'
' WPS Model 71, Measuring Device.
' Tom Jennings 
' mru 22 Nov 98
'
' Contents of this file and resulting binary are Copyright Tom Jennings, 1998.
'
' Loaded 27 Nov 98
'	Fixed divisor at ADC: such that it really runs 0..20 (21 symbols). This was
'	incorrect previously, also.
' Loaded 27 Nov 98
'	Actually found three unused bytes to remember DDU "last position" to skip
'	turning on motors when on the correct DDU symbol; eliminated motor creep!
'	Tripled the width of "zero" by adding two more symbols at ADC:.
'	Put DDU motion timeout back to 10 seconds.
'
' Loaded 22 Nov 98
'	Adjusted analog hardware (gain too low, couldn't reach +9; probe seems to have
' 	drifted).
'	Changed DONE buss slow-down cap from .01 to 1000pf; TC was far too slow, made
'	response terrible and was probably the source of the creep (eg. motor stayed on
' 	until the DONE bit charged slow-down cap and reached logic-1).
'	Added additional DONE check right after motor select in DDU:, to lessen motor-creep.
'	Increased wait to 20 seconds before errorhalt; then commented-out errohalt branch,
'	to avoid death during a demo (argh).
'
' Loaded 18 May 98
' 	Fixed the following bugs: after obtaining result, and after foo(), the contents of
' 	calV was 'restored' to the DDUs, not the actual last result (Deleted unused var
' 	sampleV, named it 'result', stored last result there.) Added kludge to DDU() so
'	that symbol 1 on LDDU (Synopsys of DAY THE EARTH STOOD STILL) is never displayed...
'	it's there as a tease.
' Loaded 14 May 98
'	Initial version.
'
' written 15 May 98
'
' The heart of the Model 71 functionality is determined by the conductivity of the sample
' in the vial. The sample is part of a voltage divider; a calibration voltage is applied, and
' the sample voltage is read; the calibration voltage required to reach a standard sample
' voltage is used to conjure up a "measurement" (result).
'
' To operate, the RUN/CAL switch is first set to the CAL position (error if not).
' The operator turns the CALIBRATE pot until the SEQUENCE display reads "0", which
' is approximately half-scale on the ADC. The calibrate pot sets a calibration voltage,
' which is applied to the sample/voltage divider; current through the sample (a few
' microamps, under 5V) induces a sample voltage, both of which are read by the PICSTIC's
' internal two-channel ADC.
' When "0" is displayed on the IEE, the RUN/CAL switch is changed to RUN (error if not 0),
' then the software conjures up some symbols and random-appearing activities. When the
' RUN/CAL switch is changed back to CAL the process is repeated.
' 
' The ADCs have 12 bit resolution; the IEE has 18 discrete symbols. Therefore the
' 12-bit range 0..4095 is normalized to the IEE range of -9..+9, with the resulting 
' loss of resolution -- which might be generally a disadvantage, but in the Model 71 it
' allows for a range of results (DDU symbol combinations) per sample measurement.
'
' Brief instructions are displayed on the DDUs; the DDUs also indicate errors and other activity
' in generally inscrutable ways.
'
' Just for increased inscrutability, after 655 seconds of inactivity, the Model 71
' generates a small burst of random DDU activity.

'
' Just to confuse things, the RUN/CAL switch is hacked into the CAL pot, such that 0V indicates
' CAL; therefore the calibration voltage cannot be read in the CAL position!
'

' GLOSSARY:
'
' DDU: Digital Display Unit, CBS Laboratories' DDU-1, electromechanical display
' device made of 12 metal flaps mounted radially around an axle; 
' driven by a synchronous motor, the flaps are arranged such that 
' a pair of flaps are folded flat, with an image printed across the
' pair of flaps. A rotary switch telegraphs the current position of
' the flaps, allowing software to turn the motor on, causing images
' (pairs of flaps) to pass by one at a time until the desired image is reached,
' then the motor is turned off.
' The DDUs are modified such that all 110AC is within each DDU; the relays were replaced
' with Crydom solid state relays (SSRs). Symbol selection is changed
' such that the 12-position switch attached to the motor shaft (and
' indicating symbol currently displayed) is driven logically "backwards";
' 1 of 12 outputs are applied to the switch, and software detects when
' the desired position has come around into view, then the motor shut off.
' (Rather than interrupting the relay that powered the motor, causing it
' to stop itself at the correct digit [and the relays caused nasty EMI
' that blew out a PICSTIC in the original design].)
' I/O: 12 output bits to drive current-symbol switch (shared by all DDUs);
' one output bit to select DDU, one input bit to detect symbol-in-position
' (pole of current-symbol switch) (input bit shared by all DDUs).
'
' IEE: Industrial Electronic Engineers, Inc, "one-plane display". Consists
' of twelve simple optical assemblies (lamp, lens, mask) that displays
' an image (digits) onto a frosted plastic screen at the front of the
' device. It is controlled simply by turning one of the lamps on. Plus and
' Minus signs are separate images (eg. displaying "-9" requires turning on
' the "9" and the "-" lamps simultaneously).
' I/O: 12 output bits to select symbols 0..9, -, +.
'
' SR: Shift Register. The WPS Model 01 CPU/IO card consumes four bits
' of the PICSTIC to provide 24 bits of output on three 8-bit shift registers.
' 12 SR bits are used by the IEE; the remaining 12 by the DDUs (shared
' symbol-select bits).
'
' The CALIBRATE pot and the RUN/CAL switch are connected to one ADC channel;
' the sample voltage is read on the other ADC channel. The RUN/CAL switch causes
' the calibration voltage to go to zero; the CALIBRATE pot is wired such that
' it cannot generate 0 volts. Therefore, 0 volts can be used to indicate RUN/CAL
' position, leaving some 4000+ other values as calibration voltage readings.
'

' Copyright Tom Jennings, 1998.
'


''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
'' file: c:\tomj\projects\model 01, CPU-IO card\model01.bas
'' General purpose definitions and functions for the Model 01
'' CPU I/O card.
''
''
'' Tom Jennings 
'' mru 2 Aug 98
'' written 17 July 98
''
'' For unrevised Model 01 card (prototype) card.
''
'' Copyright Tom Jennings, 1998.
''

'
' General purpose registers.
'
symbol	i= b0
symbol	n= b1
symbol	j= b2
symbol	D= b3				' symbol written to each DDU

''
'' The shift registers need data shifted to them MS bit first; the MS bit
'' of the value to be written is isolated by using the byte/word organization
'' of the PIC; the value to be shifted is placed in the LS byte, and added
'' to itself to shift left; the MS bit of the LS byte is thence shifted into
'' the LS bit of the MS byte. Easy, no?
''
symbol	k= w2				' {b4, b5} word reference to following bytes
symbol	k_lower= b4			' not used as-is
symbol	k_cy= b5			' carry from K after shift

symbol	l= b6				' used by low-level routines, generally SR address
'
' SR B is shared by the DDU and IEE hardware; this byte contains the last-written
' data so that the other half is preserved.
'
symbol	lastb= b7 
'
symbol	lastn= b8			' used to detect change of sample V
'symbol	lastL= b9			' last LDDU value
symbol	result= w5			' {b10, b11} result of "measurement"
symbol	calV= w6			' {b12, b13} calibration voltage
symbol	calVl= b12			' lower 8 bits of calV
symbol	calVh= b13			' upper 8 bits of calV

symbol	lastL= b9			' last LDDU value
symbol	lastM= b14			' last MDDU value
symbol	lastR= b15			' last RDDU value
'
' "Random" number generated by incrementing while waiting for the operator
' to do something, and is used to generate noise in response times. "idle" is
' a counter that increments during the loop, and is cleared when human activity
' is detected (RUN/CAL switch change or CALIBRATE pot value change).
'
symbol	rand= w8			' {b16, b17}
symbol	idle= w9			' {b18, b19}
'
' The compiler's ADC routines leave data in W10, aka V (b20, b21). The ADCs are
' 12 bits; with a 5.00V reference this is .00122V/bit.
'
symbol	V= w10				' {b20, b21}

symbol	CALSW=20			' voltage less than this is "0" ie. CAL/RUN in CAL position

symbol	REDX= 0				' big red X
symbol	DATA= 1				' (1..9) first data/symbol
symbol	INSTRUCT= 10			' instructions.
symbol	BLANK= 11			' blank (no) symbol
''
'' These definitions are hard-wired on the PCB.
'' Data is written to the shift registers by asserting data on the SD pin, and clocking
'' the appropriate SR clock pin. 8 bits per shift register, MS bit output first.
''
symbol	SDpin= 	pin0				' pin number, Serial Data to shift reg
symbol	SRC=	3				' SR C clock
symbol	SRB=	2				' SR B clock
symbol	SRA=	1				' SR A clock
symbol	DONE=	pin7				' DDU wire-OR DONE bit (true=0)

symbol	RBITS= %01000000			' right DDU select bit
symbol	MBITS= %00100000			' middle DDU select bit
symbol	LBITS= %00010000			' left DDU select bit


''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''

' Set the pin directions, and make all the displays shut up. Use the IEE
' display as POST status indicator.

main:	dirs= %01111111				' all pins outputs except PB7
	pins= 0					' deassert device selects
	lastL= 255 : lastM= 255 : lastR= 255	' set current DDU positions to impossible
	lastn= -1				' detect inactivity
	for n= 0 to 9				' test DDU lamps, mainly
		gosub IEE
		pause 100			' also sez CPU ain't ded
	next n
	n= 9 : gosub IEE : pause 600		' show plus sign
	n= 137 : gosub IEE : pause 600		' show minus 9
	n= 255 : gosub IEE			' blank the IEE display,
	D= BLANK : gosub allDDU			' blank all the DDUs;
	pause 4000				' we seem important when we inscrutably do nothing.

'
' Start of the main functionality loop. Proper operation requires starting in
' CAL mode, adjusting the calibration voltage until the sample voltage reads
' center-scale ("0" on IEE), then changing to RUN mode and generating the "result"
' on the DDUs. 
'
' Completion of startup, and operator-error entry point; if RUN/CAL is not in CAL mode,
' then display red X's, pause, then display the instruction cards.
'
m1:	gosub ADC				' if RUN/CAL switch is 
	if calV < CALSW then m1b		' in RUN mode,
merr:	n= 128 : gosub IEE			' display "-0",
	D= REDX : gosub allDDU			' set all DDUs to red X's,
	D= REDX : gosub foo			' (go check for inactivity)
	goto m1					' wait for CAL mode.
m1b:	n= 255 : gosub IEE			' blank IEE display,
	D= INSTRUCT : gosub allDDU		' display instructions,
	idle= 0					' not idle, operator changed switch.
'
' Wait until the RUN/CAL switch is set to RUN. During this, the operator
' is supposed to be turning the calibration voltage pot such that the IEE display
' reads "0", indicating the sample voltage is approximately mid-scale (see ADC code).
' We don't check for that now; all we do is read voltages, display them on the IEE,
' wait for the switch.
'
m2:	D= INSTRUCT : gosub foo			' check for inactivity, do funny things;
	gosub ADC				' read voltages,
	gosub IEE				' (N= IEE symbol) display current symbol,
	if calV < CALSW then m2			' CAL mode while calV zero,
	idle= 0					' switch changed position.
'
' Switch was changed to RUN mode; assuming goal was adjusted properly, use the
' calibration voltage to work up some symbols to apply to the DDUs. The symbol-generation
' makes no inherent sense, don't look for any.
'
m3:	if n <> 0 then merr			' operator screwed up, start over.
	n= 255 : gosub IEE			' blank IEE display,
	D= BLANK : gosub allDDU			' start with blank symbols,
	D= 10					' for..step -1 doesn't work!
m3a:		D= D - 1
		n= D : gosub IEE		' display N, (9, 8, 7, ...0)
		gosub ADC			' update random number,
		V= rand // 2000 + 750		' random pause, 
		pause V
	if D <> 0 then m3a			' to 0 inclusive.
	gosub ADC				' update rand again,
	V= rand / 91				' Generate random starting position,
	n= 255 : gosub IEE			' blank IEE,
	gosub DDUmod8				' position DDUs,
	pins= %01110000				' then turn on all DDU motors
	V= rand // 10000 + 5000			' for a random period.
	pause V
'
' We need to scale the cal. voltage (runs 0..4095) to run 0..727. Since 4096/728 is 5.6,
' and we can only do integers, we multiply by a fraction; 91/16 is pretty close. (4095 * 16
' fits in 16 bits.)
'
	result= calV * 16 / 91			' then normalize calV (0..4095) to DDU's (0..727),
	V= result : gosub DDUmod8		' display "result" on DDUs.
'
' Continue to display the current result until the RUN/CAL switch is changed to CAL.
'
m4:	D= DATA : gosub foo			' the devil finds uses for idle CPUs,
	gosub ADC				' read the pot and switch,
	if calV > CALSW then m4			' wait until RUN/CAL in CAL position,
	idle= 0					' someone was there to change the switch,
	goto m1					' start over.

'
' Called while waiting for human activity, this looks for long periods of idle
' and does odd things to the display. D contains the symbol to restore the DDUs to when
' we're done playing EXCEPT that DATA means restore the results display.
'
foo:	pause 10 				' we count 10 mS increments,
	idle= idle + 1				' increment the idle counter,
	if idle <> 0 then ret			' when/if it reaches 0 (655 secs; ~11 min)
	pins= %01110000				' turn all motors on,
	V= rand / 91 : gosub DDUmod8		' display random symbols (0..65535 --> 0..727),
	pause 1000				' pause only briefly.
	if D <> DATA then allDDU		' go display symbol D unless DATA; then
	V= result : goto DDUmod8		' re-display results data.

' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
'
' Display value V on the DDUs. V should already be normalized to 0..727, to fit the
' nine symbols per DDU. The three DDUs, A, B, C (left to right) are the mantissa of 
' the base-9 value of calV appropriately scaled. Eg. DDUs display A*9^^2 + B*9^^1 + C*9^^0.
' Maximum value for the DDU display is 8*(81) + 8*(9) + 8*(1) - 1 or 727. 
' Modifies N, J, K, L, updates LASTB.
'
DDUmod8:
	n= V / 81 + DATA : gosub MDDU		' 100's digit (0..8)
	n= V // 81 / 9 + DATA 			' 10's digit (0..8)
	if n <> DATA then du1 : n= DATA + 1	' KLUDGE prevent "DAY ..." symbol
du1:	gosub LDDU
	n= V // 9 + DATA : gosub RDDU		' 1's digit (0..8)
ret:	return

'
' Read both the calibration voltage and sample voltage and leave in memory. 
' This also calculates the goal voltage expressed ' as as -9..0..9 for the 
' IEE display; 0 == goal voltage reached, returned in N. This also detects 
' human (in)activity and runs the random number.
' NOTE that this is intentionally non-linear; it provides for a "wide" 0
' range.
'
ADC:	call ad1				' read sample voltage,
	n= V / 204				' scale 0..4095 to range 0..18 (21 symbols),
' linear (19 values)
'	lookup n, (137, 136, 135, 134, 133, 132, 131, 130, 129, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9), n
' non-linear (21 values)
	lookup n, (137, 136, 135, 134, 133, 132, 131, 130, 129, 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9), n
	if n= lastn then adc1			' if goal voltage changed,
	idle= 0					' clear inactivity indicator,
	lastn= n				' remember new symbol.
adc1:
'
' Generate a random number using the sample voltage as increment. We can use calV as a
' temporary register since we are about to set it to the value in V.
'
	calV= V 				' V= sample voltage,
	j= calVl : calVl= calVh : calVl= j	' swap upper and lower bits of sample V,
	rand= rand + calV ^ calV		' increment rand by sample V, XOR with sample V.
	call ad0				' read calibration voltage,
	calV= V					' set return value.
	return

'
' Set all DDUs to the same symbol, in D.
'
allDDU:	n= D
	gosub RDDU
	gosub MDDU
	' fall through
'
' Display symbol N on the specified DDU. Modifies J, K, L, updates LASTB and LASTx.
'
LDDU:	if n = lastL then ret			' if same symbol, no change
	lastL= n : i= LBITS : goto DDU		' K= DDU select bit
MDDU:	if n = lastM then ret
	lastM= n : i= MBITS : goto DDU
RDDU:	if n = lastR then ret
	lastR= n : i= RBITS : ' fall through
'
' Display symbol N on DDU #I. The symbol is asserted onto the DDU buss, then the
' motor driven until the selected symbol appears on the display. A timeout makes
' sure the DDU is actually moving. Terrible things happen if N is out of range.
' Modifies J, K, L. Updates LASTB.
'
DDU:	gosub DDUout				' select DDU symbol,
'
' Turn on the appropriate motor. Note that this sets all other pins to 0; this
' works OK since only pulsing a SR clock bit high would affect the hardware.
' Also note that motor-select is the logic-high source that feeds the DONE
' bit; so if we're already at the desired symbol, the motor pulses on, and sometimes
' pushes the symbol off DONE (seems to depend on DDU); so we quickly check
' for DONE before entering the loop. It also makes a little scritching sound
' which is not nice.
'
'
' Turn motor on, wait for the DONE bit to set (PB7 goes to 0). Wait up to 
' 10 seconds for this (else hardware failure).
'
	pins= i					' assert select
	for k= 1 to 1000			' 1000 * 10 mS = 10 sec
		if DONE = 0 then dd1		' if DONE asserted, exit
		rand= rand + 1717
		pause 10			' delay 10 mS,
	next k
'
' The DONE bit is asserted, ie. the switch wiper has just touched the pole.
'
dd1:	pins= 0					' motor OFF
	return

'
' Convert value in N into DDU symbol, and assert on the 12-bit DDU buss. The symbols are,
' in order: 1..9, 0, -, ., repeat. 0 is really "10", not ASCII zero.
'
DDUout:
	lookup n, (1, 2, 4, 8, 16, 32, 64, 128, 0, 0, 0, 0), k
	l= SRC : gosub shiftout			' assert LS bits,
	lookup n, (0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 4, 8), k
	lastb= lastb & $F0			' clear old DDU bits
'
' Update lastb, output SRB, the shift register that is shared by the IEE and DDUs. 
' Local subr to DDUout and IEEout.
'
outb:	lastb= lastb + k
	l= SRB : k= lastb : goto shiftout	' update DDU

'
' Display a value on the IEE. This is simply a bunch of lamps tied to
' SR outputs. + and - are also on the SR hence must be ORed in. The upper
' four bits of IEE symbols are on SR B, shared with the DDU.
' The sign is derived from N; 0 is no sign lit, + lit unless bit 7
' is set. The hardware::symbol relation is determined by the wiring,
' eg. 8="-", 4="+"; I chose to make the wiring neat, and untangle in
' software here.
'
' Symbols are: 0..9 display "0" through "9"; "+" on if n=1..9; "-" on
' if bit7 set (does allow "-0"). Blank display (no symbol) if n=255.
'
' Modifies K, L. Updates LASTB.
'
IEE:	if n = 255 then ie3			' special-case blank
	k= n & $0F				' strip "sign" bit,
	lookup k, (1, 2, 4, 8, 16, 32, 64, 128, 0, 0), k
	l= SRA : gosub shiftout			' assert LS byte,
	k= n & $0f
	lookup k, (0, 0, 0, 0, 0, 0, 0, 0, 1, 2), k
'
' This mess determines the sign bit display. Temp L holds sign bit.
'
	l= 0 : if n = 0 then ie1		' if N=0, no sign bit
	l= 8 : i= n & 128 : if i <> 0 then ie1	' L="-" if neg. N
	l= 4					' else "+"
ie1:	k= k | l * 16				' shift into upper nybble
ie2:	lastb= lastb & $0F			' strip old IEE bits
	goto outb				' output SR B
'
' Blank IEE display requires special handling.
'
ie3:	k= 0 : l= SRA : gosub shiftout		' clear these bits,
	k= 0 : goto ie2				' clear shared bits

''
'' Shift value in K out to the shift register whose address is in L. Modifies K, J.
'' Though K is a word it should be assumed to be a byte (eg. only LS byte is used.)
''
shiftout:
	for j= 1 to 8
		k= k + k			' put MS bit into LS bit of K_cy
		SDpin= k_cy			' assert data
		pulsout l, 1			' pulse correct SR clock pin
	next j
	return

'
' Report fatal hardware errors here -- error number in N. N is displayed
' on the IEE display (look ma, no moving parts) and the + and - signs are
' blinked until the power is cycled.
'
'errorhalt:
'	pins= 0					' stop motor(s),
'	gosub IEE				' display number,
'	pause 500				' delay half a second,
'	n= n + 128				' toggle sign,
'	goto errorhalt				' repeat.

END