User avatar
intric8
Seattle, WA, USA
Website

Posted Wed Nov 24, 2021 9:53 am

I shared some of my correspondence with Greg Tibbs yesterday, and I wanted to drop another one (probably the last) today. This is my absolute favorite story that he wrote down for me.

The beginning of his message references RAM chips my friend Christian made for me to use in my original 2MB Rejuvenator setup.
RAMchips.png

We'd had a separate conversation where Greg had made his own "all-in-one" design that pressed all of the RAM in via one large add-on component. He preferred the single RAM chip adapter approach Christian employed. I'd just sent him an update the day after one of my Seattle Commodore Computer Club meetings with several photos of the Rejuvenator being worked on and all of the new RAM being recognized.

It might help if you watch this short clip of him first, thankfully preserved by Amiga Bill on his YouTube channel. Once I put two and two together, the fact that this clip matches perfectly to Greg's story is nothing short of astonishing to me.

I don't think Greg had seen this archived video when he shared the story. Again, his memory was razor sharp. And you might notice Greg being a bit sweaty (look closely at his hair). The following story explains fully why.

It's amazing.

Anyway, go grab a hot beverage and maybe some popcorn. This is great stuff!

Note: Expert Services was a retail store on the Ohio/Kentucky border that funded the project and sold the boards.


Good job on the surface mount RAM chips!

The two DIP resistor packs in the photos bring back a memory. At the Wash DC show in 1990 or 1991, the first two production boards were taken with us. We waited on a vendor who populated two boards on a Friday morning in Hamilton, Ohio and we then went to Expert Services store in Florence, Ky. From there we proceeded to I-70 to Wash DC, about a 11 hour drive due to DC traffic. We got there late in the evening, after dark (11 PM). In my hotel room, I put the production board in my A1000 and it did not work! I tried the other board and it also did not work. My hand soldered (same circuit board) board worked fine. The hand-soldered board went on display while I figured things out. I spent most of the night and early morning trying to figure out what was wrong. The vendor who populated the board put in resistor packs of 33K ohm rather than 33 ohm. The 33 ohm resistors were to stop ringing on some data lines and duplicated the A500/A2000 design.

I had some wire and solder and an iron and put wire jumpers over the resistor packs to short out the 33K ohm resistors. It then worked! Then we got down on the convention floor and GVP wanted to test their video card and I put it in. Screen was somewhat working but had a red tint, if I remember correctly. I looked the wiring diagram up on my computer in the hotel room and figured out that one of the color bits was shorted to an internal layer power plane. I had mini drill bits (I came prepared) and removed the short and put the jumper on and the GVP board then worked perfectly. I took the A1000 back to the show and it was a hit. One of the writers for an Amiga magazine wanted one badly and he and I went to my room and I modded the second board the same as the first and put it into his A1000 for him so he would not have to deal with the installation hassle. He got 2MB of RAM and the 2MB Fat Agnus., and the ECS Denise. He was a happy camper. He actually bought it, he was so happy. Jay Miner came by on the show floor and I showed it to him and he was so happy that he hugged me!

The people at Expert Services then forced me to go to bed as it had been over 40 hours since I had last slept and was shaking from the stress. I think I didn’t get up until after noon on Sunday, about 24 hours later. We packed up and drove (10 hours) back to Dayton. The next show was in NYC and it went smoother. After the DC show, we called the vendor who populated the board and let them know of their error. They insisted that they put 33 ohm resistor packs in, but they misread the code on them. With that fixed, and added the hole drill and jumper wire, we were ready to sell. We sold about 800 boards to my knowledge and never reworked the artwork to remove the short. The area that the hole was located was so densely populated with traces that, due to limitations on trace size and separation in the Blackbelt software, it was impossible to fix without moving the design to another platform and software.

One of the things learned about the A1000 motherboard was that the board, a 2 layer board, had thin ground traces to the CPU and the sidecar socket. I modded many A1000s that I did the Rejuvenator installation with two or three heavy jumpers from the power supply connector to the CPU socket and the sidecar edge connector. This improved noise and made many sidecar devices work better, especially when multiple devices were stacked on one another on the side. However, there were a few sidecar devices that became unstable with that modification. The real issue was the combination of motherboard loads on the 68000 CPU data lines were beyond Motorola specification, especially when sidecar devices were added. Improving the ground to the 68000 CPU and sidecar edge connector reduced noise and strengthened the signals out of the CPU chip and made the system more reliable. Amiga Corp should have buffered the data lines, but that would have required extra parts and control logic and complicated the sidecar design.

About 4 months before the DC show, Commodore got wind of what I was doing and they paid to have me fly to West Chester NY HQ to show the two layer prototype off. While there, I met the VP (Henry Rubin) and the then president of Commodore. We worked out an agreement for them to sell me the Fat Agnus chips and the ROMs and ECS Denise chips. I came back with four (free) samples of the Fat Agnus, 2MB Agnus, the ROM chips, and the ECS Denise. About the time of the NYC show, the new president of Commodore changed his mind and he decided that the Rejuvenator was going to cost Commodore A500 sales and I was told no chips were coming. George Robbins, one of their senior engineers, told me they were working to get the president to sell the chips and to keep quiet. After four months and people calling the Rejuvenator a fake, I put word out on Compuserve that the new president decided to renege on the deal and there would be no Rejuvenators sold. Commodore received many complaints and bad press (what was new). Two weeks later I had the chips.

I told you previously about the split up between Expert Services and myself. I pretty much left the Amiga world after that except that I developed the QuickStart code that used 68020, 68030, 68040 memory virtualization to be able to load a new Kickstart ROM images from disc using virtual memory that survived reboot. I modified Commodore code for the real time clock and hard disk to detect 68040 CPU and run different 040 compatible code for OS V1.3 vs 2.0. Commodore was not happy about that as they wanted 1.3 to go away. I developed that because as a ROM beta tester, I got tired of opening my A3000 and changing ROM chips every few weeks, wearing out the ROM socket. George Robbins gave me some help with the part that redirects function calls as that area is protected via checksums and the code to checksum the function jump table was secret at that time as that would make viruses and other malicious code survive reboot.

Greg

User avatar
Beska

Posted Wed Nov 24, 2021 11:33 am

The guy sounds absolutely brilliant. And another example of Commodore shooting themselves in the foot. They should have just hired him.

What is QuickStart? This is the first I've heard of it.

-b

User avatar
A1-X1000
Toronto, Canada
Website

Posted Wed Nov 24, 2021 6:45 pm

wow what a great story Eric thanks for posting and of course for making Greg's work not be forgotten <3 :boing:





Return to “Hardware”