Historical Interviews with TI employees

[+dhe]

Intro from Michael Bunyard of the Bunyard Manual Fame.

If any additional questions come up, I will bundle them up and send them off next week.

Dano

 

The TI Home Computer System

My involvement with the TI-99/4 and A was minimal, but I was working on a sister program in the same physical area that could have been the first PC.  I knew basically what was going on with the /4, though.  While our product was being evaluated by a potential customer, the TI computer division based in Houston found out what we were doing, and they decided that they were better equipped to supply this product even though their solution would cost twice as much as ours and occupy much more space.  Our potential customers were not impressed with the new TI solution to their problem, but we had been shut down.  I chose not to transfer to the new solution group, but I chose to stay in Lubbock to work in the Test Equipment Group.   I am pleased to note that we were using a text editing program (PDWF) on the larger TI systems at this time that was a significant help for writing (and preserving) specifications, etc.  PDWS was adapted to run on the TI PCs, and I continued to use it there.   In the  Test Equipment Group I built some test equipment for the /4 Production area.

The TI99-4 was designed by new hires that had no design experience, and apparently never consulted TI device literature on the 9900 uP that indicated that a short etch across the device connection was required for the two Ground pins.  It turns out that one Ground pin served the I/O buffers, and the other served the uP logic.  The former task had momentary short term current spikes, and it required special bypassing.  Additionally, there were some race problems in the DRAM design area that were never quite resolved.  This was also true for the magnetic tape READ/WRITE circuitry.  Finally, TTL compatibility for the custom devices was never specified.  Thus, the /4 and /4A were not easy to fabricate in production.

 

The /4 physical configuration was changed to another keyboard for the /4A, but that was about all.  The product was still hard to build.

The next /4 system improvement was the Expansion Box.  Our slang for the Expansion Box was the “Johnny Box”.  We named it after the Software Manager in our area.  I was in that effort from the beginning, and did most of the design work.  I slightly modified the existing P-Code and Disk Controller designs.  I started over on the RS-232 board to comply with upcoming noise requirements, and I redesigned the timing and control of the DRAM Card.  I felt we had a decent design for the Johnny Box, and it passed EMI tests.  We also designed in plans for later expansion.

Gate Arrays were just coming out in about 1981, and I was asked if I would design one for the /4A.  I had previously provided a processor design as a trial design for Jack Kilby’s group for one of the first TI gate arrays.  I said yes, and then designed a gate array to eliminate much of the discrete /4 logic.  This was the quality improved or /4QI, but there was another problem that existed and it arose.  It seems that our purchasing group had ordered enough components to last the projected life of the /4A in order to get the lowest prices.  Thus, it didn’t make financial sense to implement this change.  I still have a /4QI that was given to me.

 

Another group was put together to design a replacement for the /4 series with much more power as well as a much better user interface.  This was the Armadillo or 99/8 and this was in the 1982-83 time frame.  It was not long before that program was shut down.  I still have a set of the specs I wrote for that machine.  About 50 of the production design /8s were built.   I used the new 16Kx4 DRAMs which eliminated the DRAM race problem.

TI and many other companies tried to set the price of a component in an effort to lower component costs.  The effects of this concept were not pleasant.  The main TI /4 problem surfaced in the transformer area.  When a vendor price was set by the user, the vendor had to cut costs to make a profit.  In a transformer, copper wire and steel “laminations” for the core are the major costs.  Decreasing the size of the copper wire produced more transformer heat generated for the same load requirements.  The same holds for decreasing the “iron” in the core.  Several product vendors had fire trouble, but to my knowledge, TI didn’t.  Corporate management came to Lubbock to make sure it didn’t.  I was not involved in this in any way.

I personally think that this problem was the straw that broke the camel’s back, for the Home Computer program was shut down shortly thereafter.  Additionally, the TI CEO had just lost his life while in Europe; so, the upper management levels had changed.

 

I will close with a bit of philosophy.  I believe that simplicity is excellence.  The best design is something that the production operation can understand and fabricate rather than the most sophisticated design.  I used this philosophy in my designs.  I just happened to be in the right place at the right time to get to do several things that nobody else got to do.

[+OLD CS1]

"Johnny Box" has a nice ring to it.

[GDMike]

Thank you for sharing some of your talent and memories.

[+FarmerPotato]

On 3/7/2020 at 7:57 AM, dhe said:

I am pleased to note that we were using a text editing program (PDWF) on the larger TI systems at this time that was a significant help for writing (and preserving) specifications, etc.  PDWS was adapted to run on the TI PCs, and I continued to use it there.

Aha, you were first to uncover the PDWS story (in this Mike Bunyard interview) before I even knew what I was looking for. 
 

Noticed a picture of the PDWS manual here:

 

https://www.computinghistory.org.uk/det/11863/PDWS-990-Writer-s-Guide/

 

Still wanting to recover or recreate PDWS. 

And look, also in 2020, a Users Guide in the CB Wilson papers: