in a member of the entire computing network of machines. The idea is being pushed by just in time production, where only what is needed is made and shipped as requested. The tying of machines together helps sales requests become actual PLC ladder changes. But that requires PLCs to talk and understand accounting data.

The answer has been putting PLC adapter cards into PC machines all talking over a common RAM area. A network based program talks to the accounting program and puts requests into the PLC's RAM area. The ladder sees these requests and knows to produce ten RED widgets next as they were just ordered. If the PLC was controlling the paint robot all is well, if the PLC was actually a machining station then you got problems.

All humor aside, the idea is not new, but just that the hardware is catching up to the desires of the companies. The power of PC's and cost of interfacing is dropped so that very powerful and cost effective solutions are now possible. Refinement and perfection of the software is what is needed next. Our application in fact might be helped by these larger pushes. Since our clients are a few years behind cutting edge, our options are limited. The more they see others having successful projects, the better our chances are for using other designs, like PC based PC104's.

PC104

PC104 format is really starting to take off. I haven't had a chance to talk to Z-World and see how their Z180 based PC104 product is doing, but I hope well for them and us. The PC104 is a small square card that has the PC bus ISA standard interface on it, in the form of a header strip. Thus it is a mini PC clone on a very small card. All designs, programming and concepts of usage can be tested and developed using regular PC clone boxes. Then when the project is ready, it gets moved to only the type and amount of very small hardware needed to run the project.

What you gain is lowered development cost and smaller overall p73 hardware cost. The packaging can be very small and yet the power of a 486 machines is possible in a four inch cube. For me, the idea is to use the PC104 Z180 platform for some 8085 projects. I should be able to move the base code from the STD BUS platform, to the PC104 with little to no program changes. A few I/O addresses and some interrupt labels would need changing, but mostly the code would not be effected. If at some later time, the Z180 is too slow, I replace it with a 386 CPU and run my code through a 8080 to 8086 converter and try again. I suppose you could use one of the Z80 emulators on a 486, but then for a single application, code conversion seems more appropriate.

The Hooks

What this boils down to is making sure you have a well rounded understanding of PC's and platforms in general. As the next few years start to unfold, we are going to find a larger mix of platforms in use, not less. We are seeing clients who want to upgrade, but have limited funds. In years past, you might just have thrown several programmers at the project and produced a new version on a new platform.

Now the user wants just some speed assist or the number of units increased, maybe several units tied in one network. The code, platforms, and interfaces of old often took a long time to develop, programmers are gone who knew how it worked, as well as fixing the old hardware platforms is a major problem. That is where I see Z-World hooking in with their PC104 Z180s. They can save the code mostly intact and yet put it on a new hardware platform that will be around for some time.

For programmers and hardware hackers, it means having skills and knowledge about these old systems and designs, as well as understanding the new ones. I work often now with, STD BUS, talking to embedded 8051's, talking to PLC's that may be feeding summary data to PC's on a network. When a problem develops you need skills and understanding of all the processes in the system. No longer can a technician or programmer just know one programming language or hardware platform. My last work involved Tandem mainframes, talking to PC's over a LAN, that passed data to 68000's on an PC expansion bus running assembly language.

READ ON

For readers of TCJ I see what we are doing as vital to you and your projects. If you look at what other magazines are doing, they have staked out some small part of the market. We on the other hand try and focus on giving you the skills to fix any problems that might arise on any platform. I am seeing the "platform independent" term everywhere these days and feel good that we were pushing it before it became the in thing.

However we look at things, one point does surface, knowledge of PC based designs is getting very important. My PC/XT person, Frank Sergeant, has been very busy paying bills and finishing his p73 master's studies. He needs help! So I am again soliciting more articles on PC/XT projects. One project in the works is an article on "watch dog timer." The author is doing some polishing up that I suggested (mainly a little explanation of what and why you need a watch dog timer) and should be available in a later issue.

Now again I must belay all fears that TCJ will become a PC only magazine. This of course will happen if everybody drops their subscription, but assuming our readers continue paying, I have plenty of articles on small systems to keep us reading. Like this issue on embedded systems, I would like to have a few specials that do fill in the gaps about how PC/XT's work and can be used for other projects. To do this however I need your articles. Please consider however when sending articles to me that my time is very limited and I often fall very far behind in catching up with you. Things can sometimes sit for several months before I can send you a note. Please be patient with me.

Lastly on the topic of articles, a few words about what I want and hopefully what you are looking for in an article from TCJ. A main feature of TCJ articles is platform specific and independence. To explain that last sentence is to say that our readers are interested in learning about the platform (or software) in greater detail, but they also want to see how that knowledge can be used on their own platform. A good example of this has been our series on the IDE drives. Here we have talked about how it is implemented on the PC/XT platform and what the protocol is all about. Then we presented hardware for interfacing to Z80 machines which allowed others to adapt for non-Z80 machines. So it goes like this: history, theory, details, and adaption of the project.

Last Word

It seems I haven't done a "what is happening" in computers column for sometime, so this was it. To review would be to say, look for sales of bigger machines to slack off as people tend to keep what they have, more inter-hooking to develop, and some big vendors to step into the only growing part of computers - embedded control where 8051's and 68HC11's will fight it out for dominance. As usual time will tell if I am near or far from the market. Till then however, keep hacking. BDK.


Kibler Electronics, PO Box 535, Lincoln, CA 95648-0535, USA. / bill@kiblerelectronics.com
Copyright © 1995, Kibler Electronics.