Dimmer

[sayzer]

Hi Steve,

As you well know, it took about a month for me to figure out that this line does not work:

@ device intrc_osc_noclkout, wdt_on, mclr_off, protect_off

It should be either like
@ device pic12F675, intrc_osc_noclkout, wdt_on, mclr_off, protect_off
or
like

@ DEVICE pic12F675, INTRC_OSC_NOCLKOUT
@ DEVICE pic12F675, WDT_ON
@ DEVICE pic12F675, PWRT_ON
@ DEVICE pic12F675, MCLR_OFF
@ DEVICE pic12F675, BOD_ON
@ DEVICE pic12F675, CPD_OFF
@ DEVICE pic12F675, PROTECT_OFF

[Atom058]

Thanks guys! I knew I would be embarrased by the solution and I was right!

OK, Now I have made the circuit change and have made the switch changes. I get the 5V outputs on pins GPIO.0 and .1 which go low when I press the up or down buttons. I do not, however, get any output from GPIO.2 which is the triac trigger. My guess is that the problem lies in the ACDetect routine. I see in previous posts that there were many references to this routine and it went through several revisions. Can someone verify that I have the correct routine?

I have attached the code as a text file and the circuit design as a pdf. Would appreciate comments.

Thanks again! Atom058

[Atom058]

Hello again - Did I stump everyone? Surely someone has a comment...

Would really like to get this working!

Thanks in advance! Atom058

[mister_e]

You have the right version.

It's suppose to work unless you have some hardware issue OR device prograemmer issue.

[Atom058]

Mister-e - Thanks! I will go over my circuit again. It helps knowing that the code I am trying to run is the correct version - I can focus my attention elsewhere. If you (or anyone else) can offer some troubleshooting ideas, they would be appreciated.

Atom058

[Atom058]

Well, I did find (part of) my problem - It was a bad solder at GPIO.4 so it was never seeing the zero crossing. Having fixed that, I now can dim the light up and down just like it is supposed to. However, (bet you saw this coming) when it is between full on and full off, it flickers randomly. The dimming works but it flickers until it gets to full on or full off. I am sure that this is a timing issue but can't see where to get at it. I am at 60 Hz, 120V. I am using the code as previously posted.

I know that some of you (Mister_e) feel like you are beating a dead horse with this dimmer project, but if you can help me get through this, I will post my final code, circuit diagrams and parts lists so that anyone can build this and you will never have to deal with it again (hopefully!).

Thanks (Mister_e)!

Atom058

[mister_e]

interesting. to me it's sounds like a an delay offset in the Zero Crossing. Done by the transformer AND the 470uF cap.

Can be solved in software.

Give me the transformer information, part #, voltage etc. I will work on this.

[Atom058]

Steve - Thanks for your reply - The transformer is from Jameco (www.jameco.com), P/N: 249412. The spec sheet says that I should be getting 8VAC @ 1.6A (I know that is a little over-kill on the amperage, but it is what I had lying around), but I am measuring 9.8VAC across the input to my bridge rectifier. Using a voltage divider (4k7 & 10K), that gets the DC voltage to GPIO.4 down to 4.33VDC. The cap is a 470 uF 25V electrolytic.

Let me know if you need anything else... Thanks!

Atom058

[Atom058]

Steve - Any luck with the info that I posted?

I've been looking at some related posts and I am wondering if the timing problem is related to the maxdelay value. If I am running at 60 hz, that means that I complete 1 cycle every 16.7 mS. Or, I cross zero every 8.35 mS (8350 uS). Is that correct? In the code, MaxDelay is 6000 uS. Is this where my problem is?

Can you tell me why the transformer and 470 uF cap is of interest to you? How does this affect my timing?

Every day is a learning experience and I'm always looking for more!

Thanks! Atom058

[Atom058]

I would love to hear from anyone that has successfully built this dimmer that is discussed on this thread. I have followed the code and schematics (as well as the changes to both) and can not achieve the end result. The circuit and code, as it stands now, will allow me to go to full on and full off by quickly tapping the inc/dec buttons. This part works perfectly. If I press and hold the inc/dec buttons, the lighting level changes gradually, just like it is supposed to, but I get flickering. Mister-e seems to think that it is a timing issue - and I am in agreement, but I do not know where to make adjustments. If anyone can give me some trouble-shooting ideas, I would greatly appreciate it.

I think this is a great design that has many uses. I would love to get it to work and post my final results!

This circuit, once I get it right, is going to be the basis for a larger design, But until I get this part working, I can not move forward.

Please help!
Atom058

[mister_e]

Yes it's a timing problem, and there's no way to determine the exact one between all transformer brand and model. Adding a calibration process is possible but... usefull?

grabbing the signal directly from the li ne is the only way i see to make it work with all transformer.

Transformerless PSU is another way.... i hate that but if everything is properly done and insulated, the user shouldn't be hurt one day.

Give me a couple of days.. i'm on a rush now.

[Atom058]

Steve - Thanks for your reply - I was actually on the path of a transformerless supply like you suggested. I have built the design that was presented by Melanie in another thread and get a nice clean 5VDC to drive my pic. I am pulling the AC for the trigger and putting it through a 20M dropping resistor (to go to pin 3 of my pic) per the schematic for the PIC REF-4 application note. However, this only drops the voltage to around 30V AC. Do I need to get it down to <= 5VAC or is 30VAC OK?

Atom058

[mister_e]

I did something like that for one user here that use a 10Meg directly from the main to a PIC input. The PIC internal diode do their job. As now it seems to works. The user is still alive... Just make sure you properly insulate your stuff.....

The 30VAC you get is without load...just your voltmeter.

[Atom058]

Thanks Steve! I will give it a try and let you know what happens!

[Atom058]

Steve! The transformerless supply did the trick! Works like a charm now. In the next couple of days, I will post everything I have so that anyone can build this (and not bug you about it anymore).

Thanks for all your help! Atom058

[mister_e]

<img src="http://www.picbasic.co.uk/forum/attachment.php?attachmentid=881&d=1148640458">

[Atom058]

Here is the artwork, BOM and code for the new-and-improved AC Lamp Dimmer. Everything you need to make one. There is only one issue that I have with this but it is not important for my use, but maybe someone else can shed some light (!) on it. When going from full bright and dimming down, it makes a small "jump" down in brightness but then dims nice and smooth the rest of the way. I am sure it's a timing thing but I could not figure it out. Anyway, enjoy! And thanks to Steve (Mister_e) for his help and patience (nice dancing letters, by the way)!

You'll be hearing more from me!

Atom058

[Pic_User]

<img src=" http://www.picbasic.co.uk/forum/attachment.php?attachmentid=1906&stc=1&d=118620155 9">

[FromTheCockpit]

Can't wait to return home... use the following ACDetect routine instead.
this should work.

disable
ACDetect:
if ACline==1 then 
    '
    '    Rising edge of AC signal
    '    ========================
	if triacdelay then              ' A delay is set by user
                                        '  
	    if FullBright==0 then       ' Full Brightness flag is not set
            pauseus maxdelay-triacdelay '    do the selected delay
            triac=1                     '    enable TRIAC	    	
            else                        '
    	        triac=1                 ' Full Brightness flag is set
                                        '    enable triac
            endif                       '
                                        '
        else                            '
    	    triac=0                     ' No Delay set by user... disable Triac
        endif                           '
    
    else                                
    '
    '    Falling edge of AC signal
    '    =========================
        if fullbright==0 then triac=0   ' Disable the Triac on falling edge of 
                                        ' Ac signal
    endif
INTCON.0=0 ' Clear GPIF (interrupt on GP4 change)
resume
enable

Yeah i know you can also use Pulsout and forget the Falling edge detection... as you wish. Both will work.

Very interesting thread I must say. I am trying to do the same and have added DT_INTERRUPTS to manage the ac detect....But my lamp is not dimming at all, also when in the code 'triacdelay=0', my lamp does not go off........ Can someone spot the error please.

INCLUDE "DT_INTS-14.bas"     ; Base Interrupt System
INCLUDE "ReEnterPBP.bas"     ; Include if using PBP interrupts
Include "modedefs.bas"
@ __Config _XT_OSC & _WDT_ON & _PWRTE_ON & _MCLRE_OFF & _BODEN & _CP & _CPD

TriacDelay                  var Word
maxdelay                    var word

Triac	var	PortA.1 ' Output to TRIAC gate
ACLine	var	PortA.2 ' Input for the FullWave rectify AC line
Buz		var PortC.2

TRISA = %001100
TRISC = 0
CMCON = 7
ANSEL = 0
OPTION_REG = %11000000  ' RAPU = off, PS WDT
PORTA=0
PORTC=0

ASM
INT_LIST  macro    ; IntSource,        Label,  Type, ResetFlag?
        INT_Handler    INT_INT,  _ACDetect,   PBP,  yes
    endm
    INT_CREATE               ; Creates the interrupt processor
ENDASM

@   INT_ENABLE   INT_INT     ; enable external (INT) interrupts

Triac=0            ' disable Triac Gate
triacdelay=0  ' Set delay to minimum

while 1

pause 3000
high buz: pause 500 : low buz
triacdelay=8000     : pause 10000 :high buz: pause 50 : low buz 
triacdelay=5000     : pause 10000 :high buz: pause 50 : low buz 
triacdelay=4500     : pause 10000 :high buz: pause 50 : low buz 
triacdelay=4000     : pause 10000 :high buz: pause 50 : low buz
triacdelay=3500     : pause 10000 :high buz: pause 50 : low buz
triacdelay=0 : pause 7000

wend

ACDetect:
	if triacdelay > 0 then
			if triacdelay=8000 then
				triac=1 ' Activate TRIAC
			else
				maxdelay=8000-triacdelay
				pauseus maxdelay
				triac=1
				pauseus 50
				triac=0
   	     	endif
    else
 		     triac=0
    endif
@ INT_RETURN

P.S. I am using full wave with few high value resistance connected to phase and zeners to drop the voltage further.

[Darrel Taylor]

P.S. I am using full wave with few high value resistance connected to phase and zeners to drop the voltage further.

What size zeners are you using?

The INT input is Schmidt trigger, so it has to get above 4.0V to trigger the interrupt.
It may not make it, or cause a delay after zero-crossing that could affect the timing.

Have you tried it without the zeners?

[FromTheCockpit]

What size zeners are you using?

The INT input is Schmidt trigger, so it has to get above 4.0V to trigger the interrupt.
It may not make it, or cause a delay after zero-crossing that could affect the timing.

Have you tried it without the zeners?

Hello Darrel, I made a slight mistake in the above post, it's not full wave rectified. The circuit goes like this - Phase->1.5Meg->12V zener->1K->4.7V zener->3.3Meg->PIC

No I have not tried it without zeners yet. Will it be OK to remove the zeners from the above equation - the place where I am gets heavy voltage fluctuations - The variation could be from 170V - 280V AC (I am serious!)

[Darrel Taylor]

According to the simulator, your resistors/zeners should give 4.4v, so that's ok. (but you shouldn't need the zeners)

mister-e was using INT ON CHANGE, so it triggered on both edges.
To get INT to do the same thing you have to toggle INTEDG on each interrupt.

This modification of the handler works in the SIM.

INTEDG  VAR OPTION_REG.6
; ...

ACDetect: 
    if INTEDG then
        if triacdelay > 0 then
            if triacdelay=8000 then
                triac=1 ' Activate TRIAC
            else
                maxdelay=8000-triacdelay
                pauseus maxdelay
                triac=1
            endif
        else
            triac=0
        endif
    else
        if triacdelay != 8000 then triac=0
    endif

    INTEDG = !INTEDG
@ INT_RETURN

In the scope image, the blue line shows the triac ON time.
It's only on during the positive half of the cycle.
Doesn't seem right, but that's the way mister-e had it.



HTH,

[Darrel Taylor]

Read the previous post first.
Then if you want both halves of the cycle, try this ...

INTEDG  VAR OPTION_REG.6
; ...

ACDetect:
    if triacdelay != 8000 then triac = 0 
    if triacdelay > 0 then
        if triacdelay=8000 then
            triac=1 ' Activate TRIAC
        else
            maxdelay=8000-triacdelay
            pauseus maxdelay
            triac=1
        endif
    else
        triac=0
    endif

    INTEDG = !INTEDG
@ INT_RETURN

[FromTheCockpit]

According to the simulator, your resistors/zeners should give 4.4v, so that's ok. (but you shouldn't need the zeners)

mister-e was using INT ON CHANGE, so it triggered on both edges.
To get INT to do the same thing you have to toggle INTEDG on each interrupt.

This modification of the handler works in the SIM.

INTEDG  VAR OPTION_REG.6
; ...

ACDetect: 
    if INTEDG then
        if triacdelay > 0 then
            if triacdelay=8000 then
                triac=1 ' Activate TRIAC
            else
                maxdelay=8000-triacdelay
                pauseus maxdelay
                triac=1
            endif
        else
            triac=0
        endif
    else
        if triacdelay != 8000 then triac=0
    endif

    INTEDG = !INTEDG
@ INT_RETURN

In the scope image, the blue line shows the triac ON time.
It's only on during the positive half of the cycle.
Doesn't seem right, but that's the way mister-e had it.

Attachment 4684

HTH,

Thanks Darrel, I just noticed that in your simulation there is no 1k between the two zeners, instead there is 3.3Meg. I will try the code and update here.

[FromTheCockpit]

Hi, I tried the following first:

INTEDG  VAR OPTION_REG.6
; ...

ACDetect:
    if triacdelay != 8000 then triac = 0 
    if triacdelay > 0 then
        if triacdelay=8000 then
            triac=1 ' Activate TRIAC
        else
            maxdelay=8000-triacdelay
            pauseus maxdelay
            triac=1
        endif
    else
        triac=0
    endif

    INTEDG = !INTEDG
@ INT_RETURN

RESULT: The bulb attached to the output did not responded to the change in triac delay value, it just kept glowing at the same intensity.

Then I experimented a little and came up with the following:

while 1

pause 3000
high buz: pause 500 : low buz
triacdelay=8000     : pause 15000 :high buz: pause 50 : low buz 
triacdelay=7998     : pause 15000 :high buz: pause 50 : low buz 
triacdelay=7950     : pause 15000 :high buz: pause 50 : low buz 
triacdelay=7900     : pause 15000 :high buz: pause 50 : low buz
triacdelay=7850     : pause 15000 :high buz: pause 500 : low buz
triacdelay=0        : pause 12000

wend


ACDetect:
;    if triacdelay != 8000 then triac = 0 
    if triacdelay > 0 then
        if triacdelay=8000 then
            triac=1 ' Activate TRIAC
        else
            maxdelay=8000-triacdelay
            pauseus maxdelay
                triac=1
                pauseus 50
                triac=0
        endif
    else
        triac=0
    endif

    INTEDG = !INTEDG
@ INT_RETURN

RESULT: The bulb responded to the changing value of triac delay, at value of 8000 it goes full bright which is normal BUT when at first lower value than 8000, it dims quiet a lot. I want it to dim a only a little at first step. I even tried 7998 after 8000 but no luck i.e. 2 uS delay.

[Darrel Taylor]

You're right, I had the resistors wrong.
Making the change brought it up to 4.6V.
But, having the zeners in there adds about 40uS to the zero-cross signal.
It takes 35uS without the zeners, and 75uS with them.

Add in the interrupt latency for PBP type interrupts, time to do math in the handler, the minimum pauseus of 24us, and taking into account that a half cycle of 60hz is 8,333uS ... I'm sure the first step was quite a bit dimmer.

But the biggest problem is delaying inside the interrupt handler.
When 90% of the time is spent in the handler, the pause 10000 in the main loop takes almost a minute.

So here's my next attempt.
It uses Timer1 for the delays, so the main program can keep running in the foreground.
The handlers have been changed to ASM type to minimize latencies.
And the math is done in a separate subroutine that only runs once when you change the brightness, instead of having to do the math on every zero-cross.

It should be pretty close to the MAX/MIN dimming, but you may need to adjust the MaxBright/MinBright constants a little.
If you adjust them too far, it will start flashing.

INCLUDE "DT_INTS-14.bas"     ; Base Interrupt System
INCLUDE "ReEnterPBP.bas"     ; Include if using PBP interrupts
Include "modedefs.bas"
@ __Config _XT_OSC & _WDT_ON & _PWRTE_ON & _MCLRE_OFF & _BODEN & _CP & _CPD

MaxBright                   CON 8250    ; maximum brightness
MinBright                   CON 200     ; minimum dimming 
Brightness                  var word
TriacDelay                  var Word
AlwaysON                    var BIT
AlwaysOFF                   var BIT
OffPeriod                   var BIT     ; used by timer1 periods
 
Triac	var	PortA.1 ' Output to TRIAC gate
ACLine	var	PortA.2 ' Input for the FullWave rectify AC line
Buz		var PortC.2

INTEDG  VAR OPTION_REG.6
TMR1ON  VAR T1CON.0
TMR1IF  VAR PIR1.0
Timer1  VAR WORD EXT
@Timer1 = TMR1L

TRISA = %001100
TRISC = 0
CMCON = 7
ANSEL = 0
PORTA=0
PORTC=0

ASM
INT_LIST  macro    ; IntSource,        Label,  Type, ResetFlag?
        INT_Handler    INT_INT,    _ACDetect,   ASM,  yes
        INT_Handler   TMR1_INT,   _T1Handler,   ASM,  yes
    endm
    INT_CREATE               ; Creates the interrupt processor
ENDASM

@   INT_ENABLE   INT_INT     ; enable external (INT) interrupts
@   INT_ENABLE  TMR1_INT     ; enable Timer1 interrupts

Triac=0            ' disable Triac Gate
brightness=0       ' Set delay to minimum
gosub SetDelay

while 1
    pause 3000
    high buz: pause 500 : low buz
    brightness=MaxBright   : gosub SetDelay : pause 10000 :high buz: pause 50 : low buz 
    brightness=MaxBright-2 : gosub SetDelay : pause 10000 :high buz: pause 50 : low buz 
    brightness=4500        : gosub SetDelay : pause 10000 :high buz: pause 50 : low buz 
    brightness=4000        : gosub SetDelay : pause 10000 :high buz: pause 50 : low buz
    brightness=3500        : gosub SetDelay : pause 10000 :high buz: pause 50 : low buz
    brightness=MinBright   : gosub SetDelay : pause 10000 :high buz: pause 50 : low buz
    brightness=0           : gosub SetDelay : pause 7000
wend

SetDelay:
    Brightness = Brightness min MaxBright                    ; limit maximum brightness
    AlwaysON  = !(Brightness != MaxBright)                   ; MaxBright is always ON
    AlwaysOFF = !(Brightness != 0)                           ; 0 is always OFF
    if !AlwaysOFF then Brightness = Brightness max MinBright ; limit minimum dimming
    triacdelay = -(MaxBright - Brightness)                   ; calc delay time
RETURN

;------------------------------------------
T1Handler:
    if OffPeriod then            ; triac delay finished
        OffPeriod = 0
        triac=1                  ; turn on triac
        Timer1 = -50             ; load timer for 50uS
    else
        triac = 0                ; turn off triac after 50uS
        TMR1ON = 0 
    endif
@ INT_RETURN

;------------------------------------------
ACDetect:
    if AlwaysON then
        triac=1 ' Activate TRIAC
    else
        if AlwaysOFF then
            triac = 0
        else
            Timer1 = triacdelay  ; load delay into timer
            TMR1IF = 0           ; clear the interrupt flag
            TMR1ON = 1           ; start the timer
            OffPeriod = 1        ; tell T1Handler this is the main delay
        endif
    endif

    INTEDG = !INTEDG             ; toggle INTEDG to get both edges
@ INT_RETURN

Hope that works better.

[FromTheCockpit]

Thanks Darrel, I am on 50Hz, do I need to change any value in the code for that??

[Darrel Taylor]

There's another reason for the first step being much dimmer.

Half a cycle of 50hz, is 10,000uS.
So take MaxBright up to 9900.