Hi! I have a problem using fractional baudrate divider on LPC2136.
When i'm not using FDR i get baudrate (for example crystal is 14.318
MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00. baudrate 55930, error
about 3% from 57600. Everything is seems to be ok, UART
errors with this setting are very rare. But when i using FDR to set
baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
baudrate = 57565, error about 0.6% i get UART transfer error on
almost every byte. This happens on both UARTS and with any baudrate
from 9600 to 115200.
Can anyone help me with this problem?
LPC2136 fractional divider for UART problem
Started by ●July 28, 2007
Reply by ●July 28, 20072007-07-28
maxmatveev wrote:
> Hi! I have a problem using fractional baudrate divider on LPC2136.
> When i'm not using FDR i get baudrate (for example crystal is 14.318
> MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00. baudrate > 55930, error about 3% from 57600. Everything is seems to be ok, UART
> errors with this setting are very rare. But when i using FDR to set
> baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
> baudrate = 57565, error about 0.6% i get UART transfer error on
> almost every byte. This happens on both UARTS and with any baudrate
> from 9600 to 115200.
> Can anyone help me with this problem?
Make sure your crystal is really 14.318MHz and not the baud rate
multiple, 14.7456MHz.
Regards
-Bill Knight
R O SoftWare
> Hi! I have a problem using fractional baudrate divider on LPC2136.
> When i'm not using FDR i get baudrate (for example crystal is 14.318
> MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00. baudrate > 55930, error about 3% from 57600. Everything is seems to be ok, UART
> errors with this setting are very rare. But when i using FDR to set
> baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
> baudrate = 57565, error about 0.6% i get UART transfer error on
> almost every byte. This happens on both UARTS and with any baudrate
> from 9600 to 115200.
> Can anyone help me with this problem?
Make sure your crystal is really 14.318MHz and not the baud rate
multiple, 14.7456MHz.
Regards
-Bill Knight
R O SoftWare
Reply by ●July 28, 20072007-07-28
maxmatveev wrote:
>
> Hi! I have a problem using fractional baudrate divider on LPC2136.
> When i'm not using FDR i get baudrate (for example crystal is 14.318
> MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00. baudrate > 55930, error about 3% from 57600. Everything is seems to be ok, UART
> errors with this setting are very rare. But when i using FDR to set
> baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
> baudrate = 57565, error about 0.6% i get UART transfer error on
> almost every byte. This happens on both UARTS and with any baudrate
> from 9600 to 115200.
> Can anyone help me with this problem?
>
I would recommend going to a 14.745600MHz crystal, if you can. That
frequency hits all the baudrates dead-on. But, you probably are using
the 14.318 somewhere else as it is a multiple of the color burst
frequency. A lot of chips used that old frequency as the crystal was
cheap (and plentiful)?
This is a perl script that I used in choosing what crystal I could use
and what accuracy that I could expect:
============ begin baudrates.pl ==============#!/usr/bin/perl
#
# this determines the degree of error between desired and actual
# baudrates on the LPC2000 processors.
#
#my $FREQ = 14745600;
my $FREQ = 14318180;
my $PLL_MUL = 4;
my $PBSD = 2;
my $CCLK = ($FREQ * $PLL_MUL);
my $PCLK = ($CCLK / $PBSD);
my @commonRates (300,600,1200,2400,4800,9600,19200,38400,57600,115200,230400);
#define UART_BAUD(baud) (uint16_t)((PCLK / ((baud) * 16.0)) + 0.5)
sub calcBaud {
my $desired = $_[0];
my $divisor = int(($PCLK / ($desired*16.0))+ 0.5);
my $actual = int(($PCLK / (16*$divisor)) * 1000) / 1000;
my $error = int((($actual - $desired) / $desired) * 1000000) / 10000;
print "divisor: ${divisor} desired: ${desired} ".
"actual: ${actual} error: ${error}%\n";
}
while (@commonRates) {
calcBaud (shift(@commonRates));
}
============ snip ======================
These are the calculations that I get from a 14.318180MHz crystal used
for an LPC2xxx processor:
============== begin rates ==============divisor: 5966 desired: 300 actual: 299.995 error: -0.0016%
divisor: 2983 desired: 600 actual: 599.99 error: -0.0016%
divisor: 1491 desired: 1200 actual: 1200.383 error: 0.0319%
divisor: 746 desired: 2400 actual: 2399.158 error: -0.035%
divisor: 373 desired: 4800 actual: 4798.317 error: -0.035%
divisor: 186 desired: 9600 actual: 9622.432 error: 0.2336%
divisor: 93 desired: 19200 actual: 19244.865 error: 0.2336%
divisor: 47 desired: 38400 actual: 38080.265 error: -0.8326%
divisor: 31 desired: 57600 actual: 57734.596 error: 0.2336%
divisor: 16 desired: 115200 actual: 111860.781 error: -2.8986%
divisor: 8 desired: 230400 actual: 223721.562 error: -2.8986%
============== snip ===================
I would not use the 115200 or 230400 rates as they are above a 2%
deviation from the desired baudrate. IIRC, 2% is the limit for baudrate
speed/bit-timing error.
TomW
--
Tom Walsh - WN3L - Embedded Systems Consultant
http://openhardware.net http://cyberiansoftware.com http://openzipit.org
"Windows? No thanks, I have work to do..."
----------------
>
> Hi! I have a problem using fractional baudrate divider on LPC2136.
> When i'm not using FDR i get baudrate (for example crystal is 14.318
> MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00. baudrate > 55930, error about 3% from 57600. Everything is seems to be ok, UART
> errors with this setting are very rare. But when i using FDR to set
> baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
> baudrate = 57565, error about 0.6% i get UART transfer error on
> almost every byte. This happens on both UARTS and with any baudrate
> from 9600 to 115200.
> Can anyone help me with this problem?
>
I would recommend going to a 14.745600MHz crystal, if you can. That
frequency hits all the baudrates dead-on. But, you probably are using
the 14.318 somewhere else as it is a multiple of the color burst
frequency. A lot of chips used that old frequency as the crystal was
cheap (and plentiful)?
This is a perl script that I used in choosing what crystal I could use
and what accuracy that I could expect:
============ begin baudrates.pl ==============#!/usr/bin/perl
#
# this determines the degree of error between desired and actual
# baudrates on the LPC2000 processors.
#
#my $FREQ = 14745600;
my $FREQ = 14318180;
my $PLL_MUL = 4;
my $PBSD = 2;
my $CCLK = ($FREQ * $PLL_MUL);
my $PCLK = ($CCLK / $PBSD);
my @commonRates (300,600,1200,2400,4800,9600,19200,38400,57600,115200,230400);
#define UART_BAUD(baud) (uint16_t)((PCLK / ((baud) * 16.0)) + 0.5)
sub calcBaud {
my $desired = $_[0];
my $divisor = int(($PCLK / ($desired*16.0))+ 0.5);
my $actual = int(($PCLK / (16*$divisor)) * 1000) / 1000;
my $error = int((($actual - $desired) / $desired) * 1000000) / 10000;
print "divisor: ${divisor} desired: ${desired} ".
"actual: ${actual} error: ${error}%\n";
}
while (@commonRates) {
calcBaud (shift(@commonRates));
}
============ snip ======================
These are the calculations that I get from a 14.318180MHz crystal used
for an LPC2xxx processor:
============== begin rates ==============divisor: 5966 desired: 300 actual: 299.995 error: -0.0016%
divisor: 2983 desired: 600 actual: 599.99 error: -0.0016%
divisor: 1491 desired: 1200 actual: 1200.383 error: 0.0319%
divisor: 746 desired: 2400 actual: 2399.158 error: -0.035%
divisor: 373 desired: 4800 actual: 4798.317 error: -0.035%
divisor: 186 desired: 9600 actual: 9622.432 error: 0.2336%
divisor: 93 desired: 19200 actual: 19244.865 error: 0.2336%
divisor: 47 desired: 38400 actual: 38080.265 error: -0.8326%
divisor: 31 desired: 57600 actual: 57734.596 error: 0.2336%
divisor: 16 desired: 115200 actual: 111860.781 error: -2.8986%
divisor: 8 desired: 230400 actual: 223721.562 error: -2.8986%
============== snip ===================
I would not use the 115200 or 230400 rates as they are above a 2%
deviation from the desired baudrate. IIRC, 2% is the limit for baudrate
speed/bit-timing error.
TomW
--
Tom Walsh - WN3L - Embedded Systems Consultant
http://openhardware.net http://cyberiansoftware.com http://openzipit.org
"Windows? No thanks, I have work to do..."
----------------
Reply by ●July 28, 20072007-07-28
--- In l..., Bill Knight wrote:
>
> maxmatveev wrote:
> > Hi! I have a problem using fractional baudrate divider on
LPC2136.
> > When i'm not using FDR i get baudrate (for example crystal is
14.318
> > MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00.
baudrate > > 55930, error about 3% from 57600. Everything is seems to be ok,
UART
> > errors with this setting are very rare. But when i using FDR to
set
> > baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
> > baudrate = 57565, error about 0.6% i get UART transfer error on
> > almost every byte. This happens on both UARTS and with any
baudrate
> > from 9600 to 115200.
> > Can anyone help me with this problem?
> Make sure your crystal is really 14.318MHz and not the baud rate
> multiple, 14.7456MHz.
>
> Regards
> -Bill Knight
> R O SoftWare
>
The crystal is 14.318. Already ordered 18.432 which is baudrate
multiple. But i want to know why fractional divider doesn't working.
I will try to check real baudrate when using fractional divider with
oscilloscope during this week.
>
> maxmatveev wrote:
> > Hi! I have a problem using fractional baudrate divider on
LPC2136.
> > When i'm not using FDR i get baudrate (for example crystal is
14.318
> > MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00.
baudrate > > 55930, error about 3% from 57600. Everything is seems to be ok,
UART
> > errors with this setting are very rare. But when i using FDR to
set
> > baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
> > baudrate = 57565, error about 0.6% i get UART transfer error on
> > almost every byte. This happens on both UARTS and with any
baudrate
> > from 9600 to 115200.
> > Can anyone help me with this problem?
> Make sure your crystal is really 14.318MHz and not the baud rate
> multiple, 14.7456MHz.
>
> Regards
> -Bill Knight
> R O SoftWare
>
The crystal is 14.318. Already ordered 18.432 which is baudrate
multiple. But i want to know why fractional divider doesn't working.
I will try to check real baudrate when using fractional divider with
oscilloscope during this week.
Reply by ●July 28, 20072007-07-28
maxmatveev wrote:
> --- In l..., Bill Knight wrote:
>> maxmatveev wrote:
>>> Hi! I have a problem using fractional baudrate divider on
> LPC2136.
>>> When i'm not using FDR i get baudrate (for example crystal is
> 14.318
>>> MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00.
> baudrate >>> 55930, error about 3% from 57600. Everything is seems to be ok,
> UART
>>> errors with this setting are very rare. But when i using FDR to
> set
>>> baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
>>> baudrate = 57565, error about 0.6% i get UART transfer error on
>>> almost every byte. This happens on both UARTS and with any
> baudrate
>>> from 9600 to 115200.
>>> Can anyone help me with this problem?
>>
>> Make sure your crystal is really 14.318MHz and not the baud rate
>> multiple, 14.7456MHz.
>>
>> Regards
>> -Bill Knight
>> R O SoftWare
>>
> The crystal is 14.318. Already ordered 18.432 which is baudrate
> multiple. But i want to know why fractional divider doesn't working.
>
> I will try to check real baudrate when using fractional divider with
> oscilloscope during this week.
>
Is it possible the crystal is series resonant? That will shift the
frequency off when used in a parallel resonant circuit.
-Bill
> --- In l..., Bill Knight wrote:
>> maxmatveev wrote:
>>> Hi! I have a problem using fractional baudrate divider on
> LPC2136.
>>> When i'm not using FDR i get baudrate (for example crystal is
> 14.318
>>> MHz, PCLK = 14.318 MHz)FDR = 0x10, DLL = 0x10, DLM = 0x00.
> baudrate >>> 55930, error about 3% from 57600. Everything is seems to be ok,
> UART
>>> errors with this setting are very rare. But when i using FDR to
> set
>>> baudrate more accurately - FDR = 0xB8, DLL = 0x09, DLM = 0x00 -
>>> baudrate = 57565, error about 0.6% i get UART transfer error on
>>> almost every byte. This happens on both UARTS and with any
> baudrate
>>> from 9600 to 115200.
>>> Can anyone help me with this problem?
>>
>> Make sure your crystal is really 14.318MHz and not the baud rate
>> multiple, 14.7456MHz.
>>
>> Regards
>> -Bill Knight
>> R O SoftWare
>>
> The crystal is 14.318. Already ordered 18.432 which is baudrate
> multiple. But i want to know why fractional divider doesn't working.
>
> I will try to check real baudrate when using fractional divider with
> oscilloscope during this week.
>
Is it possible the crystal is series resonant? That will shift the
frequency off when used in a parallel resonant circuit.
-Bill
