View previous topic :: View next topic |
Author |
Message |
Fabri
Joined: 22 Aug 2005 Posts: 275
|
PIC18F66K22 and ADC strange problem |
Posted: Fri Mar 28, 2014 10:14 am |
|
|
Hi to all,
I have problem with PIC18F66K22 and ADC. I use An0 and with higher voltage of 3Vdc I read incorrect value.
For example:
AN0 = 1 Vdc read 819 dec.
AN0 = 2 Vdc read 1630 dec. (819 x 2)
AN0 = 3 Vdc read 2445 dec. (819 x 3)
AN0 = 4 Vdc read 2273 dec. instead of 3276 !.
Voltage is on AN0 pin !
This happen in my application and also in this test program.
This is code:
Code: |
#include "18F66K22.h"
#if defined(debug)
#fuses HSM,DEBUG,NOWDT,NOPROTECT,NOPUT,NOBROWNOUT,NOCPD,BORV18,NODEBUG,NOCPB,STVREN,NOPLLEN,NOFCMEN,SOSC_DIG
#device ADC=12 icd=true
#else
#fuses HSM,NODEBUG,WDT,PROTECT,PUT,CPD,BORV18,CPB,STVREN,NOPLLEN,NOFCMEN,SOSC_DIG
#device ADC=12
#endif
//#device *=16
#BYTE PORTA = 0x0F80
#BYTE PORTB = 0x0F81
#BYTE PORTC = 0x0F82
#BYTE PORTD = 0x0F83
#BYTE PORTE = 0x0F84
#BYTE PORTF = 0x0F85
#BYTE PORTG = 0x0F86
#byte ODCON1= 0xF29
#byte ODCON2= 0xF28
#byte ODCON3= 0xF27
#byte LATC = 0xF8B
#fill_rom 0x00
#use delay(clock=64000000,RESTART_WDT)
long ad_var;
void main() {
int mm;
//Definisci I/0 in base allo schema
restart_wdt();
set_tris_a(0b00000001);
set_tris_b(0x00);
set_tris_c(0x00);
set_tris_d(0x00);
set_tris_e(0x00);
set_tris_e(0x00);
set_tris_f(0x00);
PORTA = (0x00);
PORTB = (0x00);
PORTC = (0x00);
PORTD = (0x00);
PORTE = (0x00);
PORTF = (0x00);
PORTG = (0x00);
setup_oscillator(OSC_PLL_ON);
SETUP_ADC_PORTS(sAN0||VSS_VDD);
setup_adc (ADC_CLOCK_DIV_64);
set_adc_channel(0);
delay_ms(30);
while(true) {
ad_var=read_adc(ADC_START_AND_READ);
delay_ms(100);
}
}
|
Somebody can help me ?
Thanks |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19494
|
|
Posted: Fri Mar 28, 2014 10:30 am |
|
|
Unfortunately the ADC on this chip has problems.
Quote from the errata sheet:
"The ADC will not meet the Microchip standard
ADC specification. ADC may be usable if tested
at the user end. The possible issues are high offset
error, high DNL error and multiple missing
codes. The ADC can be tested and used for
relative measurements."
Try running at 16MHz and see if things improve. I'd suspect the 'missing codes' problem.... |
|
|
Fabri
Joined: 22 Aug 2005 Posts: 275
|
|
Posted: Fri Mar 28, 2014 11:02 am |
|
|
Ok Ttelmah, but the problem occurs when input voltage is higher than 3V. With 4V I read a smaller value than 3V.
In your opinion is chip problem ? can be configuration ?
I can try at 16Mhz but I need to use PLL in future.
I call microchip support because someone suggest me this chip.
Regards,
Fabri |
|
|
Fabri
Joined: 22 Aug 2005 Posts: 275
|
|
Posted: Fri Mar 28, 2014 11:19 am |
|
|
At 16Mhz work better, at 4V I read 3260. In any case it isn't linear.
I'm waiting help from microchip support. |
|
|
temtronic
Joined: 01 Jul 2010 Posts: 9220 Location: Greensville,Ontario
|
|
Posted: Fri Mar 28, 2014 11:30 am |
|
|
I know, you've probably thought of this but...
Perhaps you can choose another PIC? I'd ask Microchip for their help on a
substitute.
hth
Jay |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19494
|
|
Posted: Fri Mar 28, 2014 11:45 am |
|
|
Agreed.
I suggested going slower, because I have seen the missing codes problem (individual bits that just don't set in some parts of the range), disappear at slower speeds.
Are you sure it is not linear at the lower rate?.
3260
2445
1630
819
4
Is linear, but with a four offset at the bottom - another part of the erratum...
I'm afraid these chips are currently ones I'd avoid if at all possible.
Best Wishes |
|
|
Fabri
Joined: 22 Aug 2005 Posts: 275
|
|
Posted: Fri Mar 28, 2014 12:00 pm |
|
|
Linear and offset is not a really problem.... I'm afraid about use of PLL because I need it to use a graphic display as I told you in other post.
I'm testing also PIC18F67K22 but it's of same family so probably has same problem.
I need another pic with same pinout because I developed PCB and I haven't time to change.
Well, thanks for your support.
Regards,
Fabri |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19494
|
|
Posted: Fri Mar 28, 2014 1:57 pm |
|
|
What smoothing have you got on the Vcore pin. This is much more fussy than it should be. I've tended to stick with the 67J94, which is 90% the same on pinout, but has a few things that are rather different (USB etc.).
Could you get away with running the ADC in 10bit mode?. Many of the problems disappear when you do, but I know I've seen quite a few other problems at 64MHz.
Best Wishes |
|
|
Fabri
Joined: 22 Aug 2005 Posts: 275
|
|
Posted: Sat Mar 29, 2014 1:36 am |
|
|
I don't use USB, just PWM for buzzer, AD, RS232 and standard I/O.
With 16Mhz and PIC18F66K22 application work fine but I need speed only to work with GLCD as I told you in other post some times ago. I already used PIC18F46K22 with PLL and 10 bit AD without problem. I'm not interest to have 12 bit resolution and I can downgrade to 10 bit.
Can I switch PIC to work wih 10 bit resolution from 0 to 5 Vdc or must I reduce input voltage in pin and use first 10 bits ?
What kind of problem you found at 64 Mhz ?
Regards,
Fabri |
|
|
temtronic
Joined: 01 Jul 2010 Posts: 9220 Location: Greensville,Ontario
|
|
Posted: Sat Mar 29, 2014 5:29 am |
|
|
Fabri
I'm curious as to why you went from a working PIC18F46K22 to the '66'? Does it have some peripheral you now must use ?
jay |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19494
|
|
Posted: Sat Mar 29, 2014 5:57 am |
|
|
Agreed with Temtronic's comments.
On these chips at 64Mhz, I found the serial, and the ADC both stopped working reliably.
Best Wishes |
|
|
Fabri
Joined: 22 Aug 2005 Posts: 275
|
|
Posted: Sat Mar 29, 2014 10:45 am |
|
|
I changed to PIC18F66K22 just to have more I/O and, in second time, with PIC67K22 more flash. Actually I'm 95% with PIC18F66K22. There's no space to use GLCD so I must have more flash. |
|
|
Fabri
Joined: 22 Aug 2005 Posts: 275
|
|
Posted: Sat Mar 29, 2014 10:47 am |
|
|
I forgot to tell you it was a suggestion from microchip !!! |
|
|
Fabri
Joined: 22 Aug 2005 Posts: 275
|
|
Posted: Fri Apr 04, 2014 10:07 am |
|
|
I solved
with PLL I must set TAD with:
Quote: |
setup_adc (ADC_CLOCK_DIV_64|ADC_TAD_MUL_20);
|
In this way ad conversion is steady. |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19494
|
|
Posted: Fri Apr 04, 2014 1:04 pm |
|
|
With your long delay, you shouldn't need any Tad time.
However there are some other chips, that won't work properly with Tad set to zero. Sounds as if this is another.
You should report this to MicroChip, since it needs adding to the errata.... |
|
|
|