My Kenbak was running at 1.6MHz using the parts Erik Klien's had. This frequency is divided by two, so my Kenbak was running at about 841kHz. This is very close to the specification of 1MHz.
But what is the fun with that? It takes 1.9 seconds to count from 0 to 255. Too Slow!
The data sheet for the memory "cans" says they can handle 5MHz. I needed a 10MHz clock.
I don't have any 10MHz crystals or oscillators! I created a 74HCT04+crystal oscillator circuit on a solderless breadboard. I tried 7.3720MHz and when it worked I got excited. That was a pretty big boost over 1.6MHz. So I tried 14.7456MHz. The buttons seem to work right, but the memory is messed up. I manage to find a 12MHz oscillator on a 286 motherboard. After liberating it I found out it was too fast also. The memory was a "little" better, but still wouldn't store reliably.
So I made a divide-by-2 circuit out of a JK Flip Flop and used a 20MHz crystal to get 10MHz. Guess what, it works perfectly! : )
Now it counts from 0-255 in 313ms. The memory cans do get hotter so I will wait until I have heat sinks for them before I run it too long. : ) The hottest IC on the board at 1.6MHz is 190F. I'm not sure if that is because of load or speed. I'm not sure the infrared temperature probe measured the cans right at 1.6MHz. It said 130-140F but the surface might be too shiny.
I have attached pictures of the logic analyzer screen for 2.000MHz and 10MHz. I generated the 2.000MHz with a function generator. While I had it hooked up to the function generator I experimented with the slowest speed it would run at. I found I couldn't get it to run under 250-270kHz (so ~125kHz). The memory is supposed to work down to 10kHz so maybe something else doesn't like it.
Since its counting from 0-255 in 313ms, and each loop requires 3 instructions then I'm getting pretty close to 1000 instructions per second. Not bad! : )
What do you think of that? : ) Its 4AM. My "lets just see" project at midnight has gone on long enough...
10MHz means the Kenbak is running at 5MHz serial rate.
2MHz
10MHz
But what is the fun with that? It takes 1.9 seconds to count from 0 to 255. Too Slow!
The data sheet for the memory "cans" says they can handle 5MHz. I needed a 10MHz clock.
I don't have any 10MHz crystals or oscillators! I created a 74HCT04+crystal oscillator circuit on a solderless breadboard. I tried 7.3720MHz and when it worked I got excited. That was a pretty big boost over 1.6MHz. So I tried 14.7456MHz. The buttons seem to work right, but the memory is messed up. I manage to find a 12MHz oscillator on a 286 motherboard. After liberating it I found out it was too fast also. The memory was a "little" better, but still wouldn't store reliably.
So I made a divide-by-2 circuit out of a JK Flip Flop and used a 20MHz crystal to get 10MHz. Guess what, it works perfectly! : )
Now it counts from 0-255 in 313ms. The memory cans do get hotter so I will wait until I have heat sinks for them before I run it too long. : ) The hottest IC on the board at 1.6MHz is 190F. I'm not sure if that is because of load or speed. I'm not sure the infrared temperature probe measured the cans right at 1.6MHz. It said 130-140F but the surface might be too shiny.
I have attached pictures of the logic analyzer screen for 2.000MHz and 10MHz. I generated the 2.000MHz with a function generator. While I had it hooked up to the function generator I experimented with the slowest speed it would run at. I found I couldn't get it to run under 250-270kHz (so ~125kHz). The memory is supposed to work down to 10kHz so maybe something else doesn't like it.
Since its counting from 0-255 in 313ms, and each loop requires 3 instructions then I'm getting pretty close to 1000 instructions per second. Not bad! : )
What do you think of that? : ) Its 4AM. My "lets just see" project at midnight has gone on long enough...
10MHz means the Kenbak is running at 5MHz serial rate.
2MHz
10MHz