XOUT and HPWM
Hi, this is about the X10 command. I would like to modulate the signal coming out of X10, at 120 KHz. I can currently do this by using the HPWM on one pin, and getting the XOUT on another, and then combining them externally. This is at the cost of 1 extra pin, and the external circuitry.
Is there any way that this can be done in software, so that one gets a composite modulated X10 signal out of a single pin?
Thanks in advance for any suggestions!
Regards,
Anand
|
Hi ardhuru,
I've never done anything like that before, but after looking at the macro's for XOUT, I think it might be possible.
But the method used depends on the chip. Which one are you using?
Darrel
|
Currently, the 16F628A for its HPWM. Being able to do it on this would be fine, but if I could achieve it on a 12F629/75, better still.
Thanks, and I'm hoping you'd have some trick to suggest!
Regards,
Anand
|
628A, OK, that should make it easy.
XOUT TRISB.3,PORTA.0,[house\lightsOff]
The idea is that the XOUT command takes a register and a bit for the output pin assignment. Normally this is something like PORTB.0 but it really doesn't matter which register and bit you use.
So by assigning TRISB.3 as the output, it should modulate the PWM by turning the pins output on and off.
PBP finds the location of the corresponding TRIS register by simply adding 80h to the PORTs address. For PORTB (06H), TRISB would be 06h + 80h = 86h. So now, if PBP thinks that TRISB.3 is a physical pin, then it will also try to set the pin to Output by clearing (106h).3 which now lines up with PORTB again in bank2. But since that pin is overriden by the Peripheral select switch, the data doesn't get thru, so all's well.
The reason I say all of that, is because if you try to do the same thing on an 18F, the registers don't line up next to each other like that, and can cause some pretty nasty problems.
As for the 12F629/75, I suppose you could set up the timer to generate the frequency using interrupts, and do the same thing with TRIS.
Should work for SEROUT(2) as well. Assuming it works at all.
HTH,
Darrel
|
Darrel, thanks for the detailed reply.
I tried this, and was at first delighted to see a strong(er) signal than what I was getting by the 2-pin method.
But, it did not work. I notice the waveform looked different from the 2 pin waveform as well (as seen after the transistor that combined the 2 signals).
I think the problem is, when you control the HWPM *before* and *after* the XOUT line, the spaces between the data bits that XOUT gives, which should remain low, are now consistently transitioning at 120 KHz. I suspect for this to work, like you pointed out earlier, the XOUT macro itself might have to be modified.
Is my analysis correct, or am I being too dense?
Regards,
Anand
|
Hmmm, what I missed was the Polarity of the XOUT signal. The PBP manual didn't really go into too much detail, but i found another document on the web that explained it better. It's description goes like this.So, 1 turns the 120khz ON, 0 turns it off.Transmit to Power Line—Output a high on this line within 200 μ of Zero Crossing
going high, and hold the line high, to output a one on the power line—that is, to turn on
the 120 kHz oscillator. Use the same timing with a low output to output a zero. The line
should be low if the interface is not to transmit data, since that keeps the oscillator off.
But with the TRIS bit, it works the other way around. 1 = INPUT (OFF), 0 = OUTPUT (ON). Just a little backwards.
I think you're right though, the only way around it is to change the XOUT command itself. Fortunately, it's an easy change if you're willing to tamper with the PBPPIC14.LIB file.
If so, be sure to make a copy of the file before editing, you might want it back the old way later on.
Your mission, should you choose to accept it, is to ...
Open the PBPPIC14.LIB file in the PBP folder, and do a search for XOUT. You should see a section that looks like thisComment out the "call LOWT". We'll handle setting the pin to the correct state manually.Code:ifdef XOUT_USED LIST XOUT2 movlw 2 ; Default to 2 cycles XOUT movwf R2 ; Save number of cycles movf RR1, W ; Get data port movwf FSR ; Put it into FSR movf RM1, W ; Get data bit mask ; call LOWT ; Set data to low, output
Then scoll down a little bit to xout1loop and change it like thisIt just reverses whether it sets the pin high or low.Code:xout1loop ; movf RM1, W ; Get data bit mask ; iorwf INDF, F ; Set the bit comf RM1, W ; Get inverted data bit mask andwf INDF, F ; Clear the bit movlw high 1000 ; Pause 1 ms movwf R0 + 1 movlw low 1000 call PAUSEUSL ; comf RM1, W ; Get inverted data bit mask ; andwf INDF, F ; Clear the bit movf RM1, W ; Get data bit mask iorwf INDF, F ; Set the bit
Now, to use it ..Fingers crossedCode:TRISB.3 = 1 ' start with output OFF XOUT TRISB.3,PORTA.0,[house\lightsOff]
P.S. Don't blame me if you don't backup the file first.
DT
|
Darrel, that was brilliant!
Worked the first time. If I wore a hat, i'd have raised it now.
Thanks and regards,
Anand
|
Darrel, whatever you suggested in the earlier messages worked perfect, and that's made me more ambitious/adventureous/greedy.
What I wonder is, can this also be achieved using your popular timer0/1 based PWM techniques (instead of the PIC's HPWM) to modulate the XOUT signal? The idea being able to use this with any pic, and on any pin.
Thanks in advance.
Regards,
Anand
|
Hi Anand,
I think at 120kz, the interrupts required to generate that frequency on ANY pin (250,000 per second) would interfere with the timing of the XOUT command.
I can't think of an easy way off hand.
DT
|
Oh, okay.
Nothing like actually trying it out, I guess.
Referring to some of your earlier posts I think I should be able to see if it flies.
Shall keep this thread posted for the benefit of others, if it works.
Thanks for all the inputs, Darrel.
Anand
|
Anand,
The 8-pin 12F683 has HWPM and only costs slightly more than the 12F629 or 12F675. I posted a simple example of turning it on/off using ~120kHz with 1ms bursts at 10ms intervals in another thread recently. Here it is again...The disadvantage of using XOUT is that it cannot send microdim/microbright and cannot send the extended commands X-10 introduced 6-7 years ago.Code:DEFINE OSC 8 @ DEVICE PIC12F683, INTRC_OSC_NOCLKOUT, MCLR_OFF TRISIO.2 = 0 'GPIO.2=Output PR2 = 17 'PWM Period 117.7kHz CCPR1L = 8 'PWM Duty-Cycle T2CON = %00000100 'Timer2=ON, 1:1 prescale OSCCON = %01110001 'INT HF OSC 8MHz WHILE OSCCON.3>0: WEND 'OSC startup timeout WHILE OSCCON.2=0: WEND 'INT HF OSC stable WHILE OSCCON.2>0 CCP1CON = %00001100 'PWM ON PauseUS 1000 'generate 1ms burst CCP1CON = 0 'PWM OFF Low GPIO.2 Pause 10 'pause 10ms WEND
|
Hey Anand,
Still using the 628A ?
and what OSC ?
I think I have something for you.
DT
Last edited by Darrel Taylor; - 5th May 2006 at 05:49. Reason: Come on... get your butt outta bed, this is Cool!
|
Hi Darrel,
Yes, still on the 16F628, with the internal 4 Mhz oscillator, although I'm hoping to migrate it to an 8 pin (or even 6?) device.
I could of course use an 12F683 as Dave pointed out, but I'm still curious to see if this can be done without the HPWM module.
Very eager to see your suggestion!
Anand
|
Dave, thanks for the inputs. I could not have designed the actual hardware interface to the mains without your invaluable pointers to powerline carrier articles.
Are you suggesting that X10 signals would be better generated in PBP without using the XOUT macro? My X10 transmitter is a very basic one, so the micro dim/brights are not an issue.
I would have loved to have some extended commands though.
Again, a heartfelt thanks.
Anand Dhuru
|
4mhz! You don't believe in making things easy do you.
I believe it can be done on any pin, without any hardware modules, CCP or Timer, and no Interrupts.but I'm still curious to see if this can be done without the HPWM module
Last time, we used the CCP module to create the 120khz, then used XOUT to turn that output on and off. But during that 1ms period that it's sending a "1", the XOUT command is just spinning it's wheels in a PAUSEUS 1000.
Instead of just sitting there letting the hardware do all the work, why not just create the frequency in software for 1ms and be done with it.
Once again, I've never done this, but maybe I can get lucky twice in a row.
For a 120khz signal of 1ms in duration, we need to create 120 cycles.
Each cycle is 8.3333 us (1/120,000).
@ 4mhz it doesn't divide perfectly, so the closest we can get is an 8us cycle which ends up at 125khz. I hope that's close enough.
First off, in your main program, make a variable like this to use for counting the cyclesCode:LoopCounter VAR BYTE $20 SYSTEM
Now let's go back to the PBPPIC14.lib file. Un-Comment the LOWT line that we had commented previously. This gives back control of the TRIS register to the XOUT command.And, change the xout1loop that we modified before, to this ...Code:ifdef XOUT_USED LIST XOUT2 movlw 2 ; Default to 2 cycles XOUT movwf R2 ; Save number of cycles movf RR1, W ; Get data port movwf FSR ; Put it into FSR movf RM1, W ; Get data bit mask call LOWT ; Set data to low, outputThen, using it should be like normal again ...Code:xout1loop movlw 120 ; 120khz for 1ms = 120 cycles movwf LoopCounter xout120loop ; ---- 1 cycle of the 120khz --------------------- movf RM1, W ; 1 Get data bit mask iorwf INDF, F ; 2 Set the bit nop ; 3 waist some time comf RM1, W ; 4 Get inverted data bit mask andwf INDF, F ; 5 Clear the bit decfsz LoopCounter, F ; 6 Are we done yet? goto xout120loop ; 8 NO. Do it again.Once again, Fingers crossedCode:XOUT PORTB.3,PORTA.0,[house\lightsOff]
<br>
DT
|
I think the best approach is to create a procedure (using ASM or using the CCP module) to generate a 1ms burst of ~120kHz and then call it as needed from PBP. That gives you the flexibility to send arbitrary patterns. If you also use Darrel's Instant Interrupt's, you can do the X-10 part in the background, sending a manchester half-bit at each zero-crossing.Originally Posted by ardhuru
Using the CCP module, you could even use a timer to end the ~120kHz burst and do single or 3-phase in the background with minimal loss of processing time. This would also allow you to listen to the powerline during manchester 0 half-bits to detect collisions. Note that it does no good to listen if using a TW523 or equivalent as it delays its output by 22 powerline half-cycles, acting like a 22-bit shift register with the powerline frequency as its clock.
|
How do you do it?!
Time to raise my hat again! That was superlative, Darrel. Worked the first time (that is, as soon as I remembered to switch off the CCP module enabled for the earlier version!). And of course, I confirmed it works as well on any other pin.4mhz! You don't believe in making things easy do you.
Once again, Fingers crossed
<br>
Thanks a ton.
Anand
|
Fantastic!
My luck seems to be getting better.
I'm thinking it's time for a trip to Vegas.
You're very welcome, and thanks for the fun project.
DT
P.S. Dave's idea sounds pretty good too. More control.
And Jeff might be happier without us messing with his macro's.
|
I would love to see an example for the 12F683.
|
So would I.I would love to see an example for the 12F683.
I'm glad it was your idea.
DT
oops, so much for the good karma from before.
Maybe I better skip Vegas.
|
Anand,
There may soon be another possibility. SmartLabs tells me they will release their powerline chips this month. They will cost less than $2 even in single quantities and can do both X-10 and Insteon.
|
Sounds very interesting
Hi Dave,
that sounds like a very useful product indeed; almost like a TW523-in-a-chip solutoin. I would look forward to its release.
In the mean time, my home brew circuit with off the shelf part also seems to be holding its own. In fact, the transformer I used (one from the set used in AM radios), gives out a signal 2 to 3 times in amplitude as the one from the RR501/TM751.
Regards,
Anand
Members who have read this thread : 1You do not have permission to view the list of names.
Posting Permissions
Bookmarks