33 replies to this topic

[Marco]

This is possible by misusing the calc's COM port.

You need no special hardware or controllers for it, what you need is nothing more than a 3.5mm to 2.5mm stereo jack adapter and a headphone. Plug the Phone into the adapter and the adapter into the calc - the COM signal is strong enough that you can hear it.

By writing defined data to the port then you can create every audio frequency you want

Try THIS, it's a simple demo program (screenshot below) that allows you to play all twelve notes of the scale over five octaves. It also supports a two channel mode, but this mode's quality isn't that good.



(Note: if it has a too low volume, take a headphone with a smaller resistor, or plug an amplifier between. I used an earphone, and it was loud enough. Btw I hope this won't demage the COM port? Naturally it shouldn't I think...)

[Skalda]

no, it shouldn't... i thought about such sounds over COM in the past, too. how do you generate the square over tx? do you use break-bit uart function (if there is one on afx) or do you just send null-bits when the edge should appear? what about stop-bits? which circuit is actually controling COM on afx?

[4nic8]

I gonna try it as soon as possible

[Bob Vila]

this is cool, now we can use music in the games

[Andy.Davies]

but surely there is not much point, it is somthing you have to carry round with you, and that defeates the point of portable....

[Marco]

True, it needs most os the CPU power, cause you have to write to the port in constant intervalls of time (according to the audio frequency you want). We can use it for games nevertheless :

- background music for low performance games (you can innterrupt sound output by int 1C for example and use it for tasks necessarry for the game; if the int takes not a too long time, it doesn't disturb sound output much). For example, I could imagine a tetris version with background music (think I'll try one)

- if it's a high performance game, you can use sound for intros or menus, and for sound effects (for example, halting the game for about 0.1s and playing a sound when shooting on the enemy or sth. like this)

how do you generate the square over tx? do you use break-bit uart function (if there is one on afx) or do you just send null-bits when the edge should appear? what about stop-bits? which circuit is actually controling COM on afx

Don't know about the circuit. Bits are putted out as square, I think, so an edge occurs when one bit is 1 and the next 0, or one 0 and the next 1. Additionally, a stop-bit seems always to cause an edge, no matter what the other bits are.

Another way we could use background music also for high performance games was, when we could prepare a buffer that is to be send through the port, not single bytes only. We could play (send) it then without having a care for it. But I think the port doesn't provide such a functionality?

[4nic8]

the 4nic8's soundsystem for an AFX:

http://www.casiopowe...oundsystem3.gif

@Marco: It sounds really good
btw. Could you give the source of the sound system library? Because then we could add simple sounds to MLC

[CrimsonCasio]

eh, id advise against it...

[4nic8]

I can feel the bit... aha aha I like it


@Marco:
http://www.ticalc.or.../338/33890.html
this program would look cool on an AFX, wouldn't it?

[huhn_m]

@4ni8:

does this speaker work? because I have the same fabric but I think it is not loud enough

[Orwell]

- if it's a high performance game, you can use sound for intros or menus, and for sound effects (for example, halting the game for about 0.1s and playing a sound when shooting on the enemy or sth. like this)
(...)
Another way we could use background music also for high performance games was, when we could prepare a buffer that is to be send through the port, not single bytes only. We could play (send) it then without having a care for it. But I think the port doesn't provide such a functionality?

<{POST_SNAPBACK}>

The problem is, that for high performance games a delay of even 0.1s is quite too long. And there is no known method to send such packs through the port while the game is still "running"...

What we could do is use the 0x1C or 0x53 interrupt, but not to interrupt the music:
We could run the game normally, and at each call of those int send a byte through the communication port to play a sound... so we could create music at a constant range of 50 Hz

Btw, i'm interested about the kind of data's that must be send to make such sounds

[4nic8]

@huhn_m: surely it's not *very* loud but it wokrs fine for me

[Mohamed]

@4ni8:

does this speaker work? because I have the same fabric but I think it is not loud enough

<{POST_SNAPBACK}>

I think you only need to connect it to an amplified speaker set, like your Hi-Fi or PC Speaker system, and you'll have loud enough system, beside not draining the calculator's battery.

[Skalda]

but surely there is not much point, it is somthing you have to carry round with you, and that defeates the point of portable....

<{POST_SNAPBACK}>

what about piezo connected to COM, inbuilt in calculator (eg behind backup battery) with simple amplifier with one transistor supplied from main batteries (you can listen even common data transfers)

[Deimos]

I was thinking about the same thing - even a piezoelectric speaker WITH a plug would be small enough to put in the battery compartment.


EDIT:

Hehe, it works, it works really good. Sound scale is much better than i expected in fact.

[Marco]

@Marco:
It sounds really good 
btw. Could you give the source of the sound system library? Because then we could add simple sounds to MLC

I'll put it to the file sharing as soon as possible (Pascal version is ready yet, currently I'm porting it to C as no one likes Pascal). Soundlib btw also has a function now for playing a kind of midi sequences.

@Marco:
http://www.ticalc.or.../338/33890.html
this program would look cool on an AFX, wouldn't it?

O yeah, it would! ==> midi tracker in progress yet (didn't know that there was a beat box for TI already- but what I read from the text I think it can play drum sounds only?)

does this speaker work? because I have the same fabric but I think it is not loud enough

I'm a bit surprised by the speaker, too - I'd not expect that you'd hear it even in a speaker loud enough, that's cool .
The thing with the piezo I also think is a good idea!

@huhn:
If your speaker is too quit, try one of these small earphones you get delivered with a discman, this should be loud enough then.

What we could do is use the 0x1C or 0x53 interrupt, but not to interrupt the music:
We could run the game normally, and at each call of those int send a byte through the communication port to play a sound... so we could create music at a constant range of 50 Hz 

Yeah, this was also the first I thought about before doing vice versa. But I guess we can't create a scale upon that (although a range from 25 to 50Hz was theoretically enough to do it) cause of the constant timer intervalls of 1/50s.

However, we could use it for 50, 25 and may be 37.5 Hz sounds, so it still will be applicable for some sound effects

[qwerty841]

for my ti calc, i just took one end of a unit to unit cable and attached a piezo buzzer.

[Deimos]

Piezo buzzer works fine - it's loud enough to be heard, but not too loud.

[Marco]

AFXSound Library V. 0.9 is up in the File Sharing right now, providing a Pascal and a C version, a doc and two demo programs.

Note, that writing the doc took as long the lib itself did (I really hate writing docs...)

Btw, C version hasn't been tested yet, altough the compiler gives no errors for it. Especially for the MIDI track routines for C I don't know if they'll work, so just try out (and tell me if or ifn't so)

[4nic8]

I've tried the C lib.

And made Sweet Sound v 0.1 (it's mostly port from snddemo.pas by you). I works almost fine. It generates sound very well. But I found no way to terminate the program (without using P button). It's really weird I guess it can be a bug in soundclose() (or in soundopen() or soundplay() ).


Sweet Sound v 0.1 (with source):
http://max.iem.pw.ed.../sweets_010.rar

btw. afxsound.c doesn't work with Digital Mars compiler

However, way to go Marco

[Marco]

Well, it doesn't work with Digital Mars? Does the compiler give any errors or warnings or is the error in runtime only?

Ah and about the termination problem - it's a mistake I made in snddemo.pas: to leave soundplay() you should prefer setting soundstop variable to 1 instead of calling closesound(), else the port will be closed while soundplay() is running and soundplay() might hang thus (I haven't thought of this when writing the demo ).

So it should work when calling soundclose() only when soundplay() has returned already after setting soundstop.

Thx for testing it btw (would have been much efforts for me to do so as I never yet wrote a C prog running on AFX)

[4nic8]

Ok! Now it works fine.

Here is the Sweet Sound 0.2 + source:
http://max.iem.pw.ed.../sweets_020.rar

btw. Marco, you can add Sweet Sound to your lib package as a C example if you want .

[4nic8]

Digital Mars Compiler gives:

---------- DM ----------
  mov word [offset channels],0xFFFF
           ^
E:\CALCS\CASIO\PROG\sound\moje\afxsound.c(161) : Error: Expected assembler directive PTR to follow assembler cast
  mov word [offset channels+5],0xFFFF
           ^
E:\CALCS\CASIO\PROG\sound\moje\afxsound.c(162) : Error: Expected assembler directive PTR to follow assembler cast
  mov word [offset channels+2],1
           ^
E:\CALCS\CASIO\PROG\sound\moje\afxsound.c(163) : Error: Expected assembler directive PTR to follow assembler cast
  mov word [offset channels+7],1
           ^
E:\CALCS\CASIO\PROG\sound\moje\afxsound.c(164) : Error: Expected assembler directive PTR to follow assembler cast
    mov [offset channels+2],dx
                             ^
E:\CALCS\CASIO\PROG\sound\moje\afxsound.c(188) : Error: Illegal type/size of operands for the mov instruction
Fatal error: too many errors
--- errorlevel 1

Output completed (0 sec consumed) - Normal Termination

[Marco]

btw. Marco, you can add Sweet Sound to your lib package as a C example if you want

That's fine, I'll add it

^^ Got a DM compiler meanwhile, hope I'll fix it soon...

[Skalda]

Digital Mars Compiler gives:
[code=auto:0]---------- DM ----------
  mov word [offset channels],0xFFFF
          ^
E:\CALCS\CASIO\PROG\sound\moje\afxsound.c(161) : Error: Expected assembler directive PTR to follow assembler cast

<{POST_SNAPBACK}>

try mov word ptr ds:[&channels+xxx], yyy
ptr is here a keyword to accent pointer; word is necessary to determinate the length of second operand yyy which is immediate (sometimes required even if yyy isn't immediate, as in case of dx), xxx is shift from base adress; perhaps, offset of channels has not to be declared with offset keyword in DMC's assembler, try &

[4nic8]

mov word ptr ds:[offset channels], 0xFFFF

DM error:

---------- DM ----------
  mov word ptr ds:[offset channels], 0xFFFF
                                          ^
E:\CALCS\CASIO\PROG\sound\moje\afxsound.c(161) : Error: Illegal type/size of operands for the mov instruction
  mov word [offset channels],0xFFFF

[Marco]

Fixed right now: In DM inline Assembler effective addresses notation is the same as in NAsm:

dw myvar
mov ax,[myvar]

You have to do a type casting similar to C however:

mov (word) [(void) channels],0xFFFF

That works so far, but it's a really strange syntax for an assembler nevertheless (and it's also incompatible with borland inline assembler, so I'll make a third version afxsound_dm.c). They could have documented it anywhere at least

[Skalda]

...but it's a really strange syntax for an assembler nevertheless (and it's also incompatible with borland inline assembler...

<{POST_SNAPBACK}>

most of programmers are used to intel asm syntax (tasm, masm, a86, nasm...) have you ever tried AT&T asm syntax, whichGCC gnu-compiler uses? my first mov was enough (you must declare all used variables, and operands are swapped and so on...)

now seriously: have you noticed (in DMC) any difference between
mov dx, myvar
mov dx, &myvar
mov dx, [myvar]
mov dx, [&myvar]
it seems that it work all the same (but logically, when brackets are used as dereference; & have to be used as reference and vice versa)

[Orwell]

That works so far, but it's a really strange syntax for an assembler nevertheless (and it's also incompatible with borland inline assembler, so I'll make a third version afxsound_dm.c). They could have documented it anywhere at least

<{POST_SNAPBACK}>

Please do not make another file, just add some

#ifdef __DMC__
... // DM special code
#else
... // other compilers
#endif

stuff in your code


But in this case, I see you don't use SI or DI... you could use it for that (of course there will be one instruction more, but its not big deal )

If you have

void soundplay() {

  mloop:
  asm {

    mov ax,[offset channels+2]
    mov bx,[offset channels]
    cmp ax,bx
    jna m1

then write

void soundplay() {

  asm lea di, channels;
  mloop:
  asm {

    mov ax,[di+2]
    mov bx,[di]
    cmp ax,bx
    jna m1

etc

[Marco]

AFXSound upgraded from V. 0.9 to 0.91 :

- works also with Digital Mars compilers now
- contains 4nics C demo
- the closesound bug in the other two demos has been fixed

[kucalc]

It will be possible when CASIO adds serial communication functions to the SDK for the fx-9860.

[kucalc]

Yes maybe on the fx-9860SD. I asked CASIO for serial communications support. We will see what happens...