Model Train controller - revisited

[malc-c]

Guys, a while ago Darrel and Steve (and a few others) helped me out with the development of a PWM pic based train controller. However although the prototype worked fine on the breadboard, when I re-designed the board to include the powersupplies (5v and 12v) I got some strange behaviour from the resulting PCB. Sometimes the PWM would stop and the PIC needed resetting, but I put this down to the noise these models cause (RF etc) and I also noted that whilst the slow speed was exceptional, the range of control was very fine, so I went back to the breadboard.

I've breadboarded the schematic attached, but with the 20 Mhz xtal. The code in the PIC is as follows

'****************************************************************
'*  Name    : Twin Track Controller.BAS                         *
'*  Author  : M. Crabbe                                         *
'*  Notice  : Copyright (c) 2007 M. Crabbe                      *
'*          : All Rights Reserved                               *
'*  Date    : 17/02/2007                                        *
'*  Version : 1.0                                               *
'*  Notes   : 16F767 using 20 Mhz xtal                          *
'*          :                                                   *
'****************************************************************



@ __CONFIG  _HS_OSC & _WDT_ON & _PWRTE_ON & _MCLRE_OFF 

DEFINE OSC 20

ANSEL = %00010111    			        ' Set which ports are digital and which are analogue
CMCON=7                                 ' Turm Comparitors off
TRISA=%11111111        			        ' set PORTA as all input 
TRISC=%00000001        			        ' set PORTC as all output apart from RC0

DEFINE ADC_BITS  8     			        ' ADCIN resolution  (Bits)
DEFINE ADC_CLOCK 2     			        ' ADC clock source  (Fosc/32)
DEFINE ADC_SAMPLEUS 11 			        ' ADC sampling time (uSec)

INCLUDE "DT_INTS-14.bas"            	' Base Interrupt System
INCLUDE "SPWM_INT.bas"              	' Software PWM module

DEFINE SPWM_FREQ  150                	' SPWM Frequency (Hz)
DEFINE SPWM_RES   254               	' SPWM Resolution (0 - 255)

DutyVar VAR byte[4]                     ' Array used to store ADC [channels]

ASM
SPWM_LIST  macro                    	; Define Pin's to use for SPWM
    SPWM_PIN  PORTC, 2, _DutyVar        ; and the associated DutyCycle variables
    SPWM_PIN  PORTC, 3, _DutyVar+1      ; (Additional channels DutyVar+2, +3 etc)
  endm
  SPWM_INIT  SPWM_LIST              	; Initialize the Pins
ENDASM

ASM
INT_LIST  macro                         ; IntSource,        Label,  Type, ResetFlag?
        INT_Handler   TMR1_INT,  SPWMhandler,  ASM,  yes
    endm
    INT_CREATE                      	; Creates the interrupt processor
ENDASM

@ INT_ENABLE  TMR1_INT              	; enable Timer 1 interrupts

Main:                               
 ADCIN 0, DUTYVAR[0]                    ; if having more channels add ADCIN 2, DUTYVAR[2] etc
 ADCIN 1, DUTYVAR[1]
GOTO Main                               ; program simply loops round reading ADC and sending corresponding PWM out

The result of this code can be seen in the avi (4MB DIVX) http://www.micro-heli.co.uk/scope2.avi which you will see and hear is a good variable square wave. This drives the 2N3904 and darlington pair well and the resulting in fantastic slow speed crawl as can be seen in the avi http://www.micro-heli.co.uk/controller.avi

However, the amount of rotation of the pot is quite small before the loco is up to a medium speed. I've tried changing the pot for 100K, 22K and even 2k, and I've played about with different values for R1, R2 and R3 and it still makes no difference.

So my question... Is there a way to "scale" or range the ADC so that the PWM performs is less proportional, and more log like in action. Ie instead of the pulse being 50% when the pot is 50% travel (and presumably the value is 127 on the ADC) its say 25%, but when the pot is at 75% travel the value is 75% if you follow my drift ? This way it will provide better slow motion control and still give excellenf high speed performance as well.

Also, what would be the best way to protect the PIC from interfearence, placing 0.1uf caps between inputs and gnd or simply take all unsused pins to gnd ?

[skimask]

However, the amount of rotation of the pot is quite small before the loco is up to a medium speed. I've tried changing the pot for 100K, 22K and even 2k, and I've played about with different values for R1, R2 and R3 and it still makes no difference.

Are you using a linear pot or a 'log' ('audio') pot? I've been tripped up by those before, thought I had a linear type, ended up being a log type.

Is there a way to "scale" or range the ADC so that the PWM performs is less proportional, and more log like in action

Maybe use PBP's built in SIN function. It won't be LOG-like, but if you use the first 90 degrees...upside down (if that makes any sense) and a bit of programming, at least it won't be a linear function and should be a bit smoother at the low end.

Also, what would be the best way to protect the PIC from interfearence, placing 0.1uf caps between inputs and gnd or simply take all unsused pins to gnd ?

Pulling all unused inputs to either Vdd or Vdd is never a bad idea.
Putting .1uf across all the inputs might 'slow down' the inputs, you might miss a pulse for something here and there. Maybe you need more than .1uf across Vdd & Vss on your PIC, 'cause the .1uf tends to only decouple one frequency range, generally the higher ones. Maybe you need to add a big 'ol 10uf or something in parallel with that .1uf, to help get some of the lower frequencies. And maybe make sure you've got good quality cap's doing that job, low ESR types.

[Acetronics2]

Hi, Malc

Try this:

TEMP var WORD

Main:

ADCIN 0, DUTYVAR[0] ; if having more channels add ADCIN 2, DUTYVAR[2] etc

TEMP = DUTYVAR[0] * DUTYVAR[0] / 256 '********************

IF DUTYVAR[0] < 10 THEN ' 10 = value for a neat motor start , if needed !

DUTYVAR[0] = 0
GOTO Jump0

ENDIF

DUTYVAR[0] = TEMP MAX 10

Jump0:


ADCIN 1, DUTYVAR[1]

TEMP = DUTYVAR[1] * DUTYVAR[1] / 256 '********************

IF DUTYVAR[1] < 10 THEN ' 10 = value for a neat motor start ,if needed !


DUTYVAR[1] = 0
GOTO Jump1

ENDIF

DUTYVAR[1] = TEMP MAX 10

Jump1:

GOTO Main ; program simply loops round reading ADC and sending corresponding PWM out

That what I use for my boat speed controllers ... works really fine !!!

Regards
Alain

PS: here is the original

'************************************************* *****************************
'Mise à l'echelle 0 - 255 de la sortie PWM pour 0 - 96 % de la course
'************************************************* *****************************

' 265 = 255 / 0.96 ... CQFD

'************************************************* *****************************
'Moteur Gauche
'************************************************* *****************************

EcartGazG = ( ABS ( PulsGazG - MidGaz)) ' for fwd/rev on the same stick

IF EcartGazG < 55 THEN 'Center Deadband
PowerG = 0
GOTO MotD
ENDIF

dummy = EcartGazG * 265
PowerG = DIV32 CourseGaz

PowerG = (PowerG + 1 ) MIN 255 ' 255 value as MAXI for PWM

'************************************************* *****************************
'Variation exponentielle
'************************************************* *****************************

dummy = PowerG * PowerG
PowerG = DIV32 255

'************************************************* *****************************
'Puissance mini appliquee au moteur - démarrage franc ( 20/255 )
'************************************************* *****************************

PowerG = PowerG MAX 20

[malc-c]

Guys, thanks for the replies.

To answer some of the questions, yes the pots are lin, or at least they are marked as lin and purchased as such, and using a DVM it seem that they are liniar. I'll look at taking all the unused pins to +ve, possibly via 10k resistors.

Alain,

I think that the original code above basically works on the same principle of taking the 0 - 254 from the adc and sending that out as the pulse. However it all got a bit complicated when Darrel's routines were used as all I know is that i need to do is change then two values

DEFINE SPWM_FREQ  150                	' SPWM Frequency (Hz)
DEFINE SPWM_RES   254               	' SPWM Resolution (0 - 255)

I'll try adding your code and will report back

[mister_e]

The LOG pot is a nice idea, Alain's routine too. You can also build a huge LOOKUP table with excel and read from it. If you have some VCA (like ssm2164) you can use a Linear pot and get a LOG curve.

Many different ways.

At least, there's some progress

PS: you can trick the curve a little bit by adding a resistor in parrallel with the wiper... far to be LOG, but could be interesting to trace the curve...

THE PWM freq shouldn't affect the results... shouldn't..

[malc-c]

Well it compiles Ok and it does indeed work, providing a better range of travel for the lower end of the POT. It produces an interesting wave form

The only problem is that oscilatons produce noise in the loco's armature. I tried the old trick of using a capacitor across the output which worked to some degree, but as it effectivly smoothed the pulses it defeats the object of using PWM (or PPM)

Steve, I'll try your suggestion

[malc-c]

Oh and one other thing I've noticed,which may be causingn the PIC to bomb is when the when the wiper is at the top and thus connected to the 5.21v supply the pulses stop and the voltage out drops, and on some occasions this has caused the driver to sort of lock as moving the pot after the effect has no effect as the PMW generation has stopped. Is there a way of setting the voltage ref or ADC value so that a PMW signal is always produced ?

Edit:
maybe this video might help explain (2.2 MB Divx format) http://www.micro-heli.co.uk/scope4.avi