tomcw / mb-audit Goto Github PK

Hi,

I have a stereo issue on IIe (enh), the reactivemicro phazor card gives me the following result with mb-audit .

$00 = 1 in slot 4 $80 nothing SP nothing 65C02 detected.

Original reference of the issue :
a2-4am/4cade#538

One immediate issue is that the //e ROM's 6502 IRQ (=BRK) vector at $FFFE will vector to $C3FA.
(RESET goes to $FA62 and NMI goes to $3FB)

Not an issue with the II/II+, since IRQ goes to $FA40 (there is in C3 ROM).

Perhaps the best solution is to enable the LC's HIGH RAM and change IRQ to vector to my code.
(Fine for IIe/IIc/IIgs since they all have LC's)
NB. After each RESET then re-enable LC and slot 3 ROM.

On cea2, ("mb-audit: a Mockingboard test suite for emulators") Kent Dickey said:

I have a suggestion for 11:0E:00: you call the monitor routine WAIT to wait
a number of cycles--and you care about the exact number of cycles elapsed
(well, really you care about cycles mod 2, so if it's off by a multiple of
2 it's fine). WAIT on a IIgs is not cycle accurate, and it might not be
on other monitor ROMs either since the WAIT description says it waits a
minimum number of microseconds. So the IIgs ROM has extra code to ensure
the CPU is running at slow speed, and this adds extra time. You would be
better off creating your own WAIT in your code so that it's consistent
regardless of the monitor code in ROM.

Repro:

• mb-audit v1.1 (preview 0.2)
• do 'Music Test' for a few seconds (perhaps until it uses the AY noise channel?)
• ESC to exit, then do 'Audit & Diagnostics'
• the MB-C (s4) makes no extra noise
• the Phasor (s1) make ~1s of noise before saying 'Classic Adventure'
• NB. doesn't happen everytime

A simple repro for this is:

• use a Phasor card
• CALL-151
• C0C5 ; select Phasor native
• 2000G

Problems:

1. Detect6522 only detects 6522-B at $80 and not 6522-A.
   • This is because 6522-A is mapped to $10.
2. After calling DetectSSI263 the code hangs due to infinite IRQs.
   • This is because the SSI263 generates a direct IRQ to the 6502, and not via the 6522-B.

NB. On a setup with Slot-1=Phasor, Slot-4=MB-C, then after the MB-C detection is done first, you'll get the following msg:
Slot #1 : Warning: Unknown card (no AYs)
This is because DisplayCardUpdate gets called after each card has had sub-system detection done.

From the MB4c User Manual (https://quick09.tistory.com/1385):

Mockingboard 4c needs to use Slot#4, but, APPLE //c already occupied and mouse firmware there.
It shouldn’t be removed some cases it need for software.
For this, I used an switching function to make it vacant Slot#4 for Mockingboard 4c.
Basically, If some data wrote to $C400(In fact C400-C4ff) the CPLD make vacant the Slot#4 and set the Mockingboard
to use it.
Many of cases it works properly, but, some of Mockingboard detecting S/W it failed to detect. Hence, You may need to
Turn it ON manually by put data as below.

]CALL-151
*C403:FF FF

And French Touch code here shows some magic writes to $C403 & $C404 to "wake up MB4C for IIc".

1. For mba: at the "Tone" test stage:
   • state that keys 1,2,3... (and Q,W,E...) toggle the AY channels
   • visually show when a channel is on (INVERSE) or off (NORMAL)
2. For "(C) SPEECH" test show a message if there are no speech chips + "Press any key"
   • currently it just returns immediately back to the main menu
3. For "(E) 6522 T2C_H" test show a message if it doesn't run the test + "Press any key"
   • currently it just returns immediately back to the main menu

Since I can't read the AY regs, then I don't even know the AY writes are working!
EDIT: I can read the Echo+ AY regs (and this is now implemented in mb-audit).

NB. the Echo+ has a single 6522 and 2x AY's.

• try to play a tone on the AY channels
• try to read AY regs using different four MSB bits...

From Wikipedia:

Although there are only 16 registers, the four MSB bits of the 8-bit bus must be set to the factory default '0000' value when selecting a register. Incorrectly setting the MSB bits causes the chip to ignore the register change. General Instruments did take orders for customized MSB bits (factory set to other than '0000'). The chips made with customize-set MSB register bits allow the same processor to control more than one AY chip on the same bus (e.g. Mockingboard sound card for Apple or TurboSound for ZX Spectrum). There are many new-old-stock (NOS) chips on the secondary market with MSB bits factory set to a non-'0000' value. The non-0000 value can cause significant developmental troubles for designers and repair technicians. Software must be written to identify the correct value of the MSB bits on any given chip. Also, software must be changed or hardware added to allow these factory set MSB chips to be used in place of the default '0000' chips.

From cea2, Kent Dickey said: (after running against v0.7-beta)

I have two more requests, both low priority: Can you change T6522_1
so that it doesn't hang if the interrupt doesn't arrive? Just put in a
timeout of some sort?

And second, can you change the T6522_1 Timer1 value from $0303 to $0404?
I like to run KEGS in a special forced 1MHz mode when running Apple II
programs, where it ignores the SPEEDREG and always runs at 1MHz and never
2.8MHz. And when I do that with mb-audit v0.7-beta, then T6522_1
hangs since I'm slowing down the ROM interrupt handler (which manipulates the
speed register and so would normally run at 2.8MHz) and so the interrupt
occurs before the T6522_1 code wants it to, and mb-audit hangs.

MEGA Audio configured just for slot 4:
NB. Register $E (IER) reads back with b7=0
(Also: Timer1-Latch is $0000, but the Timer1 count isn't constrained & Timer2 seems to always read as $00FF!)

After patching Detect6522() to ignore the check for IER.b7==1 and use Timer1 (not Timer2):
Test: 10:02:00
Expected:E0 Actual:60
The test is failing, as again, register $E reads back with b7=0

Using version v0.2, and a Sound Speech I in a IIe, exits to monitor after the initial check. I get this output:

mb-audit v0.2-beta, 2021
                    1  2  3  4  5  6  7
              $00:           ?
              $80:           ?
               SP:
65C02 detected

*

Hi @tomcw

is there a reference that describes which mockingboards used NMI?
How was a programmer to know which boards used only IRQ?

It hangs in this code:

+	lda		#0
	sta		isrIFR

-		lda		#IxR_TIMER1			; T1 int from one-shot mode
		and		isrIFR
		beq		-

I think the IIgs interrupt handling code is longer than 0x0202 cycles, so the timer has already expired... and so the timer1 interrupt never fires after setting isrIFR = 0.

On cea2, Johnson reported:

my card is SD Music Deluxe [...] from Ian Kim that only compatible with Mockingboard, this is a good test to see how good it is, this card can play SkyFox but the tempo or sound not perfect in Music Construction Set.
:
Since it's OPL3 emulate AY8913, it fail is reasonable, maybe you can add this to the record later, the card is using W65C22S (PLCC).

It currently detects Echo+ as an SDM, and on real h/w the playback is silent.

With a Mockingboard C and a Sound Speech I installed at the same time. The detection works correctly and the multi card tests pass. When mb-audit gets to the tone test, the tone test does not work for the card in the highest slot. It does work ok for the next card. If I swap the cards around, the card in the highest slot is always the one that does not work.

This work ok in AppleWin with its MB C pair, so feels like its something to do with the S/SI and MB C combo. I have another MB A, but its failing the 6522 test 10 (10:00:00) so can't test this with another combo at the moment.

tested with v0.4-beta.

Sorry, this is a bit of an edge case, but had me puzzled for awhile when trying the tone tests until I worked out what was happening.