Deglitching the write output of the 1130 for Virtual 2315 Cartridge Facility - part 2

RESULTS OF TESTING A WRITE TO THE V2315CF USING MY NEW DETECTION METHOD

I loaded a test 2315 cartridge into the Virtual 2315 Cartridge Facility (V2315CF) in virtual mode, so that the IBM 1130 saw the disk drive as ready for activity about 90 seconds after loading. I had a few hand inserted instructions in memory to be able to test the disk.

I coded an XIO Start Read command to the disk for a full sector of 321 words, bringing the current data on the mini cartridge into memory. I then modifed a few words of the memory and changed the instruction to an XIO Start Write. This will capture the changed data from memory and replace the contents of that sector inside the V2315CF RAM. 

I can then zero out the memory location before doing another XIO Start Read, ensuring that what comes back from RAM (from the disk) is what I intended to write. If not, I can see how close or far it is from properly capturing the stream of data coming from the IBM 1130. The flag from my logic when it detects an error in the error checking (ECC) bits was routed out to a connector so that I could capture it with the logic analyzer and oscilloscope. That would tell me where in the sector it first detected invalid data. 

RESULT OF FIRST TESTS

My logic analyzer is watching the incoming data from the 1130 on -Write Clock and Data and observing the signal that is raised if the four error checking code bits at the end of each word indicate some error grabbing the word. The error signal was never raised - my logic is grabbing the data properly, I believe, and never sees an ECC error. 

However, when I read back the sector that I wrote, I get random garbage. I used my utility box to open the mini cartridge on my laptop and then examined the sector with the showsector.exe program I developed. The data matches - it is the same garbage that I see when I read the sector. The data that is being carefully assembled and error checked is somehow not getting put into RAM correctly. 

This leads to to suspect the interaction of the logic processing the write from the 1130 and the logic that updates the RAM. I will do some careful examination - the likelihood is that I am not holding the data value steady long enough for my RAM module to lock it in. I will do some simulation aimed at testing this theory and look for ways to trigger the RAM module earlier. 

BUILT ANOTHER COPY OF THE V2315CF FOR THE SYSTEM SOURCE MUSEUM 1130

Since I intend to install another instance of the V2315CF on the 1130 system up in Maryland, I wired together the entire power system, built another 2310 Interface Board and have the second main unit ready to transport once I have all the debugging wrapped up.