Update sd.asm

Optimize SD Card protocol timing delay for CSIO interface.

Source/HBIOS/sd.asm

@@ -866,7 +866,7 @@ SD_INITCARD1:
 	; MAKE SURE WE FINISH SENDING
 #IF ((SDMODE == SDMODE_CSIO) | (SDMODE == SDMODE_MK4) | (SDMODE == SDMODE_SC))
 	CALL	SD_WAITTX		; WAIT FOR TE TO CLEAR
-	CALL	DLY4			; WAIT A BIT MORE FOR FINAL BIT
+	CALL	DLY32			; WAIT A BIT MORE FOR FINAL BIT
 #ENDIF
 ;
 	; PUT CARD IN IDLE STATE
@@ -1025,7 +1025,7 @@ SD_INITCARD5:
 	; PER SPEC, THE CARD SHOULD NOW BE ABLE TO HANDLE FULL SPEED OPERATION
 	; SO, FOR CSIO OPERATION, WE SET CSIO TO MAXIMUM SPEED
 	CALL	SD_WAITTX		; MAKE SURE WE ARE DONE SENDING
-	CALL	DLY4			; WAIT A BIT MORE FOR FINAL BIT
+	CALL	DLY32			; WAIT A BIT MORE FOR FINAL BIT
 	XOR	A			; ZERO MEANS MAX SPEED
 	OUT	(Z180_CNTR),A		; NOW SET CSIO PORT
 #ENDIF
@@ -1674,12 +1674,10 @@ SD_DESELECT:
 	; FINISH SENDING AFTER TE IS CLEARED.  THE DELAY BELOW WILL
 	; DO THIS FOR THE SLOWEST POSSIBLE SEND RATE WHICH IS
 	; CLK / 1320, SO DELAY AT LEAST 1320 T-STATES
-	;CALL	DLY64			; DELAY FOR FINAL BIT
 ;
-	; IN PRACTICE, IT LOOKS LIKE THIS WORST CASE SCENARIO NEVER
-	; OCCURS.  FOR NOW, USE A SMALL DELAY WHICH SEEMS TO BE MORE
-	; THAN ADEQUATE BASED ON LOGIC ANALYZER TRACES.
-	CALL	DLY4			; DELAY FOR FINAL BIT
+	; IN PRACTICE, A SMALLER DELAY IS FINE BASED ON LOGIC ANALYZER
+	; TRACES.
+	CALL	DLY32			; DELAY FOR FINAL BIT
 #ENDIF
 ;
 	LD	A,(SD_OPRVAL)