Parent 

IBM 5161  -  Extender Card  -  Switch Settings


NOTE:   IBM's switch setting table for the extender card indicates that wait state generation is limited to only a segment, the particular segment chosen by the switches.
That conflicts with certain other IBM documention, that suggests that more than one segment is affected.

My analysis (here) of the published circuit diagram of the extender card, indicates that more than one segment is affected; the wait state generation scope is larger than a segment.
Unfortunately, I do not have an extender card, and so I cannot verify my analysis.
The switch settings below reflect my analysis.


Memory (ROM/RAM) operations between the CPU (in host computer) and the 5161, are slowed by the extender card and receiver card and cable.  Because of that, IBM require that a wait state be invoked for operations involving any memory (ROM/RAM) in the 5161.  The switches on the extender card control the memory address range in which the extender card signals the CPU to add a wait state.

  1 = ON
  0 = OFF


Switches
1234		   Address range in which extender card
  will get the CPU to add a wait state		  
Comment
0000   Disable all wait state generation by the extender card  
0001   E0000 to EFFFF   (896 K to 960-1 K)  
0010   D0000 to EFFFF   (832 K to 960-1 K)  
0011   C0000 to EFFFF   (768 K to 960-1 K)   Warning: See note 1 at page bottom.
0100   B0000 to EFFFF   (704 K to 960-1 K)  
0101   A0000 to EFFFF   (640 K to 960-1 K)  
0110   90000 to EFFFF   (576 K to 960-1 K)  
0111   80000 to EFFFF   (521 K to 960-1 K)  
1000   70000 to EFFFF   (448 K to 960-1 K)  
1001   60000 to EFFFF   (384 K to 960-1 K)  
1010   50000 to EFFFF   (320 K to 960-1 K)  
1011   40000 to EFFFF   (256 K to 960-1 K)  
1100   30000 to EFFFF   (192 K to 960-1 K)  
1101   20000 to EFFFF   (128 K to 960-1 K)  
1110   10000 to EFFFF   (  64 K to 960-1 K)  
1111   00000 to EFFFF   (    0 K to 960-1 K)  



NOTE:  The EFFFF upper limit is designed into the extender card.
            It is so that CPU reads of the BIOS ROM's on the 5160 motherboard do not result in a wait state.


EXAMPLE #1:

There is an XT-class hard disk controller in the 5161.
The controller's ROM is addressed C8000 to C9FFF, and is the only memory in the 5161.
Therefore you need to have the extender card generate a wait state for address range C8000 to C9FFF.
The appropriate extender card switch setting would be 0011, i.e. address range C0000 to EFFFF.   <----- However, read note 1 below. 

Note that in this example, although not required, address range C0000 to C7FFF and address range CA000 to EFFFF, also invoke a wait state.
So, if there is an expansion card in the host computer that has a ROM that starts at address C0000, reads of that ROM will be unnecessary slowed by a wait state.
An unfortunate side effect.


EXAMPLE #2:

You have 256 KB of RAM on the 5160 motherboard (addressed 0 to 3FFFF), and 384 KB of RAM in the 5161 (addressed 40000 to 9FFFF).
Therefore you need to have the extender card generate a wait state for address range 40000 to 9FFFF.
The appropriate extender card switch setting would be 1011, i.e. address range 40000 to EFFFF.

Note that RAM in the 5161 is a poor example, because it is to be avoided, for performance reasons (in the 5161, RAM will be slowed by the wait state).





Note 1 0011 is the expected setting that would be made if the 5161 contains an 'IBM Fixed Disk Adapter'  (the adapter having a BIOS ROM that starts at address C8000)
However, it has been discovered (by at least three people) that if the host computer is an IBM 5150, the 5150 will start, display the flashing cursor, then no progress past that.
•  A workaround is to use the 0100 setting.
•  Presently unknown if the problem affects the IBM 5160.
•  An XT-IDE card configured for C8000 works at the 0011 setting (and does not work for settings that result in no wait state).