[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [N8VEM-S100:6351] Re: 80386 CPU Board - A possible solution

One more bit of follow-up

I had mentioned that I could not get the Texas Instruments version (TX486DLC/E-40GA) to work.  Turns out that the problem was that for some reason the TX486DLC/E needs more wait states than the AMD386DX to work properly in my system.  I needed 7 wait states for both IO and RAM at 33 mhz to allow DOS to boot reliably.

The TX486DLC/E may be a useful option for some folks as it seems more common and inexpensive than the Cyrix CX486DLC/E versions. 

- Gary

On Friday, February 20, 2015 at 12:21:49 PM UTC-5, Gary Kaufman wrote:
John -

I added the inverter and the problem is certainly better, but not totally fixed.

I can now properly switch 80386<-->Z80 using QOEE,01/00 as long as U27 is a 74F138 or 74S138 and the clock speed of the 80386 is 16mhz or less.  At 18mhz or > it still fails.

If I switch to 74LS138 it won't work above 6mhz. 

I/O wait state setting on the 80386 doesn't affect the problem.  CPU is an AMD80386-40

I'm comfortable we've isolated the area of the problem.

- Gary

On 2/19/2015 1:51 PM, John Monahan wrote:

Gary, took another look at this V2 SMB,  U33, the 74LS175 is clock triggered on the rising edge. The signal coming from U31B (pin 6) may be arriving too late and somehow the data bus has lost it by the time it rises into U33.   I’m seriously considering doing a V3 SMB using a few GAL’s and more functionality, since I don’t have your problem could you check if this is the problem.


We will invert the signal coming from U31 pin 6, so the start of the rising pulse will latch the data lines into U33.  Fortunately the is a spare gate on the board U10B,  a 74LS00.  Jumper its pin 4 to Vcc. Bend out pin 6 of U31B. Jumper this pin to pin 5 of U10. Jumper U10 pin 6 to pin 9 of U33.   Then see if you can get home with the 80386.