 |
 |
| View previous topic :: View next topic |
| Author |
Message |
andyd
Joined: 08 Mar 2007
Posts: 30
|
| Weird problem with values |
 Posted: Wed May 02, 2007 9:39 am |
 |
|
I've recently changed PICs from an F88 to an F886 as I needed hardware I2C and now code I had running fine on the 88 doesn't want to work on the 886.
Basically, I have a function which decodes compressed audio (4 bit ADPCM to 16 bit PCM), but when I encode the audio using a function written in MATLAB, it decodes fine and the PIC produces a nice range of values from -32000 to + 32000, when I use my encoder it produces values that just hover around the same low value the whole time. Now I know this would point to my encoder being the problem, but the two encoders produce incredibly similar values and, when using the F88, it decoded the same files produced by my encoder flawlessly. Are there any changes in the 886's architecture compared to the 88 which would result in something like this happening?
This is the decoding routine:
| Code: | signed long ADPCMDecoder(char code) // ADPCM decoding routine
{
// Set previous values of predicted sample and step size index point
predsample = prevsample;
index = previndex;
step = StepSizeTable[index]; // Find quantiser step size
// Inverse quantise ADPCM code
diffq = step >> 3;
if(code & 4) diffq += step;
if(code & 2) diffq += step>>1;
if(code & 1) diffq += step>>2;
// Add the difference to the predicted sample
if( code & 8 ) predsample -= diffq;
else predsample += diffq;
// Check for and stop overflow of the new predicted sample
if(predsample > 30000) predsample = 30000;
else if(predsample < -30000) predsample = -30000;
index += IndexTable[ code]; // Find new quantiser step size
// Check for and stop overflow of the new quantizer step size index
if( index < 0 ) index = 0;
if( index > 88 ) index = 88;
// Save predicted sample and quantiser step size index point
prevsample = predsample;
previndex = index;
return((long)predsample); // Return decoded sample
} |
I've tried pasting in the values produced by decoding each encoder's audio, but the formatting is terrible unfortunately. |
|
 |
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Wed May 02, 2007 10:57 am |
|
|
Post the declarations of the global variables that are used in your routine.
Also post your compiler version. (NB: We don't remember your compiler
version from one post to the next. Always include it in your posts). |
|
|
andyd
Joined: 08 Mar 2007
Posts: 30
|
|
| Posted: Wed May 02, 2007 1:11 pm |
|
|
Apologies, compiler version is 4.013 (I think, that's what PCM, PCB and PCH are listed as, IDE is 4.010). MPLAB version is 7.52 if that's relevant, thinking about it I did have to upgrade it from 7.50 so I could program this new PIC.
Global variable declarations are:
| Code: | // Index changes
const int IndexTable[16] =
{
0xff, 0xff, 0xff, 0xff, 2, 4, 6, 8,
0xff, 0xff, 0xff, 0xff, 2, 4, 6, 8
};
// Quantiser step sizes
const long StepSizeTable[89] =
{
7, 8, 9, 10, 11, 12, 13, 14, 16, 17,
19, 21, 23, 25, 28, 31, 34, 37, 41, 45,
50, 55, 60, 66, 73, 80, 88, 97, 107, 118,
130, 143, 157, 173, 190, 209, 230, 253, 279, 307,
337, 371, 408, 449, 494, 544, 598, 658, 724, 796,
876, 963, 1060, 1166, 1282, 1411, 1552, 1707, 1878, 2066,
2272, 2499, 2749, 3024, 3327, 3660, 4026, 4428, 4871, 5358,
5894, 6484, 7132, 7845, 8630, 9493, 10442, 11487, 12635, 13899,
15289, 16818, 18500, 20350, 22385, 24623, 27086, 29794, 32767
};
// ADPCM decoder variables
long step; // Quantiser step size
signed int32 predsample; // Predicted ADPCM sample (decoder output)
signed int32 prevsample; // Previous sample
signed long diffq; // Dequantised predicted difference
signed int index; // Step size table index point
signed int previndex; // Previous step size table index point
|
Many thanks,
Andy |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Wed May 02, 2007 1:22 pm |
|
|
It's probably going to be a problem in 4.013.
If you can combine the code that you posted into a complete test
program that can be dropped into MPLAB and compiled, and run
in the MPLAB simulator (with printf output going to 'UART1' in the
output window), then I can test it. I don't have PCM 4.013.
The closest I have is 4.014. But I can test it with that, and also
with 4.033 and tell you the results. Post a test program that
prints out some easily verifiable results, so that you can tell at
a glance whether the program is running properly or not. |
|
|
andyd
Joined: 08 Mar 2007
Posts: 30
|
|
| Posted: Wed May 02, 2007 5:18 pm |
|
|
Thanks for your help 
I've modified my code so it should hopefully compile first time for you (it does for me) but was unsure as to how to get the printf statements to print to UART1, so haven't changed those... I've put the first 200 "code" values my encoder produces in an array just to test, as they're normally read from an EEPROM. The program should read in a code byte from the array, print its value, decode the upper nibble and print that value, then decode the lower nibble and print that value.
Thanks again for your help,
Andy
| Code: | #include <16f886.h>
#include <stdio.h>
#include <stdlib.h>
#fuses INTRC_IO, NOWDT, NOLVP // PIC setup
#use delay(clock = 8000000) // Delay setup
#use rs232(baud = 9600, xmit = PIN_C6, rcv = PIN_C7, stream = PC) // RS232 setup
#use I2C(Master, sda = PIN_C4, scl = PIN_C3, FAST = 400000, FORCE_HW) // I2C setup
// Define registers & bits
#byte PORTA = 0x05
#byte PORTB = 0x06
#byte CCP2CON = 0x1D
#byte CCPR2L = 0x1B
#byte OSCCON = 0x8F
#byte ANSEL = 0x188
#byte T1CON = 0x10
#byte T2CON = 0x12
#byte PR2 = 0x92
// ADPCM code bytes
const char codeTable[200] =
{
167, 119, 255, 195, 53, 10, 202, 131, 53, 10, 202, 131,
53, 137, 217, 130, 51, 11, 233, 130, 66, 11, 189, 1, 81,
10, 187, 2, 112, 26, 186, 131, 67, 11, 233, 130, 66, 11,
189, 1, 81, 10, 187, 2, 112, 26, 186, 131, 67, 138, 248,
146, 50, 11, 203, 2, 112, 26, 186, 131, 67, 11, 233, 130,
66, 11, 189, 1, 81, 10, 187, 2, 112, 26, 186, 131, 67, 11,
233, 130, 66, 11, 189, 1, 81, 10, 187, 2, 112, 26, 186, 131,
67, 11, 233, 130, 66, 11, 189, 1, 81, 10, 187, 2, 112, 26, 186,
131, 67, 11, 233, 130, 66, 11, 189, 1, 81, 10, 187, 2, 112, 26,
186, 131, 67, 11, 233, 130, 66, 11, 189, 1, 81, 10, 187, 2, 112,
26, 186, 131, 67, 138, 248, 146, 50, 11, 203, 2, 112, 26, 186, 131,
67, 11, 233, 130, 66, 11, 189, 1, 81, 10, 187, 2, 112, 26, 186, 131,
67, 11, 233, 130, 66, 11, 189, 1, 81, 10, 187, 2, 112, 26, 186, 131,
67, 11, 233, 130, 66, 11, 189, 1, 81, 10, 187, 2, 112, 26, 186, 131
};
// Index changes
const int IndexTable[16] =
{
0xff, 0xff, 0xff, 0xff, 2, 4, 6, 8,
0xff, 0xff, 0xff, 0xff, 2, 4, 6, 8
};
// Quantiser step sizes
const long StepSizeTable[89] =
{
7, 8, 9, 10, 11, 12, 13, 14, 16, 17,
19, 21, 23, 25, 28, 31, 34, 37, 41, 45,
50, 55, 60, 66, 73, 80, 88, 97, 107, 118,
130, 143, 157, 173, 190, 209, 230, 253, 279, 307,
337, 371, 408, 449, 494, 544, 598, 658, 724, 796,
876, 963, 1060, 1166, 1282, 1411, 1552, 1707, 1878, 2066,
2272, 2499, 2749, 3024, 3327, 3660, 4026, 4428, 4871, 5358,
5894, 6484, 7132, 7845, 8630, 9493, 10442, 11487, 12635, 13899,
15289, 16818, 18500, 20350, 22385, 24623, 27086, 29794, 32767
};
// ADPCM decoder variables
long step; // Quantiser step size
signed int32 predsample; // Predicted ADPCM sample (decoder output)
signed int32 prevsample; // Previous sample
signed long diffq; // Dequantised predicted difference
signed int index; // Step size table index point
signed int previndex; // Previous step size table index point
signed long ADPCMDecoder(char code) // ADPCM decoding routine
{
// Set previous values of predicted sample and step size index point
predsample = prevsample;
index = previndex;
step = StepSizeTable[index]; // Find quantiser step size
// Inverse quantise ADPCM code
diffq = step >> 3;
if(code & 4) diffq += step;
if(code & 2) diffq += (step / 2);
if(code & 1) diffq += (step / 4);
// Add the difference to the predicted sample
if( code & 8 ) predsample -= diffq;
else predsample += diffq;
// Check for and stop overflow of the new predicted sample
if(predsample > 32767) predsample = 32767;
else if(predsample < -32768) predsample = -32768;
index += IndexTable[ code]; // Find new quantiser step size
// Check for and stop overflow of the new quantizer step size index
if( index < 0 ) index = 0;
if( index > 88 ) index = 88;
// Save predicted sample and quantiser step size index point
prevsample = predsample;
previndex = index;
return((long)predsample); // Return decoded sample
}
void main(void)
{
int16 i; // Counter variable
char code; // ADPCM byte from EEPROM
signed long sample; // Decoded sample
OSCCON = 0x71; // Use internal 8 MHz oscillator
set_tris_c(0xFC); // Setup PORTC
ANSEL = 0x00; // Turn off analogue inputs
ANSEL = 0x00; // Turn off analogue inputs
T1CON = 0x0; // Disable TIMER1
T2CON = 0x4; // Turn on TIMER2 and set pre- and postscalers to 1
PR2 = 0xF9; // Set PR2 value to 249 to give PWM frequency of 8 kHz
CCP2CON = 0xC; // Setup CCP2 as PWM
setup_uart(9600, PC); // Setup hardware serial port
prevsample = 0; // Clear ADPCM previous sample
previndex = 0; // Clear ADPCM previous index
index = 0;
diffq = 0;
step = StepSizeTable[index];
for(i=0;i<200;i++)
{
code = codeTable[i];
fprintf(PC, "%d\t", code);
sample = ADPCMDecoder((code>>4) & 0x0f); // Decode upper nibble
fprintf(PC, "%ld\t", sample);
sample = ADPCMDecoder(code & 0x0f); // Decode lower nibble
fprintf(PC, "%ld\n", sample);
}
} |
|
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Wed May 02, 2007 5:47 pm |
|
|
Here's the first part of output when compiled with vs. 4.033.
Does it look correct ? What do you get ?
I changed your printf statements slightly by inserting a field
width of six ("%6ld"), so it would be nicely formatted below.
| Code: |
-89 -3 8
119 38 101
-1 -35 -328
-61 -707 -350
53 -27 436
10 497 217
-54 -242 -550
-125 -606 -249
53 74 537
10 598 318
-54 -141 -449
-125 -505 -148
53 175 638
-119 577 409
-39 -152 -375
-126 -443 -135
51 257 614
11 660 366
-23 -132 -336
-126 -397 -117
66 342 650
11 706 349
-67 26 -437
1 -376 -208
81 353 576
10 644 336
-69 -56 -413
2 -367 -157
112 417 499
26 722 382
-70 -49 -329
-125 -380 -57
67 322 679
11 725 431
-23 -67 -271
-126 -332 -52
66 407 715
11 771 414
-67 91 -372
1 -311 -143
81 418 641
10 709 401
-69 9 -348
2 -302 -92
112 482 564 |
|
|
|
andyd
Joined: 08 Mar 2007
Posts: 30
|
|
| Posted: Thu May 03, 2007 5:38 am |
|
|
Thanks for taking the time to do that! Unfortunately it would seem it's not down to the compiler as those are exactly the same values that I get. Could you possibly try it again with these values (produced by the other encoder) and if they give a nice range of values, can you think of anything which could be causing the difference?
Many thanks,
Andy
| Code: | // ADPCM code bytes
const char codeTable[200] =
{
7, 119, 223, 255, 87, 116, 252, 128, 67, 9, 188, 145, 67, 25, 204,
145, 67, 25, 218, 145, 53, 25, 203, 129, 52, 25, 203, 145, 68, 25,
203, 145, 82, 25, 189, 128, 66, 25, 188, 145, 67, 25, 204, 145, 67,
25, 218, 145, 53, 25, 203, 129, 52, 25, 203, 145, 68, 25, 203, 145,
82, 25, 189, 128, 66, 25, 188, 145, 67, 25, 204, 145, 67, 25, 218,
145, 53, 25, 203, 129, 52, 25, 203, 145, 68, 25, 203, 145, 82, 25,
189, 128, 66, 25, 188, 145, 67, 25, 204, 145, 67, 25, 218, 145, 53,
25, 203, 129, 52, 25, 203, 145, 68, 25, 203, 145, 82, 25, 189, 128,
66, 25, 188, 145, 67, 25, 204, 145, 67, 25, 218, 145, 53, 25, 203,
129, 52, 25, 203, 145, 68, 25, 203, 145, 82, 25, 189, 128, 66, 25,
188, 145, 67, 25, 204, 145, 67, 25, 218, 145, 53, 25, 203, 129, 52,
25, 203, 145, 68, 25, 203, 145, 82, 25, 189, 128, 66, 25, 188, 145,
67, 25, 204, 145, 67, 25, 218, 145, 53, 25, 203, 129, 52, 25, 203,
145, 68, 25, 203, 145
}; |
|
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Thu May 03, 2007 11:19 am |
|
|
Here's results for vs. 4.033 with your latest table:
| Code: | 7 0 11
119 41 104
-33 4 -195
-1 -625 -1550
87 -93 2817
116 9053 17076
-4 896 -19916
-128 -22714 -20171
67 641 20227
9 22770 15833
-68 1118 -16082
-111 -23019 -16713
67 487 16674
25 22980 17247
-52 1611 -17309
-111 -24939 -18002
67 918 18723
25 25660 19354
-38 -1668 -15658
-111 -23288 -16351 |
If you think there's a bug, then upgrade the compiler. 4.013 is such
an early version, that there could easily be some bug that affects
your code. |
|
|
andyd
Joined: 08 Mar 2007
Posts: 30
|
|
| Posted: Thu May 03, 2007 5:37 pm |
|
|
| Unfortunately I don't think it is a compiler issue as you get the same values I do when it's compiled with 4.033, I just find it really strange that it all went wrong when I changed PICs. I guess if you can't think of anything else then I'll have to change my encoder so it produces exactly the same values as the MATLAB one, though I'm not sure exactly how I'll do that as it seems to be following exactly the same steps...! |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Thu May 03, 2007 5:41 pm |
|
|
I don't know what values you're expecting to get.
What do you consider to be a "good" output ? |
|
|
andyd
Joined: 08 Mar 2007
Posts: 30
|
|
| Posted: Fri May 04, 2007 2:59 am |
|
|
| The ones I got from the second set - a nice range of values in the +/- tens of thousands, rather than hovering around a couple of hundred like the first ones do |
|
|
andyd
Joined: 08 Mar 2007
Posts: 30
|
|
| Posted: Fri May 04, 2007 8:45 am |
|
|
Also, please could you show me how you modified the printf statement to print into MPLAB so that I don't have to stick the code onto a chip everytime I change something?
Andy |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
